2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* mm/mmap.c
|
|
|
|
|
*
|
|
|
|
|
* Written by obz.
|
|
|
|
|
*
|
2009-01-05 22:06:29 +08:00
|
|
|
* Address space accounting code <alan@lxorguk.ukuu.org.uk>
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#include <linux/slab.h>
|
2007-10-17 14:29:23 +08:00
|
|
|
#include <linux/backing-dev.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <linux/mm.h>
|
|
|
|
|
#include <linux/shm.h>
|
|
|
|
|
#include <linux/mman.h>
|
|
|
|
|
#include <linux/pagemap.h>
|
|
|
|
|
#include <linux/swap.h>
|
|
|
|
|
#include <linux/syscalls.h>
|
2006-01-12 04:17:46 +08:00
|
|
|
#include <linux/capability.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <linux/init.h>
|
|
|
|
|
#include <linux/file.h>
|
|
|
|
|
#include <linux/fs.h>
|
|
|
|
|
#include <linux/personality.h>
|
|
|
|
|
#include <linux/security.h>
|
|
|
|
|
#include <linux/hugetlb.h>
|
|
|
|
|
#include <linux/profile.h>
|
2011-10-16 14:01:52 +08:00
|
|
|
#include <linux/export.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
#include <linux/mount.h>
|
|
|
|
|
#include <linux/mempolicy.h>
|
|
|
|
|
#include <linux/rmap.h>
|
mmu-notifiers: core
With KVM/GFP/XPMEM there isn't just the primary CPU MMU pointing to pages.
There are secondary MMUs (with secondary sptes and secondary tlbs) too.
sptes in the kvm case are shadow pagetables, but when I say spte in
mmu-notifier context, I mean "secondary pte". In GRU case there's no
actual secondary pte and there's only a secondary tlb because the GRU
secondary MMU has no knowledge about sptes and every secondary tlb miss
event in the MMU always generates a page fault that has to be resolved by
the CPU (this is not the case of KVM where the a secondary tlb miss will
walk sptes in hardware and it will refill the secondary tlb transparently
to software if the corresponding spte is present). The same way
zap_page_range has to invalidate the pte before freeing the page, the spte
(and secondary tlb) must also be invalidated before any page is freed and
reused.
Currently we take a page_count pin on every page mapped by sptes, but that
means the pages can't be swapped whenever they're mapped by any spte
because they're part of the guest working set. Furthermore a spte unmap
event can immediately lead to a page to be freed when the pin is released
(so requiring the same complex and relatively slow tlb_gather smp safe
logic we have in zap_page_range and that can be avoided completely if the
spte unmap event doesn't require an unpin of the page previously mapped in
the secondary MMU).
The mmu notifiers allow kvm/GRU/XPMEM to attach to the tsk->mm and know
when the VM is swapping or freeing or doing anything on the primary MMU so
that the secondary MMU code can drop sptes before the pages are freed,
avoiding all page pinning and allowing 100% reliable swapping of guest
physical address space. Furthermore it avoids the code that teardown the
mappings of the secondary MMU, to implement a logic like tlb_gather in
zap_page_range that would require many IPI to flush other cpu tlbs, for
each fixed number of spte unmapped.
To make an example: if what happens on the primary MMU is a protection
downgrade (from writeable to wrprotect) the secondary MMU mappings will be
invalidated, and the next secondary-mmu-page-fault will call
get_user_pages and trigger a do_wp_page through get_user_pages if it
called get_user_pages with write=1, and it'll re-establishing an updated
spte or secondary-tlb-mapping on the copied page. Or it will setup a
readonly spte or readonly tlb mapping if it's a guest-read, if it calls
get_user_pages with write=0. This is just an example.
This allows to map any page pointed by any pte (and in turn visible in the
primary CPU MMU), into a secondary MMU (be it a pure tlb like GRU, or an
full MMU with both sptes and secondary-tlb like the shadow-pagetable layer
with kvm), or a remote DMA in software like XPMEM (hence needing of
schedule in XPMEM code to send the invalidate to the remote node, while no
need to schedule in kvm/gru as it's an immediate event like invalidating
primary-mmu pte).
At least for KVM without this patch it's impossible to swap guests
reliably. And having this feature and removing the page pin allows
several other optimizations that simplify life considerably.
Dependencies:
1) mm_take_all_locks() to register the mmu notifier when the whole VM
isn't doing anything with "mm". This allows mmu notifier users to keep
track if the VM is in the middle of the invalidate_range_begin/end
critical section with an atomic counter incraese in range_begin and
decreased in range_end. No secondary MMU page fault is allowed to map
any spte or secondary tlb reference, while the VM is in the middle of
range_begin/end as any page returned by get_user_pages in that critical
section could later immediately be freed without any further
->invalidate_page notification (invalidate_range_begin/end works on
ranges and ->invalidate_page isn't called immediately before freeing
the page). To stop all page freeing and pagetable overwrites the
mmap_sem must be taken in write mode and all other anon_vma/i_mmap
locks must be taken too.
2) It'd be a waste to add branches in the VM if nobody could possibly
run KVM/GRU/XPMEM on the kernel, so mmu notifiers will only enabled if
CONFIG_KVM=m/y. In the current kernel kvm won't yet take advantage of
mmu notifiers, but this already allows to compile a KVM external module
against a kernel with mmu notifiers enabled and from the next pull from
kvm.git we'll start using them. And GRU/XPMEM will also be able to
continue the development by enabling KVM=m in their config, until they
submit all GRU/XPMEM GPLv2 code to the mainline kernel. Then they can
also enable MMU_NOTIFIERS in the same way KVM does it (even if KVM=n).
This guarantees nobody selects MMU_NOTIFIER=y if KVM and GRU and XPMEM
are all =n.
The mmu_notifier_register call can fail because mm_take_all_locks may be
interrupted by a signal and return -EINTR. Because mmu_notifier_reigster
is used when a driver startup, a failure can be gracefully handled. Here
an example of the change applied to kvm to register the mmu notifiers.
Usually when a driver startups other allocations are required anyway and
-ENOMEM failure paths exists already.
struct kvm *kvm_arch_create_vm(void)
{
struct kvm *kvm = kzalloc(sizeof(struct kvm), GFP_KERNEL);
+ int err;
if (!kvm)
return ERR_PTR(-ENOMEM);
INIT_LIST_HEAD(&kvm->arch.active_mmu_pages);
+ kvm->arch.mmu_notifier.ops = &kvm_mmu_notifier_ops;
+ err = mmu_notifier_register(&kvm->arch.mmu_notifier, current->mm);
+ if (err) {
+ kfree(kvm);
+ return ERR_PTR(err);
+ }
+
return kvm;
}
mmu_notifier_unregister returns void and it's reliable.
The patch also adds a few needed but missing includes that would prevent
kernel to compile after these changes on non-x86 archs (x86 didn't need
them by luck).
[akpm@linux-foundation.org: coding-style fixes]
[akpm@linux-foundation.org: fix mm/filemap_xip.c build]
[akpm@linux-foundation.org: fix mm/mmu_notifier.c build]
Signed-off-by: Andrea Arcangeli <andrea@qumranet.com>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Jack Steiner <steiner@sgi.com>
Cc: Robin Holt <holt@sgi.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Kanoj Sarcar <kanojsarcar@yahoo.com>
Cc: Roland Dreier <rdreier@cisco.com>
Cc: Steve Wise <swise@opengridcomputing.com>
Cc: Avi Kivity <avi@qumranet.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Anthony Liguori <aliguori@us.ibm.com>
Cc: Chris Wright <chrisw@redhat.com>
Cc: Marcelo Tosatti <marcelo@kvack.org>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: Izik Eidus <izike@qumranet.com>
Cc: Anthony Liguori <aliguori@us.ibm.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-29 06:46:29 +08:00
|
|
|
#include <linux/mmu_notifier.h>
|
perf: Do the big rename: Performance Counters -> Performance Events
Bye-bye Performance Counters, welcome Performance Events!
In the past few months the perfcounters subsystem has grown out its
initial role of counting hardware events, and has become (and is
becoming) a much broader generic event enumeration, reporting, logging,
monitoring, analysis facility.
Naming its core object 'perf_counter' and naming the subsystem
'perfcounters' has become more and more of a misnomer. With pending
code like hw-breakpoints support the 'counter' name is less and
less appropriate.
All in one, we've decided to rename the subsystem to 'performance
events' and to propagate this rename through all fields, variables
and API names. (in an ABI compatible fashion)
The word 'event' is also a bit shorter than 'counter' - which makes
it slightly more convenient to write/handle as well.
Thanks goes to Stephane Eranian who first observed this misnomer and
suggested a rename.
User-space tooling and ABI compatibility is not affected - this patch
should be function-invariant. (Also, defconfigs were not touched to
keep the size down.)
This patch has been generated via the following script:
FILES=$(find * -type f | grep -vE 'oprofile|[^K]config')
sed -i \
-e 's/PERF_EVENT_/PERF_RECORD_/g' \
-e 's/PERF_COUNTER/PERF_EVENT/g' \
-e 's/perf_counter/perf_event/g' \
-e 's/nb_counters/nb_events/g' \
-e 's/swcounter/swevent/g' \
-e 's/tpcounter_event/tp_event/g' \
$FILES
for N in $(find . -name perf_counter.[ch]); do
M=$(echo $N | sed 's/perf_counter/perf_event/g')
mv $N $M
done
FILES=$(find . -name perf_event.*)
sed -i \
-e 's/COUNTER_MASK/REG_MASK/g' \
-e 's/COUNTER/EVENT/g' \
-e 's/\<event\>/event_id/g' \
-e 's/counter/event/g' \
-e 's/Counter/Event/g' \
$FILES
... to keep it as correct as possible. This script can also be
used by anyone who has pending perfcounters patches - it converts
a Linux kernel tree over to the new naming. We tried to time this
change to the point in time where the amount of pending patches
is the smallest: the end of the merge window.
Namespace clashes were fixed up in a preparatory patch - and some
stylistic fallout will be fixed up in a subsequent patch.
( NOTE: 'counters' are still the proper terminology when we deal
with hardware registers - and these sed scripts are a bit
over-eager in renaming them. I've undone some of that, but
in case there's something left where 'counter' would be
better than 'event' we can undo that on an individual basis
instead of touching an otherwise nicely automated patch. )
Suggested-by: Stephane Eranian <eranian@google.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Paul Mackerras <paulus@samba.org>
Reviewed-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: <linux-arch@vger.kernel.org>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-21 18:02:48 +08:00
|
|
|
#include <linux/perf_event.h>
|
2010-10-30 14:54:44 +08:00
|
|
|
#include <linux/audit.h>
|
2011-01-14 07:46:59 +08:00
|
|
|
#include <linux/khugepaged.h>
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
#include <linux/uprobes.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
#include <asm/uaccess.h>
|
|
|
|
|
#include <asm/cacheflush.h>
|
|
|
|
|
#include <asm/tlb.h>
|
2007-05-03 01:27:14 +08:00
|
|
|
#include <asm/mmu_context.h>
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2008-07-24 12:27:10 +08:00
|
|
|
#include "internal.h"
|
|
|
|
|
|
2006-09-07 18:17:04 +08:00
|
|
|
#ifndef arch_mmap_check
|
|
|
|
|
#define arch_mmap_check(addr, len, flags) (0)
|
|
|
|
|
#endif
|
|
|
|
|
|
2008-02-05 14:29:16 +08:00
|
|
|
#ifndef arch_rebalance_pgtables
|
|
|
|
|
#define arch_rebalance_pgtables(addr, len) (addr)
|
|
|
|
|
#endif
|
|
|
|
|
|
[PATCH] freepgt: free_pgtables use vma list
Recent woes with some arches needing their own pgd_addr_end macro; and 4-level
clear_page_range regression since 2.6.10's clear_page_tables; and its
long-standing well-known inefficiency in searching throughout the higher-level
page tables for those few entries to clear and free: all can be blamed on
ignoring the list of vmas when we free page tables.
Replace exit_mmap's clear_page_range of the total user address space by
free_pgtables operating on the mm's vma list; unmap_region use it in the same
way, giving floor and ceiling beyond which it may not free tables. This
brings lmbench fork/exec/sh numbers back to 2.6.10 (unless preempt is enabled,
in which case latency fixes spoil unmap_vmas throughput).
Beware: the do_mmap_pgoff driver failure case must now use unmap_region
instead of zap_page_range, since a page table might have been allocated, and
can only be freed while it is touched by some vma.
Move free_pgtables from mmap.c to memory.c, where its lower levels are adapted
from the clear_page_range levels. (Most of free_pgtables' old code was
actually for a non-existent case, prev not properly set up, dating from before
hch gave us split_vma.) Pass mmu_gather** in the public interfaces, since we
might want to add latency lockdrops later; but no attempt to do so yet, going
by vma should itself reduce latency.
But what if is_hugepage_only_range? Those ia64 and ppc64 cases need careful
examination: put that off until a later patch of the series.
What of x86_64's 32bit vdso page __map_syscall32 maps outside any vma?
And the range to sparc64's flush_tlb_pgtables? It's less clear to me now that
we need to do more than is done here - every PMD_SIZE ever occupied will be
flushed, do we really have to flush every PGDIR_SIZE ever partially occupied?
A shame to complicate it unnecessarily.
Special thanks to David Miller for time spent repairing my ceilings.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-20 04:29:15 +08:00
|
|
|
static void unmap_region(struct mm_struct *mm,
|
|
|
|
|
struct vm_area_struct *vma, struct vm_area_struct *prev,
|
|
|
|
|
unsigned long start, unsigned long end);
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* WARNING: the debugging will use recursive algorithms so never enable this
|
|
|
|
|
* unless you know what you are doing.
|
|
|
|
|
*/
|
|
|
|
|
#undef DEBUG_MM_RB
|
|
|
|
|
|
|
|
|
|
/* description of effects of mapping type and prot in current implementation.
|
|
|
|
|
* this is due to the limited x86 page protection hardware. The expected
|
|
|
|
|
* behavior is in parens:
|
|
|
|
|
*
|
|
|
|
|
* map_type prot
|
|
|
|
|
* PROT_NONE PROT_READ PROT_WRITE PROT_EXEC
|
|
|
|
|
* MAP_SHARED r: (no) no r: (yes) yes r: (no) yes r: (no) yes
|
|
|
|
|
* w: (no) no w: (no) no w: (yes) yes w: (no) no
|
|
|
|
|
* x: (no) no x: (no) yes x: (no) yes x: (yes) yes
|
|
|
|
|
*
|
|
|
|
|
* MAP_PRIVATE r: (no) no r: (yes) yes r: (no) yes r: (no) yes
|
|
|
|
|
* w: (no) no w: (no) no w: (copy) copy w: (no) no
|
|
|
|
|
* x: (no) no x: (no) yes x: (no) yes x: (yes) yes
|
|
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
pgprot_t protection_map[16] = {
|
|
|
|
|
__P000, __P001, __P010, __P011, __P100, __P101, __P110, __P111,
|
|
|
|
|
__S000, __S001, __S010, __S011, __S100, __S101, __S110, __S111
|
|
|
|
|
};
|
|
|
|
|
|
2006-07-27 04:39:49 +08:00
|
|
|
pgprot_t vm_get_page_prot(unsigned long vm_flags)
|
|
|
|
|
{
|
2008-07-07 22:28:51 +08:00
|
|
|
return __pgprot(pgprot_val(protection_map[vm_flags &
|
|
|
|
|
(VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)]) |
|
|
|
|
|
pgprot_val(arch_vm_get_page_prot(vm_flags)));
|
2006-07-27 04:39:49 +08:00
|
|
|
}
|
|
|
|
|
EXPORT_SYMBOL(vm_get_page_prot);
|
|
|
|
|
|
2011-05-25 08:11:18 +08:00
|
|
|
int sysctl_overcommit_memory __read_mostly = OVERCOMMIT_GUESS; /* heuristic overcommit */
|
|
|
|
|
int sysctl_overcommit_ratio __read_mostly = 50; /* default is 50% */
|
2005-09-07 06:16:33 +08:00
|
|
|
int sysctl_max_map_count __read_mostly = DEFAULT_MAX_MAP_COUNT;
|
2011-05-25 08:11:18 +08:00
|
|
|
/*
|
|
|
|
|
* Make sure vm_committed_as in one cacheline and not cacheline shared with
|
|
|
|
|
* other variables. It can be updated by several CPUs frequently.
|
|
|
|
|
*/
|
|
|
|
|
struct percpu_counter vm_committed_as ____cacheline_aligned_in_smp;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Check that a process has enough memory to allocate a new virtual
|
|
|
|
|
* mapping. 0 means there is enough memory for the allocation to
|
|
|
|
|
* succeed and -ENOMEM implies there is not.
|
|
|
|
|
*
|
|
|
|
|
* We currently support three overcommit policies, which are set via the
|
|
|
|
|
* vm.overcommit_memory sysctl. See Documentation/vm/overcommit-accounting
|
|
|
|
|
*
|
|
|
|
|
* Strict overcommit modes added 2002 Feb 26 by Alan Cox.
|
|
|
|
|
* Additional code 2002 Jul 20 by Robert Love.
|
|
|
|
|
*
|
|
|
|
|
* cap_sys_admin is 1 if the process has admin privileges, 0 otherwise.
|
|
|
|
|
*
|
|
|
|
|
* Note this is a helper function intended to be used by LSMs which
|
|
|
|
|
* wish to use this logic.
|
|
|
|
|
*/
|
2007-08-23 05:01:28 +08:00
|
|
|
int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
unsigned long free, allowed;
|
|
|
|
|
|
|
|
|
|
vm_acct_memory(pages);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Sometimes we want to use more memory than we have
|
|
|
|
|
*/
|
|
|
|
|
if (sysctl_overcommit_memory == OVERCOMMIT_ALWAYS)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
if (sysctl_overcommit_memory == OVERCOMMIT_GUESS) {
|
2011-07-26 08:12:19 +08:00
|
|
|
free = global_page_state(NR_FREE_PAGES);
|
|
|
|
|
free += global_page_state(NR_FILE_PAGES);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* shmem pages shouldn't be counted as free in this
|
|
|
|
|
* case, they can't be purged, only swapped out, and
|
|
|
|
|
* that won't affect the overall amount of available
|
|
|
|
|
* memory in the system.
|
|
|
|
|
*/
|
|
|
|
|
free -= global_page_state(NR_SHMEM);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
free += nr_swap_pages;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Any slabs which are created with the
|
|
|
|
|
* SLAB_RECLAIM_ACCOUNT flag claim to have contents
|
|
|
|
|
* which are reclaimable, under pressure. The dentry
|
|
|
|
|
* cache and most inode caches should fall into this
|
|
|
|
|
*/
|
2006-09-26 14:31:51 +08:00
|
|
|
free += global_page_state(NR_SLAB_RECLAIMABLE);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2006-04-11 13:53:00 +08:00
|
|
|
/*
|
|
|
|
|
* Leave reserved pages. The pages are not for anonymous pages.
|
|
|
|
|
*/
|
2011-07-26 08:12:19 +08:00
|
|
|
if (free <= totalreserve_pages)
|
2006-04-11 13:53:00 +08:00
|
|
|
goto error;
|
|
|
|
|
else
|
2011-07-26 08:12:19 +08:00
|
|
|
free -= totalreserve_pages;
|
2006-04-11 13:53:00 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Leave the last 3% for root
|
|
|
|
|
*/
|
2005-04-17 06:20:36 +08:00
|
|
|
if (!cap_sys_admin)
|
2011-07-26 08:12:19 +08:00
|
|
|
free -= free / 32;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
if (free > pages)
|
|
|
|
|
return 0;
|
2006-04-11 13:53:00 +08:00
|
|
|
|
|
|
|
|
goto error;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
allowed = (totalram_pages - hugetlb_total_pages())
|
|
|
|
|
* sysctl_overcommit_ratio / 100;
|
|
|
|
|
/*
|
|
|
|
|
* Leave the last 3% for root
|
|
|
|
|
*/
|
|
|
|
|
if (!cap_sys_admin)
|
|
|
|
|
allowed -= allowed / 32;
|
|
|
|
|
allowed += total_swap_pages;
|
|
|
|
|
|
|
|
|
|
/* Don't let a single process grow too big:
|
|
|
|
|
leave 3% of the size of this process for other processes */
|
2008-10-30 05:01:20 +08:00
|
|
|
if (mm)
|
|
|
|
|
allowed -= mm->total_vm / 32;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2009-05-01 06:08:51 +08:00
|
|
|
if (percpu_counter_read_positive(&vm_committed_as) < allowed)
|
2005-04-17 06:20:36 +08:00
|
|
|
return 0;
|
2006-04-11 13:53:00 +08:00
|
|
|
error:
|
2005-04-17 06:20:36 +08:00
|
|
|
vm_unacct_memory(pages);
|
|
|
|
|
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
2011-05-25 08:12:06 +08:00
|
|
|
* Requires inode->i_mapping->i_mmap_mutex
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
|
|
|
|
static void __remove_shared_vm_struct(struct vm_area_struct *vma,
|
|
|
|
|
struct file *file, struct address_space *mapping)
|
|
|
|
|
{
|
|
|
|
|
if (vma->vm_flags & VM_DENYWRITE)
|
2006-12-08 18:36:44 +08:00
|
|
|
atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (vma->vm_flags & VM_SHARED)
|
|
|
|
|
mapping->i_mmap_writable--;
|
|
|
|
|
|
|
|
|
|
flush_dcache_mmap_lock(mapping);
|
|
|
|
|
if (unlikely(vma->vm_flags & VM_NONLINEAR))
|
|
|
|
|
list_del_init(&vma->shared.vm_set.list);
|
|
|
|
|
else
|
|
|
|
|
vma_prio_tree_remove(vma, &mapping->i_mmap);
|
|
|
|
|
flush_dcache_mmap_unlock(mapping);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
2005-10-30 09:15:57 +08:00
|
|
|
* Unlink a file-based vm structure from its prio_tree, to hide
|
|
|
|
|
* vma from rmap and vmtruncate before freeing its page tables.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2005-10-30 09:15:57 +08:00
|
|
|
void unlink_file_vma(struct vm_area_struct *vma)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
struct file *file = vma->vm_file;
|
|
|
|
|
|
|
|
|
|
if (file) {
|
|
|
|
|
struct address_space *mapping = file->f_mapping;
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_lock(&mapping->i_mmap_mutex);
|
2005-04-17 06:20:36 +08:00
|
|
|
__remove_shared_vm_struct(vma, file, mapping);
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_unlock(&mapping->i_mmap_mutex);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
2005-10-30 09:15:57 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Close a vm structure and free it, returning the next.
