Ingo recently introduced a great speedup for allocating new mmaps using the
free_area_cache pointer which boosts the specweb SSL benchmark by 4-5% and
causes huge performance increases in thread creation.
The downside of this patch is that it does lead to fragmentation in the
mmap-ed areas (visible via /proc/self/maps), such that some applications
that work fine under 2.4 kernels quickly run out of memory on any 2.6
kernel.
The problem is twofold:
1) the free_area_cache is used to continue a search for memory where
the last search ended. Before the change new areas were always
searched from the base address on.
So now new small areas are cluttering holes of all sizes
throughout the whole mmap-able region whereas before small holes
tended to close holes near the base leaving holes far from the base
large and available for larger requests.
2) the free_area_cache also is set to the location of the last
munmap-ed area so in scenarios where we allocate e.g. five regions of
1K each, then free regions 4 2 3 in this order the next request for 1K
will be placed in the position of the old region 3, whereas before we
appended it to the still active region 1, placing it at the location
of the old region 2. Before we had 1 free region of 2K, now we only
get two free regions of 1K -> fragmentation.
The patch addresses thes issues by introducing yet another cache descriptor
cached_hole_size that contains the largest known hole size below the
current free_area_cache. If a new request comes in the size is compared
against the cached_hole_size and if the request can be filled with a hole
below free_area_cache the search is started from the base instead.
The results look promising: Whereas 2.6.12-rc4 fragments quickly and my
(earlier posted) leakme.c test program terminates after 50000+ iterations
with 96 distinct and fragmented maps in /proc/self/maps it performs nicely
(as expected) with thread creation, Ingo's test_str02 with 20000 threads
requires 0.7s system time.
Taking out Ingo's patch (un-patch available per request) by basically
deleting all mentions of free_area_cache from the kernel and starting the
search for new memory always at the respective bases we observe: leakme
terminates successfully with 11 distinctive hardly fragmented areas in
/proc/self/maps but thread creating is gringdingly slow: 30+s(!) system
time for Ingo's test_str02 with 20000 threads.
Now - drumroll ;-) the appended patch works fine with leakme: it ends with
only 7 distinct areas in /proc/self/maps and also thread creation seems
sufficiently fast with 0.71s for 20000 threads.
Signed-off-by: Wolfgang Wander <wwc@rentec.com>
Credit-to: "Richard Purdie" <rpurdie@rpsys.net>
Signed-off-by: Ken Chen <kenneth.w.chen@intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu> (partly)
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
A lot of the code in arch/*/mm/hugetlbpage.c is quite similar. This patch
attempts to consolidate a lot of the code across the arch's, putting the
combined version in mm/hugetlb.c. There are a couple of uglyish hacks in
order to covert all the hugepage archs, but the result is a very large
reduction in the total amount of code. It also means things like hugepage
lazy allocation could be implemented in one place, instead of six.
Tested, at least a little, on ppc64, i386 and x86_64.
Notes:
- this patch changes the meaning of set_huge_pte() to be more
analagous to set_pte()
- does SH4 need s special huge_ptep_get_and_clear()??
Acked-by: William Lee Irwin <wli@holomorphy.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch implements a number of smp_processor_id() cleanup ideas that
Arjan van de Ven and I came up with.
The previous __smp_processor_id/_smp_processor_id/smp_processor_id API
spaghetti was hard to follow both on the implementational and on the
usage side.
Some of the complexity arose from picking wrong names, some of the
complexity comes from the fact that not all architectures defined
__smp_processor_id.
In the new code, there are two externally visible symbols:
- smp_processor_id(): debug variant.
- raw_smp_processor_id(): nondebug variant. Replaces all existing
uses of _smp_processor_id() and __smp_processor_id(). Defined
by every SMP architecture in include/asm-*/smp.h.
There is one new internal symbol, dependent on DEBUG_PREEMPT:
- debug_smp_processor_id(): internal debug variant, mapped to
smp_processor_id().