|
|
|
|
|
*/
|
|
|
|
|
static struct vm_area_struct *remove_vma(struct vm_area_struct *vma)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct *next = vma->vm_next;
|
|
|
|
|
|
|
|
|
|
might_sleep();
|
2005-04-17 06:20:36 +08:00
|
|
|
if (vma->vm_ops && vma->vm_ops->close)
|
|
|
|
|
vma->vm_ops->close(vma);
|
2008-04-29 16:01:36 +08:00
|
|
|
if (vma->vm_file) {
|
2005-10-30 09:15:57 +08:00
|
|
|
fput(vma->vm_file);
|
2008-04-29 16:01:36 +08:00
|
|
|
if (vma->vm_flags & VM_EXECUTABLE)
|
|
|
|
|
removed_exe_file_vma(vma->vm_mm);
|
|
|
|
|
}
|
2008-04-28 17:13:08 +08:00
|
|
|
mpol_put(vma_policy(vma));
|
2005-04-17 06:20:36 +08:00
|
|
|
kmem_cache_free(vm_area_cachep, vma);
|
2005-10-30 09:15:57 +08:00
|
|
|
return next;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
2012-04-21 06:35:40 +08:00
|
|
|
static unsigned long do_brk(unsigned long addr, unsigned long len);
|
|
|
|
|
|
2009-01-14 21:14:15 +08:00
|
|
|
SYSCALL_DEFINE1(brk, unsigned long, brk)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
unsigned long rlim, retval;
|
|
|
|
|
unsigned long newbrk, oldbrk;
|
|
|
|
|
struct mm_struct *mm = current->mm;
|
2008-06-06 13:46:05 +08:00
|
|
|
unsigned long min_brk;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
down_write(&mm->mmap_sem);
|
|
|
|
|
|
2008-06-06 13:46:05 +08:00
|
|
|
#ifdef CONFIG_COMPAT_BRK
|
2011-01-14 07:47:23 +08:00
|
|
|
/*
|
|
|
|
|
* CONFIG_COMPAT_BRK can still be overridden by setting
|
|
|
|
|
* randomize_va_space to 2, which will still cause mm->start_brk
|
|
|
|
|
* to be arbitrarily shifted
|
|
|
|
|
*/
|
2011-04-15 06:22:09 +08:00
|
|
|
if (current->brk_randomized)
|
2011-01-14 07:47:23 +08:00
|
|
|
min_brk = mm->start_brk;
|
|
|
|
|
else
|
|
|
|
|
min_brk = mm->end_data;
|
2008-06-06 13:46:05 +08:00
|
|
|
#else
|
|
|
|
|
min_brk = mm->start_brk;
|
|
|
|
|
#endif
|
|
|
|
|
if (brk < min_brk)
|
2005-04-17 06:20:36 +08:00
|
|
|
goto out;
|
2006-04-11 13:52:57 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Check against rlimit here. If this check is done later after the test
|
|
|
|
|
* of oldbrk with newbrk then it can escape the test and let the data
|
|
|
|
|
* segment grow beyond its set limit the in case where the limit is
|
|
|
|
|
* not page aligned -Ram Gupta
|
|
|
|
|
*/
|
2010-03-06 05:41:44 +08:00
|
|
|
rlim = rlimit(RLIMIT_DATA);
|
2008-01-30 20:30:40 +08:00
|
|
|
if (rlim < RLIM_INFINITY && (brk - mm->start_brk) +
|
|
|
|
|
(mm->end_data - mm->start_data) > rlim)
|
2006-04-11 13:52:57 +08:00
|
|
|
goto out;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
newbrk = PAGE_ALIGN(brk);
|
|
|
|
|
oldbrk = PAGE_ALIGN(mm->brk);
|
|
|
|
|
if (oldbrk == newbrk)
|
|
|
|
|
goto set_brk;
|
|
|
|
|
|
|
|
|
|
/* Always allow shrinking brk. */
|
|
|
|
|
if (brk <= mm->brk) {
|
|
|
|
|
if (!do_munmap(mm, newbrk, oldbrk-newbrk))
|
|
|
|
|
goto set_brk;
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Check against existing mmap mappings. */
|
|
|
|
|
if (find_vma_intersection(mm, oldbrk, newbrk+PAGE_SIZE))
|
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
/* Ok, looks good - let it rip. */
|
|
|
|
|
if (do_brk(oldbrk, newbrk-oldbrk) != oldbrk)
|
|
|
|
|
goto out;
|
|
|
|
|
set_brk:
|
|
|
|
|
mm->brk = brk;
|
|
|
|
|
out:
|
|
|
|
|
retval = mm->brk;
|
|
|
|
|
up_write(&mm->mmap_sem);
|
|
|
|
|
return retval;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#ifdef DEBUG_MM_RB
|
|
|
|
|
static int browse_rb(struct rb_root *root)
|
|
|
|
|
{
|
|
|
|
|
int i = 0, j;
|
|
|
|
|
struct rb_node *nd, *pn = NULL;
|
|
|
|
|
unsigned long prev = 0, pend = 0;
|
|
|
|
|
|
|
|
|
|
for (nd = rb_first(root); nd; nd = rb_next(nd)) {
|
|
|
|
|
struct vm_area_struct *vma;
|
|
|
|
|
vma = rb_entry(nd, struct vm_area_struct, vm_rb);
|
|
|
|
|
if (vma->vm_start < prev)
|
|
|
|
|
printk("vm_start %lx prev %lx\n", vma->vm_start, prev), i = -1;
|
|
|
|
|
if (vma->vm_start < pend)
|
|
|
|
|
printk("vm_start %lx pend %lx\n", vma->vm_start, pend);
|
|
|
|
|
if (vma->vm_start > vma->vm_end)
|
|
|
|
|
printk("vm_end %lx < vm_start %lx\n", vma->vm_end, vma->vm_start);
|
|
|
|
|
i++;
|
|
|
|
|
pn = nd;
|
2007-03-01 12:13:13 +08:00
|
|
|
prev = vma->vm_start;
|
|
|
|
|
pend = vma->vm_end;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
j = 0;
|
|
|
|
|
for (nd = pn; nd; nd = rb_prev(nd)) {
|
|
|
|
|
j++;
|
|
|
|
|
}
|
|
|
|
|
if (i != j)
|
|
|
|
|
printk("backwards %d, forwards %d\n", j, i), i = 0;
|
|
|
|
|
return i;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void validate_mm(struct mm_struct *mm)
|
|
|
|
|
{
|
|
|
|
|
int bug = 0;
|
|
|
|
|
int i = 0;
|
|
|
|
|
struct vm_area_struct *tmp = mm->mmap;
|
|
|
|
|
while (tmp) {
|
|
|
|
|
tmp = tmp->vm_next;
|
|
|
|
|
i++;
|
|
|
|
|
}
|
|
|
|
|
if (i != mm->map_count)
|
|
|
|
|
printk("map_count %d vm_next %d\n", mm->map_count, i), bug = 1;
|
|
|
|
|
i = browse_rb(&mm->mm_rb);
|
|
|
|
|
if (i != mm->map_count)
|
|
|
|
|
printk("map_count %d rb %d\n", mm->map_count, i), bug = 1;
|
2006-04-01 07:23:29 +08:00
|
|
|
BUG_ON(bug);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
#else
|
|
|
|
|
#define validate_mm(mm) do { } while (0)
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
static struct vm_area_struct *
|
|
|
|
|
find_vma_prepare(struct mm_struct *mm, unsigned long addr,
|
|
|
|
|
struct vm_area_struct **pprev, struct rb_node ***rb_link,
|
|
|
|
|
struct rb_node ** rb_parent)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct * vma;
|
|
|
|
|
struct rb_node ** __rb_link, * __rb_parent, * rb_prev;
|
|
|
|
|
|
|
|
|
|
__rb_link = &mm->mm_rb.rb_node;
|
|
|
|
|
rb_prev = __rb_parent = NULL;
|
|
|
|
|
vma = NULL;
|
|
|
|
|
|
|
|
|
|
while (*__rb_link) {
|
|
|
|
|
struct vm_area_struct *vma_tmp;
|
|
|
|
|
|
|
|
|
|
__rb_parent = *__rb_link;
|
|
|
|
|
vma_tmp = rb_entry(__rb_parent, struct vm_area_struct, vm_rb);
|
|
|
|
|
|
|
|
|
|
if (vma_tmp->vm_end > addr) {
|
|
|
|
|
vma = vma_tmp;
|
|
|
|
|
if (vma_tmp->vm_start <= addr)
|
2008-08-06 04:01:41 +08:00
|
|
|
break;
|
2005-04-17 06:20:36 +08:00
|
|
|
__rb_link = &__rb_parent->rb_left;
|
|
|
|
|
} else {
|
|
|
|
|
rb_prev = __rb_parent;
|
|
|
|
|
__rb_link = &__rb_parent->rb_right;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
*pprev = NULL;
|
|
|
|
|
if (rb_prev)
|
|
|
|
|
*pprev = rb_entry(rb_prev, struct vm_area_struct, vm_rb);
|
|
|
|
|
*rb_link = __rb_link;
|
|
|
|
|
*rb_parent = __rb_parent;
|
|
|
|
|
return vma;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void __vma_link_rb(struct mm_struct *mm, struct vm_area_struct *vma,
|
|
|
|
|
struct rb_node **rb_link, struct rb_node *rb_parent)
|
|
|
|
|
{
|
|
|
|
|
rb_link_node(&vma->vm_rb, rb_parent, rb_link);
|
|
|
|
|
rb_insert_color(&vma->vm_rb, &mm->mm_rb);
|
|
|
|
|
}
|
|
|
|
|
|
2008-10-19 11:27:01 +08:00
|
|
|
static void __vma_link_file(struct vm_area_struct *vma)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2009-01-07 06:40:21 +08:00
|
|
|
struct file *file;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
file = vma->vm_file;
|
|
|
|
|
if (file) {
|
|
|
|
|
struct address_space *mapping = file->f_mapping;
|
|
|
|
|
|
|
|
|
|
if (vma->vm_flags & VM_DENYWRITE)
|
2006-12-08 18:36:44 +08:00
|
|
|
atomic_dec(&file->f_path.dentry->d_inode->i_writecount);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (vma->vm_flags & VM_SHARED)
|
|
|
|
|
mapping->i_mmap_writable++;
|
|
|
|
|
|
|
|
|
|
flush_dcache_mmap_lock(mapping);
|
|
|
|
|
if (unlikely(vma->vm_flags & VM_NONLINEAR))
|
|
|
|
|
vma_nonlinear_insert(vma, &mapping->i_mmap_nonlinear);
|
|
|
|
|
else
|
|
|
|
|
vma_prio_tree_insert(vma, &mapping->i_mmap);
|
|
|
|
|
flush_dcache_mmap_unlock(mapping);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void
|
|
|
|
|
__vma_link(struct mm_struct *mm, struct vm_area_struct *vma,
|
|
|
|
|
struct vm_area_struct *prev, struct rb_node **rb_link,
|
|
|
|
|
struct rb_node *rb_parent)
|
|
|
|
|
{
|
|
|
|
|
__vma_link_list(mm, vma, prev, rb_parent);
|
|
|
|
|
__vma_link_rb(mm, vma, rb_link, rb_parent);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void vma_link(struct mm_struct *mm, struct vm_area_struct *vma,
|
|
|
|
|
struct vm_area_struct *prev, struct rb_node **rb_link,
|
|
|
|
|
struct rb_node *rb_parent)
|
|
|
|
|
{
|
|
|
|
|
struct address_space *mapping = NULL;
|
|
|
|
|
|
|
|
|
|
if (vma->vm_file)
|
|
|
|
|
mapping = vma->vm_file->f_mapping;
|
|
|
|
|
|
2011-05-25 08:12:04 +08:00
|
|
|
if (mapping)
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_lock(&mapping->i_mmap_mutex);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
__vma_link(mm, vma, prev, rb_link, rb_parent);
|
|
|
|
|
__vma_link_file(vma);
|
|
|
|
|
|
|
|
|
|
if (mapping)
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_unlock(&mapping->i_mmap_mutex);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
mm->map_count++;
|
|
|
|
|
validate_mm(mm);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
2012-03-22 07:34:16 +08:00
|
|
|
* Helper for vma_adjust() in the split_vma insert case: insert a vma into the
|
|
|
|
|
* mm's list and rbtree. It has already been inserted into the prio_tree.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2009-01-07 06:40:21 +08:00
|
|
|
static void __insert_vm_struct(struct mm_struct *mm, struct vm_area_struct *vma)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2009-01-07 06:40:21 +08:00
|
|
|
struct vm_area_struct *__vma, *prev;
|
|
|
|
|
struct rb_node **rb_link, *rb_parent;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
__vma = find_vma_prepare(mm, vma->vm_start,&prev, &rb_link, &rb_parent);
|
2006-04-01 07:23:29 +08:00
|
|
|
BUG_ON(__vma && __vma->vm_start < vma->vm_end);
|
2005-04-17 06:20:36 +08:00
|
|
|
__vma_link(mm, vma, prev, rb_link, rb_parent);
|
|
|
|
|
mm->map_count++;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
|
__vma_unlink(struct mm_struct *mm, struct vm_area_struct *vma,
|
|
|
|
|
struct vm_area_struct *prev)
|
|
|
|
|
{
|
2010-08-21 07:24:55 +08:00
|
|
|
struct vm_area_struct *next = vma->vm_next;
|
|
|
|
|
|
|
|
|
|
prev->vm_next = next;
|
|
|
|
|
if (next)
|
|
|
|
|
next->vm_prev = prev;
|
2005-04-17 06:20:36 +08:00
|
|
|
rb_erase(&vma->vm_rb, &mm->mm_rb);
|
|
|
|
|
if (mm->mmap_cache == vma)
|
|
|
|
|
mm->mmap_cache = prev;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We cannot adjust vm_start, vm_end, vm_pgoff fields of a vma that
|
|
|
|
|
* is already present in an i_mmap tree without adjusting the tree.
|
|
|
|
|
* The following helper function should be used when such adjustments
|
|
|
|
|
* are necessary. The "insert" vma (if any) is to be inserted
|
|
|
|
|
* before we drop the necessary locks.
|
|
|
|
|
*/
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
int vma_adjust(struct vm_area_struct *vma, unsigned long start,
|
2005-04-17 06:20:36 +08:00
|
|
|
unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert)
|
|
|
|
|
{
|
|
|
|
|
struct mm_struct *mm = vma->vm_mm;
|
|
|
|
|
struct vm_area_struct *next = vma->vm_next;
|
|
|
|
|
struct vm_area_struct *importer = NULL;
|
|
|
|
|
struct address_space *mapping = NULL;
|
|
|
|
|
struct prio_tree_root *root = NULL;
|
2010-08-10 08:18:40 +08:00
|
|
|
struct anon_vma *anon_vma = NULL;
|
2005-04-17 06:20:36 +08:00
|
|
|
struct file *file = vma->vm_file;
|
|
|
|
|
long adjust_next = 0;
|
|
|
|
|
int remove_next = 0;
|
|
|
|
|
|
|
|
|
|
if (next && !insert) {
|
2010-04-11 06:22:30 +08:00
|
|
|
struct vm_area_struct *exporter = NULL;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (end >= next->vm_end) {
|
|
|
|
|
/*
|
|
|
|
|
* vma expands, overlapping all the next, and
|
|
|
|
|
* perhaps the one after too (mprotect case 6).
|
|
|
|
|
*/
|
|
|
|
|
again: remove_next = 1 + (end > next->vm_end);
|
|
|
|
|
end = next->vm_end;
|
2010-04-11 06:22:30 +08:00
|
|
|
exporter = next;
|
2005-04-17 06:20:36 +08:00
|
|
|
importer = vma;
|
|
|
|
|
} else if (end > next->vm_start) {
|
|
|
|
|
/*
|
|
|
|
|
* vma expands, overlapping part of the next:
|
|
|
|
|
* mprotect case 5 shifting the boundary up.
|
|
|
|
|
*/
|
|
|
|
|
adjust_next = (end - next->vm_start) >> PAGE_SHIFT;
|
2010-04-11 06:22:30 +08:00
|
|
|
exporter = next;
|
2005-04-17 06:20:36 +08:00
|
|
|
importer = vma;
|
|
|
|
|
} else if (end < vma->vm_end) {
|
|
|
|
|
/*
|
|
|
|
|
* vma shrinks, and !insert tells it's not
|
|
|
|
|
* split_vma inserting another: so it must be
|
|
|
|
|
* mprotect case 4 shifting the boundary down.
|
|
|
|
|
*/
|
|
|
|
|
adjust_next = - ((vma->vm_end - end) >> PAGE_SHIFT);
|
2010-04-11 06:22:30 +08:00
|
|
|
exporter = vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
importer = next;
|
|
|
|
|
}
|
|
|
|
|
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
/*
|
|
|
|
|
* Easily overlooked: when mprotect shifts the boundary,
|
|
|
|
|
* make sure the expanding vma has anon_vma set if the
|
|
|
|
|
* shrinking vma had, to cover any anon pages imported.
|
|
|
|
|
*/
|
2010-04-11 06:22:30 +08:00
|
|
|
if (exporter && exporter->anon_vma && !importer->anon_vma) {
|
|
|
|
|
if (anon_vma_clone(importer, exporter))
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
return -ENOMEM;
|
2010-04-11 06:22:30 +08:00
|
|
|
importer->anon_vma = exporter->anon_vma;
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (file) {
|
|
|
|
|
mapping = file->f_mapping;
|
2012-03-31 02:26:46 +08:00
|
|
|
if (!(vma->vm_flags & VM_NONLINEAR)) {
|
2005-04-17 06:20:36 +08:00
|
|
|
root = &mapping->i_mmap;
|
2012-04-11 18:35:27 +08:00
|
|
|
uprobe_munmap(vma, vma->vm_start, vma->vm_end);
|
2012-03-31 02:26:46 +08:00
|
|
|
|
|
|
|
|
if (adjust_next)
|
2012-04-11 18:35:27 +08:00
|
|
|
uprobe_munmap(next, next->vm_start,
|
|
|
|
|
next->vm_end);
|
2012-03-31 02:26:46 +08:00
|
|
|
}
|
|
|
|
|
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_lock(&mapping->i_mmap_mutex);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (insert) {
|
|
|
|
|
/*
|
|
|
|
|
* Put into prio_tree now, so instantiated pages
|
|
|
|
|
* are visible to arm/parisc __flush_dcache_page
|
|
|
|
|
* throughout; but we cannot insert into address
|
|
|
|
|
* space until vma start or end is updated.
|
|
|
|
|
*/
|
|
|
|
|
__vma_link_file(insert);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2011-01-14 07:47:08 +08:00
|
|
|
vma_adjust_trans_huge(vma, start, end, adjust_next);
|
|
|
|
|
|
2010-08-10 08:18:40 +08:00
|
|
|
/*
|
|
|
|
|
* When changing only vma->vm_end, we don't really need anon_vma
|
|
|
|
|
* lock. This is a fairly rare case by itself, but the anon_vma
|
|
|
|
|
* lock may be shared between many sibling processes. Skipping
|
|
|
|
|
* the lock for brk adjustments makes a difference sometimes.
|
|
|
|
|
*/
|
2011-05-25 08:11:19 +08:00
|
|
|
if (vma->anon_vma && (importer || start != vma->vm_start)) {
|
2010-08-10 08:18:40 +08:00
|
|
|
anon_vma = vma->anon_vma;
|
|
|
|
|
anon_vma_lock(anon_vma);
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (root) {
|
|
|
|
|
flush_dcache_mmap_lock(mapping);
|
|
|
|
|
vma_prio_tree_remove(vma, root);
|
|
|
|
|
if (adjust_next)
|
|
|
|
|
vma_prio_tree_remove(next, root);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
vma->vm_start = start;
|
|
|
|
|
vma->vm_end = end;
|
|
|
|
|
vma->vm_pgoff = pgoff;
|
|
|
|
|
if (adjust_next) {
|
|
|
|
|
next->vm_start += adjust_next << PAGE_SHIFT;
|
|
|
|
|
next->vm_pgoff += adjust_next;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (root) {
|
|
|
|
|
if (adjust_next)
|
|
|
|
|
vma_prio_tree_insert(next, root);
|
|
|
|
|
vma_prio_tree_insert(vma, root);
|
|
|
|
|
flush_dcache_mmap_unlock(mapping);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (remove_next) {
|
|
|
|
|
/*
|
|
|
|
|
* vma_merge has merged next into vma, and needs
|
|
|
|
|
* us to remove next before dropping the locks.
|
|
|
|
|
*/
|
|
|
|
|
__vma_unlink(mm, next, vma);
|
|
|
|
|
if (file)
|
|
|
|
|
__remove_shared_vm_struct(next, file, mapping);
|
|
|
|
|
} else if (insert) {
|
|
|
|
|
/*
|
|
|
|
|
* split_vma has split insert from vma, and needs
|
|
|
|
|
* us to insert it before dropping the locks
|
|
|
|
|
* (it may either follow vma or precede it).
|
|
|
|
|
*/
|
|
|
|
|
__insert_vm_struct(mm, insert);
|
|
|
|
|
}
|
|
|
|
|
|
2010-08-10 08:18:40 +08:00
|
|
|
if (anon_vma)
|
|
|
|
|
anon_vma_unlock(anon_vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (mapping)
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_unlock(&mapping->i_mmap_mutex);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
if (root) {
|
2012-02-17 16:27:41 +08:00
|
|
|
uprobe_mmap(vma);
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
|
|
|
|
|
if (adjust_next)
|
2012-02-17 16:27:41 +08:00
|
|
|
uprobe_mmap(next);
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (remove_next) {
|
2008-04-29 16:01:36 +08:00
|
|
|
if (file) {
|
2012-04-11 18:35:27 +08:00
|
|
|
uprobe_munmap(next, next->vm_start, next->vm_end);
|
2005-04-17 06:20:36 +08:00
|
|
|
fput(file);
|
2008-04-29 16:01:36 +08:00
|
|
|
if (next->vm_flags & VM_EXECUTABLE)
|
|
|
|
|
removed_exe_file_vma(mm);
|
|
|
|
|
}
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
if (next->anon_vma)
|
|
|
|
|
anon_vma_merge(vma, next);
|
2005-04-17 06:20:36 +08:00
|
|
|
mm->map_count--;
|
2008-04-28 17:13:08 +08:00
|
|
|
mpol_put(vma_policy(next));
|
2005-04-17 06:20:36 +08:00
|
|
|
kmem_cache_free(vm_area_cachep, next);
|
|
|
|
|
/*
|
|
|
|
|
* In mprotect's case 6 (see comments on vma_merge),
|
|
|
|
|
* we must remove another next too. It would clutter
|
|
|
|
|
* up the code too much to do both in one go.
|
|
|
|
|
*/
|
|
|
|
|
if (remove_next == 2) {
|
|
|
|
|
next = vma->vm_next;
|
|
|
|
|
goto again;
|
|
|
|
|
}
|
|
|
|
|
}
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
if (insert && file)
|
2012-02-17 16:27:41 +08:00
|
|
|
uprobe_mmap(insert);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
validate_mm(mm);
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
|
|
|
|
|
return 0;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If the vma has a ->close operation then the driver probably needs to release
|
|
|
|
|
* per-vma resources, so we don't attempt to merge those.
|
|
|
|
|
*/
|
|
|
|
|
static inline int is_mergeable_vma(struct vm_area_struct *vma,
|
|
|
|
|
struct file *file, unsigned long vm_flags)
|
|
|
|
|
{
|
2009-09-22 08:02:25 +08:00
|
|
|
/* VM_CAN_NONLINEAR may get set later by f_op->mmap() */
|
|
|
|
|
if ((vma->vm_flags ^ vm_flags) & ~VM_CAN_NONLINEAR)
|
2005-04-17 06:20:36 +08:00
|
|
|
return 0;
|
|
|
|
|
if (vma->vm_file != file)
|
|
|
|
|
return 0;
|
|
|
|
|
if (vma->vm_ops && vma->vm_ops->close)
|
|
|
|
|
return 0;
|
|
|
|
|
return 1;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static inline int is_mergeable_anon_vma(struct anon_vma *anon_vma1,
|
2011-05-25 08:11:20 +08:00
|
|
|
struct anon_vma *anon_vma2,
|
|
|
|
|
struct vm_area_struct *vma)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2011-05-25 08:11:20 +08:00
|
|
|
/*
|
|
|
|
|
* The list_is_singular() test is to avoid merging VMA cloned from
|
|
|
|
|
* parents. This can improve scalability caused by anon_vma lock.
|
|
|
|
|
*/
|
|
|
|
|
if ((!anon_vma1 || !anon_vma2) && (!vma ||
|
|
|
|
|
list_is_singular(&vma->anon_vma_chain)))
|
|
|
|
|
return 1;
|
|
|
|
|
return anon_vma1 == anon_vma2;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Return true if we can merge this (vm_flags,anon_vma,file,vm_pgoff)
|
|
|
|
|
* in front of (at a lower virtual address and file offset than) the vma.
|
|
|
|
|
*
|
|
|
|
|
* We cannot merge two vmas if they have differently assigned (non-NULL)
|
|
|
|
|
* anon_vmas, nor if same anon_vma is assigned but offsets incompatible.
|
|
|
|
|
*
|
|
|
|
|
* We don't check here for the merged mmap wrapping around the end of pagecache
|
|
|
|
|
* indices (16TB on ia32) because do_mmap_pgoff() does not permit mmap's which
|
|
|
|
|
* wrap, nor mmaps which cover the final page at index -1UL.
|
|
|
|
|
*/
|
|
|
|
|
static int
|
|
|
|
|
can_vma_merge_before(struct vm_area_struct *vma, unsigned long vm_flags,
|
|
|
|
|
struct anon_vma *anon_vma, struct file *file, pgoff_t vm_pgoff)
|
|
|
|
|
{
|
|
|
|
|
if (is_mergeable_vma(vma, file, vm_flags) &&
|
2011-05-25 08:11:20 +08:00
|
|
|
is_mergeable_anon_vma(anon_vma, vma->anon_vma, vma)) {
|
2005-04-17 06:20:36 +08:00
|
|
|
if (vma->vm_pgoff == vm_pgoff)
|
|
|
|
|
return 1;
|
|
|
|
|
}
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Return true if we can merge this (vm_flags,anon_vma,file,vm_pgoff)
|
|
|
|
|
* beyond (at a higher virtual address and file offset than) the vma.
|
|
|
|
|
*
|
|
|
|
|
* We cannot merge two vmas if they have differently assigned (non-NULL)
|
|
|
|
|
* anon_vmas, nor if same anon_vma is assigned but offsets incompatible.
|
|
|
|
|
*/
|
|
|
|
|
static int
|
|
|
|
|
can_vma_merge_after(struct vm_area_struct *vma, unsigned long vm_flags,
|
|
|
|
|
struct anon_vma *anon_vma, struct file *file, pgoff_t vm_pgoff)
|
|
|
|
|
{
|
|
|
|
|
if (is_mergeable_vma(vma, file, vm_flags) &&
|
2011-05-25 08:11:20 +08:00
|
|
|
is_mergeable_anon_vma(anon_vma, vma->anon_vma, vma)) {
|
2005-04-17 06:20:36 +08:00
|
|
|
pgoff_t vm_pglen;
|
|
|
|
|
vm_pglen = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
|
|
|
|
|
if (vma->vm_pgoff + vm_pglen == vm_pgoff)
|
|
|
|
|
return 1;
|
|
|
|
|
}
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Given a mapping request (addr,end,vm_flags,file,pgoff), figure out
|
|
|
|
|
* whether that can be merged with its predecessor or its successor.
|
|
|
|
|
* Or both (it neatly fills a hole).
|
|
|
|
|
*
|
|
|
|
|
* In most cases - when called for mmap, brk or mremap - [addr,end) is
|
|
|
|
|
* certain not to be mapped by the time vma_merge is called; but when
|
|
|
|
|
* called for mprotect, it is certain to be already mapped (either at
|
|
|
|
|
* an offset within prev, or at the start of next), and the flags of
|
|
|
|
|
* this area are about to be changed to vm_flags - and the no-change
|
|
|
|
|
* case has already been eliminated.
|
|
|
|
|
*
|
|
|
|
|
* The following mprotect cases have to be considered, where AAAA is
|
|
|
|
|
* the area passed down from mprotect_fixup, never extending beyond one
|
|
|
|
|
* vma, PPPPPP is the prev vma specified, and NNNNNN the next vma after:
|
|
|
|
|
*
|
|
|
|
|
* AAAA AAAA AAAA AAAA
|
|
|
|
|
* PPPPPPNNNNNN PPPPPPNNNNNN PPPPPPNNNNNN PPPPNNNNXXXX
|
|
|
|
|
* cannot merge might become might become might become
|
|
|
|
|
* PPNNNNNNNNNN PPPPPPPPPPNN PPPPPPPPPPPP 6 or
|
|
|
|
|
* mmap, brk or case 4 below case 5 below PPPPPPPPXXXX 7 or
|
|
|
|
|
* mremap move: PPPPNNNNNNNN 8
|
|
|
|
|
* AAAA
|
|
|
|
|
* PPPP NNNN PPPPPPPPPPPP PPPPPPPPNNNN PPPPNNNNNNNN
|
|
|
|
|
* might become case 1 below case 2 below case 3 below
|
|
|
|
|
*
|
|
|
|
|
* Odd one out? Case 8, because it extends NNNN but needs flags of XXXX:
|
|
|
|
|
* mprotect_fixup updates vm_flags & vm_page_prot on successful return.
|
|
|
|
|
*/
|
|
|
|
|
struct vm_area_struct *vma_merge(struct mm_struct *mm,
|
|
|
|
|
struct vm_area_struct *prev, unsigned long addr,
|
|
|
|
|
unsigned long end, unsigned long vm_flags,
|
|
|
|
|
struct anon_vma *anon_vma, struct file *file,
|
|
|
|
|
pgoff_t pgoff, struct mempolicy *policy)
|
|
|
|
|
{
|
|
|
|
|
pgoff_t pglen = (end - addr) >> PAGE_SHIFT;
|
|
|
|
|
struct vm_area_struct *area, *next;
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
int err;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We later require that vma->vm_flags == vm_flags,
|
|
|
|
|
* so this tests vma->vm_flags & VM_SPECIAL, too.
|
|
|
|
|
*/
|
|
|
|
|
if (vm_flags & VM_SPECIAL)
|
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
|
|
if (prev)
|
|
|
|
|
next = prev->vm_next;
|
|
|
|
|
else
|
|
|
|
|
next = mm->mmap;
|
|
|
|
|
area = next;
|
|
|
|
|
if (next && next->vm_end == end) /* cases 6, 7, 8 */
|
|
|
|
|
next = next->vm_next;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Can it merge with the predecessor?
|
|
|
|
|
*/
|
|
|
|
|
if (prev && prev->vm_end == addr &&
|
|
|
|
|
mpol_equal(vma_policy(prev), policy) &&
|
|
|
|
|
can_vma_merge_after(prev, vm_flags,
|
|
|
|
|
anon_vma, file, pgoff)) {
|
|
|
|
|
/*
|
|
|
|
|
* OK, it can. Can we now merge in the successor as well?
|
|
|
|
|
*/
|
|
|
|
|
if (next && end == next->vm_start &&
|
|
|
|
|
mpol_equal(policy, vma_policy(next)) &&
|
|
|
|
|
can_vma_merge_before(next, vm_flags,
|
|
|
|
|
anon_vma, file, pgoff+pglen) &&
|
|
|
|
|
is_mergeable_anon_vma(prev->anon_vma,
|
2011-05-25 08:11:20 +08:00
|
|
|
next->anon_vma, NULL)) {
|
2005-04-17 06:20:36 +08:00
|
|
|
/* cases 1, 6 */
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
err = vma_adjust(prev, prev->vm_start,
|
2005-04-17 06:20:36 +08:00
|
|
|
next->vm_end, prev->vm_pgoff, NULL);
|
|
|
|
|
} else /* cases 2, 5, 7 */
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
err = vma_adjust(prev, prev->vm_start,
|
2005-04-17 06:20:36 +08:00
|
|
|
end, prev->vm_pgoff, NULL);
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
if (err)
|
|
|
|
|
return NULL;
|
2011-01-14 07:46:59 +08:00
|
|
|
khugepaged_enter_vma_merge(prev);
|
2005-04-17 06:20:36 +08:00
|
|
|
return prev;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Can this new request be merged in front of next?