Also, i moved debug_smp_processor_id() from lib/kernel_lock.c into a new
lib/smp_processor_id.c file. All related comments got updated and/or
clarified.
I have build/boot tested the following 8 .config combinations on x86:
{SMP,UP} x {PREEMPT,!PREEMPT} x {DEBUG_PREEMPT,!DEBUG_PREEMPT}
I have also build/boot tested x64 on UP/PREEMPT/DEBUG_PREEMPT. (Other
architectures are untested, but should work just fine.)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
A fairly recent platform requirement states that the OS must clear the
whole TCE table at setup time, in case firmware left any active
mappings in it. Without this initialization, dynamic bus removes can
fail. Firmware rejects these requests if active mappings still exist
for a slot that has been deallocated by the OS.
Signed-off-by: John Rose <johnrose@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Use the new cpu_has_feature macros instead of open coding it.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Some rtasd printks were too loud. They would appear on a quiet boot
even though they were only informational.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
enable cpusets
enable new lpfc and jsm drivers
enable new dm-multipath
leave new agp disabled
disable rivafb, it does not handle the cards in G5 models (FX5200 as example)
the new nvidiafb doesnt work on bigendian, yet
Signed-off-by: Olaf Hering <olh@suse.de>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Here is a patch to update the example configs in arch/ppc64/configs.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch fixes some bugs in the ppc64 PER_LINUX32 implementation,
noted by Juergen Kreileder:
* uname(2) doesn't respect PER_LINUX32, it returns 'ppc64' instead of 'ppc'
* Child processes of a PER_LINUX32 process don't inherit PER_LINUX32
Along the way I took the opportunity to move things around so that
sys_ppc32.c only has 32-bit syscall emulation functions and to remove
the obsolete "fakeppc" command line option.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Remove spurious MSR_SE reset during kprobe processing.
single_step_exception() already does it for us. Reset it to be safe when
executing the fault_handler.
Signed-off-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Add stricter checks during kprobe registration. Return correct error value so
insmod doesn't succeed. Also printk reason for registration failure.
Signed-off-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Kprobes was eating the hardware instruction and data address
breakpoint exceptions. This patch fixes it; kprobes doesn't use those
exceptions at all and should ignore them.
Signed-off-by: Ananth N Mavinakayanahalli <amavin@redhat.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
if num has a value of -1, accessing the digits[] array will fail and the
format string will be printed in funny way, or not at all. This happens if
one prints negative numbers.
Just change the code to match lib/vsprintf.c
asm/div64.h cant be used because u64 maps to u32 for this build.
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
A typo in prom_find_machine_type from Ben's recent patch "ppc64: Fix
result code handling in prom_init" prevents pSeries LPAR systems from
booting.
Tested on a pSeries 570 and OpenPower 720 (both Power5 LPAR).
Signed-off-by: Nathan Lynch <ntl@pobox.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Now that we have HZ=1000 there is much less of a need for decr_overclock.
Remove it.
Leave spread_lpevents but move it into iSeries_setup.c. We should look at
making event spreading the default some day.
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The iseries has a bar graph on the front panel that shows how busy it is.
The operating system sets and clears a bit in the CTRL register to control
it.
Instead of going to the complexity of using a thread info bit, just set and
clear it in the idle loop.
Also create two helper functions, ppc64_runlatch_on and ppc64_runlatch_off.
Finally don't use the short form of the SPR defines.
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
prom_init(), the trampoline code that "talks" to Open Firmware during
early boot, has various issues with managing OF result codes. Some of my
recent fixups in fact made the problem worse on some platforms.
This patch reworks it all. Tested on g5, Maple, POWER3 and POWER5.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This cleans up the /proc/device-tree representation of the Open Firmware
device-tree on ppc and ppc64. It does the following things:
- Workaround an issue in some Apple device-trees where a property may
exist with the same name as a child node of the parent. We now
simply "drop" the property instead of creating duplicate entries in
/proc with random result...
- Do not try to chop off the "@0" at the end of a node name whose unit
address is 0. This is not useful, inconsistent, and the code was
buggy and didn't always work anyway.
- Do not create symlinks for the short name and unit address parts of a
node. These were never really used, bloated the memory footprint of
the device-tree with useless struct proc_dir_entry and their matching
dentry and inode cache bloat.
This results in smaller code, smaller memory footprint, and a more
accurate view of the tree presented to userland.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Apple's Open Firmware has a funny bug when creating the /cpus nodes
where it leaves a dangling '\0' character in the CPU name which ends up
appearing in the full path of the node. This is bogus and
confuses /proc/device-tree badly.
This patch strips those bogus zero's from the node full path when
reading the device-tree from Open Firmware. The "name" property is not
modified and still contains the spurrious 0 (it basically contains 0
tailing 0 instead of one) but that shouldn't be a problem.
An equivalent patch for ppc32 will follow shortly
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
We used to have an iseries specific profiler that used /proc/profile. Now
thats gone we can use the generic timer based stuff.
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
When I sent in the patch adding the code for the kernel to tell the
firmware about its capabilities on pSeries machines, I included the
function to give the capabilities to firmware but somehow forgot the
hunk that adds the call to the new function. This patch adds the
call.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
For quite a while, there has existed a hypervisor bug on legacy iSeries
which means that we do not get the boot time set in the kernel. This
patch works around that bug. This was most noticable when the root
partition needed to be checked at every boot as the kernel thought it
was some time in 1905 until user mode reset the time correctly.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
On PPC64, we keep track of when we need to update jiffies (and the
variables used to calculate the time of day) based on the time base.
If the time base frequence is sufficiently high compared to the
processor clock frequency, then it is possible for the time of day
variables to be corrupted at the time of the first decrementer interrupt
we take. This became obvious on a legacy iSeries where the time base
frequency is the same as the processor clock.
This one line patch fixes the initialisation so that the time of day
variables and the indicator we use to tell when updates are due are
better synchronised.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The hardware sync of the timebase on SMP G5s uses a black magic
incantation to the i2c clock chip that was inspired from what Darwin
does.
However, this was an earlier version of Darwin that was ... buggy !
heh. This causes the latest models to break though when starting SMP,
so it's worth fixing.
Here's a new version of the incantation based on careful transcription
of the said incantations as found in the latest version of apple's
temple.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The latest speedbumped Apple G5 models have a "bug" in the Open Firmware
device tree that lacks the proper interrupt routing information for the
northbridge i2c controller. Apple's driver silently falls back into a
sub-optimal "polled" mode (heh, maybe they didn't even notice the bug
because of that :), our driver didn't properly check and crashes :(
This patch fixes our driver to not crash, and adds code to the
prom_init() OF trampoline code that detects the "bug" and adds the
missing information back for this chipset revision. This fixes booting
and thermal control on these models.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch enables CONFIG_RTAS_PROC by default on pSeries. This will
preserve /proc/ppc64/rtas/rmo_buffer, which is needed by librtas.
Signed-off-by: John Rose <johnrose@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Move the code to set global interrupt queue membership to xics.c,
and remove no longer needed extern declarations. Also call it on
all cpus (even the boot cpu) to prepare for kexec.
Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: R Sharada <sharada@in.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Trivial patch to remove our last direct reference to contig_page_data.
This will make it just that much less hard to seperate NUMA and
DISCONTIG. Please forward on. Against 2.6.12-rc1
Signed-off-by: Joel Schopp <jschopp@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
start.c is not referenced in the arch/ppc64/boot/Makefile
compile tested with the defconfig.
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The defines in bootinfo.h are not used, so the include can be removed.
According to Ben, birecs are not used on ppc64:
on ppc64, we made the decision of enforcing the presence of an
OF device-tree and either an OF-like client interface or a kexec
like flattened tree.
so if your bootloader want to say things to the kernel,
it can do so by adding properties to the device-tree
compile-tested with defconfig
Signed-off-by: Olaf Hering <olh@suse.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The code in reloc_offset is actually subtracting the address in the link
register from the address calculated by the linker. Perhaps the
extended mnemonic `sub' replaced an original `subf' and the comment just
did not get updated.
bl 1f
1: mflr r3
LOADADDR(r4,1b)
sub r3,r4,r3
Signed-off-by: Amos Waterland <apw@us.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The code in unflatten_device_tree knows that get_property is written to
only return with lenp equal to 1 when also returning a valid pointer.