|
|
|
|
|
*/
|
|
|
|
|
if (next && end == next->vm_start &&
|
|
|
|
|
mpol_equal(policy, vma_policy(next)) &&
|
|
|
|
|
can_vma_merge_before(next, vm_flags,
|
|
|
|
|
anon_vma, file, pgoff+pglen)) {
|
|
|
|
|
if (prev && addr < prev->vm_end) /* case 4 */
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
err = vma_adjust(prev, prev->vm_start,
|
2005-04-17 06:20:36 +08:00
|
|
|
addr, prev->vm_pgoff, NULL);
|
|
|
|
|
else /* cases 3, 8 */
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
err = vma_adjust(area, addr, next->vm_end,
|
2005-04-17 06:20:36 +08:00
|
|
|
next->vm_pgoff - pglen, NULL);
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
if (err)
|
|
|
|
|
return NULL;
|
2011-01-14 07:46:59 +08:00
|
|
|
khugepaged_enter_vma_merge(area);
|
2005-04-17 06:20:36 +08:00
|
|
|
return area;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
2010-04-11 01:36:19 +08:00
|
|
|
/*
|
|
|
|
|
* Rough compatbility check to quickly see if it's even worth looking
|
|
|
|
|
* at sharing an anon_vma.
|
|
|
|
|
*
|
|
|
|
|
* They need to have the same vm_file, and the flags can only differ
|
|
|
|
|
* in things that mprotect may change.
|
|
|
|
|
*
|
|
|
|
|
* NOTE! The fact that we share an anon_vma doesn't _have_ to mean that
|
|
|
|
|
* we can merge the two vma's. For example, we refuse to merge a vma if
|
|
|
|
|
* there is a vm_ops->close() function, because that indicates that the
|
|
|
|
|
* driver is doing some kind of reference counting. But that doesn't
|
|
|
|
|
* really matter for the anon_vma sharing case.
|
|
|
|
|
*/
|
|
|
|
|
static int anon_vma_compatible(struct vm_area_struct *a, struct vm_area_struct *b)
|
|
|
|
|
{
|
|
|
|
|
return a->vm_end == b->vm_start &&
|
|
|
|
|
mpol_equal(vma_policy(a), vma_policy(b)) &&
|
|
|
|
|
a->vm_file == b->vm_file &&
|
|
|
|
|
!((a->vm_flags ^ b->vm_flags) & ~(VM_READ|VM_WRITE|VM_EXEC)) &&
|
|
|
|
|
b->vm_pgoff == a->vm_pgoff + ((b->vm_start - a->vm_start) >> PAGE_SHIFT);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Do some basic sanity checking to see if we can re-use the anon_vma
|
|
|
|
|
* from 'old'. The 'a'/'b' vma's are in VM order - one of them will be
|
|
|
|
|
* the same as 'old', the other will be the new one that is trying
|
|
|
|
|
* to share the anon_vma.
|
|
|
|
|
*
|
|
|
|
|
* NOTE! This runs with mm_sem held for reading, so it is possible that
|
|
|
|
|
* the anon_vma of 'old' is concurrently in the process of being set up
|
|
|
|
|
* by another page fault trying to merge _that_. But that's ok: if it
|
|
|
|
|
* is being set up, that automatically means that it will be a singleton
|
|
|
|
|
* acceptable for merging, so we can do all of this optimistically. But
|
|
|
|
|
* we do that ACCESS_ONCE() to make sure that we never re-load the pointer.
|
|
|
|
|
*
|
|
|
|
|
* IOW: that the "list_is_singular()" test on the anon_vma_chain only
|
|
|
|
|
* matters for the 'stable anon_vma' case (ie the thing we want to avoid
|
|
|
|
|
* is to return an anon_vma that is "complex" due to having gone through
|
|
|
|
|
* a fork).
|
|
|
|
|
*
|
|
|
|
|
* We also make sure that the two vma's are compatible (adjacent,
|
|
|
|
|
* and with the same memory policies). That's all stable, even with just
|
|
|
|
|
* a read lock on the mm_sem.
|
|
|
|
|
*/
|
|
|
|
|
static struct anon_vma *reusable_anon_vma(struct vm_area_struct *old, struct vm_area_struct *a, struct vm_area_struct *b)
|
|
|
|
|
{
|
|
|
|
|
if (anon_vma_compatible(a, b)) {
|
|
|
|
|
struct anon_vma *anon_vma = ACCESS_ONCE(old->anon_vma);
|
|
|
|
|
|
|
|
|
|
if (anon_vma && list_is_singular(&old->anon_vma_chain))
|
|
|
|
|
return anon_vma;
|
|
|
|
|
}
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* find_mergeable_anon_vma is used by anon_vma_prepare, to check
|
|
|
|
|
* neighbouring vmas for a suitable anon_vma, before it goes off
|
|
|
|
|
* to allocate a new anon_vma. It checks because a repetitive
|
|
|
|
|
* sequence of mprotects and faults may otherwise lead to distinct
|
|
|
|
|
* anon_vmas being allocated, preventing vma merge in subsequent
|
|
|
|
|
* mprotect.
|
|
|
|
|
*/
|
|
|
|
|
struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *vma)
|
|
|
|
|
{
|
2010-04-11 01:36:19 +08:00
|
|
|
struct anon_vma *anon_vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
struct vm_area_struct *near;
|
|
|
|
|
|
|
|
|
|
near = vma->vm_next;
|
|
|
|
|
if (!near)
|
|
|
|
|
goto try_prev;
|
|
|
|
|
|
2010-04-11 01:36:19 +08:00
|
|
|
anon_vma = reusable_anon_vma(near, vma, near);
|
|
|
|
|
if (anon_vma)
|
|
|
|
|
return anon_vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
try_prev:
|
2011-06-16 15:35:09 +08:00
|
|
|
near = vma->vm_prev;
|
2005-04-17 06:20:36 +08:00
|
|
|
if (!near)
|
|
|
|
|
goto none;
|
|
|
|
|
|
2010-04-11 01:36:19 +08:00
|
|
|
anon_vma = reusable_anon_vma(near, near, vma);
|
|
|
|
|
if (anon_vma)
|
|
|
|
|
return anon_vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
none:
|
|
|
|
|
/*
|
|
|
|
|
* There's no absolute need to look only at touching neighbours:
|
|
|
|
|
* we could search further afield for "compatible" anon_vmas.
|
|
|
|
|
* But it would probably just be a waste of time searching,
|
|
|
|
|
* or lead to too many vmas hanging off the same anon_vma.
|
|
|
|
|
* We're trying to allow mprotect remerging later on,
|
|
|
|
|
* not trying to minimize memory used for anon_vmas.
|
|
|
|
|
*/
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#ifdef CONFIG_PROC_FS
|
2005-10-30 09:15:56 +08:00
|
|
|
void vm_stat_account(struct mm_struct *mm, unsigned long flags,
|
2005-04-17 06:20:36 +08:00
|
|
|
struct file *file, long pages)
|
|
|
|
|
{
|
|
|
|
|
const unsigned long stack_flags
|
|
|
|
|
= VM_STACK_FLAGS & (VM_GROWSUP|VM_GROWSDOWN);
|
|
|
|
|
|
|
|
|
|
if (file) {
|
|
|
|
|
mm->shared_vm += pages;
|
|
|
|
|
if ((flags & (VM_EXEC|VM_WRITE)) == VM_EXEC)
|
|
|
|
|
mm->exec_vm += pages;
|
|
|
|
|
} else if (flags & stack_flags)
|
|
|
|
|
mm->stack_vm += pages;
|
|
|
|
|
if (flags & (VM_RESERVED|VM_IO))
|
|
|
|
|
mm->reserved_vm += pages;
|
|
|
|
|
}
|
|
|
|
|
#endif /* CONFIG_PROC_FS */
|
|
|
|
|
|
2012-02-13 11:58:52 +08:00
|
|
|
/*
|
|
|
|
|
* If a hint addr is less than mmap_min_addr change hint to be as
|
|
|
|
|
* low as possible but still greater than mmap_min_addr
|
|
|
|
|
*/
|
|
|
|
|
static inline unsigned long round_hint_to_min(unsigned long hint)
|
|
|
|
|
{
|
|
|
|
|
hint &= PAGE_MASK;
|
|
|
|
|
if (((void *)hint != NULL) &&
|
|
|
|
|
(hint < mmap_min_addr))
|
|
|
|
|
return PAGE_ALIGN(mmap_min_addr);
|
|
|
|
|
return hint;
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
2009-09-18 10:26:26 +08:00
|
|
|
* The caller must hold down_write(¤t->mm->mmap_sem).
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
|
|
|
|
|
2012-04-21 08:13:58 +08:00
|
|
|
static unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
|
2005-04-17 06:20:36 +08:00
|
|
|
unsigned long len, unsigned long prot,
|
|
|
|
|
unsigned long flags, unsigned long pgoff)
|
|
|
|
|
{
|
|
|
|
|
struct mm_struct * mm = current->mm;
|
|
|
|
|
struct inode *inode;
|
2011-05-26 18:16:19 +08:00
|
|
|
vm_flags_t vm_flags;
|
2005-04-17 06:20:36 +08:00
|
|
|
int error;
|
2007-07-16 14:38:26 +08:00
|
|
|
unsigned long reqprot = prot;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Does the application expect PROT_READ to imply PROT_EXEC?
|
|
|
|
|
*
|
|
|
|
|
* (the exception is when the underlying filesystem is noexec
|
|
|
|
|
* mounted, in which case we dont add PROT_EXEC.)
|
|
|
|
|
*/
|
|
|
|
|
if ((prot & PROT_READ) && (current->personality & READ_IMPLIES_EXEC))
|
2006-12-08 18:36:44 +08:00
|
|
|
if (!(file && (file->f_path.mnt->mnt_flags & MNT_NOEXEC)))
|
2005-04-17 06:20:36 +08:00
|
|
|
prot |= PROT_EXEC;
|
|
|
|
|
|
|
|
|
|
if (!len)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
2007-11-27 07:47:40 +08:00
|
|
|
if (!(flags & MAP_FIXED))
|
|
|
|
|
addr = round_hint_to_min(addr);
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* Careful about overflows.. */
|
|
|
|
|
len = PAGE_ALIGN(len);
|
2009-12-04 04:23:11 +08:00
|
|
|
if (!len)
|
2005-04-17 06:20:36 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
/* offset overflow? */
|
|
|
|
|
if ((pgoff + (len >> PAGE_SHIFT)) < pgoff)
|
|
|
|
|
return -EOVERFLOW;
|
|
|
|
|
|
|
|
|
|
/* Too many mappings? */
|
|
|
|
|
if (mm->map_count > sysctl_max_map_count)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
/* Obtain the address to map to. we verify (or select) it and ensure
|
|
|
|
|
* that it represents a valid section of the address space.
|
|
|
|
|
*/
|
|
|
|
|
addr = get_unmapped_area(file, addr, len, pgoff, flags);
|
|
|
|
|
if (addr & ~PAGE_MASK)
|
|
|
|
|
return addr;
|
|
|
|
|
|
|
|
|
|
/* Do simple checking here so the lower-level routines won't have
|
|
|
|
|
* to. we assume access permissions have been handled by the open
|
|
|
|
|
* of the memory object, so we don't do any here.
|
|
|
|
|
*/
|
|
|
|
|
vm_flags = calc_vm_prot_bits(prot) | calc_vm_flag_bits(flags) |
|
|
|
|
|
mm->def_flags | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC;
|
|
|
|
|
|
2009-09-22 08:03:36 +08:00
|
|
|
if (flags & MAP_LOCKED)
|
2005-04-17 06:20:36 +08:00
|
|
|
if (!can_do_mlock())
|
|
|
|
|
return -EPERM;
|
2008-10-19 11:26:50 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* mlock MCL_FUTURE? */
|
|
|
|
|
if (vm_flags & VM_LOCKED) {
|
|
|
|
|
unsigned long locked, lock_limit;
|
2005-05-01 23:58:38 +08:00
|
|
|
locked = len >> PAGE_SHIFT;
|
|
|
|
|
locked += mm->locked_vm;
|
2010-03-06 05:41:44 +08:00
|
|
|
lock_limit = rlimit(RLIMIT_MEMLOCK);
|
2005-05-01 23:58:38 +08:00
|
|
|
lock_limit >>= PAGE_SHIFT;
|
2005-04-17 06:20:36 +08:00
|
|
|
if (locked > lock_limit && !capable(CAP_IPC_LOCK))
|
|
|
|
|
return -EAGAIN;
|
|
|
|
|
}
|
|
|
|
|
|
2006-12-08 18:36:44 +08:00
|
|
|
inode = file ? file->f_path.dentry->d_inode : NULL;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
if (file) {
|
|
|
|
|
switch (flags & MAP_TYPE) {
|
|
|
|
|
case MAP_SHARED:
|
|
|
|
|
if ((prot&PROT_WRITE) && !(file->f_mode&FMODE_WRITE))
|
|
|
|
|
return -EACCES;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Make sure we don't allow writing to an append-only
|
|
|
|
|
* file..
|
|
|
|
|
*/
|
|
|
|
|
if (IS_APPEND(inode) && (file->f_mode & FMODE_WRITE))
|
|
|
|
|
return -EACCES;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Make sure there are no mandatory locks on the file.
|
|
|
|
|
*/
|
|
|
|
|
if (locks_verify_locked(inode))
|
|
|
|
|
return -EAGAIN;
|
|
|
|
|
|
|
|
|
|
vm_flags |= VM_SHARED | VM_MAYSHARE;
|
|
|
|
|
if (!(file->f_mode & FMODE_WRITE))
|
|
|
|
|
vm_flags &= ~(VM_MAYWRITE | VM_SHARED);
|
|
|
|
|
|
|
|
|
|
/* fall through */
|
|
|
|
|
case MAP_PRIVATE:
|
|
|
|
|
if (!(file->f_mode & FMODE_READ))
|
|
|
|
|
return -EACCES;
|
2006-12-08 18:36:44 +08:00
|
|
|
if (file->f_path.mnt->mnt_flags & MNT_NOEXEC) {
|
2006-10-16 05:09:55 +08:00
|
|
|
if (vm_flags & VM_EXEC)
|
|
|
|
|
return -EPERM;
|
|
|
|
|
vm_flags &= ~VM_MAYEXEC;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (!file->f_op || !file->f_op->mmap)
|
|
|
|
|
return -ENODEV;
|
2005-04-17 06:20:36 +08:00
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
default:
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
switch (flags & MAP_TYPE) {
|
|
|
|
|
case MAP_SHARED:
|
2008-09-03 22:09:47 +08:00
|
|
|
/*
|
|
|
|
|
* Ignore pgoff.
|
|
|
|
|
*/
|
|
|
|
|
pgoff = 0;
|
2005-04-17 06:20:36 +08:00
|
|
|
vm_flags |= VM_SHARED | VM_MAYSHARE;
|
|
|
|
|
break;
|
|
|
|
|
case MAP_PRIVATE:
|
|
|
|
|
/*
|
|
|
|
|
* Set pgoff according to addr for anon_vma.
|
|
|
|
|
*/
|
|
|
|
|
pgoff = addr >> PAGE_SHIFT;
|
|
|
|
|
break;
|
|
|
|
|
default:
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2007-06-29 03:55:21 +08:00
|
|
|
error = security_file_mmap(file, reqprot, prot, flags, addr, 0);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (error)
|
|
|
|
|
return error;
|
2007-06-29 03:55:21 +08:00
|
|
|
|
2009-02-10 22:02:27 +08:00
|
|
|
return mmap_region(file, addr, len, flags, vm_flags, pgoff);
|
2007-07-16 14:38:26 +08:00
|
|
|
}
|
2012-04-21 08:13:58 +08:00
|
|
|
|
|
|
|
|
unsigned long do_mmap(struct file *file, unsigned long addr,
|
|
|
|
|
unsigned long len, unsigned long prot,
|
|
|
|
|
unsigned long flag, unsigned long offset)
|
|
|
|
|
{
|
|
|
|
|
if (unlikely(offset + PAGE_ALIGN(len) < offset))
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
if (unlikely(offset & ~PAGE_MASK))
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
return do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
|
|
|
|
|
}
|
|
|
|
|
EXPORT_SYMBOL(do_mmap);
|
|
|
|
|
|
|
|
|
|
unsigned long vm_mmap(struct file *file, unsigned long addr,
|
|
|
|
|
unsigned long len, unsigned long prot,
|
|
|
|
|
unsigned long flag, unsigned long offset)
|
|
|
|
|
{
|
|
|
|
|
unsigned long ret;
|
|
|
|
|
struct mm_struct *mm = current->mm;
|
|
|
|
|
|
|
|
|
|
down_write(&mm->mmap_sem);
|
|
|
|
|
ret = do_mmap(file, addr, len, prot, flag, offset);
|
|
|
|
|
up_write(&mm->mmap_sem);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
EXPORT_SYMBOL(vm_mmap);
|
2007-07-16 14:38:26 +08:00
|
|
|
|
2009-12-31 04:17:34 +08:00
|
|
|
SYSCALL_DEFINE6(mmap_pgoff, unsigned long, addr, unsigned long, len,
|
|
|
|
|
unsigned long, prot, unsigned long, flags,
|
|
|
|
|
unsigned long, fd, unsigned long, pgoff)
|
|
|
|
|
{
|
|
|
|
|
struct file *file = NULL;
|
|
|
|
|
unsigned long retval = -EBADF;
|
|
|
|
|
|
|
|
|
|
if (!(flags & MAP_ANONYMOUS)) {
|
2010-10-30 14:54:44 +08:00
|
|
|
audit_mmap_fd(fd, flags);
|
2009-12-31 04:17:34 +08:00
|
|
|
if (unlikely(flags & MAP_HUGETLB))
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
file = fget(fd);
|
|
|
|
|
if (!file)
|
|
|
|
|
goto out;
|
|
|
|
|
} else if (flags & MAP_HUGETLB) {
|
|
|
|
|
struct user_struct *user = NULL;
|
|
|
|
|
/*
|
|
|
|
|
* VM_NORESERVE is used because the reservations will be
|
|
|
|
|
* taken when vm_ops->mmap() is called
|
|
|
|
|
* A dummy user value is used because we are not locking
|
|
|
|
|
* memory so no accounting is necessary
|
|
|
|
|
*/
|
2012-03-22 07:34:14 +08:00
|
|
|
file = hugetlb_file_setup(HUGETLB_ANON_FILE, addr, len,
|
|
|
|
|
VM_NORESERVE, &user,
|
|
|
|
|
HUGETLB_ANONHUGE_INODE);
|
2009-12-31 04:17:34 +08:00
|
|
|
if (IS_ERR(file))
|
|
|
|
|
return PTR_ERR(file);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
flags &= ~(MAP_EXECUTABLE | MAP_DENYWRITE);
|
|
|
|
|
|
|
|
|
|
down_write(¤t->mm->mmap_sem);
|
|
|
|
|
retval = do_mmap_pgoff(file, addr, len, prot, flags, pgoff);
|
|
|
|
|
up_write(¤t->mm->mmap_sem);
|
|
|
|
|
|
|
|
|
|
if (file)
|
|
|
|
|
fput(file);
|
|
|
|
|
out:
|
|
|
|
|
return retval;
|
|
|
|
|
}
|
|
|
|
|
|
2010-03-11 07:21:15 +08:00
|
|
|
#ifdef __ARCH_WANT_SYS_OLD_MMAP
|
|
|
|
|
struct mmap_arg_struct {
|
|
|
|
|
unsigned long addr;
|
|
|
|
|
unsigned long len;
|
|
|
|
|
unsigned long prot;
|
|
|
|
|
unsigned long flags;
|
|
|
|
|
unsigned long fd;
|
|
|
|
|
unsigned long offset;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
SYSCALL_DEFINE1(old_mmap, struct mmap_arg_struct __user *, arg)
|
|
|
|
|
{
|
|
|
|
|
struct mmap_arg_struct a;
|
|
|
|
|
|
|
|
|
|
if (copy_from_user(&a, arg, sizeof(a)))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
if (a.offset & ~PAGE_MASK)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
|
|
return sys_mmap_pgoff(a.addr, a.len, a.prot, a.flags, a.fd,
|
|
|
|
|
a.offset >> PAGE_SHIFT);
|
|
|
|
|
}
|
|
|
|
|
#endif /* __ARCH_WANT_SYS_OLD_MMAP */
|
|
|
|
|
|
2007-07-30 06:36:13 +08:00
|
|
|
/*
|
|
|
|
|
* Some shared mappigns will want the pages marked read-only
|
|
|
|
|
* to track write events. If so, we'll downgrade vm_page_prot
|
|
|
|
|
* to the private version (using protection_map[] without the
|
|
|
|
|
* VM_SHARED bit).
|
|
|
|
|
*/
|
|
|
|
|
int vma_wants_writenotify(struct vm_area_struct *vma)
|
|
|
|
|
{
|
2011-05-26 18:16:19 +08:00
|
|
|
vm_flags_t vm_flags = vma->vm_flags;
|
2007-07-30 06:36:13 +08:00
|
|
|
|
|
|
|
|
/* If it was private or non-writable, the write bit is already clear */
|
|
|
|
|
if ((vm_flags & (VM_WRITE|VM_SHARED)) != ((VM_WRITE|VM_SHARED)))
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
/* The backer wishes to know when pages are first written to? */
|
|
|
|
|
if (vma->vm_ops && vma->vm_ops->page_mkwrite)
|
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
|
|
/* The open routine did something to the protections already? */
|
|
|
|
|
if (pgprot_val(vma->vm_page_prot) !=
|
2007-10-19 14:39:15 +08:00
|
|
|
pgprot_val(vm_get_page_prot(vm_flags)))
|
2007-07-30 06:36:13 +08:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
/* Specialty mapping? */
|
|
|
|
|
if (vm_flags & (VM_PFNMAP|VM_INSERTPAGE))
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
/* Can the mapping track the dirty pages? */
|
|
|
|
|
return vma->vm_file && vma->vm_file->f_mapping &&
|
|
|
|
|
mapping_cap_account_dirty(vma->vm_file->f_mapping);
|
|
|
|
|
}
|
|
|
|
|
|
2009-02-01 07:08:56 +08:00
|
|
|
/*
|
|
|
|
|
* We account for memory if it's a private writeable mapping,
|
2009-02-10 22:02:27 +08:00
|
|
|
* not hugepages and VM_NORESERVE wasn't set.
|
2009-02-01 07:08:56 +08:00
|
|
|
*/
|
2011-05-26 18:16:19 +08:00
|
|
|
static inline int accountable_mapping(struct file *file, vm_flags_t vm_flags)
|
2009-02-01 07:08:56 +08:00
|
|
|
{
|
2009-02-10 22:02:27 +08:00
|
|
|
/*
|
|
|
|
|
* hugetlb has its own accounting separate from the core VM
|
|
|
|
|
* VM_HUGETLB may not be set yet so we cannot check for that flag.
|
|
|
|
|
*/
|
|
|
|
|
if (file && is_file_hugepages(file))
|
|
|
|
|
return 0;
|
|
|
|
|
|
2009-02-01 07:08:56 +08:00
|
|
|
return (vm_flags & (VM_NORESERVE | VM_SHARED | VM_WRITE)) == VM_WRITE;
|
|
|
|
|
}
|
|
|
|
|
|
2007-07-16 14:38:26 +08:00
|
|
|
unsigned long mmap_region(struct file *file, unsigned long addr,
|
|
|
|
|
unsigned long len, unsigned long flags,
|
2011-05-26 18:16:19 +08:00
|
|
|
vm_flags_t vm_flags, unsigned long pgoff)
|
2007-07-16 14:38:26 +08:00
|
|
|
{
|
|
|
|
|
struct mm_struct *mm = current->mm;
|
|
|
|
|
struct vm_area_struct *vma, *prev;
|
|
|
|
|
int correct_wcount = 0;
|
|
|
|
|
int error;
|
|
|
|
|
struct rb_node **rb_link, *rb_parent;
|
|
|
|
|
unsigned long charged = 0;
|
|
|
|
|
struct inode *inode = file ? file->f_path.dentry->d_inode : NULL;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* Clear old maps */
|
|
|
|
|
error = -ENOMEM;
|
|
|
|
|
munmap_back:
|
|
|
|
|
vma = find_vma_prepare(mm, addr, &prev, &rb_link, &rb_parent);
|
|
|
|
|
if (vma && vma->vm_start < addr + len) {
|
|
|
|
|
if (do_munmap(mm, addr, len))
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
goto munmap_back;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Check against address space limit. */
|
2005-05-01 23:58:35 +08:00
|
|
|
if (!may_expand_vm(mm, len >> PAGE_SHIFT))
|
2005-04-17 06:20:36 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2009-02-01 07:08:56 +08:00
|
|
|
/*
|
|
|
|
|
* Set 'VM_NORESERVE' if we should not account for the
|
2009-02-10 22:02:27 +08:00
|
|
|
* memory use of this mapping.
|
2009-02-01 07:08:56 +08:00
|
|
|
*/
|
2009-02-10 22:02:27 +08:00
|
|
|
if ((flags & MAP_NORESERVE)) {
|
|
|
|
|
/* We honor MAP_NORESERVE if allowed to overcommit */
|
|
|
|
|
if (sysctl_overcommit_memory != OVERCOMMIT_NEVER)
|
|
|
|
|
vm_flags |= VM_NORESERVE;
|
|
|
|
|
|
|
|
|
|
/* hugetlb applies strict overcommit unless MAP_NORESERVE */
|
|
|
|
|
if (file && is_file_hugepages(file))
|
|
|
|
|
vm_flags |= VM_NORESERVE;
|
|
|
|
|
}
|
2008-07-24 12:27:28 +08:00
|
|
|
|
2009-02-01 07:08:56 +08:00
|
|
|
/*
|
|
|
|
|
* Private writable mapping: check memory availability
|
|
|
|
|
*/
|
2009-02-10 22:02:27 +08:00
|
|
|
if (accountable_mapping(file, vm_flags)) {
|
2009-02-01 07:08:56 +08:00
|
|
|
charged = len >> PAGE_SHIFT;
|
2012-02-13 11:58:52 +08:00
|
|
|
if (security_vm_enough_memory_mm(mm, charged))
|
2009-02-01 07:08:56 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
vm_flags |= VM_ACCOUNT;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
2009-01-30 09:46:42 +08:00
|
|
|
* Can we just expand an old mapping?