The gcc 3.3.3 compiler is not able to prove this to itself, so it warns
about a possible uninitialized pointer dereference:
.../arch/ppc64/kernel/prom.c: In function `unflatten_device_tree':
.../arch/ppc64/kernel/prom.c:828:
warning: `p' might be used uninitialized in this function
Unless it is desired to rework the interaction between the two
functions, this will keep the existing behavior but quiet the compiler.
Signed-off-by: Amos Waterland <apw@us.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Replace a custom MIN() macro with the min() macro from kernel.h
This patch removes 4 lines of redundant code.
Signed-off-by: Tobias Klauser <tklauser@nuerscht.ch>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch started as simply removing a few never-used macros from
asm-ppc64/pgtable.h, then kind of grew. It now makes a bunch of
cleanups to the ppc64 low-level header files (with corresponding
changes to .c files where necessary) such as:
- Abolishing never-used macros
- Eliminating multiple #defines with the same purpose
- Removing pointless macros (cases where just expanding the
macro everywhere turns out clearer and more sensible)
- Removing some cases where macros which could be defined in
terms of each other weren't
- Moving imalloc() related definitions from pgtable.h to their
own header file (imalloc.h)
- Re-arranging headers to group things more logically
- Moving all VSID allocation related things to mmu.h, instead
of being split between mmu.h and mmu_context.h
- Removing some reserved space for flags from the PMD - we're
not using it.
- Fix some bugs which broke compile with STRICT_MM_TYPECHECKS.
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
There's no help text for CONFIG_DEBUG_STACKOVERFLOW - add one.
Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
While looking at code generated by gcc4.0 I noticed some functions still
had frame pointers, even after we stopped ppc64 from defining
CONFIG_FRAME_POINTER. It turns out kernel/Makefile hardwires
-fno-omit-frame-pointer on when compiling schedule.c.
Create CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER and define it on architectures
that dont require frame pointers in sched.c code.
(akpm: blame me for the name)
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
A bunch of drivers use ISA DMA helpers or their equivalents for
platforms that have ISA with different DMA controller (a lot of ARM
boxen). Currently there is no way to put such dependency in Kconfig -
CONFIG_ISA is not it (e.g. it is not set on platforms that have no ISA
slots, but have on-board devices that pretend to be ISA ones).
New symbol added - ISA_DMA_API. Set when we have functional
enable_dma()/set_dma_mode()/etc. set of helpers. Next patches in the
series will add missing dependencies for drivers that need them.
I'm very carefully staying the hell out of the recurring flamefest on
what exactly CONFIG_ISA would mean in ideal world - added symbol has a
well-defined meaning and for now I really want to treat it as completely
independent from the mess around CONFIG_ISA.
Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
gcc-4.0 generates altivec code implicitly when -mcpu indicates an
altivec capable CPU which is not suitable for the kernel. However, we
used to set -mcpu=970 when CONFIG_ALTIVEC was set because a gcc-3.x bug
prevented from using -maltivec along with -mcpu=power4, thus prevented
building the RAID6 altivec code.
This patch fixes all of this by testing for the gcc version. If 4.0 or
later, just normally use -mcpu=power4 and let the RAID6 code add
-maltivec to the few files it needs to be compiled with altivec support.
For 3.x, we still use -mcpu=970 to work around the above problem, which
is fine as 3.x will never implicitly generate altivec code.
The Makefile hackery may not be the most lovely, I welcome anybody more
skilled than me to improve it.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Convert most of the current code that uses _NSIG directly to instead use
valid_signal(). This avoids gcc -W warnings and off-by-one errors.
Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch changes calls to synchronize_kernel(), deprecated in the earlier
"Deprecate synchronize_kernel, GPL replacement" patch to instead call the new
synchronize_rcu() and synchronize_sched() APIs.
Signed-off-by: Paul E. McKenney <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Use smp_mb and smp_wmb. In particular smp_wmb is lighter weight than wmb.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Calls into the hypervisor do not raise the thread priority. Ensure we are
running at medium priority upon entry to the hypervisor.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Recent gcc 4.0 testing uncovered a firmware issue. Some properties are larger
than 31 bytes and due to gcc 4.0s better stack allocation this overflow ran
over non volatile register storage.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
It turns out that our current __hash_page code will do a very hot busy-wait
loop waiting on _PAGE_BUSY to be cleared. It even does ldarx/stdcx in the
loop, which will bounce reservations around like crazy if there's more than
one CPU spinning on the same PTE (or even another PTE in the same
reservation granule). The end result is that each fault takes longer when
there's contention, which in turn increases the chance of another thread
hitting the same fault and also piling up. Not pretty.
There's two options here:
1. Do an out-of-line busy loop a'la spinlocks with just loads (no
reserves)
2. Just bail and refault if needed.
(2) makes sense here: If the PTE is busy, chances are it's in flux anyway
and the other code path making a change might just be ready to hash it.
This fixes a stampede seen on a large-ish system where a multithreaded
HPC app faults in the same text pages on several cpus at the same time.
Signed-off-by: Olof Johansson <olof@lixom.net>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
On pSeries systems, according to the platform architecture specs, we are
supposed to be supplying a structure to firmware that tells firmware about
our capabilities, such as which version of the data structures that
describe available memory we are expecting to see. The way we end up
having to supply this data structure is a bit gross, since it was designed
for AIX and doesn't suit us very well. This patch adds the code to supply
this data structure to the firmware.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch converts ppc64 to use the generic pgtable-nopud.h instead of the
"fixup" header.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
- Fix
arch/ppc64/kernel/nvram.c:342: warning: `part' might be used uninitialized in this function
- Various codingstyle tweaks.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
When I tried Ben's patches to the powermac sound driver on my G5, I found
that it was taking enormous numbers of sound DMA transmit interrupts. This
turned out to be because it was incorrectly configured as level-sensitive
instead of edge-sensitive, which in turn was because the code that parses
the interrupt tree that Open Firmware gives us was incorrectly assigning
another device the same irq number as the sound DMA transmit interrupt
(i.e. 1).
This patch fixes the problem, in a somewhat quick and dirty way for now,
but one which will work for all the machines we currently run on.
Ultimately Ben and I want to do something more general and robust, but this
should go in for 2.6.12.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Remove vsid argument to create_slbe, since it's no longer used.
Spotted by R Sharada.
Signed-off-by: Olof Johansson <olof@austin.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch from Roland adds a PT_NOTE section to both 32 and 64 bits vDSOs
to expose the kernel version to glibc, thus avoiding a uname syscall on
every launch. This is equivalent to the patches Roland posted already for
x86 and x86-64.
Note: the 64 bits .note is actually using the 32 bits format. This is
normal. The ELF spec specifies a different format for 64 bits .note, but
for some reason, this was never properly implemented, the core dumps for
example are all using 32 bits format .note, and binutils cannot even read a
64 bits format .note. Talking to our toolchain folks, they think we'd
rather stick to 32 bits format .note everywhere and get the spec fixed some
day ...
Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
When the kernel creates a signal frame on the user stack, it puts the
old stack pointer value at the beginning so that the signal frame is
linked into the chain of stack frames like any other frame.
Unfortunately, for 32-bit processes we are writing the old stack
pointer as a 64-bit value rather than a 32-bit value, and the process
sees that as a null pointer, since it only looks at the first 32 bits,
which are zero since ppc is bigendian and the stack pointer is below
4GB. This bug is in SLES9 and RHEL4 too, hence the ccs.