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2009-01-30 09:46:42 +08:00
|
|
|
vma = vma_merge(mm, prev, addr, addr + len, vm_flags, NULL, file, pgoff, NULL);
|
|
|
|
|
if (vma)
|
|
|
|
|
goto out;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Determine the object being mapped and call the appropriate
|
|
|
|
|
* specific mapper. the address has already been validated, but
|
|
|
|
|
* not unmapped, but the maps are removed from the list.
|
|
|
|
|
*/
|
2006-03-25 19:06:43 +08:00
|
|
|
vma = kmem_cache_zalloc(vm_area_cachep, GFP_KERNEL);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (!vma) {
|
|
|
|
|
error = -ENOMEM;
|
|
|
|
|
goto unacct_error;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
vma->vm_mm = mm;
|
|
|
|
|
vma->vm_start = addr;
|
|
|
|
|
vma->vm_end = addr + len;
|
|
|
|
|
vma->vm_flags = vm_flags;
|
2007-10-19 14:39:15 +08:00
|
|
|
vma->vm_page_prot = vm_get_page_prot(vm_flags);
|
2005-04-17 06:20:36 +08:00
|
|
|
vma->vm_pgoff = pgoff;
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
INIT_LIST_HEAD(&vma->anon_vma_chain);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2012-03-07 04:28:52 +08:00
|
|
|
error = -EINVAL; /* when rejecting VM_GROWSDOWN|VM_GROWSUP */
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (file) {
|
|
|
|
|
if (vm_flags & (VM_GROWSDOWN|VM_GROWSUP))
|
|
|
|
|
goto free_vma;
|
|
|
|
|
if (vm_flags & VM_DENYWRITE) {
|
|
|
|
|
error = deny_write_access(file);
|
|
|
|
|
if (error)
|
|
|
|
|
goto free_vma;
|
|
|
|
|
correct_wcount = 1;
|
|
|
|
|
}
|
|
|
|
|
vma->vm_file = file;
|
|
|
|
|
get_file(file);
|
|
|
|
|
error = file->f_op->mmap(file, vma);
|
|
|
|
|
if (error)
|
|
|
|
|
goto unmap_and_free_vma;
|
2008-04-29 16:01:36 +08:00
|
|
|
if (vm_flags & VM_EXECUTABLE)
|
|
|
|
|
added_exe_file_vma(mm);
|
2009-09-22 08:03:41 +08:00
|
|
|
|
|
|
|
|
/* Can addr have changed??
|
|
|
|
|
*
|
|
|
|
|
* Answer: Yes, several device drivers can do it in their
|
|
|
|
|
* f_op->mmap method. -DaveM
|
|
|
|
|
*/
|
|
|
|
|
addr = vma->vm_start;
|
|
|
|
|
pgoff = vma->vm_pgoff;
|
|
|
|
|
vm_flags = vma->vm_flags;
|
2005-04-17 06:20:36 +08:00
|
|
|
} else if (vm_flags & VM_SHARED) {
|
2012-03-05 14:39:47 +08:00
|
|
|
if (unlikely(vm_flags & (VM_GROWSDOWN|VM_GROWSUP)))
|
|
|
|
|
goto free_vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
error = shmem_zero_setup(vma);
|
|
|
|
|
if (error)
|
|
|
|
|
goto free_vma;
|
|
|
|
|
}
|
|
|
|
|
|
2009-12-15 09:59:49 +08:00
|
|
|
if (vma_wants_writenotify(vma)) {
|
|
|
|
|
pgprot_t pprot = vma->vm_page_prot;
|
|
|
|
|
|
|
|
|
|
/* Can vma->vm_page_prot have changed??
|
|
|
|
|
*
|
|
|
|
|
* Answer: Yes, drivers may have changed it in their
|
|
|
|
|
* f_op->mmap method.
|
|
|
|
|
*
|
|
|
|
|
* Ensures that vmas marked as uncached stay that way.
|
|
|
|
|
*/
|
2007-10-23 11:45:12 +08:00
|
|
|
vma->vm_page_prot = vm_get_page_prot(vm_flags & ~VM_SHARED);
|
2009-12-15 09:59:49 +08:00
|
|
|
if (pgprot_val(pprot) == pgprot_val(pgprot_noncached(pprot)))
|
|
|
|
|
vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
|
|
|
|
|
}
|
2006-09-26 14:30:57 +08:00
|
|
|
|
2009-01-30 09:46:42 +08:00
|
|
|
vma_link(mm, vma, prev, rb_link, rb_parent);
|
|
|
|
|
file = vma->vm_file;
|
2008-04-28 17:12:10 +08:00
|
|
|
|
|
|
|
|
/* Once vma denies write, undo our temporary denial count */
|
|
|
|
|
if (correct_wcount)
|
|
|
|
|
atomic_inc(&inode->i_writecount);
|
|
|
|
|
out:
|
perf: Do the big rename: Performance Counters -> Performance Events
Bye-bye Performance Counters, welcome Performance Events!
In the past few months the perfcounters subsystem has grown out its
initial role of counting hardware events, and has become (and is
becoming) a much broader generic event enumeration, reporting, logging,
monitoring, analysis facility.
Naming its core object 'perf_counter' and naming the subsystem
'perfcounters' has become more and more of a misnomer. With pending
code like hw-breakpoints support the 'counter' name is less and
less appropriate.
All in one, we've decided to rename the subsystem to 'performance
events' and to propagate this rename through all fields, variables
and API names. (in an ABI compatible fashion)
The word 'event' is also a bit shorter than 'counter' - which makes
it slightly more convenient to write/handle as well.
Thanks goes to Stephane Eranian who first observed this misnomer and
suggested a rename.
User-space tooling and ABI compatibility is not affected - this patch
should be function-invariant. (Also, defconfigs were not touched to
keep the size down.)
This patch has been generated via the following script:
FILES=$(find * -type f | grep -vE 'oprofile|[^K]config')
sed -i \
-e 's/PERF_EVENT_/PERF_RECORD_/g' \
-e 's/PERF_COUNTER/PERF_EVENT/g' \
-e 's/perf_counter/perf_event/g' \
-e 's/nb_counters/nb_events/g' \
-e 's/swcounter/swevent/g' \
-e 's/tpcounter_event/tp_event/g' \
$FILES
for N in $(find . -name perf_counter.[ch]); do
M=$(echo $N | sed 's/perf_counter/perf_event/g')
mv $N $M
done
FILES=$(find . -name perf_event.*)
sed -i \
-e 's/COUNTER_MASK/REG_MASK/g' \
-e 's/COUNTER/EVENT/g' \
-e 's/\<event\>/event_id/g' \
-e 's/counter/event/g' \
-e 's/Counter/Event/g' \
$FILES
... to keep it as correct as possible. This script can also be
used by anyone who has pending perfcounters patches - it converts
a Linux kernel tree over to the new naming. We tried to time this
change to the point in time where the amount of pending patches
is the smallest: the end of the merge window.
Namespace clashes were fixed up in a preparatory patch - and some
stylistic fallout will be fixed up in a subsequent patch.
( NOTE: 'counters' are still the proper terminology when we deal
with hardware registers - and these sed scripts are a bit
over-eager in renaming them. I've undone some of that, but
in case there's something left where 'counter' would be
better than 'event' we can undo that on an individual basis
instead of touching an otherwise nicely automated patch. )
Suggested-by: Stephane Eranian <eranian@google.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Paul Mackerras <paulus@samba.org>
Reviewed-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: <linux-arch@vger.kernel.org>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-21 18:02:48 +08:00
|
|
|
perf_event_mmap(vma);
|
2009-03-31 01:07:05 +08:00
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
mm->total_vm += len >> PAGE_SHIFT;
|
2005-10-30 09:15:56 +08:00
|
|
|
vm_stat_account(mm, vm_flags, file, len >> PAGE_SHIFT);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (vm_flags & VM_LOCKED) {
|
2010-03-06 05:41:43 +08:00
|
|
|
if (!mlock_vma_pages_range(vma, addr, addr + len))
|
|
|
|
|
mm->locked_vm += (len >> PAGE_SHIFT);
|
2008-10-19 11:26:50 +08:00
|
|
|
} else if ((flags & MAP_POPULATE) && !(flags & MAP_NONBLOCK))
|
2007-07-19 16:46:59 +08:00
|
|
|
make_pages_present(addr, addr + len);
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
|
2012-02-17 16:27:41 +08:00
|
|
|
if (file && uprobe_mmap(vma))
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
/* matching probes but cannot insert */
|
|
|
|
|
goto unmap_and_free_vma;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
return addr;
|
|
|
|
|
|
|
|
|
|
unmap_and_free_vma:
|
|
|
|
|
if (correct_wcount)
|
|
|
|
|
atomic_inc(&inode->i_writecount);
|
|
|
|
|
vma->vm_file = NULL;
|
|
|
|
|
fput(file);
|
|
|
|
|
|
|
|
|
|
/* Undo any partial mapping done by a device driver. */
|
[PATCH] freepgt: free_pgtables use vma list
Recent woes with some arches needing their own pgd_addr_end macro; and 4-level
clear_page_range regression since 2.6.10's clear_page_tables; and its
long-standing well-known inefficiency in searching throughout the higher-level
page tables for those few entries to clear and free: all can be blamed on
ignoring the list of vmas when we free page tables.
Replace exit_mmap's clear_page_range of the total user address space by
free_pgtables operating on the mm's vma list; unmap_region use it in the same
way, giving floor and ceiling beyond which it may not free tables. This
brings lmbench fork/exec/sh numbers back to 2.6.10 (unless preempt is enabled,
in which case latency fixes spoil unmap_vmas throughput).
Beware: the do_mmap_pgoff driver failure case must now use unmap_region
instead of zap_page_range, since a page table might have been allocated, and
can only be freed while it is touched by some vma.
Move free_pgtables from mmap.c to memory.c, where its lower levels are adapted
from the clear_page_range levels. (Most of free_pgtables' old code was
actually for a non-existent case, prev not properly set up, dating from before
hch gave us split_vma.) Pass mmu_gather** in the public interfaces, since we
might want to add latency lockdrops later; but no attempt to do so yet, going
by vma should itself reduce latency.
But what if is_hugepage_only_range? Those ia64 and ppc64 cases need careful
examination: put that off until a later patch of the series.
What of x86_64's 32bit vdso page __map_syscall32 maps outside any vma?
And the range to sparc64's flush_tlb_pgtables? It's less clear to me now that
we need to do more than is done here - every PMD_SIZE ever occupied will be
flushed, do we really have to flush every PGDIR_SIZE ever partially occupied?
A shame to complicate it unnecessarily.
Special thanks to David Miller for time spent repairing my ceilings.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-20 04:29:15 +08:00
|
|
|
unmap_region(mm, vma, prev, vma->vm_start, vma->vm_end);
|
|
|
|
|
charged = 0;
|
2005-04-17 06:20:36 +08:00
|
|
|
free_vma:
|
|
|
|
|
kmem_cache_free(vm_area_cachep, vma);
|
|
|
|
|
unacct_error:
|
|
|
|
|
if (charged)
|
|
|
|
|
vm_unacct_memory(charged);
|
|
|
|
|
return error;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Get an address range which is currently unmapped.
|
|
|
|
|
* For shmat() with addr=0.
|
|
|
|
|
*
|
|
|
|
|
* Ugly calling convention alert:
|
|
|
|
|
* Return value with the low bits set means error value,
|
|
|
|
|
* ie
|
|
|
|
|
* if (ret & ~PAGE_MASK)
|
|
|
|
|
* error = ret;
|
|
|
|
|
*
|
|
|
|
|
* This function "knows" that -ENOMEM has the bits set.
|
|
|
|
|
*/
|
|
|
|
|
#ifndef HAVE_ARCH_UNMAPPED_AREA
|
|
|
|
|
unsigned long
|
|
|
|
|
arch_get_unmapped_area(struct file *filp, unsigned long addr,
|
|
|
|
|
unsigned long len, unsigned long pgoff, unsigned long flags)
|
|
|
|
|
{
|
|
|
|
|
struct mm_struct *mm = current->mm;
|
|
|
|
|
struct vm_area_struct *vma;
|
|
|
|
|
unsigned long start_addr;
|
|
|
|
|
|
|
|
|
|
if (len > TASK_SIZE)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2007-05-07 05:50:13 +08:00
|
|
|
if (flags & MAP_FIXED)
|
|
|
|
|
return addr;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (addr) {
|
|
|
|
|
addr = PAGE_ALIGN(addr);
|
|
|
|
|
vma = find_vma(mm, addr);
|
|
|
|
|
if (TASK_SIZE - len >= addr &&
|
|
|
|
|
(!vma || addr + len <= vma->vm_start))
|
|
|
|
|
return addr;
|
|
|
|
|
}
|
2005-06-22 08:14:49 +08:00
|
|
|
if (len > mm->cached_hole_size) {
|
|
|
|
|
start_addr = addr = mm->free_area_cache;
|
|
|
|
|
} else {
|
|
|
|
|
start_addr = addr = TASK_UNMAPPED_BASE;
|
|
|
|
|
mm->cached_hole_size = 0;
|
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
full_search:
|
|
|
|
|
for (vma = find_vma(mm, addr); ; vma = vma->vm_next) {
|
|
|
|
|
/* At this point: (!vma || addr < vma->vm_end). */
|
|
|
|
|
if (TASK_SIZE - len < addr) {
|
|
|
|
|
/*
|
|
|
|
|
* Start a new search - just in case we missed
|
|
|
|
|
* some holes.
|
|
|
|
|
*/
|
|
|
|
|
if (start_addr != TASK_UNMAPPED_BASE) {
|
2005-06-22 08:14:49 +08:00
|
|
|
addr = TASK_UNMAPPED_BASE;
|
|
|
|
|
start_addr = addr;
|
|
|
|
|
mm->cached_hole_size = 0;
|
2005-04-17 06:20:36 +08:00
|
|
|
goto full_search;
|
|
|
|
|
}
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
}
|
|
|
|
|
if (!vma || addr + len <= vma->vm_start) {
|
|
|
|
|
/*
|
|
|
|
|
* Remember the place where we stopped the search:
|
|
|
|
|
*/
|
|
|
|
|
mm->free_area_cache = addr + len;
|
|
|
|
|
return addr;
|
|
|
|
|
}
|
2005-06-22 08:14:49 +08:00
|
|
|
if (addr + mm->cached_hole_size < vma->vm_start)
|
|
|
|
|
mm->cached_hole_size = vma->vm_start - addr;
|
2005-04-17 06:20:36 +08:00
|
|
|
addr = vma->vm_end;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
2005-06-22 08:14:49 +08:00
|
|
|
void arch_unmap_area(struct mm_struct *mm, unsigned long addr)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
/*
|
|
|
|
|
* Is this a new hole at the lowest possible address?
|
|
|
|
|
*/
|
2012-03-22 07:33:56 +08:00
|
|
|
if (addr >= TASK_UNMAPPED_BASE && addr < mm->free_area_cache)
|
2005-06-22 08:14:49 +08:00
|
|
|
mm->free_area_cache = addr;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* This mmap-allocator allocates new areas top-down from below the
|
|
|
|
|
* stack's low limit (the base):
|
|
|
|
|
*/
|
|
|
|
|
#ifndef HAVE_ARCH_UNMAPPED_AREA_TOPDOWN
|
|
|
|
|
unsigned long
|
|
|
|
|
arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
|
|
|
|
|
const unsigned long len, const unsigned long pgoff,
|
|
|
|
|
const unsigned long flags)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct *vma;
|
|
|
|
|
struct mm_struct *mm = current->mm;
|
2012-03-22 07:33:56 +08:00
|
|
|
unsigned long addr = addr0, start_addr;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/* requested length too big for entire address space */
|
|
|
|
|
if (len > TASK_SIZE)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2007-05-07 05:50:13 +08:00
|
|
|
if (flags & MAP_FIXED)
|
|
|
|
|
return addr;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* requesting a specific address */
|
|
|
|
|
if (addr) {
|
|
|
|
|
addr = PAGE_ALIGN(addr);
|
|
|
|
|
vma = find_vma(mm, addr);
|
|
|
|
|
if (TASK_SIZE - len >= addr &&
|
|
|
|
|
(!vma || addr + len <= vma->vm_start))
|
|
|
|
|
return addr;
|
|
|
|
|
}
|
|
|
|
|
|
2005-06-22 08:14:49 +08:00
|
|
|
/* check if free_area_cache is useful for us */
|
|
|
|
|
if (len <= mm->cached_hole_size) {
|
|
|
|
|
mm->cached_hole_size = 0;
|
|
|
|
|
mm->free_area_cache = mm->mmap_base;
|
|
|
|
|
}
|
|
|
|
|
|
2012-03-22 07:33:56 +08:00
|
|
|
try_again:
|
2005-04-17 06:20:36 +08:00
|
|
|
/* either no address requested or can't fit in requested address hole */
|
2012-03-22 07:33:56 +08:00
|
|
|
start_addr = addr = mm->free_area_cache;
|
2005-06-22 08:14:52 +08:00
|
|
|
|
2012-03-22 07:33:56 +08:00
|
|
|
if (addr < len)
|
|
|
|
|
goto fail;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2012-03-22 07:33:56 +08:00
|
|
|
addr -= len;
|
2005-04-17 06:20:36 +08:00
|
|
|
do {
|
|
|
|
|
/*
|
|
|
|
|
* Lookup failure means no vma is above this address,
|
|
|
|
|
* else if new region fits below vma->vm_start,
|
|
|
|
|
* return with success:
|
|
|
|
|
*/
|
|
|
|
|
vma = find_vma(mm, addr);
|
|
|
|
|
if (!vma || addr+len <= vma->vm_start)
|
|
|
|
|
/* remember the address as a hint for next time */
|
|
|
|
|
return (mm->free_area_cache = addr);
|
|
|
|
|
|
2005-06-22 08:14:49 +08:00
|
|
|
/* remember the largest hole we saw so far */
|
|
|
|
|
if (addr + mm->cached_hole_size < vma->vm_start)
|
|
|
|
|
mm->cached_hole_size = vma->vm_start - addr;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* try just below the current vma->vm_start */
|
|
|
|
|
addr = vma->vm_start-len;
|
2005-05-19 06:39:33 +08:00
|
|
|
} while (len < vma->vm_start);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2012-03-22 07:33:56 +08:00
|
|
|
fail:
|
|
|
|
|
/*
|
|
|
|
|
* if hint left us with no space for the requested
|
|
|
|
|
* mapping then try again:
|
|
|
|
|
*
|
|
|
|
|
* Note: this is different with the case of bottomup
|
|
|
|
|
* which does the fully line-search, but we use find_vma
|
|
|
|
|
* here that causes some holes skipped.
|
|
|
|
|
*/
|
|
|
|
|
if (start_addr != mm->mmap_base) {
|
|
|
|
|
mm->free_area_cache = mm->mmap_base;
|
|
|
|
|
mm->cached_hole_size = 0;
|
|
|
|
|
goto try_again;
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* A failed mmap() very likely causes application failure,
|
|
|
|
|
* so fall back to the bottom-up function here. This scenario
|
|
|
|
|
* can happen with large stack limits and large mmap()
|
|
|
|
|
* allocations.
|
|
|
|
|
*/
|
2005-06-22 08:14:49 +08:00
|
|
|
mm->cached_hole_size = ~0UL;
|
|
|
|
|
mm->free_area_cache = TASK_UNMAPPED_BASE;
|
2005-04-17 06:20:36 +08:00
|
|
|
addr = arch_get_unmapped_area(filp, addr0, len, pgoff, flags);
|
|
|
|
|
/*
|
|
|
|
|
* Restore the topdown base:
|
|
|
|
|
*/
|
|
|
|
|
mm->free_area_cache = mm->mmap_base;
|
2005-06-22 08:14:49 +08:00
|
|
|
mm->cached_hole_size = ~0UL;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
return addr;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
2005-06-22 08:14:49 +08:00
|
|
|
void arch_unmap_area_topdown(struct mm_struct *mm, unsigned long addr)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
/*
|
|
|
|
|
* Is this a new hole at the highest possible address?
|
|
|
|
|
*/
|
2005-06-22 08:14:49 +08:00
|
|
|
if (addr > mm->free_area_cache)
|
|
|
|
|
mm->free_area_cache = addr;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/* dont allow allocations above current base */
|
2005-06-22 08:14:49 +08:00
|
|
|
if (mm->free_area_cache > mm->mmap_base)
|
|
|
|
|
mm->free_area_cache = mm->mmap_base;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
unsigned long
|
|
|
|
|
get_unmapped_area(struct file *file, unsigned long addr, unsigned long len,
|
|
|
|
|
unsigned long pgoff, unsigned long flags)
|
|
|
|
|
{
|
2007-05-07 05:50:13 +08:00
|
|
|
unsigned long (*get_area)(struct file *, unsigned long,
|
|
|
|
|
unsigned long, unsigned long, unsigned long);
|
|
|
|
|
|
2009-12-04 04:23:11 +08:00
|
|
|
unsigned long error = arch_mmap_check(addr, len, flags);
|
|
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
|
|
|
|
|
/* Careful about overflows.. */
|
|
|
|
|
if (len > TASK_SIZE)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2007-05-07 05:50:13 +08:00
|
|
|
get_area = current->mm->get_unmapped_area;
|
|
|
|
|
if (file && file->f_op && file->f_op->get_unmapped_area)
|
|
|
|
|
get_area = file->f_op->get_unmapped_area;
|
|
|
|
|
addr = get_area(file, addr, len, pgoff, flags);
|
|
|
|
|
if (IS_ERR_VALUE(addr))
|
|
|
|
|
return addr;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2005-05-20 13:43:37 +08:00
|
|
|
if (addr > TASK_SIZE - len)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
if (addr & ~PAGE_MASK)
|
|
|
|
|
return -EINVAL;
|
2007-05-07 05:50:13 +08:00
|
|
|
|
2008-02-05 14:29:16 +08:00
|
|
|
return arch_rebalance_pgtables(addr, len);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(get_unmapped_area);
|
|
|
|
|
|
|
|
|
|
/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
|
2009-01-07 06:40:21 +08:00
|
|
|
struct vm_area_struct *find_vma(struct mm_struct *mm, unsigned long addr)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
struct vm_area_struct *vma = NULL;
|
|
|
|
|
|
|
|
|
|
if (mm) {
|
|
|
|
|
/* Check the cache first. */
|
|
|
|
|
/* (Cache hit rate is typically around 35%.) */
|
|
|
|
|
vma = mm->mmap_cache;
|
|
|
|
|
if (!(vma && vma->vm_end > addr && vma->vm_start <= addr)) {
|
|
|
|
|
struct rb_node * rb_node;
|
|
|
|
|
|
|
|
|
|
rb_node = mm->mm_rb.rb_node;
|
|
|
|
|
vma = NULL;
|
|
|
|
|
|
|
|
|
|
while (rb_node) {
|
|
|
|
|
struct vm_area_struct * vma_tmp;
|
|
|
|
|
|
|
|
|
|
vma_tmp = rb_entry(rb_node,
|
|
|
|
|
struct vm_area_struct, vm_rb);
|
|
|
|
|
|
|
|
|
|
if (vma_tmp->vm_end > addr) {
|
|
|
|
|
vma = vma_tmp;
|
|
|
|
|
if (vma_tmp->vm_start <= addr)
|
|
|
|
|
break;
|
|
|
|
|
rb_node = rb_node->rb_left;
|
|
|
|
|
} else
|
|
|
|
|
rb_node = rb_node->rb_right;
|
|
|
|
|
}
|
|
|
|
|
if (vma)
|
|
|
|
|
mm->mmap_cache = vma;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return vma;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EXPORT_SYMBOL(find_vma);
|
|
|
|
|
|
2012-01-11 07:08:07 +08:00
|
|
|
/*
|
|
|
|
|
* Same as find_vma, but also return a pointer to the previous VMA in *pprev.
|
|
|
|
|
*/
|
2005-04-17 06:20:36 +08:00
|
|
|
struct vm_area_struct *
|
|
|
|
|
find_vma_prev(struct mm_struct *mm, unsigned long addr,
|
|
|
|
|
struct vm_area_struct **pprev)
|
|
|
|
|
{
|
2012-01-11 07:08:07 +08:00
|
|
|
struct vm_area_struct *vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2012-01-11 07:08:07 +08:00
|
|
|
vma = find_vma(mm, addr);
|
2012-03-05 08:52:03 +08:00
|
|
|
if (vma) {
|
|
|
|
|
*pprev = vma->vm_prev;
|
|
|
|
|
} else {
|
|
|
|
|
struct rb_node *rb_node = mm->mm_rb.rb_node;
|
|
|
|
|
*pprev = NULL;
|
|
|
|
|
while (rb_node) {
|
|
|
|
|
*pprev = rb_entry(rb_node, struct vm_area_struct, vm_rb);
|
|
|
|
|
rb_node = rb_node->rb_right;
|
|
|
|
|
}
|
|
|
|
|
}
|
2012-01-11 07:08:07 +08:00
|
|
|
return vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Verify that the stack growth is acceptable and
|
|
|
|
|
* update accounting. This is shared with both the
|
|
|
|
|
* grow-up and grow-down cases.
|
|
|
|
|
*/
|
2009-01-07 06:40:21 +08:00
|
|
|
static int acct_stack_growth(struct vm_area_struct *vma, unsigned long size, unsigned long grow)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
struct mm_struct *mm = vma->vm_mm;
|
|
|
|
|
struct rlimit *rlim = current->signal->rlim;
|
2007-01-31 06:35:39 +08:00
|
|
|
unsigned long new_start;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/* address space limit tests */
|
2005-05-01 23:58:35 +08:00
|
|
|
if (!may_expand_vm(mm, grow))
|
2005-04-17 06:20:36 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
/* Stack limit test */
|
2010-03-06 05:41:44 +08:00
|
|
|
if (size > ACCESS_ONCE(rlim[RLIMIT_STACK].rlim_cur))
|
2005-04-17 06:20:36 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
/* mlock limit tests */
|
|
|
|
|
if (vma->vm_flags & VM_LOCKED) {
|
|
|
|
|
unsigned long locked;
|
|
|
|
|
unsigned long limit;
|
|
|
|
|
locked = mm->locked_vm + grow;
|
2010-03-06 05:41:44 +08:00
|
|
|
limit = ACCESS_ONCE(rlim[RLIMIT_MEMLOCK].rlim_cur);
|
|
|
|
|
limit >>= PAGE_SHIFT;
|
2005-04-17 06:20:36 +08:00
|
|
|
if (locked > limit && !capable(CAP_IPC_LOCK))
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
}
|
|
|
|
|
|
2007-01-31 06:35:39 +08:00
|
|
|
/* Check to ensure the stack will not grow into a hugetlb-only region */
|
|
|
|
|
new_start = (vma->vm_flags & VM_GROWSUP) ? vma->vm_start :
|
|
|
|
|
vma->vm_end - size;
|
|
|
|
|
if (is_hugepage_only_range(vma->vm_mm, new_start, size))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* Overcommit.. This must be the final test, as it will
|
|
|
|
|
* update security statistics.
|
|
|
|
|
*/
|
2009-04-17 04:58:12 +08:00
|
|
|
if (security_vm_enough_memory_mm(mm, grow))
|
2005-04-17 06:20:36 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
/* Ok, everything looks good - let it rip */
|
|
|
|
|
mm->total_vm += grow;
|
|
|
|
|
if (vma->vm_flags & VM_LOCKED)
|
|
|
|
|
mm->locked_vm += grow;
|
2005-10-30 09:15:56 +08:00
|
|
|
vm_stat_account(mm, vma->vm_flags, vma->vm_file, grow);
|
2005-04-17 06:20:36 +08:00
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
2005-10-30 09:16:20 +08:00
|
|
|
#if defined(CONFIG_STACK_GROWSUP) || defined(CONFIG_IA64)
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
2005-10-30 09:16:20 +08:00
|
|
|
* PA-RISC uses this for its stack; IA64 for its Register Backing Store.
|
|
|
|
|
* vma is the last one with address > vma->vm_end. Have to extend vma.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2005-10-30 09:16:20 +08:00
|
|
|
int expand_upwards(struct vm_area_struct *vma, unsigned long address)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
int error;
|
|
|
|
|
|
|
|
|
|
if (!(vma->vm_flags & VM_GROWSUP))
|
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We must make sure the anon_vma is allocated
|
|
|
|
|
* so that the anon_vma locking is not a noop.
|
|
|
|
|
*/
|
|
|
|
|
if (unlikely(anon_vma_prepare(vma)))
|
|
|
|
|
return -ENOMEM;
|
2010-08-10 08:18:37 +08:00
|
|
|
vma_lock_anon_vma(vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* vma->vm_start/vm_end cannot change under us because the caller
|
|
|
|
|
* is required to hold the mmap_sem in read mode. We need the
|
|
|
|
|
* anon_vma lock to serialize against concurrent expand_stacks.
|
2006-12-20 02:28:33 +08:00
|
|
|
* Also guard against wrapping around to address 0.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2006-12-20 02:28:33 +08:00
|
|
|
if (address < PAGE_ALIGN(address+4))
|
|
|
|
|
address = PAGE_ALIGN(address+4);
|
|
|
|
|
else {
|
2010-08-10 08:18:37 +08:00
|
|
|
vma_unlock_anon_vma(vma);
|
2006-12-20 02:28:33 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
error = 0;
|
|
|
|
|
|
|
|
|
|
/* Somebody else might have raced and expanded it already */
|
|
|
|
|
if (address > vma->vm_end) {
|
|
|
|
|
unsigned long size, grow;
|
|
|
|
|
|
|
|
|
|
size = address - vma->vm_start;
|
|
|
|
|
grow = (address - vma->vm_end) >> PAGE_SHIFT;
|
|
|
|
|
|
2011-05-10 08:44:42 +08:00
|
|
|
error = -ENOMEM;
|
|
|
|
|
if (vma->vm_pgoff + (size >> PAGE_SHIFT) >= vma->vm_pgoff) {
|
|
|
|
|
error = acct_stack_growth(vma, size, grow);
|
|
|
|
|
if (!error) {
|
|
|
|
|
vma->vm_end = address;
|
|
|
|
|
perf_event_mmap(vma);
|
|
|
|
|
}
|
2010-05-18 22:30:49 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
2010-08-10 08:18:37 +08:00
|
|
|
vma_unlock_anon_vma(vma);
|
2011-01-14 07:46:59 +08:00
|
|
|
khugepaged_enter_vma_merge(vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
return error;
|
|
|
|
|
}
|
2005-10-30 09:16:20 +08:00
|
|
|
#endif /* CONFIG_STACK_GROWSUP || CONFIG_IA64 */
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* vma is the first one with address < vma->vm_start. Have to extend vma.
|
|
|
|
|
*/
|
2011-05-25 08:11:44 +08:00
|
|
|
int expand_downwards(struct vm_area_struct *vma,
|
2007-07-19 16:48:16 +08:00
|
|
|
unsigned long address)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
int error;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* We must make sure the anon_vma is allocated
|
|
|
|
|
* so that the anon_vma locking is not a noop.
|
|
|
|
|
*/
|
|
|
|
|
if (unlikely(anon_vma_prepare(vma)))
|
|
|
|
|
return -ENOMEM;
|
2007-11-27 07:47:26 +08:00
|
|
|
|
|
|
|
|
address &= PAGE_MASK;
|
2007-12-08 19:02:48 +08:00
|
|
|
error = security_file_mmap(NULL, 0, 0, 0, address, 1);
|
2007-11-27 07:47:26 +08:00
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
|
2010-08-10 08:18:37 +08:00
|
|
|
vma_lock_anon_vma(vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* vma->vm_start/vm_end cannot change under us because the caller
|
|
|
|
|
* is required to hold the mmap_sem in read mode. We need the
|
|
|
|
|
* anon_vma lock to serialize against concurrent expand_stacks.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* Somebody else might have raced and expanded it already */
|
|
|
|
|
if (address < vma->vm_start) {
|
|
|
|
|
unsigned long size, grow;
|
|
|
|
|
|
|
|
|
|
size = vma->vm_end - address;
|
|
|
|
|
grow = (vma->vm_start - address) >> PAGE_SHIFT;
|
|
|
|
|
|
2011-04-13 23:07:28 +08:00
|
|
|
error = -ENOMEM;
|
|
|
|
|
if (grow <= vma->vm_pgoff) {
|
|
|
|
|
error = acct_stack_growth(vma, size, grow);
|
|
|
|
|
if (!error) {
|
|
|
|
|
vma->vm_start = address;
|
|
|
|
|
vma->vm_pgoff -= grow;
|
|
|
|
|
perf_event_mmap(vma);
|
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
}
|
2010-08-10 08:18:37 +08:00
|
|
|
vma_unlock_anon_vma(vma);
|
2011-01-14 07:46:59 +08:00
|
|
|
khugepaged_enter_vma_merge(vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
return error;
|
|
|
|
|
}
|
|
|
|
|
|
2007-07-19 16:48:16 +08:00
|
|
|
#ifdef CONFIG_STACK_GROWSUP
|
|
|
|
|
int expand_stack(struct vm_area_struct *vma, unsigned long address)
|
|
|
|
|
{
|
|
|
|
|
return expand_upwards(vma, address);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct vm_area_struct *
|
|
|
|
|
find_extend_vma(struct mm_struct *mm, unsigned long addr)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct *vma, *prev;
|
|
|
|
|
|
|
|
|
|
addr &= PAGE_MASK;
|
|
|
|
|
vma = find_vma_prev(mm, addr, &prev);
|
|
|
|
|
if (vma && (vma->vm_start <= addr))
|
|
|
|
|
return vma;
|
2008-11-12 08:24:41 +08:00
|
|
|
if (!prev || expand_stack(prev, addr))
|
2007-07-19 16:48:16 +08:00
|
|
|
return NULL;
|
2008-10-19 11:26:50 +08:00
|
|
|
if (prev->vm_flags & VM_LOCKED) {
|
2010-03-06 05:41:43 +08:00
|
|
|
mlock_vma_pages_range(prev, addr, prev->vm_end);
|
2008-10-19 11:26:50 +08:00
|
|
|
}
|
2007-07-19 16:48:16 +08:00
|
|
|
return prev;
|
|
|
|
|
}
|
|
|
|
|
#else
|
|
|
|
|
int expand_stack(struct vm_area_struct *vma, unsigned long address)
|
|
|
|
|
{
|
|
|
|
|
return expand_downwards(vma, address);
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
struct vm_area_struct *
|
|
|
|
|
find_extend_vma(struct mm_struct * mm, unsigned long addr)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct * vma;
|
|
|
|
|
unsigned long start;
|
|
|
|
|
|
|
|
|
|
addr &= PAGE_MASK;
|
|
|
|
|
vma = find_vma(mm,addr);
|
|
|
|
|
if (!vma)
|
|
|
|
|
return NULL;
|
|
|
|
|
if (vma->vm_start <= addr)
|
|
|
|
|
return vma;
|
|
|
|
|
if (!(vma->vm_flags & VM_GROWSDOWN))
|
|
|
|
|
return NULL;
|
|
|
|
|
start = vma->vm_start;
|
|
|
|
|
if (expand_stack(vma, addr))
|
|
|
|
|
return NULL;
|
2008-10-19 11:26:50 +08:00
|
|
|
if (vma->vm_flags & VM_LOCKED) {
|
2010-03-06 05:41:43 +08:00
|
|
|
mlock_vma_pages_range(vma, addr, start);
|
2008-10-19 11:26:50 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
return vma;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
/*
|
2005-10-30 09:15:56 +08:00
|
|
|
* Ok - we have the memory areas we should free on the vma list,
|
2005-04-17 06:20:36 +08:00
|
|
|
* so release them, and do the vma updates.
|
2005-10-30 09:15:56 +08:00
|
|
|
*
|
|
|
|
|
* Called with the mm semaphore held.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2005-10-30 09:15:56 +08:00
|
|
|
static void remove_vma_list(struct mm_struct *mm, struct vm_area_struct *vma)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
2012-05-07 04:54:06 +08:00
|
|
|
unsigned long nr_accounted = 0;
|
|
|
|
|
|
[PATCH] mm: update_hiwaters just in time
update_mem_hiwater has attracted various criticisms, in particular from those
concerned with mm scalability. Originally it was called whenever rss or
total_vm got raised. Then many of those callsites were replaced by a timer
tick call from account_system_time. Now Frank van Maarseveen reports that to
be found inadequate. How about this? Works for Frank.
Replace update_mem_hiwater, a poor combination of two unrelated ops, by macros
update_hiwater_rss and update_hiwater_vm. Don't attempt to keep
mm->hiwater_rss up to date at timer tick, nor every time we raise rss (usually
by 1): those are hot paths. Do the opposite, update only when about to lower
rss (usually by many), or just before final accounting in do_exit. Handle
mm->hiwater_vm in the same way, though it's much less of an issue. Demand
that whoever collects these hiwater statistics do the work of taking the
maximum with rss or total_vm.
And there has been no collector of these hiwater statistics in the tree. The
new convention needs an example, so match Frank's usage by adding a VmPeak
line above VmSize to /proc/<pid>/status, and also a VmHWM line above VmRSS
(High-Water-Mark or High-Water-Memory).
There was a particular anomaly during mremap move, that hiwater_vm might be
captured too high. A fleeting such anomaly remains, but it's quickly
corrected now, whereas before it would stick.
What locking? None: if the app is racy then these statistics will be racy,
it's not worth any overhead to make them exact. But whenever it suits,
hiwater_vm is updated under exclusive mmap_sem, and hiwater_rss under
page_table_lock (for now) or with preemption disabled (later on): without
going to any trouble, minimize the time between reading current values and
updating, to minimize those occasions when a racing thread bumps a count up
and back down in between.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 09:16:18 +08:00
|
|
|
/* Update high watermark before we lower total_vm */
|
|
|
|
|
update_hiwater_vm(mm);
|
2005-04-17 06:20:36 +08:00
|
|
|
do {
|
2005-10-30 09:15:56 +08:00
|
|
|
long nrpages = vma_pages(vma);
|
|
|
|
|
|
2012-05-07 04:54:06 +08:00
|
|
|
if (vma->vm_flags & VM_ACCOUNT)
|
|
|
|
|
nr_accounted += nrpages;
|
2005-10-30 09:15:56 +08:00
|
|
|
mm->total_vm -= nrpages;
|
|
|
|
|
vm_stat_account(mm, vma->vm_flags, vma->vm_file, -nrpages);
|
2005-10-30 09:15:57 +08:00
|
|
|
vma = remove_vma(vma);
|
2005-04-20 04:29:18 +08:00
|
|
|
} while (vma);
|
2012-05-07 04:54:06 +08:00
|
|
|
vm_unacct_memory(nr_accounted);
|
2005-04-17 06:20:36 +08:00
|
|
|
validate_mm(mm);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Get rid of page table information in the indicated region.
|
|
|
|
|
*
|
2005-09-22 00:55:37 +08:00
|
|
|
* Called with the mm semaphore held.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
|
|
|
|
static void unmap_region(struct mm_struct *mm,
|
[PATCH] freepgt: free_pgtables use vma list
Recent woes with some arches needing their own pgd_addr_end macro; and 4-level
clear_page_range regression since 2.6.10's clear_page_tables; and its
long-standing well-known inefficiency in searching throughout the higher-level
page tables for those few entries to clear and free: all can be blamed on
ignoring the list of vmas when we free page tables.
Replace exit_mmap's clear_page_range of the total user address space by
free_pgtables operating on the mm's vma list; unmap_region use it in the same
way, giving floor and ceiling beyond which it may not free tables. This
brings lmbench fork/exec/sh numbers back to 2.6.10 (unless preempt is enabled,
in which case latency fixes spoil unmap_vmas throughput).
Beware: the do_mmap_pgoff driver failure case must now use unmap_region
instead of zap_page_range, since a page table might have been allocated, and
can only be freed while it is touched by some vma.
Move free_pgtables from mmap.c to memory.c, where its lower levels are adapted
from the clear_page_range levels. (Most of free_pgtables' old code was
actually for a non-existent case, prev not properly set up, dating from before
hch gave us split_vma.) Pass mmu_gather** in the public interfaces, since we
might want to add latency lockdrops later; but no attempt to do so yet, going
by vma should itself reduce latency.
But what if is_hugepage_only_range? Those ia64 and ppc64 cases need careful
examination: put that off until a later patch of the series.
What of x86_64's 32bit vdso page __map_syscall32 maps outside any vma?
And the range to sparc64's flush_tlb_pgtables? It's less clear to me now that
we need to do more than is done here - every PMD_SIZE ever occupied will be
flushed, do we really have to flush every PGDIR_SIZE ever partially occupied?
A shame to complicate it unnecessarily.
Special thanks to David Miller for time spent repairing my ceilings.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-20 04:29:15 +08:00
|
|
|
struct vm_area_struct *vma, struct vm_area_struct *prev,
|
|
|
|
|
unsigned long start, unsigned long end)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
[PATCH] freepgt: free_pgtables use vma list
Recent woes with some arches needing their own pgd_addr_end macro; and 4-level
clear_page_range regression since 2.6.10's clear_page_tables; and its
long-standing well-known inefficiency in searching throughout the higher-level
page tables for those few entries to clear and free: all can be blamed on
ignoring the list of vmas when we free page tables.
Replace exit_mmap's clear_page_range of the total user address space by
free_pgtables operating on the mm's vma list; unmap_region use it in the same
way, giving floor and ceiling beyond which it may not free tables. This
brings lmbench fork/exec/sh numbers back to 2.6.10 (unless preempt is enabled,
in which case latency fixes spoil unmap_vmas throughput).
Beware: the do_mmap_pgoff driver failure case must now use unmap_region
instead of zap_page_range, since a page table might have been allocated, and
can only be freed while it is touched by some vma.
Move free_pgtables from mmap.c to memory.c, where its lower levels are adapted
from the clear_page_range levels. (Most of free_pgtables' old code was
actually for a non-existent case, prev not properly set up, dating from before
hch gave us split_vma.) Pass mmu_gather** in the public interfaces, since we
might want to add latency lockdrops later; but no attempt to do so yet, going
by vma should itself reduce latency.
But what if is_hugepage_only_range? Those ia64 and ppc64 cases need careful
examination: put that off until a later patch of the series.
What of x86_64's 32bit vdso page __map_syscall32 maps outside any vma?
And the range to sparc64's flush_tlb_pgtables? It's less clear to me now that
we need to do more than is done here - every PMD_SIZE ever occupied will be
flushed, do we really have to flush every PGDIR_SIZE ever partially occupied?
A shame to complicate it unnecessarily.
Special thanks to David Miller for time spent repairing my ceilings.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-20 04:29:15 +08:00
|
|
|
struct vm_area_struct *next = prev? prev->vm_next: mm->mmap;
|
2011-05-25 08:11:45 +08:00
|
|
|
struct mmu_gather tlb;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
lru_add_drain();
|
2011-05-25 08:11:45 +08:00
|
|
|
tlb_gather_mmu(&tlb, mm, 0);
|
[PATCH] mm: update_hiwaters just in time
update_mem_hiwater has attracted various criticisms, in particular from those
concerned with mm scalability. Originally it was called whenever rss or
total_vm got raised. Then many of those callsites were replaced by a timer
tick call from account_system_time. Now Frank van Maarseveen reports that to
be found inadequate. How about this? Works for Frank.
Replace update_mem_hiwater, a poor combination of two unrelated ops, by macros
update_hiwater_rss and update_hiwater_vm. Don't attempt to keep
mm->hiwater_rss up to date at timer tick, nor every time we raise rss (usually
by 1): those are hot paths. Do the opposite, update only when about to lower
rss (usually by many), or just before final accounting in do_exit. Handle
mm->hiwater_vm in the same way, though it's much less of an issue. Demand
that whoever collects these hiwater statistics do the work of taking the
maximum with rss or total_vm.
And there has been no collector of these hiwater statistics in the tree. The
new convention needs an example, so match Frank's usage by adding a VmPeak
line above VmSize to /proc/<pid>/status, and also a VmHWM line above VmRSS
(High-Water-Mark or High-Water-Memory).
There was a particular anomaly during mremap move, that hiwater_vm might be
captured too high. A fleeting such anomaly remains, but it's quickly
corrected now, whereas before it would stick.
What locking? None: if the app is racy then these statistics will be racy,
it's not worth any overhead to make them exact. But whenever it suits,
hiwater_vm is updated under exclusive mmap_sem, and hiwater_rss under
page_table_lock (for now) or with preemption disabled (later on): without
going to any trouble, minimize the time between reading current values and
updating, to minimize those occasions when a racing thread bumps a count up
and back down in between.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 09:16:18 +08:00
|
|
|
update_hiwater_rss(mm);
|
2012-05-07 04:54:06 +08:00
|
|
|
unmap_vmas(&tlb, vma, start, end);
|
2011-05-25 08:11:45 +08:00
|
|
|
free_pgtables(&tlb, vma, prev ? prev->vm_end : FIRST_USER_ADDRESS,
|
|
|
|
|
next ? next->vm_start : 0);
|
|
|
|
|
tlb_finish_mmu(&tlb, start, end);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Create a list of vma's touched by the unmap, removing them from the mm's
|
|
|
|
|
* vma list as we go..
|
|
|
|
|
*/
|
|
|
|
|
static void
|
|
|
|
|
detach_vmas_to_be_unmapped(struct mm_struct *mm, struct vm_area_struct *vma,
|
|
|
|
|
struct vm_area_struct *prev, unsigned long end)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct **insertion_point;
|
|
|
|
|
struct vm_area_struct *tail_vma = NULL;
|
2005-06-22 08:14:49 +08:00
|
|
|
unsigned long addr;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
insertion_point = (prev ? &prev->vm_next : &mm->mmap);
|
2010-08-21 07:24:55 +08:00
|
|
|
vma->vm_prev = NULL;
|
2005-04-17 06:20:36 +08:00
|
|
|
do {
|
|
|
|
|
rb_erase(&vma->vm_rb, &mm->mm_rb);
|
|
|
|
|
mm->map_count--;
|
|
|
|
|
tail_vma = vma;
|
|
|
|
|
vma = vma->vm_next;
|
|
|
|
|
} while (vma && vma->vm_start < end);
|
|
|
|
|
*insertion_point = vma;
|
2010-08-21 07:24:55 +08:00
|
|
|
if (vma)
|
|
|
|
|
vma->vm_prev = prev;
|
2005-04-17 06:20:36 +08:00
|
|
|
tail_vma->vm_next = NULL;
|
2005-06-22 08:14:49 +08:00
|
|
|
if (mm->unmap_area == arch_unmap_area)
|
|
|
|
|
addr = prev ? prev->vm_end : mm->mmap_base;
|
|
|
|
|
else
|
|
|
|
|
addr = vma ? vma->vm_start : mm->mmap_base;
|
|
|
|
|
mm->unmap_area(mm, addr);
|
2005-04-17 06:20:36 +08:00
|
|
|
mm->mmap_cache = NULL; /* Kill the cache. */
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
mmap: don't return ENOMEM when mapcount is temporarily exceeded in munmap()
On ia64, the following test program exit abnormally, because glibc thread
library called abort().
========================================================
(gdb) bt
#0 0xa000000000010620 in __kernel_syscall_via_break ()
#1 0x20000000003208e0 in raise () from /lib/libc.so.6.1
#2 0x2000000000324090 in abort () from /lib/libc.so.6.1
#3 0x200000000027c3e0 in __deallocate_stack () from /lib/libpthread.so.0
#4 0x200000000027f7c0 in start_thread () from /lib/libpthread.so.0
#5 0x200000000047ef60 in __clone2 () from /lib/libc.so.6.1
========================================================
The fact is, glibc call munmap() when thread exitng time for freeing
stack, and it assume munlock() never fail. However, munmap() often make
vma splitting and it with many mapcount make -ENOMEM.
Oh well, that's crazy, because stack unmapping never increase mapcount.
The maxcount exceeding is only temporary. internal temporary exceeding
shouldn't make ENOMEM.
This patch does it.
test_max_mapcount.c
==================================================================
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<pthread.h>
#include<errno.h>
#include<unistd.h>
#define THREAD_NUM 30000
#define MAL_SIZE (8*1024*1024)
void *wait_thread(void *args)
{
void *addr;
addr = malloc(MAL_SIZE);
sleep(10);
return NULL;
}
void *wait_thread2(void *args)
{
sleep(60);
return NULL;
}
int main(int argc, char *argv[])
{
int i;
pthread_t thread[THREAD_NUM], th;
int ret, count = 0;
pthread_attr_t attr;
ret = pthread_attr_init(&attr);
if(ret) {
perror("pthread_attr_init");
}
ret = pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
if(ret) {
perror("pthread_attr_setdetachstate");
}
for (i = 0; i < THREAD_NUM; i++) {
ret = pthread_create(&th, &attr, wait_thread, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
ret = pthread_create(&thread[i], &attr, wait_thread2, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
}
sleep(3600);
return 0;
}
==================================================================
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 09:57:56 +08:00
|
|
|
* __split_vma() bypasses sysctl_max_map_count checking. We use this on the
|
|
|
|
|
* munmap path where it doesn't make sense to fail.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
mmap: don't return ENOMEM when mapcount is temporarily exceeded in munmap()
On ia64, the following test program exit abnormally, because glibc thread
library called abort().
========================================================
(gdb) bt
#0 0xa000000000010620 in __kernel_syscall_via_break ()
#1 0x20000000003208e0 in raise () from /lib/libc.so.6.1
#2 0x2000000000324090 in abort () from /lib/libc.so.6.1
#3 0x200000000027c3e0 in __deallocate_stack () from /lib/libpthread.so.0
#4 0x200000000027f7c0 in start_thread () from /lib/libpthread.so.0
#5 0x200000000047ef60 in __clone2 () from /lib/libc.so.6.1
========================================================
The fact is, glibc call munmap() when thread exitng time for freeing
stack, and it assume munlock() never fail. However, munmap() often make
vma splitting and it with many mapcount make -ENOMEM.
Oh well, that's crazy, because stack unmapping never increase mapcount.
The maxcount exceeding is only temporary. internal temporary exceeding
shouldn't make ENOMEM.
This patch does it.
test_max_mapcount.c
==================================================================
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<pthread.h>
#include<errno.h>
#include<unistd.h>
#define THREAD_NUM 30000
#define MAL_SIZE (8*1024*1024)
void *wait_thread(void *args)
{
void *addr;
addr = malloc(MAL_SIZE);
sleep(10);
return NULL;
}
void *wait_thread2(void *args)
{
sleep(60);
return NULL;
}
int main(int argc, char *argv[])
{
int i;
pthread_t thread[THREAD_NUM], th;
int ret, count = 0;
pthread_attr_t attr;
ret = pthread_attr_init(&attr);
if(ret) {
perror("pthread_attr_init");
}
ret = pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
if(ret) {
perror("pthread_attr_setdetachstate");
}
for (i = 0; i < THREAD_NUM; i++) {
ret = pthread_create(&th, &attr, wait_thread, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
ret = pthread_create(&thread[i], &attr, wait_thread2, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
}
sleep(3600);
return 0;
}
==================================================================
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 09:57:56 +08:00
|
|
|
static int __split_vma(struct mm_struct * mm, struct vm_area_struct * vma,
|
2005-04-17 06:20:36 +08:00
|
|
|
unsigned long addr, int new_below)
|
|
|
|
|
{
|
|
|
|
|
struct mempolicy *pol;
|
|
|
|
|
struct vm_area_struct *new;
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
int err = -ENOMEM;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2008-07-24 12:27:41 +08:00
|
|
|
if (is_vm_hugetlb_page(vma) && (addr &
|
|
|
|
|
~(huge_page_mask(hstate_vma(vma)))))
|
2005-04-17 06:20:36 +08:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
2006-12-07 12:33:17 +08:00
|
|
|
new = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (!new)
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
goto out_err;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/* most fields are the same, copy all, and then fixup */
|
|
|
|
|
*new = *vma;
|
|
|
|
|
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
INIT_LIST_HEAD(&new->anon_vma_chain);
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
if (new_below)
|
|
|
|
|
new->vm_end = addr;
|
|
|
|
|
else {
|
|
|
|
|
new->vm_start = addr;
|
|
|
|
|
new->vm_pgoff += ((addr - vma->vm_start) >> PAGE_SHIFT);
|
|
|
|
|
}
|
|
|
|
|
|
2008-04-28 17:13:09 +08:00
|
|
|
pol = mpol_dup(vma_policy(vma));
|
2005-04-17 06:20:36 +08:00
|
|
|
if (IS_ERR(pol)) {
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
err = PTR_ERR(pol);
|
|
|
|
|
goto out_free_vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
vma_set_policy(new, pol);
|
|
|
|
|
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
if (anon_vma_clone(new, vma))
|
|
|
|
|
goto out_free_mpol;
|
|
|
|
|
|
2008-04-29 16:01:36 +08:00
|
|
|
if (new->vm_file) {
|
2005-04-17 06:20:36 +08:00
|
|
|
get_file(new->vm_file);
|
2008-04-29 16:01:36 +08:00
|
|
|
if (vma->vm_flags & VM_EXECUTABLE)
|
|
|
|
|
added_exe_file_vma(mm);
|
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
if (new->vm_ops && new->vm_ops->open)
|
|
|
|
|
new->vm_ops->open(new);
|
|
|
|
|
|
|
|
|
|
if (new_below)
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
err = vma_adjust(vma, addr, vma->vm_end, vma->vm_pgoff +
|
2005-04-17 06:20:36 +08:00
|
|
|
((addr - new->vm_start) >> PAGE_SHIFT), new);
|
|
|
|
|
else
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
err = vma_adjust(vma, vma->vm_start, addr, vma->vm_pgoff, new);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
/* Success. */
|
|
|
|
|
if (!err)
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
/* Clean everything up if vma_adjust failed. */
|
2010-04-27 00:33:03 +08:00
|
|
|
if (new->vm_ops && new->vm_ops->close)
|
|
|
|
|
new->vm_ops->close(new);
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
if (new->vm_file) {
|
|
|
|
|
if (vma->vm_flags & VM_EXECUTABLE)
|
|
|
|
|
removed_exe_file_vma(mm);
|
|
|
|
|
fput(new->vm_file);
|
|
|
|
|
}
|
2010-09-23 04:05:12 +08:00
|
|
|
unlink_anon_vmas(new);
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
out_free_mpol:
|
|
|
|
|
mpol_put(pol);
|
|
|
|
|
out_free_vma:
|
|
|
|
|
kmem_cache_free(vm_area_cachep, new);
|
|
|
|
|
out_err:
|
|
|
|
|
return err;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
mmap: don't return ENOMEM when mapcount is temporarily exceeded in munmap()
On ia64, the following test program exit abnormally, because glibc thread
library called abort().
========================================================
(gdb) bt
#0 0xa000000000010620 in __kernel_syscall_via_break ()
#1 0x20000000003208e0 in raise () from /lib/libc.so.6.1
#2 0x2000000000324090 in abort () from /lib/libc.so.6.1
#3 0x200000000027c3e0 in __deallocate_stack () from /lib/libpthread.so.0
#4 0x200000000027f7c0 in start_thread () from /lib/libpthread.so.0
#5 0x200000000047ef60 in __clone2 () from /lib/libc.so.6.1
========================================================
The fact is, glibc call munmap() when thread exitng time for freeing
stack, and it assume munlock() never fail. However, munmap() often make
vma splitting and it with many mapcount make -ENOMEM.
Oh well, that's crazy, because stack unmapping never increase mapcount.
The maxcount exceeding is only temporary. internal temporary exceeding
shouldn't make ENOMEM.