This patch fixes the bug by making the signal code write the old stack
pointer as a u32 instead of an unsigned long.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
We were calling ptrace_notify() after auditing the syscall and arguments,
but the debugger could have _changed_ them before the syscall was actually
invoked. Reorder the calls to fix that.
While we're touching ever call to audit_syscall_entry(), we also make it
take an extra argument: the architecture of the syscall which was made,
because some architectures allow more than one type of syscall.
Also add an explicit success/failure flag to audit_syscall_exit(), for
the benefit of architectures which return that in a condition register
rather than only returning a single register.
Change type of syscall return value to 'long' not 'int'.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
The ppc vDSO would not properly clear the return value for some calls,
which will be a problem when interfacing those calls with glibc. This
should be fixed before 2.6.12 is released (as it is the first kernel
with the ppc vDSO) so that we don't have to play with symbol versioning
and ugly workarounds.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Once we're strict about clearing away page tables, hugetlb_prefault can assume
there are no page tables left within its range. Since the other arches
continue if !pte_none here, let i386 do the same.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
ia64 and ppc64 had hugetlb_free_pgtables functions which were no longer being
called, and it wasn't obvious what to do about them.
The ppc64 case turns out to be easy: the associated tables are noted elsewhere
and freed later, safe to either skip its hugetlb areas or go through the
motions of freeing nothing. Since ia64 does need a special case, restore to
ppc64 the special case of skipping them.
The ia64 hugetlb case has been broken since pgd_addr_end went in, though it
probably appeared to work okay if you just had one such area; in fact it's
been broken much longer if you consider a long munmap spanning from another
region into the hugetlb region.
In the ia64 hugetlb region, more virtual address bits are available than in
the other regions, yet the page tables are structured the same way: the page
at the bottom is larger. Here we need to scale down each addr before passing
it to the standard free_pgd_range. Was about to write a hugely_scaled_down
macro, but found htlbpage_to_page already exists for just this purpose. Fixed
off-by-one in ia64 is_hugepage_only_range.
Uninline free_pgd_range to make it available to ia64. Make sure the
vma-gathering loop in free_pgtables cannot join a hugepage_only_range to any
other (safe to join huges? probably but don't bother).
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This fixes pm_message_t vs. u32 confusion in ppc and aty (I *hope* that's
basically radeon code...). I was not able to test most of these, but I'm
not really changing anything, so it should be okay.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
During some code inspection using gcc 4.0 I noticed a stack frame was being
created for a number of functions that didnt require it. For example:
c0000000000df944 <._spin_unlock>:
c0000000000df944: fb e1 ff f0 std r31,-16(r1)
c0000000000df948: f8 21 ff c1 stdu r1,-64(r1)
c0000000000df94c: 7c 3f 0b 78 mr r31,r1
c0000000000df950: 7c 20 04 ac lwsync
c0000000000df954: e8 21 00 00 ld r1,0(r1)
c0000000000df958: 38 00 00 00 li r0,0
c0000000000df95c: 90 03 00 00 stw r0,0(r3)
c0000000000df960: eb e1 ff f0 ld r31,-16(r1)
c0000000000df964: 4e 80 00 20 blr
It turns out we are adding -fno-omit-frame-pointer to ppc64 which is
causing the above behaviour. Removing that flag results in much better
code:
c0000000000d5b30 <._spin_unlock>:
c0000000000d5b30: 7c 20 04 ac lwsync
c0000000000d5b34: 38 00 00 00 li r0,0
c0000000000d5b38: 90 03 00 00 stw r0,0(r3)
c0000000000d5b3c: 4e 80 00 20 blr
We dont require a frame pointer to debug on ppc64, so remove it.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
The code that parses the OF device tree contains an old bogus hack which
was killed a long time ago on ppc32, but survived in ppc64. It was
supposed to help with a problem on the f50 which is ... a 32 bits machine
:) Additionally, that hack is causing problems, so let's just get rid of
it.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch adds detection of the Altivec capability of the CPU via the
firmware in addition to the cpu table. This allows newer CPUs that aren't
in the table to still have working altivec support in the kernel.