This patch does it.
test_max_mapcount.c
==================================================================
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<pthread.h>
#include<errno.h>
#include<unistd.h>
#define THREAD_NUM 30000
#define MAL_SIZE (8*1024*1024)
void *wait_thread(void *args)
{
void *addr;
addr = malloc(MAL_SIZE);
sleep(10);
return NULL;
}
void *wait_thread2(void *args)
{
sleep(60);
return NULL;
}
int main(int argc, char *argv[])
{
int i;
pthread_t thread[THREAD_NUM], th;
int ret, count = 0;
pthread_attr_t attr;
ret = pthread_attr_init(&attr);
if(ret) {
perror("pthread_attr_init");
}
ret = pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
if(ret) {
perror("pthread_attr_setdetachstate");
}
for (i = 0; i < THREAD_NUM; i++) {
ret = pthread_create(&th, &attr, wait_thread, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
ret = pthread_create(&thread[i], &attr, wait_thread2, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
}
sleep(3600);
return 0;
}
==================================================================
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 09:57:56 +08:00
|
|
|
/*
|
|
|
|
|
* Split a vma into two pieces at address 'addr', a new vma is allocated
|
|
|
|
|
* either for the first part or the tail.
|
|
|
|
|
*/
|
|
|
|
|
int split_vma(struct mm_struct *mm, struct vm_area_struct *vma,
|
|
|
|
|
unsigned long addr, int new_below)
|
|
|
|
|
{
|
|
|
|
|
if (mm->map_count >= sysctl_max_map_count)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
return __split_vma(mm, vma, addr, new_below);
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/* Munmap is split into 2 main parts -- this part which finds
|
|
|
|
|
* what needs doing, and the areas themselves, which do the
|
|
|
|
|
* work. This now handles partial unmappings.
|
|
|
|
|
* Jeremy Fitzhardinge <jeremy@goop.org>
|
|
|
|
|
*/
|
|
|
|
|
int do_munmap(struct mm_struct *mm, unsigned long start, size_t len)
|
|
|
|
|
{
|
|
|
|
|
unsigned long end;
|
2005-04-20 04:29:18 +08:00
|
|
|
struct vm_area_struct *vma, *prev, *last;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
if ((start & ~PAGE_MASK) || start > TASK_SIZE || len > TASK_SIZE-start)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
|
|
if ((len = PAGE_ALIGN(len)) == 0)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
|
|
/* Find the first overlapping VMA */
|
2011-06-16 15:35:09 +08:00
|
|
|
vma = find_vma(mm, start);
|
2005-04-20 04:29:18 +08:00
|
|
|
if (!vma)
|
2005-04-17 06:20:36 +08:00
|
|
|
return 0;
|
2011-06-16 15:35:09 +08:00
|
|
|
prev = vma->vm_prev;
|
2005-04-20 04:29:18 +08:00
|
|
|
/* we have start < vma->vm_end */
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/* if it doesn't overlap, we have nothing.. */
|
|
|
|
|
end = start + len;
|
2005-04-20 04:29:18 +08:00
|
|
|
if (vma->vm_start >= end)
|
2005-04-17 06:20:36 +08:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If we need to split any vma, do it now to save pain later.
|
|
|
|
|
*
|
|
|
|
|
* Note: mremap's move_vma VM_ACCOUNT handling assumes a partially
|
|
|
|
|
* unmapped vm_area_struct will remain in use: so lower split_vma
|
|
|
|
|
* places tmp vma above, and higher split_vma places tmp vma below.
|
|
|
|
|
*/
|
2005-04-20 04:29:18 +08:00
|
|
|
if (start > vma->vm_start) {
|
mmap: don't return ENOMEM when mapcount is temporarily exceeded in munmap()
On ia64, the following test program exit abnormally, because glibc thread
library called abort().
========================================================
(gdb) bt
#0 0xa000000000010620 in __kernel_syscall_via_break ()
#1 0x20000000003208e0 in raise () from /lib/libc.so.6.1
#2 0x2000000000324090 in abort () from /lib/libc.so.6.1
#3 0x200000000027c3e0 in __deallocate_stack () from /lib/libpthread.so.0
#4 0x200000000027f7c0 in start_thread () from /lib/libpthread.so.0
#5 0x200000000047ef60 in __clone2 () from /lib/libc.so.6.1
========================================================
The fact is, glibc call munmap() when thread exitng time for freeing
stack, and it assume munlock() never fail. However, munmap() often make
vma splitting and it with many mapcount make -ENOMEM.
Oh well, that's crazy, because stack unmapping never increase mapcount.
The maxcount exceeding is only temporary. internal temporary exceeding
shouldn't make ENOMEM.
This patch does it.
test_max_mapcount.c
==================================================================
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<pthread.h>
#include<errno.h>
#include<unistd.h>
#define THREAD_NUM 30000
#define MAL_SIZE (8*1024*1024)
void *wait_thread(void *args)
{
void *addr;
addr = malloc(MAL_SIZE);
sleep(10);
return NULL;
}
void *wait_thread2(void *args)
{
sleep(60);
return NULL;
}
int main(int argc, char *argv[])
{
int i;
pthread_t thread[THREAD_NUM], th;
int ret, count = 0;
pthread_attr_t attr;
ret = pthread_attr_init(&attr);
if(ret) {
perror("pthread_attr_init");
}
ret = pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
if(ret) {
perror("pthread_attr_setdetachstate");
}
for (i = 0; i < THREAD_NUM; i++) {
ret = pthread_create(&th, &attr, wait_thread, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
ret = pthread_create(&thread[i], &attr, wait_thread2, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
}
sleep(3600);
return 0;
}
==================================================================
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 09:57:56 +08:00
|
|
|
int error;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Make sure that map_count on return from munmap() will
|
|
|
|
|
* not exceed its limit; but let map_count go just above
|
|
|
|
|
* its limit temporarily, to help free resources as expected.
|
|
|
|
|
*/
|
|
|
|
|
if (end < vma->vm_end && mm->map_count >= sysctl_max_map_count)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
error = __split_vma(mm, vma, start, 0);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (error)
|
|
|
|
|
return error;
|
2005-04-20 04:29:18 +08:00
|
|
|
prev = vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Does it split the last one? */
|
|
|
|
|
last = find_vma(mm, end);
|
|
|
|
|
if (last && end > last->vm_start) {
|
mmap: don't return ENOMEM when mapcount is temporarily exceeded in munmap()
On ia64, the following test program exit abnormally, because glibc thread
library called abort().
========================================================
(gdb) bt
#0 0xa000000000010620 in __kernel_syscall_via_break ()
#1 0x20000000003208e0 in raise () from /lib/libc.so.6.1
#2 0x2000000000324090 in abort () from /lib/libc.so.6.1
#3 0x200000000027c3e0 in __deallocate_stack () from /lib/libpthread.so.0
#4 0x200000000027f7c0 in start_thread () from /lib/libpthread.so.0
#5 0x200000000047ef60 in __clone2 () from /lib/libc.so.6.1
========================================================
The fact is, glibc call munmap() when thread exitng time for freeing
stack, and it assume munlock() never fail. However, munmap() often make
vma splitting and it with many mapcount make -ENOMEM.
Oh well, that's crazy, because stack unmapping never increase mapcount.
The maxcount exceeding is only temporary. internal temporary exceeding
shouldn't make ENOMEM.
This patch does it.
test_max_mapcount.c
==================================================================
#include<stdio.h>
#include<stdlib.h>
#include<string.h>
#include<pthread.h>
#include<errno.h>
#include<unistd.h>
#define THREAD_NUM 30000
#define MAL_SIZE (8*1024*1024)
void *wait_thread(void *args)
{
void *addr;
addr = malloc(MAL_SIZE);
sleep(10);
return NULL;
}
void *wait_thread2(void *args)
{
sleep(60);
return NULL;
}
int main(int argc, char *argv[])
{
int i;
pthread_t thread[THREAD_NUM], th;
int ret, count = 0;
pthread_attr_t attr;
ret = pthread_attr_init(&attr);
if(ret) {
perror("pthread_attr_init");
}
ret = pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED);
if(ret) {
perror("pthread_attr_setdetachstate");
}
for (i = 0; i < THREAD_NUM; i++) {
ret = pthread_create(&th, &attr, wait_thread, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
ret = pthread_create(&thread[i], &attr, wait_thread2, NULL);
if(ret) {
fprintf(stderr, "[%d] ", count);
perror("pthread_create");
} else {
printf("[%d] create OK.\n", count);
}
count++;
}
sleep(3600);
return 0;
}
==================================================================
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-15 09:57:56 +08:00
|
|
|
int error = __split_vma(mm, last, end, 1);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
}
|
2005-04-20 04:29:18 +08:00
|
|
|
vma = prev? prev->vm_next: mm->mmap;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
2008-10-19 11:26:50 +08:00
|
|
|
/*
|
|
|
|
|
* unlock any mlock()ed ranges before detaching vmas
|
|
|
|
|
*/
|
|
|
|
|
if (mm->locked_vm) {
|
|
|
|
|
struct vm_area_struct *tmp = vma;
|
|
|
|
|
while (tmp && tmp->vm_start < end) {
|
|
|
|
|
if (tmp->vm_flags & VM_LOCKED) {
|
|
|
|
|
mm->locked_vm -= vma_pages(tmp);
|
|
|
|
|
munlock_vma_pages_all(tmp);
|
|
|
|
|
}
|
|
|
|
|
tmp = tmp->vm_next;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* Remove the vma's, and unmap the actual pages
|
|
|
|
|
*/
|
2005-04-20 04:29:18 +08:00
|
|
|
detach_vmas_to_be_unmapped(mm, vma, prev, end);
|
|
|
|
|
unmap_region(mm, vma, prev, start, end);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/* Fix up all other VM information */
|
2005-10-30 09:15:56 +08:00
|
|
|
remove_vma_list(mm, vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
EXPORT_SYMBOL(do_munmap);
|
|
|
|
|
|
2012-04-21 09:57:04 +08:00
|
|
|
int vm_munmap(unsigned long start, size_t len)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
int ret;
|
2012-04-21 09:57:04 +08:00
|
|
|
struct mm_struct *mm = current->mm;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
down_write(&mm->mmap_sem);
|
2012-04-21 07:20:01 +08:00
|
|
|
ret = do_munmap(mm, start, len);
|
2005-04-17 06:20:36 +08:00
|
|
|
up_write(&mm->mmap_sem);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
2012-04-21 07:20:01 +08:00
|
|
|
EXPORT_SYMBOL(vm_munmap);
|
|
|
|
|
|
|
|
|
|
SYSCALL_DEFINE2(munmap, unsigned long, addr, size_t, len)
|
|
|
|
|
{
|
|
|
|
|
profile_munmap(addr);
|
2012-04-21 09:57:04 +08:00
|
|
|
return vm_munmap(addr, len);
|
2012-04-21 07:20:01 +08:00
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
static inline void verify_mm_writelocked(struct mm_struct *mm)
|
|
|
|
|
{
|
2005-10-31 07:03:12 +08:00
|
|
|
#ifdef CONFIG_DEBUG_VM
|
2005-04-17 06:20:36 +08:00
|
|
|
if (unlikely(down_read_trylock(&mm->mmap_sem))) {
|
|
|
|
|
WARN_ON(1);
|
|
|
|
|
up_read(&mm->mmap_sem);
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* this is really a simplified "do_mmap". it only handles
|
|
|
|
|
* anonymous maps. eventually we may be able to do some
|
|
|
|
|
* brk-specific accounting here.
|
|
|
|
|
*/
|
2012-04-21 06:35:40 +08:00
|
|
|
static unsigned long do_brk(unsigned long addr, unsigned long len)
|
2005-04-17 06:20:36 +08:00
|
|
|
{
|
|
|
|
|
struct mm_struct * mm = current->mm;
|
|
|
|
|
struct vm_area_struct * vma, * prev;
|
|
|
|
|
unsigned long flags;
|
|
|
|
|
struct rb_node ** rb_link, * rb_parent;
|
|
|
|
|
pgoff_t pgoff = addr >> PAGE_SHIFT;
|
2006-09-07 18:17:04 +08:00
|
|
|
int error;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
len = PAGE_ALIGN(len);
|
|
|
|
|
if (!len)
|
|
|
|
|
return addr;
|
|
|
|
|
|
2007-12-08 19:02:48 +08:00
|
|
|
error = security_file_mmap(NULL, 0, 0, 0, addr, 1);
|
2007-12-05 00:06:55 +08:00
|
|
|
if (error)
|
|
|
|
|
return error;
|
|
|
|
|
|
2006-09-07 18:17:04 +08:00
|
|
|
flags = VM_DATA_DEFAULT_FLAGS | VM_ACCOUNT | mm->def_flags;
|
|
|
|
|
|
2009-12-04 08:40:46 +08:00
|
|
|
error = get_unmapped_area(NULL, addr, len, 0, MAP_FIXED);
|
|
|
|
|
if (error & ~PAGE_MASK)
|
2006-09-07 18:17:04 +08:00
|
|
|
return error;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
/*
|
|
|
|
|
* mlock MCL_FUTURE?
|
|
|
|
|
*/
|
|
|
|
|
if (mm->def_flags & VM_LOCKED) {
|
|
|
|
|
unsigned long locked, lock_limit;
|
2005-05-01 23:58:38 +08:00
|
|
|
locked = len >> PAGE_SHIFT;
|
|
|
|
|
locked += mm->locked_vm;
|
2010-03-06 05:41:44 +08:00
|
|
|
lock_limit = rlimit(RLIMIT_MEMLOCK);
|
2005-05-01 23:58:38 +08:00
|
|
|
lock_limit >>= PAGE_SHIFT;
|
2005-04-17 06:20:36 +08:00
|
|
|
if (locked > lock_limit && !capable(CAP_IPC_LOCK))
|
|
|
|
|
return -EAGAIN;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* mm->mmap_sem is required to protect against another thread
|
|
|
|
|
* changing the mappings in case we sleep.
|
|
|
|
|
*/
|
|
|
|
|
verify_mm_writelocked(mm);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Clear old maps. this also does some error checking for us
|
|
|
|
|
*/
|
|
|
|
|
munmap_back:
|
|
|
|
|
vma = find_vma_prepare(mm, addr, &prev, &rb_link, &rb_parent);
|
|
|
|
|
if (vma && vma->vm_start < addr + len) {
|
|
|
|
|
if (do_munmap(mm, addr, len))
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
goto munmap_back;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Check against address space limits *after* clearing old maps... */
|
2005-05-01 23:58:35 +08:00
|
|
|
if (!may_expand_vm(mm, len >> PAGE_SHIFT))
|
2005-04-17 06:20:36 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
if (mm->map_count > sysctl_max_map_count)
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
2012-02-13 11:58:52 +08:00
|
|
|
if (security_vm_enough_memory_mm(mm, len >> PAGE_SHIFT))
|
2005-04-17 06:20:36 +08:00
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
|
|
/* Can we just expand an old private anonymous mapping? */
|
2008-10-19 11:26:50 +08:00
|
|
|
vma = vma_merge(mm, prev, addr, addr + len, flags,
|
|
|
|
|
NULL, NULL, pgoff, NULL);
|
|
|
|
|
if (vma)
|
2005-04-17 06:20:36 +08:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* create a vma struct for an anonymous mapping
|
|
|
|
|
*/
|
2006-03-25 19:06:43 +08:00
|
|
|
vma = kmem_cache_zalloc(vm_area_cachep, GFP_KERNEL);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (!vma) {
|
|
|
|
|
vm_unacct_memory(len >> PAGE_SHIFT);
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
}
|
|
|
|
|
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
INIT_LIST_HEAD(&vma->anon_vma_chain);
|
2005-04-17 06:20:36 +08:00
|
|
|
vma->vm_mm = mm;
|
|
|
|
|
vma->vm_start = addr;
|
|
|
|
|
vma->vm_end = addr + len;
|
|
|
|
|
vma->vm_pgoff = pgoff;
|
|
|
|
|
vma->vm_flags = flags;
|
2007-10-19 14:39:15 +08:00
|
|
|
vma->vm_page_prot = vm_get_page_prot(flags);
|
2005-04-17 06:20:36 +08:00
|
|
|
vma_link(mm, vma, prev, rb_link, rb_parent);
|
|
|
|
|
out:
|
2010-05-18 22:30:49 +08:00
|
|
|
perf_event_mmap(vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
mm->total_vm += len >> PAGE_SHIFT;
|
|
|
|
|
if (flags & VM_LOCKED) {
|
2008-10-19 11:26:50 +08:00
|
|
|
if (!mlock_vma_pages_range(vma, addr, addr + len))
|
|
|
|
|
mm->locked_vm += (len >> PAGE_SHIFT);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
return addr;
|
|
|
|
|
}
|
|
|
|
|
|
2012-04-21 06:35:40 +08:00
|
|
|
unsigned long vm_brk(unsigned long addr, unsigned long len)
|
|
|
|
|
{
|
|
|
|
|
struct mm_struct *mm = current->mm;
|
|
|
|
|
unsigned long ret;
|
|
|
|
|
|
|
|
|
|
down_write(&mm->mmap_sem);
|
|
|
|
|
ret = do_brk(addr, len);
|
|
|
|
|
up_write(&mm->mmap_sem);
|
|
|
|
|
return ret;
|
|
|
|
|
}
|
|
|
|
|
EXPORT_SYMBOL(vm_brk);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/* Release all mmaps. */
|
|
|
|
|
void exit_mmap(struct mm_struct *mm)
|
|
|
|
|
{
|
2011-05-25 08:11:45 +08:00
|
|
|
struct mmu_gather tlb;
|
2008-10-19 11:26:50 +08:00
|
|
|
struct vm_area_struct *vma;
|
2005-04-17 06:20:36 +08:00
|
|
|
unsigned long nr_accounted = 0;
|
|
|
|
|
|
2007-05-03 01:27:14 +08:00
|
|
|
/* mm's last user has gone, and its about to be pulled down */
|
mmu-notifiers: core
With KVM/GFP/XPMEM there isn't just the primary CPU MMU pointing to pages.
There are secondary MMUs (with secondary sptes and secondary tlbs) too.
sptes in the kvm case are shadow pagetables, but when I say spte in
mmu-notifier context, I mean "secondary pte". In GRU case there's no
actual secondary pte and there's only a secondary tlb because the GRU
secondary MMU has no knowledge about sptes and every secondary tlb miss
event in the MMU always generates a page fault that has to be resolved by
the CPU (this is not the case of KVM where the a secondary tlb miss will
walk sptes in hardware and it will refill the secondary tlb transparently
to software if the corresponding spte is present). The same way
zap_page_range has to invalidate the pte before freeing the page, the spte
(and secondary tlb) must also be invalidated before any page is freed and
reused.
Currently we take a page_count pin on every page mapped by sptes, but that
means the pages can't be swapped whenever they're mapped by any spte
because they're part of the guest working set. Furthermore a spte unmap
event can immediately lead to a page to be freed when the pin is released
(so requiring the same complex and relatively slow tlb_gather smp safe
logic we have in zap_page_range and that can be avoided completely if the
spte unmap event doesn't require an unpin of the page previously mapped in
the secondary MMU).
The mmu notifiers allow kvm/GRU/XPMEM to attach to the tsk->mm and know
when the VM is swapping or freeing or doing anything on the primary MMU so
that the secondary MMU code can drop sptes before the pages are freed,
avoiding all page pinning and allowing 100% reliable swapping of guest
physical address space. Furthermore it avoids the code that teardown the
mappings of the secondary MMU, to implement a logic like tlb_gather in
zap_page_range that would require many IPI to flush other cpu tlbs, for
each fixed number of spte unmapped.
To make an example: if what happens on the primary MMU is a protection
downgrade (from writeable to wrprotect) the secondary MMU mappings will be
invalidated, and the next secondary-mmu-page-fault will call
get_user_pages and trigger a do_wp_page through get_user_pages if it
called get_user_pages with write=1, and it'll re-establishing an updated
spte or secondary-tlb-mapping on the copied page. Or it will setup a
readonly spte or readonly tlb mapping if it's a guest-read, if it calls
get_user_pages with write=0. This is just an example.
This allows to map any page pointed by any pte (and in turn visible in the
primary CPU MMU), into a secondary MMU (be it a pure tlb like GRU, or an
full MMU with both sptes and secondary-tlb like the shadow-pagetable layer
with kvm), or a remote DMA in software like XPMEM (hence needing of
schedule in XPMEM code to send the invalidate to the remote node, while no
need to schedule in kvm/gru as it's an immediate event like invalidating
primary-mmu pte).
At least for KVM without this patch it's impossible to swap guests
reliably. And having this feature and removing the page pin allows
several other optimizations that simplify life considerably.
Dependencies:
1) mm_take_all_locks() to register the mmu notifier when the whole VM
isn't doing anything with "mm". This allows mmu notifier users to keep
track if the VM is in the middle of the invalidate_range_begin/end
critical section with an atomic counter incraese in range_begin and
decreased in range_end. No secondary MMU page fault is allowed to map
any spte or secondary tlb reference, while the VM is in the middle of
range_begin/end as any page returned by get_user_pages in that critical
section could later immediately be freed without any further
->invalidate_page notification (invalidate_range_begin/end works on
ranges and ->invalidate_page isn't called immediately before freeing
the page). To stop all page freeing and pagetable overwrites the
mmap_sem must be taken in write mode and all other anon_vma/i_mmap
locks must be taken too.
2) It'd be a waste to add branches in the VM if nobody could possibly
run KVM/GRU/XPMEM on the kernel, so mmu notifiers will only enabled if
CONFIG_KVM=m/y. In the current kernel kvm won't yet take advantage of
mmu notifiers, but this already allows to compile a KVM external module
against a kernel with mmu notifiers enabled and from the next pull from
kvm.git we'll start using them. And GRU/XPMEM will also be able to
continue the development by enabling KVM=m in their config, until they
submit all GRU/XPMEM GPLv2 code to the mainline kernel. Then they can
also enable MMU_NOTIFIERS in the same way KVM does it (even if KVM=n).
This guarantees nobody selects MMU_NOTIFIER=y if KVM and GRU and XPMEM
are all =n.
The mmu_notifier_register call can fail because mm_take_all_locks may be
interrupted by a signal and return -EINTR. Because mmu_notifier_reigster
is used when a driver startup, a failure can be gracefully handled. Here
an example of the change applied to kvm to register the mmu notifiers.
Usually when a driver startups other allocations are required anyway and
-ENOMEM failure paths exists already.
struct kvm *kvm_arch_create_vm(void)
{
struct kvm *kvm = kzalloc(sizeof(struct kvm), GFP_KERNEL);
+ int err;
if (!kvm)
return ERR_PTR(-ENOMEM);
INIT_LIST_HEAD(&kvm->arch.active_mmu_pages);
+ kvm->arch.mmu_notifier.ops = &kvm_mmu_notifier_ops;
+ err = mmu_notifier_register(&kvm->arch.mmu_notifier, current->mm);
+ if (err) {
+ kfree(kvm);
+ return ERR_PTR(err);
+ }
+
return kvm;
}
mmu_notifier_unregister returns void and it's reliable.
The patch also adds a few needed but missing includes that would prevent
kernel to compile after these changes on non-x86 archs (x86 didn't need
them by luck).
[akpm@linux-foundation.org: coding-style fixes]
[akpm@linux-foundation.org: fix mm/filemap_xip.c build]
[akpm@linux-foundation.org: fix mm/mmu_notifier.c build]
Signed-off-by: Andrea Arcangeli <andrea@qumranet.com>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Jack Steiner <steiner@sgi.com>
Cc: Robin Holt <holt@sgi.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Kanoj Sarcar <kanojsarcar@yahoo.com>
Cc: Roland Dreier <rdreier@cisco.com>
Cc: Steve Wise <swise@opengridcomputing.com>
Cc: Avi Kivity <avi@qumranet.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Anthony Liguori <aliguori@us.ibm.com>
Cc: Chris Wright <chrisw@redhat.com>
Cc: Marcelo Tosatti <marcelo@kvack.org>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: Izik Eidus <izike@qumranet.com>
Cc: Anthony Liguori <aliguori@us.ibm.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-07-29 06:46:29 +08:00
|
|
|
mmu_notifier_release(mm);
|
2007-05-03 01:27:14 +08:00
|
|
|
|
2008-10-19 11:26:50 +08:00
|
|
|
if (mm->locked_vm) {
|
|
|
|
|
vma = mm->mmap;
|
|
|
|
|
while (vma) {
|
|
|
|
|
if (vma->vm_flags & VM_LOCKED)
|
|
|
|
|
munlock_vma_pages_all(vma);
|
|
|
|
|
vma = vma->vm_next;
|
|
|
|
|
}
|
|
|
|
|
}
|
mm: rearrange exit_mmap() to unlock before arch_exit_mmap
Christophe Saout reported [in precursor to:
http://marc.info/?l=linux-kernel&m=123209902707347&w=4]:
> Note that I also some a different issue with CONFIG_UNEVICTABLE_LRU.
> Seems like Xen tears down current->mm early on process termination, so
> that __get_user_pages in exit_mmap causes nasty messages when the
> process had any mlocked pages. (in fact, it somehow manages to get into
> the swapping code and produces a null pointer dereference trying to get
> a swap token)
Jeremy explained:
Yes. In the normal case under Xen, an in-use pagetable is "pinned",
meaning that it is RO to the kernel, and all updates must go via hypercall
(or writes are trapped and emulated, which is much the same thing). An
unpinned pagetable is not currently in use by any process, and can be
directly accessed as normal RW pages.
As an optimisation at process exit time, we unpin the pagetable as early
as possible (switching the process to init_mm), so that all the normal
pagetable teardown can happen with direct memory accesses.
This happens in exit_mmap() -> arch_exit_mmap(). The munlocking happens
a few lines below. The obvious thing to do would be to move
arch_exit_mmap() to below the munlock code, but I think we'd want to
call it even if mm->mmap is NULL, just to be on the safe side.
Thus, this patch:
exit_mmap() needs to unlock any locked vmas before calling arch_exit_mmap,
as the latter may switch the current mm to init_mm, which would cause the
former to fail.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christophe Saout <christophe@saout.de>
Cc: Keir Fraser <keir.fraser@eu.citrix.com>
Cc: Christophe Saout <christophe@saout.de>
Cc: Alex Williamson <alex.williamson@hp.com>
Cc: <stable@kernel.org> [2.6.28.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-02-12 05:04:41 +08:00
|
|
|
|
|
|
|
|
arch_exit_mmap(mm);
|
|
|
|
|
|
2008-10-19 11:26:50 +08:00
|
|
|
vma = mm->mmap;
|
mm: rearrange exit_mmap() to unlock before arch_exit_mmap
Christophe Saout reported [in precursor to:
http://marc.info/?l=linux-kernel&m=123209902707347&w=4]:
> Note that I also some a different issue with CONFIG_UNEVICTABLE_LRU.
> Seems like Xen tears down current->mm early on process termination, so
> that __get_user_pages in exit_mmap causes nasty messages when the
> process had any mlocked pages. (in fact, it somehow manages to get into
> the swapping code and produces a null pointer dereference trying to get
> a swap token)
Jeremy explained:
Yes. In the normal case under Xen, an in-use pagetable is "pinned",
meaning that it is RO to the kernel, and all updates must go via hypercall
(or writes are trapped and emulated, which is much the same thing). An
unpinned pagetable is not currently in use by any process, and can be
directly accessed as normal RW pages.
As an optimisation at process exit time, we unpin the pagetable as early
as possible (switching the process to init_mm), so that all the normal
pagetable teardown can happen with direct memory accesses.
This happens in exit_mmap() -> arch_exit_mmap(). The munlocking happens
a few lines below. The obvious thing to do would be to move
arch_exit_mmap() to below the munlock code, but I think we'd want to
call it even if mm->mmap is NULL, just to be on the safe side.
Thus, this patch:
exit_mmap() needs to unlock any locked vmas before calling arch_exit_mmap,
as the latter may switch the current mm to init_mm, which would cause the
former to fail.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christophe Saout <christophe@saout.de>
Cc: Keir Fraser <keir.fraser@eu.citrix.com>
Cc: Christophe Saout <christophe@saout.de>
Cc: Alex Williamson <alex.williamson@hp.com>
Cc: <stable@kernel.org> [2.6.28.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-02-12 05:04:41 +08:00
|
|
|
if (!vma) /* Can happen if dup_mmap() received an OOM */
|
|
|
|
|
return;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
lru_add_drain();
|
|
|
|
|
flush_cache_mm(mm);
|
2011-05-25 08:11:45 +08:00
|
|
|
tlb_gather_mmu(&tlb, mm, 1);
|
2009-01-07 06:40:29 +08:00
|
|
|
/* update_hiwater_rss(mm) here? but nobody should be looking */
|
[PATCH] freepgt: free_pgtables use vma list
Recent woes with some arches needing their own pgd_addr_end macro; and 4-level
clear_page_range regression since 2.6.10's clear_page_tables; and its
long-standing well-known inefficiency in searching throughout the higher-level
page tables for those few entries to clear and free: all can be blamed on
ignoring the list of vmas when we free page tables.