It also fixes a problem where if a CPU isn't recognized as having altivec
features, and takes an altivec unavailable exception due to userland
issuing altivec instructions, the kernel would happily enable it and
context switch the registers ... but not all of them (it would basically
forget vrsave). With this patch, the kernel will refuse to enable altivec
when the feature isn't detected for the CPU (SIGILL).
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch reworks the way the ppc64 is mapped in user memory by the kernel
to make it more robust against possible collisions with executable
segments. Instead of just whacking a VMA at 1Mb, I now use
get_unmapped_area() with a hint, and I moved the mapping of the vDSO to
after the mapping of the various ELF segments and of the interpreter, so
that conflicts get caught properly (it still has to be before
create_elf_tables since the later will fill the AT_SYSINFO_EHDR with the
proper address).
While I was at it, I also changed the 32 and 64 bits vDSO's to link at
their "natural" address of 1Mb instead of 0. This is the address where
they are normally mapped in absence of conflict. By doing so, it should be
possible to properly prelink one it's been verified to work on glibc.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
In arch/ppc64/kernel/ppc_ksyms.c, we are still exporting
flush_icache_range, but that has been changed to be an inline in
include/asm-ppc64/cacheflush.h which calls __flush_icache_range (defined in
arch/ppc64/kernel/misc.S).
This patch changes the export to __flush_icache_range, thus allowing
modules to use the inline flush_icache_range.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch fixes ppc64 __ioremap() so that it stops adding implicitely
_PAGE_GUARDED when the cache is not writeback, and instead, let the callers
provide the flag they want here. This allows things like framebuffers to
explicitely request a non-cacheable and non-guarded mapping which is more
efficient for that type of memory without side effects. The patch also
fixes all current callers to add _PAGE_GUARDED except btext, which is fine
without it.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This patch hacks the current PowerMac Alsa driver to add some basic support
of analog sound output to some desktop G5s. It has severe limitations
though:
- Only 44100Khz 16 bits
- Only work on G5 models using a TAS3004 analog code, that is early
single CPU desktops and all dual CPU desktops at this date, but none
of the more recent ones like iMac G5.
- It does analog only, no digital/SPDIF support at all, no native
AC3 support
Better support would require a complete rewrite of the driver (which I am
working on, but don't hold your breath), to properly support the diversity
of apple sound HW setup, including dual codecs, several i2s busses, all the
new codecs used in the new machines, proper clock switching with digital,
etc etc etc...
This patch applies on top of the other PowerMac sound patches I posted in
the past couple of days (new powerbook support and sleep fixes).
Note: This is a FAQ entry for PowerMac sound support with TI codecs: They
have a feature called "DRC" which is automatically enabled for the internal
speaker (at least when auto mute control is enabled) which will cause your
sound to fade out to nothing after half a second of playback if you don't
set a proper "DRC Range" in the mixer. So if you have a problem like that,
check alsamixer and raise your DRC Range to something reasonable.
Note2: This patch will also add auto-mute of the speaker when line-out jack
is used on some earlier desktop G4s (and on the G5) in addition to the
headphone jack. If that behaviour isn't what you want, just disable
auto-muting and use the manual mute controls in alsamixer.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
My previous patch that added sleep support for uninorth-agp and some AGP
"off" stuff in radeonfb and aty128fb is breaking some configs. More
specifically, it has problems with rage128 setups since the DRI code for
these in X doesn't properly re-enable AGP on wakeup or console switch
(unlike the radeon DRM).
This patch fixes the problem for pmac once for all by using a different
approach. The AGP driver "registers" special suspend/resume callbacks with
some arch code that the fbdev's can later on call to suspend and resume
AGP, making sure it's resumed back in the same state it was when suspended.
This is platform specific for now. It would be too complicated to try to
do a generic implementation of this at this point due to all sort of weird
things going on with AGP on other architectures. We'll re-work that whole
problem cleanly once we finally merge fbdev's and DRI.
In the meantime, please apply this patch which brings back some r128 based
laptops into working condition as far as system sleep is concerned.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