Replace exit_mmap's clear_page_range of the total user address space by
free_pgtables operating on the mm's vma list; unmap_region use it in the same
way, giving floor and ceiling beyond which it may not free tables. This
brings lmbench fork/exec/sh numbers back to 2.6.10 (unless preempt is enabled,
in which case latency fixes spoil unmap_vmas throughput).
Beware: the do_mmap_pgoff driver failure case must now use unmap_region
instead of zap_page_range, since a page table might have been allocated, and
can only be freed while it is touched by some vma.
Move free_pgtables from mmap.c to memory.c, where its lower levels are adapted
from the clear_page_range levels. (Most of free_pgtables' old code was
actually for a non-existent case, prev not properly set up, dating from before
hch gave us split_vma.) Pass mmu_gather** in the public interfaces, since we
might want to add latency lockdrops later; but no attempt to do so yet, going
by vma should itself reduce latency.
But what if is_hugepage_only_range? Those ia64 and ppc64 cases need careful
examination: put that off until a later patch of the series.
What of x86_64's 32bit vdso page __map_syscall32 maps outside any vma?
And the range to sparc64's flush_tlb_pgtables? It's less clear to me now that
we need to do more than is done here - every PMD_SIZE ever occupied will be
flushed, do we really have to flush every PGDIR_SIZE ever partially occupied?
A shame to complicate it unnecessarily.
Special thanks to David Miller for time spent repairing my ceilings.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-20 04:29:15 +08:00
|
|
|
/* Use -1 here to ensure all VMAs in the mm are unmapped */
|
2012-05-07 04:54:06 +08:00
|
|
|
unmap_vmas(&tlb, vma, 0, -1);
|
ksm: fix oom deadlock
There's a now-obvious deadlock in KSM's out-of-memory handling:
imagine ksmd or KSM_RUN_UNMERGE handling, holding ksm_thread_mutex,
trying to allocate a page to break KSM in an mm which becomes the
OOM victim (quite likely in the unmerge case): it's killed and goes
to exit, and hangs there waiting to acquire ksm_thread_mutex.
Clearly we must not require ksm_thread_mutex in __ksm_exit, simple
though that made everything else: perhaps use mmap_sem somehow?
And part of the answer lies in the comments on unmerge_ksm_pages:
__ksm_exit should also leave all the rmap_item removal to ksmd.
But there's a fundamental problem, that KSM relies upon mmap_sem to
guarantee the consistency of the mm it's dealing with, yet exit_mmap
tears down an mm without taking mmap_sem. And bumping mm_users won't
help at all, that just ensures that the pages the OOM killer assumes
are on their way to being freed will not be freed.
The best answer seems to be, to move the ksm_exit callout from just
before exit_mmap, to the middle of exit_mmap: after the mm's pages
have been freed (if the mmu_gather is flushed), but before its page
tables and vma structures have been freed; and down_write,up_write
mmap_sem there to serialize with KSM's own reliance on mmap_sem.
But KSM then needs to be careful, whenever it downs mmap_sem, to
check that the mm is not already exiting: there's a danger of using
find_vma on a layout that's being torn apart, or writing into page
tables which have been freed for reuse; and even do_anonymous_page
and __do_fault need to check they're not being called by break_ksm
to reinstate a pte after zap_pte_range has zapped that page table.
Though it might be clearer to add an exiting flag, set while holding
mmap_sem in __ksm_exit, that wouldn't cover the issue of reinstating
a zapped pte. All we need is to check whether mm_users is 0 - but
must remember that ksmd may detect that before __ksm_exit is reached.
So, ksm_test_exit(mm) added to comment such checks on mm->mm_users.
__ksm_exit now has to leave clearing up the rmap_items to ksmd,
that needs ksm_thread_mutex; but shift the exiting mm just after the
ksm_scan cursor so that it will soon be dealt with. __ksm_enter raise
mm_count to hold the mm_struct, ksmd's exit processing (exactly like
its processing when it finds all VM_MERGEABLEs unmapped) mmdrop it,
similar procedure for KSM_RUN_UNMERGE (which has stopped ksmd).
But also give __ksm_exit a fast path: when there's no complication
(no rmap_items attached to mm and it's not at the ksm_scan cursor),
it can safely do all the exiting work itself. This is not just an
optimization: when ksmd is not running, the raised mm_count would
otherwise leak mm_structs.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Acked-by: Izik Eidus <ieidus@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-09-22 08:02:20 +08:00
|
|
|
|
2011-05-25 08:11:45 +08:00
|
|
|
free_pgtables(&tlb, vma, FIRST_USER_ADDRESS, 0);
|
2012-03-06 03:03:47 +08:00
|
|
|
tlb_finish_mmu(&tlb, 0, -1);
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
[PATCH] mm: unlink vma before pagetables
In most places the descent from pgd to pud to pmd to pte holds mmap_sem
(exclusively or not), which ensures that free_pgtables cannot be freeing page
tables from any level at the same time. But truncation and reverse mapping
descend without mmap_sem.
No problem: just make sure that a vma is unlinked from its prio_tree (or
nonlinear list) and from its anon_vma list, after zapping the vma, but before
freeing its page tables. Then neither vmtruncate nor rmap can reach that vma
whose page tables are now volatile (nor do they need to reach it, since all
its page entries have been zapped by this stage).
The i_mmap_lock and anon_vma->lock already serialize this correctly; but the
locking hierarchy is such that we cannot take them while holding
page_table_lock. Well, we're trying to push that down anyway. So in this
patch, move anon_vma_unlink and unlink_file_vma into free_pgtables, at the
same time as moving page_table_lock around calls to unmap_vmas.
tlb_gather_mmu and tlb_finish_mmu then fall outside the page_table_lock, but
we made them preempt_disable and preempt_enable earlier; and a long source
audit of all the architectures has shown no problem with removing
page_table_lock from them. free_pgtables doesn't need page_table_lock for
itself, nor for what it calls; tlb->mm->nr_ptes is usually protected by
page_table_lock, but partly by non-exclusive mmap_sem - here it's decremented
with exclusive mmap_sem, or mm_users 0. update_hiwater_rss and
vm_unacct_memory don't need page_table_lock either.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 09:16:29 +08:00
|
|
|
* Walk the list again, actually closing and freeing it,
|
|
|
|
|
* with preemption enabled, without holding any MM locks.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
2012-05-07 04:54:06 +08:00
|
|
|
while (vma) {
|
|
|
|
|
if (vma->vm_flags & VM_ACCOUNT)
|
|
|
|
|
nr_accounted += vma_pages(vma);
|
2005-10-30 09:15:57 +08:00
|
|
|
vma = remove_vma(vma);
|
2012-05-07 04:54:06 +08:00
|
|
|
}
|
|
|
|
|
vm_unacct_memory(nr_accounted);
|
[PATCH] freepgt: free_pgtables use vma list
Recent woes with some arches needing their own pgd_addr_end macro; and 4-level
clear_page_range regression since 2.6.10's clear_page_tables; and its
long-standing well-known inefficiency in searching throughout the higher-level
page tables for those few entries to clear and free: all can be blamed on
ignoring the list of vmas when we free page tables.
Replace exit_mmap's clear_page_range of the total user address space by
free_pgtables operating on the mm's vma list; unmap_region use it in the same
way, giving floor and ceiling beyond which it may not free tables. This
brings lmbench fork/exec/sh numbers back to 2.6.10 (unless preempt is enabled,
in which case latency fixes spoil unmap_vmas throughput).
Beware: the do_mmap_pgoff driver failure case must now use unmap_region
instead of zap_page_range, since a page table might have been allocated, and
can only be freed while it is touched by some vma.
Move free_pgtables from mmap.c to memory.c, where its lower levels are adapted
from the clear_page_range levels. (Most of free_pgtables' old code was
actually for a non-existent case, prev not properly set up, dating from before
hch gave us split_vma.) Pass mmu_gather** in the public interfaces, since we
might want to add latency lockdrops later; but no attempt to do so yet, going
by vma should itself reduce latency.
But what if is_hugepage_only_range? Those ia64 and ppc64 cases need careful
examination: put that off until a later patch of the series.
What of x86_64's 32bit vdso page __map_syscall32 maps outside any vma?
And the range to sparc64's flush_tlb_pgtables? It's less clear to me now that
we need to do more than is done here - every PMD_SIZE ever occupied will be
flushed, do we really have to flush every PGDIR_SIZE ever partially occupied?
A shame to complicate it unnecessarily.
Special thanks to David Miller for time spent repairing my ceilings.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-04-20 04:29:15 +08:00
|
|
|
|
2005-04-20 04:29:19 +08:00
|
|
|
BUG_ON(mm->nr_ptes > (FIRST_USER_ADDRESS+PMD_SIZE-1)>>PMD_SHIFT);
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Insert vm structure into process list sorted by address
|
|
|
|
|
* and into the inode's i_mmap tree. If vm_file is non-NULL
|
2011-05-25 08:12:06 +08:00
|
|
|
* then i_mmap_mutex is taken here.
|
2005-04-17 06:20:36 +08:00
|
|
|
*/
|
|
|
|
|
int insert_vm_struct(struct mm_struct * mm, struct vm_area_struct * vma)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct * __vma, * prev;
|
|
|
|
|
struct rb_node ** rb_link, * rb_parent;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* The vm_pgoff of a purely anonymous vma should be irrelevant
|
|
|
|
|
* until its first write fault, when page's anon_vma and index
|
|
|
|
|
* are set. But now set the vm_pgoff it will almost certainly
|
|
|
|
|
* end up with (unless mremap moves it elsewhere before that
|
|
|
|
|
* first wfault), so /proc/pid/maps tells a consistent story.
|
|
|
|
|
*
|
|
|
|
|
* By setting it to reflect the virtual start address of the
|
|
|
|
|
* vma, merges and splits can happen in a seamless way, just
|
|
|
|
|
* using the existing file pgoff checks and manipulations.
|
|
|
|
|
* Similarly in do_mmap_pgoff and in do_brk.
|
|
|
|
|
*/
|
|
|
|
|
if (!vma->vm_file) {
|
|
|
|
|
BUG_ON(vma->anon_vma);
|
|
|
|
|
vma->vm_pgoff = vma->vm_start >> PAGE_SHIFT;
|
|
|
|
|
}
|
|
|
|
|
__vma = find_vma_prepare(mm,vma->vm_start,&prev,&rb_link,&rb_parent);
|
|
|
|
|
if (__vma && __vma->vm_start < vma->vm_end)
|
|
|
|
|
return -ENOMEM;
|
2005-09-14 13:13:02 +08:00
|
|
|
if ((vma->vm_flags & VM_ACCOUNT) &&
|
2007-08-23 05:01:28 +08:00
|
|
|
security_vm_enough_memory_mm(mm, vma_pages(vma)))
|
2005-09-14 13:13:02 +08:00
|
|
|
return -ENOMEM;
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
|
2012-02-17 16:27:41 +08:00
|
|
|
if (vma->vm_file && uprobe_mmap(vma))
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
2005-04-17 06:20:36 +08:00
|
|
|
vma_link(mm, vma, prev, rb_link, rb_parent);
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Copy the vma structure to a new location in the same mm,
|
|
|
|
|
* prior to moving page table entries, to effect an mremap move.
|
|
|
|
|
*/
|
|
|
|
|
struct vm_area_struct *copy_vma(struct vm_area_struct **vmap,
|
|
|
|
|
unsigned long addr, unsigned long len, pgoff_t pgoff)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct *vma = *vmap;
|
|
|
|
|
unsigned long vma_start = vma->vm_start;
|
|
|
|
|
struct mm_struct *mm = vma->vm_mm;
|
|
|
|
|
struct vm_area_struct *new_vma, *prev;
|
|
|
|
|
struct rb_node **rb_link, *rb_parent;
|
|
|
|
|
struct mempolicy *pol;
|
mremap: enforce rmap src/dst vma ordering in case of vma_merge() succeeding in copy_vma()
migrate was doing an rmap_walk with speculative lock-less access on
pagetables. That could lead it to not serializing properly against mremap
PT locks. But a second problem remains in the order of vmas in the
same_anon_vma list used by the rmap_walk.
If vma_merge succeeds in copy_vma, the src vma could be placed after the
dst vma in the same_anon_vma list. That could still lead to migrate
missing some pte.
This patch adds an anon_vma_moveto_tail() function to force the dst vma at
the end of the list before mremap starts to solve the problem.
If the mremap is very large and there are a lots of parents or childs
sharing the anon_vma root lock, this should still scale better than taking
the anon_vma root lock around every pte copy practically for the whole
duration of mremap.
Update: Hugh noticed special care is needed in the error path where
move_page_tables goes in the reverse direction, a second
anon_vma_moveto_tail() call is needed in the error path.
This program exercises the anon_vma_moveto_tail:
===
int main()
{
static struct timeval oldstamp, newstamp;
long diffsec;
char *p, *p2, *p3, *p4;
if (posix_memalign((void **)&p, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p2, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p3, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
memset(p, 0xff, SIZE);
printf("%p\n", p);
memset(p2, 0xff, SIZE);
memset(p3, 0x77, 4096);
if (memcmp(p, p2, SIZE))
printf("error\n");
p4 = mremap(p+SIZE/2, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p3);
if (p4 != p3)
perror("mremap"), exit(1);
p4 = mremap(p4, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p+SIZE/2);
if (p4 != p+SIZE/2)
perror("mremap"), exit(1);
if (memcmp(p, p2, SIZE))
printf("error\n");
printf("ok\n");
return 0;
}
===
$ perf probe -a anon_vma_moveto_tail
Add new event:
probe:anon_vma_moveto_tail (on anon_vma_moveto_tail)
You can now use it on all perf tools, such as:
perf record -e probe:anon_vma_moveto_tail -aR sleep 1
$ perf record -e probe:anon_vma_moveto_tail -aR ./anon_vma_moveto_tail
0x7f2ca2800000
ok
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.043 MB perf.data (~1860 samples) ]
$ perf report --stdio
100.00% anon_vma_moveto [kernel.kallsyms] [k] anon_vma_moveto_tail
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Nai Xia <nai.xia@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pawel Sikora <pluto@agmk.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-01-11 07:08:05 +08:00
|
|
|
bool faulted_in_anon_vma = true;
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If anonymous vma has not yet been faulted, update new pgoff
|
|
|
|
|
* to match new location, to increase its chance of merging.
|
|
|
|
|
*/
|
mremap: enforce rmap src/dst vma ordering in case of vma_merge() succeeding in copy_vma()
migrate was doing an rmap_walk with speculative lock-less access on
pagetables. That could lead it to not serializing properly against mremap
PT locks. But a second problem remains in the order of vmas in the
same_anon_vma list used by the rmap_walk.
If vma_merge succeeds in copy_vma, the src vma could be placed after the
dst vma in the same_anon_vma list. That could still lead to migrate
missing some pte.
This patch adds an anon_vma_moveto_tail() function to force the dst vma at
the end of the list before mremap starts to solve the problem.
If the mremap is very large and there are a lots of parents or childs
sharing the anon_vma root lock, this should still scale better than taking
the anon_vma root lock around every pte copy practically for the whole
duration of mremap.
Update: Hugh noticed special care is needed in the error path where
move_page_tables goes in the reverse direction, a second
anon_vma_moveto_tail() call is needed in the error path.
This program exercises the anon_vma_moveto_tail:
===
int main()
{
static struct timeval oldstamp, newstamp;
long diffsec;
char *p, *p2, *p3, *p4;
if (posix_memalign((void **)&p, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p2, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p3, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
memset(p, 0xff, SIZE);
printf("%p\n", p);
memset(p2, 0xff, SIZE);
memset(p3, 0x77, 4096);
if (memcmp(p, p2, SIZE))
printf("error\n");
p4 = mremap(p+SIZE/2, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p3);
if (p4 != p3)
perror("mremap"), exit(1);
p4 = mremap(p4, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p+SIZE/2);
if (p4 != p+SIZE/2)
perror("mremap"), exit(1);
if (memcmp(p, p2, SIZE))
printf("error\n");
printf("ok\n");
return 0;
}
===
$ perf probe -a anon_vma_moveto_tail
Add new event:
probe:anon_vma_moveto_tail (on anon_vma_moveto_tail)
You can now use it on all perf tools, such as:
perf record -e probe:anon_vma_moveto_tail -aR sleep 1
$ perf record -e probe:anon_vma_moveto_tail -aR ./anon_vma_moveto_tail
0x7f2ca2800000
ok
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.043 MB perf.data (~1860 samples) ]
$ perf report --stdio
100.00% anon_vma_moveto [kernel.kallsyms] [k] anon_vma_moveto_tail
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Nai Xia <nai.xia@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pawel Sikora <pluto@agmk.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-01-11 07:08:05 +08:00
|
|
|
if (unlikely(!vma->vm_file && !vma->anon_vma)) {
|
2005-04-17 06:20:36 +08:00
|
|
|
pgoff = addr >> PAGE_SHIFT;
|
mremap: enforce rmap src/dst vma ordering in case of vma_merge() succeeding in copy_vma()
migrate was doing an rmap_walk with speculative lock-less access on
pagetables. That could lead it to not serializing properly against mremap
PT locks. But a second problem remains in the order of vmas in the
same_anon_vma list used by the rmap_walk.
If vma_merge succeeds in copy_vma, the src vma could be placed after the
dst vma in the same_anon_vma list. That could still lead to migrate
missing some pte.
This patch adds an anon_vma_moveto_tail() function to force the dst vma at
the end of the list before mremap starts to solve the problem.
If the mremap is very large and there are a lots of parents or childs
sharing the anon_vma root lock, this should still scale better than taking
the anon_vma root lock around every pte copy practically for the whole
duration of mremap.
Update: Hugh noticed special care is needed in the error path where
move_page_tables goes in the reverse direction, a second
anon_vma_moveto_tail() call is needed in the error path.
This program exercises the anon_vma_moveto_tail:
===
int main()
{
static struct timeval oldstamp, newstamp;
long diffsec;
char *p, *p2, *p3, *p4;
if (posix_memalign((void **)&p, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p2, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p3, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
memset(p, 0xff, SIZE);
printf("%p\n", p);
memset(p2, 0xff, SIZE);
memset(p3, 0x77, 4096);
if (memcmp(p, p2, SIZE))
printf("error\n");
p4 = mremap(p+SIZE/2, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p3);
if (p4 != p3)
perror("mremap"), exit(1);
p4 = mremap(p4, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p+SIZE/2);
if (p4 != p+SIZE/2)
perror("mremap"), exit(1);
if (memcmp(p, p2, SIZE))
printf("error\n");
printf("ok\n");
return 0;
}
===
$ perf probe -a anon_vma_moveto_tail
Add new event:
probe:anon_vma_moveto_tail (on anon_vma_moveto_tail)
You can now use it on all perf tools, such as:
perf record -e probe:anon_vma_moveto_tail -aR sleep 1
$ perf record -e probe:anon_vma_moveto_tail -aR ./anon_vma_moveto_tail
0x7f2ca2800000
ok
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.043 MB perf.data (~1860 samples) ]
$ perf report --stdio
100.00% anon_vma_moveto [kernel.kallsyms] [k] anon_vma_moveto_tail
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Nai Xia <nai.xia@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pawel Sikora <pluto@agmk.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-01-11 07:08:05 +08:00
|
|
|
faulted_in_anon_vma = false;
|
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
|
|
|
|
|
find_vma_prepare(mm, addr, &prev, &rb_link, &rb_parent);
|
|
|
|
|
new_vma = vma_merge(mm, prev, addr, addr + len, vma->vm_flags,
|
|
|
|
|
vma->anon_vma, vma->vm_file, pgoff, vma_policy(vma));
|
|
|
|
|
if (new_vma) {
|
|
|
|
|
/*
|
|
|
|
|
* Source vma may have been merged into new_vma
|
|
|
|
|
*/
|
mremap: enforce rmap src/dst vma ordering in case of vma_merge() succeeding in copy_vma()
migrate was doing an rmap_walk with speculative lock-less access on
pagetables. That could lead it to not serializing properly against mremap
PT locks. But a second problem remains in the order of vmas in the
same_anon_vma list used by the rmap_walk.
If vma_merge succeeds in copy_vma, the src vma could be placed after the
dst vma in the same_anon_vma list. That could still lead to migrate
missing some pte.
This patch adds an anon_vma_moveto_tail() function to force the dst vma at
the end of the list before mremap starts to solve the problem.
If the mremap is very large and there are a lots of parents or childs
sharing the anon_vma root lock, this should still scale better than taking
the anon_vma root lock around every pte copy practically for the whole
duration of mremap.
Update: Hugh noticed special care is needed in the error path where
move_page_tables goes in the reverse direction, a second
anon_vma_moveto_tail() call is needed in the error path.
This program exercises the anon_vma_moveto_tail:
===
int main()
{
static struct timeval oldstamp, newstamp;
long diffsec;
char *p, *p2, *p3, *p4;
if (posix_memalign((void **)&p, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p2, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p3, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
memset(p, 0xff, SIZE);
printf("%p\n", p);
memset(p2, 0xff, SIZE);
memset(p3, 0x77, 4096);
if (memcmp(p, p2, SIZE))
printf("error\n");
p4 = mremap(p+SIZE/2, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p3);
if (p4 != p3)
perror("mremap"), exit(1);
p4 = mremap(p4, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p+SIZE/2);
if (p4 != p+SIZE/2)
perror("mremap"), exit(1);
if (memcmp(p, p2, SIZE))
printf("error\n");
printf("ok\n");
return 0;
}
===
$ perf probe -a anon_vma_moveto_tail
Add new event:
probe:anon_vma_moveto_tail (on anon_vma_moveto_tail)
You can now use it on all perf tools, such as:
perf record -e probe:anon_vma_moveto_tail -aR sleep 1
$ perf record -e probe:anon_vma_moveto_tail -aR ./anon_vma_moveto_tail
0x7f2ca2800000
ok
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.043 MB perf.data (~1860 samples) ]
$ perf report --stdio
100.00% anon_vma_moveto [kernel.kallsyms] [k] anon_vma_moveto_tail
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Nai Xia <nai.xia@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pawel Sikora <pluto@agmk.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-01-11 07:08:05 +08:00
|
|
|
if (unlikely(vma_start >= new_vma->vm_start &&
|
|
|
|
|
vma_start < new_vma->vm_end)) {
|
|
|
|
|
/*
|
|
|
|
|
* The only way we can get a vma_merge with
|
|
|
|
|
* self during an mremap is if the vma hasn't
|
|
|
|
|
* been faulted in yet and we were allowed to
|
|
|
|
|
* reset the dst vma->vm_pgoff to the
|
|
|
|
|
* destination address of the mremap to allow
|
|
|
|
|
* the merge to happen. mremap must change the
|
|
|
|
|
* vm_pgoff linearity between src and dst vmas
|
|
|
|
|
* (in turn preventing a vma_merge) to be
|
|
|
|
|
* safe. It is only safe to keep the vm_pgoff
|
|
|
|
|
* linear if there are no pages mapped yet.
|
|
|
|
|
*/
|
|
|
|
|
VM_BUG_ON(faulted_in_anon_vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
*vmap = new_vma;
|
mremap: enforce rmap src/dst vma ordering in case of vma_merge() succeeding in copy_vma()
migrate was doing an rmap_walk with speculative lock-less access on
pagetables. That could lead it to not serializing properly against mremap
PT locks. But a second problem remains in the order of vmas in the
same_anon_vma list used by the rmap_walk.
If vma_merge succeeds in copy_vma, the src vma could be placed after the
dst vma in the same_anon_vma list. That could still lead to migrate
missing some pte.
This patch adds an anon_vma_moveto_tail() function to force the dst vma at
the end of the list before mremap starts to solve the problem.
If the mremap is very large and there are a lots of parents or childs
sharing the anon_vma root lock, this should still scale better than taking
the anon_vma root lock around every pte copy practically for the whole
duration of mremap.
Update: Hugh noticed special care is needed in the error path where
move_page_tables goes in the reverse direction, a second
anon_vma_moveto_tail() call is needed in the error path.
This program exercises the anon_vma_moveto_tail:
===
int main()
{
static struct timeval oldstamp, newstamp;
long diffsec;
char *p, *p2, *p3, *p4;
if (posix_memalign((void **)&p, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p2, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
if (posix_memalign((void **)&p3, 2*1024*1024, SIZE))
perror("memalign"), exit(1);
memset(p, 0xff, SIZE);
printf("%p\n", p);
memset(p2, 0xff, SIZE);
memset(p3, 0x77, 4096);
if (memcmp(p, p2, SIZE))
printf("error\n");
p4 = mremap(p+SIZE/2, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p3);
if (p4 != p3)
perror("mremap"), exit(1);
p4 = mremap(p4, SIZE/2, SIZE/2, MREMAP_FIXED|MREMAP_MAYMOVE, p+SIZE/2);
if (p4 != p+SIZE/2)
perror("mremap"), exit(1);
if (memcmp(p, p2, SIZE))
printf("error\n");
printf("ok\n");
return 0;
}
===
$ perf probe -a anon_vma_moveto_tail
Add new event:
probe:anon_vma_moveto_tail (on anon_vma_moveto_tail)
You can now use it on all perf tools, such as:
perf record -e probe:anon_vma_moveto_tail -aR sleep 1
$ perf record -e probe:anon_vma_moveto_tail -aR ./anon_vma_moveto_tail
0x7f2ca2800000
ok
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.043 MB perf.data (~1860 samples) ]
$ perf report --stdio
100.00% anon_vma_moveto [kernel.kallsyms] [k] anon_vma_moveto_tail
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Nai Xia <nai.xia@gmail.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pawel Sikora <pluto@agmk.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-01-11 07:08:05 +08:00
|
|
|
} else
|
|
|
|
|
anon_vma_moveto_tail(new_vma);
|
2005-04-17 06:20:36 +08:00
|
|
|
} else {
|
2006-12-07 12:33:17 +08:00
|
|
|
new_vma = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL);
|
2005-04-17 06:20:36 +08:00
|
|
|
if (new_vma) {
|
|
|
|
|
*new_vma = *vma;
|
2008-04-28 17:13:09 +08:00
|
|
|
pol = mpol_dup(vma_policy(vma));
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
if (IS_ERR(pol))
|
|
|
|
|
goto out_free_vma;
|
|
|
|
|
INIT_LIST_HEAD(&new_vma->anon_vma_chain);
|
|
|
|
|
if (anon_vma_clone(new_vma, vma))
|
|
|
|
|
goto out_free_mempol;
|
2005-04-17 06:20:36 +08:00
|
|
|
vma_set_policy(new_vma, pol);
|
|
|
|
|
new_vma->vm_start = addr;
|
|
|
|
|
new_vma->vm_end = addr + len;
|
|
|
|
|
new_vma->vm_pgoff = pgoff;
|
2008-04-29 16:01:36 +08:00
|
|
|
if (new_vma->vm_file) {
|
2005-04-17 06:20:36 +08:00
|
|
|
get_file(new_vma->vm_file);
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
|
2012-02-17 16:27:41 +08:00
|
|
|
if (uprobe_mmap(new_vma))
|
uprobes, mm, x86: Add the ability to install and remove uprobes breakpoints
Add uprobes support to the core kernel, with x86 support.
This commit adds the kernel facilities, the actual uprobes
user-space ABI and perf probe support comes in later commits.
General design:
Uprobes are maintained in an rb-tree indexed by inode and offset
(the offset here is from the start of the mapping). For a unique
(inode, offset) tuple, there can be at most one uprobe in the
rb-tree.
Since the (inode, offset) tuple identifies a unique uprobe, more
than one user may be interested in the same uprobe. This provides
the ability to connect multiple 'consumers' to the same uprobe.
Each consumer defines a handler and a filter (optional). The
'handler' is run every time the uprobe is hit, if it matches the
'filter' criteria.
The first consumer of a uprobe causes the breakpoint to be
inserted at the specified address and subsequent consumers are
appended to this list. On subsequent probes, the consumer gets
appended to the existing list of consumers. The breakpoint is
removed when the last consumer unregisters. For all other
unregisterations, the consumer is removed from the list of
consumers.
Given a inode, we get a list of the mms that have mapped the
inode. Do the actual registration if mm maps the page where a
probe needs to be inserted/removed.
We use a temporary list to walk through the vmas that map the
inode.
- The number of maps that map the inode, is not known before we
walk the rmap and keeps changing.
- extending vm_area_struct wasn't recommended, it's a
size-critical data structure.
- There can be more than one maps of the inode in the same mm.
We add callbacks to the mmap methods to keep an eye on text vmas
that are of interest to uprobes. When a vma of interest is mapped,
we insert the breakpoint at the right address.
Uprobe works by replacing the instruction at the address defined
by (inode, offset) with the arch specific breakpoint
instruction. We save a copy of the original instruction at the
uprobed address.
This is needed for:
a. executing the instruction out-of-line (xol).
b. instruction analysis for any subsequent fixups.
c. restoring the instruction back when the uprobe is unregistered.
We insert or delete a breakpoint instruction, and this
breakpoint instruction is assumed to be the smallest instruction
available on the platform. For fixed size instruction platforms
this is trivially true, for variable size instruction platforms
the breakpoint instruction is typically the smallest (often a
single byte).
Writing the instruction is done by COWing the page and changing
the instruction during the copy, this even though most platforms
allow atomic writes of the breakpoint instruction. This also
mirrors the behaviour of a ptrace() memory write to a PRIVATE
file map.
The core worker is derived from KSM's replace_page() logic.
In essence, similar to KSM:
a. allocate a new page and copy over contents of the page that
has the uprobed vaddr
b. modify the copy and insert the breakpoint at the required
address
c. switch the original page with the copy containing the
breakpoint
d. flush page tables.
replace_page() is being replicated here because of some minor
changes in the type of pages and also because Hugh Dickins had
plans to improve replace_page() for KSM specific work.
Instruction analysis on x86 is based on instruction decoder and
determines if an instruction can be probed and determines the
necessary fixups after singlestep. Instruction analysis is done
at probe insertion time so that we avoid having to repeat the
same analysis every time a probe is hit.
A lot of code here is due to the improvement/suggestions/inputs
from Peter Zijlstra.
Changelog:
(v10):
- Add code to clear REX.B prefix as suggested by Denys Vlasenko
and Masami Hiramatsu.
(v9):
- Use insn_offset_modrm as suggested by Masami Hiramatsu.
(v7):
Handle comments from Peter Zijlstra:
- Dont take reference to inode. (expect inode to uprobe_register to be sane).
- Use PTR_ERR to set the return value.
- No need to take reference to inode.
- use PTR_ERR to return error value.
- register and uprobe_unregister share code.
(v5):
- Modified del_consumer as per comments from Peter.
- Drop reference to inode before dropping reference to uprobe.
- Use i_size_read(inode) instead of inode->i_size.
- Ensure uprobe->consumers is NULL, before __uprobe_unregister() is called.
- Includes errno.h as recommended by Stephen Rothwell to fix a build issue
on sparc defconfig
- Remove restrictions while unregistering.
- Earlier code leaked inode references under some conditions while
registering/unregistering.
- Continue the vma-rmap walk even if the intermediate vma doesnt
meet the requirements.
- Validate the vma found by find_vma before inserting/removing the
breakpoint
- Call del_consumer under mutex_lock.
- Use hash locks.
- Handle mremap.
- Introduce find_least_offset_node() instead of close match logic in
find_uprobe
- Uprobes no more depends on MM_OWNER; No reference to task_structs
while inserting/removing a probe.
- Uses read_mapping_page instead of grab_cache_page so that the pages
have valid content.
- pass NULL to get_user_pages for the task parameter.
- call SetPageUptodate on the new page allocated in write_opcode.
- fix leaking a reference to the new page under certain conditions.
- Include Instruction Decoder if Uprobes gets defined.
- Remove const attributes for instruction prefix arrays.
- Uses mm_context to know if the application is 32 bit.
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Also-written-by: Jim Keniston <jkenisto@us.ibm.com>
Reviewed-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Roland McGrath <roland@hack.frob.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Anton Arapov <anton@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux-mm <linux-mm@kvack.org>
Link: http://lkml.kernel.org/r/20120209092642.GE16600@linux.vnet.ibm.com
[ Made various small edits to the commit log ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2012-02-09 17:26:42 +08:00
|
|
|
goto out_free_mempol;
|
|
|
|
|
|
2008-04-29 16:01:36 +08:00
|
|
|
if (vma->vm_flags & VM_EXECUTABLE)
|
|
|
|
|
added_exe_file_vma(mm);
|
|
|
|
|
}
|
2005-04-17 06:20:36 +08:00
|
|
|
if (new_vma->vm_ops && new_vma->vm_ops->open)
|
|
|
|
|
new_vma->vm_ops->open(new_vma);
|
|
|
|
|
vma_link(mm, new_vma, prev, rb_link, rb_parent);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return new_vma;
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
|
|
|
|
|
out_free_mempol:
|
|
|
|
|
mpol_put(pol);
|
|
|
|
|
out_free_vma:
|
|
|
|
|
kmem_cache_free(vm_area_cachep, new_vma);
|
|
|
|
|
return NULL;
|
2005-04-17 06:20:36 +08:00
|
|
|
}
|
2005-05-01 23:58:35 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Return true if the calling process may expand its vm space by the passed
|
|
|
|
|
* number of pages
|
|
|
|
|
*/
|
|
|
|
|
int may_expand_vm(struct mm_struct *mm, unsigned long npages)
|
|
|
|
|
{
|
|
|
|
|
unsigned long cur = mm->total_vm; /* pages */
|
|
|
|
|
unsigned long lim;
|
|
|
|
|
|
2010-03-06 05:41:44 +08:00
|
|
|
lim = rlimit(RLIMIT_AS) >> PAGE_SHIFT;
|
2005-05-01 23:58:35 +08:00
|
|
|
|
|
|
|
|
if (cur + npages > lim)
|
|
|
|
|
return 0;
|
|
|
|
|
return 1;
|
|
|
|
|
}
|
2007-02-09 06:20:41 +08:00
|
|
|
|
|
|
|
|
|
2008-02-09 08:15:19 +08:00
|
|
|
static int special_mapping_fault(struct vm_area_struct *vma,
|
|
|
|
|
struct vm_fault *vmf)
|
2007-02-09 06:20:41 +08:00
|
|
|
{
|
2008-02-09 08:15:19 +08:00
|
|
|
pgoff_t pgoff;
|
2007-02-09 06:20:41 +08:00
|
|
|
struct page **pages;
|
|
|
|
|
|
2008-02-09 08:15:19 +08:00
|
|
|
/*
|
|
|
|
|
* special mappings have no vm_file, and in that case, the mm
|
|
|
|
|
* uses vm_pgoff internally. So we have to subtract it from here.
|
|
|
|
|
* We are allowed to do this because we are the mm; do not copy
|
|
|
|
|
* this code into drivers!
|
|
|
|
|
*/
|
|
|
|
|
pgoff = vmf->pgoff - vma->vm_pgoff;
|
2007-02-09 06:20:41 +08:00
|
|
|
|
2008-02-09 08:15:19 +08:00
|
|
|
for (pages = vma->vm_private_data; pgoff && *pages; ++pages)
|
|
|
|
|
pgoff--;
|
2007-02-09 06:20:41 +08:00
|
|
|
|
|
|
|
|
if (*pages) {
|
|
|
|
|
struct page *page = *pages;
|
|
|
|
|
get_page(page);
|
2008-02-09 08:15:19 +08:00
|
|
|
vmf->page = page;
|
|
|
|
|
return 0;
|
2007-02-09 06:20:41 +08:00
|
|
|
}
|
|
|
|
|
|
2008-02-09 08:15:19 +08:00
|
|
|
return VM_FAULT_SIGBUS;
|
2007-02-09 06:20:41 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Having a close hook prevents vma merging regardless of flags.
|
|
|
|
|
*/
|
|
|
|
|
static void special_mapping_close(struct vm_area_struct *vma)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
2009-09-28 02:29:37 +08:00
|
|
|
static const struct vm_operations_struct special_mapping_vmops = {
|
2007-02-09 06:20:41 +08:00
|
|
|
.close = special_mapping_close,
|
2008-02-09 08:15:19 +08:00
|
|
|
.fault = special_mapping_fault,
|
2007-02-09 06:20:41 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Called with mm->mmap_sem held for writing.
|
|
|
|
|
* Insert a new vma covering the given region, with the given flags.
|
|
|
|
|
* Its pages are supplied by the given array of struct page *.
|
|
|
|
|
* The array can be shorter than len >> PAGE_SHIFT if it's null-terminated.
|
|
|
|
|
* The region past the last page supplied will always produce SIGBUS.
|
|
|
|
|
* The array pointer and the pages it points to are assumed to stay alive
|
|
|
|
|
* for as long as this mapping might exist.
|
|
|
|
|
*/
|
|
|
|
|
int install_special_mapping(struct mm_struct *mm,
|
|
|
|
|
unsigned long addr, unsigned long len,
|
|
|
|
|
unsigned long vm_flags, struct page **pages)
|
|
|
|
|
{
|
2010-12-09 22:29:42 +08:00
|
|
|
int ret;
|
2007-02-09 06:20:41 +08:00
|
|
|
struct vm_area_struct *vma;
|
|
|
|
|
|
|
|
|
|
vma = kmem_cache_zalloc(vm_area_cachep, GFP_KERNEL);
|
|
|
|
|
if (unlikely(vma == NULL))
|
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
INIT_LIST_HEAD(&vma->anon_vma_chain);
|
2007-02-09 06:20:41 +08:00
|
|
|
vma->vm_mm = mm;
|
|
|
|
|
vma->vm_start = addr;
|
|
|
|
|
vma->vm_end = addr + len;
|
|
|
|
|
|
2008-02-02 10:08:53 +08:00
|
|
|
vma->vm_flags = vm_flags | mm->def_flags | VM_DONTEXPAND;
|
2007-10-19 14:39:15 +08:00
|
|
|
vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
|
2007-02-09 06:20:41 +08:00
|
|
|
|
|
|
|
|
vma->vm_ops = &special_mapping_vmops;
|
|
|
|
|
vma->vm_private_data = pages;
|
|
|
|
|
|
2010-12-09 22:29:42 +08:00
|
|
|
ret = security_file_mmap(NULL, 0, 0, 0, vma->vm_start, 1);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
|
|
ret = insert_vm_struct(mm, vma);
|
|
|
|
|
if (ret)
|
|
|
|
|
goto out;
|
2007-02-09 06:20:41 +08:00
|
|
|
|
|
|
|
|
mm->total_vm += len >> PAGE_SHIFT;
|
|
|
|
|
|
perf: Do the big rename: Performance Counters -> Performance Events
Bye-bye Performance Counters, welcome Performance Events!
In the past few months the perfcounters subsystem has grown out its
initial role of counting hardware events, and has become (and is
becoming) a much broader generic event enumeration, reporting, logging,
monitoring, analysis facility.
Naming its core object 'perf_counter' and naming the subsystem
'perfcounters' has become more and more of a misnomer. With pending
code like hw-breakpoints support the 'counter' name is less and
less appropriate.
All in one, we've decided to rename the subsystem to 'performance
events' and to propagate this rename through all fields, variables
and API names. (in an ABI compatible fashion)
The word 'event' is also a bit shorter than 'counter' - which makes
it slightly more convenient to write/handle as well.
Thanks goes to Stephane Eranian who first observed this misnomer and
suggested a rename.
User-space tooling and ABI compatibility is not affected - this patch
should be function-invariant. (Also, defconfigs were not touched to
keep the size down.)
This patch has been generated via the following script:
FILES=$(find * -type f | grep -vE 'oprofile|[^K]config')
sed -i \
-e 's/PERF_EVENT_/PERF_RECORD_/g' \
-e 's/PERF_COUNTER/PERF_EVENT/g' \
-e 's/perf_counter/perf_event/g' \
-e 's/nb_counters/nb_events/g' \
-e 's/swcounter/swevent/g' \
-e 's/tpcounter_event/tp_event/g' \
$FILES
for N in $(find . -name perf_counter.[ch]); do
M=$(echo $N | sed 's/perf_counter/perf_event/g')
mv $N $M
done
FILES=$(find . -name perf_event.*)
sed -i \
-e 's/COUNTER_MASK/REG_MASK/g' \
-e 's/COUNTER/EVENT/g' \
-e 's/\<event\>/event_id/g' \
-e 's/counter/event/g' \
-e 's/Counter/Event/g' \
$FILES
... to keep it as correct as possible. This script can also be
used by anyone who has pending perfcounters patches - it converts
a Linux kernel tree over to the new naming. We tried to time this
change to the point in time where the amount of pending patches
is the smallest: the end of the merge window.
Namespace clashes were fixed up in a preparatory patch - and some
stylistic fallout will be fixed up in a subsequent patch.
( NOTE: 'counters' are still the proper terminology when we deal
with hardware registers - and these sed scripts are a bit
over-eager in renaming them. I've undone some of that, but
in case there's something left where 'counter' would be
better than 'event' we can undo that on an individual basis
instead of touching an otherwise nicely automated patch. )
Suggested-by: Stephane Eranian <eranian@google.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Paul Mackerras <paulus@samba.org>
Reviewed-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: <linux-arch@vger.kernel.org>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-21 18:02:48 +08:00
|
|
|
perf_event_mmap(vma);
|
2009-06-05 20:04:55 +08:00
|
|
|
|
2007-02-09 06:20:41 +08:00
|
|
|
return 0;
|
2010-12-09 22:29:42 +08:00
|
|
|
|
|
|
|
|
out:
|
|
|
|
|
kmem_cache_free(vm_area_cachep, vma);
|
|
|
|
|
return ret;
|
2007-02-09 06:20:41 +08:00
|
|
|
}
|
2008-07-29 06:46:26 +08:00
|
|
|
|
|
|
|
|
static DEFINE_MUTEX(mm_all_locks_mutex);
|
|
|
|
|
|
2008-08-11 15:30:25 +08:00
|
|
|
static void vm_lock_anon_vma(struct mm_struct *mm, struct anon_vma *anon_vma)
|
2008-07-29 06:46:26 +08:00
|
|
|
{
|
2010-08-10 08:18:40 +08:00
|
|
|
if (!test_bit(0, (unsigned long *) &anon_vma->root->head.next)) {
|
2008-07-29 06:46:26 +08:00
|
|
|
/*
|
|
|
|
|
* The LSB of head.next can't change from under us
|
|
|
|
|
* because we hold the mm_all_locks_mutex.
|
|
|
|
|
*/
|
2011-05-25 08:12:11 +08:00
|
|
|
mutex_lock_nest_lock(&anon_vma->root->mutex, &mm->mmap_sem);
|
2008-07-29 06:46:26 +08:00
|
|
|
/*
|
|
|
|
|
* We can safely modify head.next after taking the
|
2011-05-25 08:12:11 +08:00
|
|
|
* anon_vma->root->mutex. If some other vma in this mm shares
|
2008-07-29 06:46:26 +08:00
|
|
|
* the same anon_vma we won't take it again.
|
|
|
|
|
*
|
|
|
|
|
* No need of atomic instructions here, head.next
|
|
|
|
|
* can't change from under us thanks to the
|
2011-05-25 08:12:11 +08:00
|
|
|
* anon_vma->root->mutex.
|
2008-07-29 06:46:26 +08:00
|
|
|
*/
|
|
|
|
|
if (__test_and_set_bit(0, (unsigned long *)
|
2010-08-10 08:18:40 +08:00
|
|
|
&anon_vma->root->head.next))
|
2008-07-29 06:46:26 +08:00
|
|
|
BUG();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2008-08-11 15:30:25 +08:00
|
|
|
static void vm_lock_mapping(struct mm_struct *mm, struct address_space *mapping)
|
2008-07-29 06:46:26 +08:00
|
|
|
{
|
|
|
|
|
if (!test_bit(AS_MM_ALL_LOCKS, &mapping->flags)) {
|
|
|
|
|
/*
|
|
|
|
|
* AS_MM_ALL_LOCKS can't change from under us because
|
|
|
|
|
* we hold the mm_all_locks_mutex.
|
|
|
|
|
*
|
|
|
|
|
* Operations on ->flags have to be atomic because
|
|
|
|
|
* even if AS_MM_ALL_LOCKS is stable thanks to the
|
|
|
|
|
* mm_all_locks_mutex, there may be other cpus
|
|
|
|
|
* changing other bitflags in parallel to us.
|
|
|
|
|
*/
|
|
|
|
|
if (test_and_set_bit(AS_MM_ALL_LOCKS, &mapping->flags))
|
|
|
|
|
BUG();
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_lock_nest_lock(&mapping->i_mmap_mutex, &mm->mmap_sem);
|
2008-07-29 06:46:26 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* This operation locks against the VM for all pte/vma/mm related
|
|
|
|
|
* operations that could ever happen on a certain mm. This includes
|
|
|
|
|
* vmtruncate, try_to_unmap, and all page faults.
|
|
|
|
|
*
|
|
|
|
|
* The caller must take the mmap_sem in write mode before calling
|
|
|
|
|
* mm_take_all_locks(). The caller isn't allowed to release the
|
|
|
|
|
* mmap_sem until mm_drop_all_locks() returns.
|
|
|
|
|
*
|
|
|
|
|
* mmap_sem in write mode is required in order to block all operations
|
|
|
|
|
* that could modify pagetables and free pages without need of
|
|
|
|
|
* altering the vma layout (for example populate_range() with
|
|
|
|
|
* nonlinear vmas). It's also needed in write mode to avoid new
|
|
|
|
|
* anon_vmas to be associated with existing vmas.
|
|
|
|
|
*
|
|
|
|
|
* A single task can't take more than one mm_take_all_locks() in a row
|
|
|
|
|
* or it would deadlock.
|
|
|
|
|
*
|
|
|
|
|
* The LSB in anon_vma->head.next and the AS_MM_ALL_LOCKS bitflag in
|
|
|
|
|
* mapping->flags avoid to take the same lock twice, if more than one
|
|
|
|
|
* vma in this mm is backed by the same anon_vma or address_space.
|
|
|
|
|
*
|
|
|
|
|
* We can take all the locks in random order because the VM code
|
2011-05-25 08:12:11 +08:00
|
|
|
* taking i_mmap_mutex or anon_vma->mutex outside the mmap_sem never
|
2008-07-29 06:46:26 +08:00
|
|
|
* takes more than one of them in a row. Secondly we're protected
|
|
|
|
|
* against a concurrent mm_take_all_locks() by the mm_all_locks_mutex.
|
|
|
|
|
*
|
|
|
|
|
* mm_take_all_locks() and mm_drop_all_locks are expensive operations
|
|
|
|
|
* that may have to take thousand of locks.
|
|
|
|
|
*
|
|
|
|
|
* mm_take_all_locks() can fail if it's interrupted by signals.
|
|
|
|
|
*/
|
|
|
|
|
int mm_take_all_locks(struct mm_struct *mm)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct *vma;
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
struct anon_vma_chain *avc;
|
2008-07-29 06:46:26 +08:00
|
|
|
|
|
|
|
|
BUG_ON(down_read_trylock(&mm->mmap_sem));
|
|
|
|
|
|
|
|
|
|
mutex_lock(&mm_all_locks_mutex);
|
|
|
|
|
|
|
|
|
|
for (vma = mm->mmap; vma; vma = vma->vm_next) {
|
|
|
|
|
if (signal_pending(current))
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
if (vma->vm_file && vma->vm_file->f_mapping)
|
2008-08-11 15:30:25 +08:00
|
|
|
vm_lock_mapping(mm, vma->vm_file->f_mapping);
|
2008-07-29 06:46:26 +08:00
|
|
|
}
|
2008-08-11 15:30:25 +08:00
|
|
|
|
|
|
|
|
for (vma = mm->mmap; vma; vma = vma->vm_next) {
|
|
|
|
|
if (signal_pending(current))
|
|
|
|
|
goto out_unlock;
|
|
|
|
|
if (vma->anon_vma)
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
list_for_each_entry(avc, &vma->anon_vma_chain, same_vma)
|
|
|
|
|
vm_lock_anon_vma(mm, avc->anon_vma);
|
2008-07-29 06:46:26 +08:00
|
|
|
}
|
2008-08-11 15:30:25 +08:00
|
|
|
|
2011-11-01 08:08:59 +08:00
|
|
|
return 0;
|
2008-07-29 06:46:26 +08:00
|
|
|
|
|
|
|
|
out_unlock:
|
2011-11-01 08:08:59 +08:00
|
|
|
mm_drop_all_locks(mm);
|
|
|
|
|
return -EINTR;
|
2008-07-29 06:46:26 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void vm_unlock_anon_vma(struct anon_vma *anon_vma)
|
|
|
|
|
{
|
2010-08-10 08:18:40 +08:00
|
|
|
if (test_bit(0, (unsigned long *) &anon_vma->root->head.next)) {
|
2008-07-29 06:46:26 +08:00
|
|
|
/*
|
|
|
|
|
* The LSB of head.next can't change to 0 from under
|
|
|
|
|
* us because we hold the mm_all_locks_mutex.
|
|
|
|
|
*
|
|
|
|
|
* We must however clear the bitflag before unlocking
|
|
|
|
|
* the vma so the users using the anon_vma->head will
|
|
|
|
|
* never see our bitflag.
|
|
|
|
|
*
|
|
|
|
|
* No need of atomic instructions here, head.next
|
|
|
|
|
* can't change from under us until we release the
|
2011-05-25 08:12:11 +08:00
|
|
|
* anon_vma->root->mutex.
|
2008-07-29 06:46:26 +08:00
|
|
|
*/
|
|
|
|
|
if (!__test_and_clear_bit(0, (unsigned long *)
|
2010-08-10 08:18:40 +08:00
|
|
|
&anon_vma->root->head.next))
|
2008-07-29 06:46:26 +08:00
|
|
|
BUG();
|
2010-08-10 08:18:38 +08:00
|
|
|
anon_vma_unlock(anon_vma);
|
2008-07-29 06:46:26 +08:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void vm_unlock_mapping(struct address_space *mapping)
|
|
|
|
|
{
|
|
|
|
|
if (test_bit(AS_MM_ALL_LOCKS, &mapping->flags)) {
|
|
|
|
|
/*
|
|
|
|
|
* AS_MM_ALL_LOCKS can't change to 0 from under us
|
|
|
|
|
* because we hold the mm_all_locks_mutex.
|
|
|
|
|
*/
|
2011-05-25 08:12:06 +08:00
|
|
|
mutex_unlock(&mapping->i_mmap_mutex);
|
2008-07-29 06:46:26 +08:00
|
|
|
if (!test_and_clear_bit(AS_MM_ALL_LOCKS,
|
|
|
|
|
&mapping->flags))
|
|
|
|
|
BUG();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* The mmap_sem cannot be released by the caller until
|
|
|
|
|
* mm_drop_all_locks() returns.
|
|
|
|
|
*/
|
|
|
|
|
void mm_drop_all_locks(struct mm_struct *mm)
|
|
|
|
|
{
|
|
|
|
|
struct vm_area_struct *vma;
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
struct anon_vma_chain *avc;
|
2008-07-29 06:46:26 +08:00
|
|
|
|
|
|
|
|
BUG_ON(down_read_trylock(&mm->mmap_sem));
|
|
|
|
|
BUG_ON(!mutex_is_locked(&mm_all_locks_mutex));
|
|
|
|
|
|
|
|
|
|
for (vma = mm->mmap; vma; vma = vma->vm_next) {
|
|
|
|
|
if (vma->anon_vma)
|
mm: change anon_vma linking to fix multi-process server scalability issue
The old anon_vma code can lead to scalability issues with heavily forking
workloads. Specifically, each anon_vma will be shared between the parent
process and all its child processes.
In a workload with 1000 child processes and a VMA with 1000 anonymous
pages per process that get COWed, this leads to a system with a million
anonymous pages in the same anon_vma, each of which is mapped in just one
of the 1000 processes. However, the current rmap code needs to walk them
all, leading to O(N) scanning complexity for each page.
This can result in systems where one CPU is walking the page tables of
1000 processes in page_referenced_one, while all other CPUs are stuck on
the anon_vma lock. This leads to catastrophic failure for a benchmark
like AIM7, where the total number of processes can reach in the tens of
thousands. Real workloads are still a factor 10 less process intensive
than AIM7, but they are catching up.
This patch changes the way anon_vmas and VMAs are linked, which allows us
to associate multiple anon_vmas with a VMA. At fork time, each child
process gets its own anon_vmas, in which its COWed pages will be
instantiated. The parents' anon_vma is also linked to the VMA, because
non-COWed pages could be present in any of the children.
This reduces rmap scanning complexity to O(1) for the pages of the 1000
child processes, with O(N) complexity for at most 1/N pages in the system.
This reduces the average scanning cost in heavily forking workloads from
O(N) to 2.
The only real complexity in this patch stems from the fact that linking a
VMA to anon_vmas now involves memory allocations. This means vma_adjust
can fail, if it needs to attach a VMA to anon_vma structures. This in
turn means error handling needs to be added to the calling functions.
A second source of complexity is that, because there can be multiple
anon_vmas, the anon_vma linking in vma_adjust can no longer be done under
"the" anon_vma lock. To prevent the rmap code from walking up an
incomplete VMA, this patch introduces the VM_LOCK_RMAP VMA flag. This bit
flag uses the same slot as the NOMMU VM_MAPPED_COPY, with an ifdef in mm.h
to make sure it is impossible to compile a kernel that needs both symbolic
values for the same bitflag.
Some test results:
Without the anon_vma changes, when AIM7 hits around 9.7k users (on a test
box with 16GB RAM and not quite enough IO), the system ends up running
>99% in system time, with every CPU on the same anon_vma lock in the
pageout code.
With these changes, AIM7 hits the cross-over point around 29.7k users.
This happens with ~99% IO wait time, there never seems to be any spike in
system time. The anon_vma lock contention appears to be resolved.
[akpm@linux-foundation.org: cleanups]
Signed-off-by: Rik van Riel <riel@redhat.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-06 05:42:07 +08:00
|
|
|
list_for_each_entry(avc, &vma->anon_vma_chain, same_vma)
|
|
|
|
|
vm_unlock_anon_vma(avc->anon_vma);
|
2008-07-29 06:46:26 +08:00
|
|
|
if (vma->vm_file && vma->vm_file->f_mapping)
|
|
|
|
|
vm_unlock_mapping(vma->vm_file->f_mapping);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
mutex_unlock(&mm_all_locks_mutex);
|
|
|
|
|
}
|
2009-01-08 20:04:47 +08:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* initialise the VMA slab
|
|
|
|
|
*/
|
|
|
|
|
void __init mmap_init(void)
|
|
|
|
|
{
|
2009-05-01 06:08:51 +08:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
|
|
ret = percpu_counter_init(&vm_committed_as, 0);
|
|
|
|
|
VM_BUG_ON(ret);
|
2009-01-08 20:04:47 +08:00
|
|
|
}
|
