linux/arch/x86/xen/mmu.c

2365 lines
58 KiB
C
Raw Normal View History

xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
/*
* Xen mmu operations
*
* This file contains the various mmu fetch and update operations.
* The most important job they must perform is the mapping between the
* domain's pfn and the overall machine mfns.
*
* Xen allows guests to directly update the pagetable, in a controlled
* fashion. In other words, the guest modifies the same pagetable
* that the CPU actually uses, which eliminates the overhead of having
* a separate shadow pagetable.
*
* In order to allow this, it falls on the guest domain to map its
* notion of a "physical" pfn - which is just a domain-local linear
* address - into a real "machine address" which the CPU's MMU can
* use.
*
* A pgd_t/pmd_t/pte_t will typically contain an mfn, and so can be
* inserted directly into the pagetable. When creating a new
* pte/pmd/pgd, it converts the passed pfn into an mfn. Conversely,
* when reading the content back with __(pgd|pmd|pte)_val, it converts
* the mfn back into a pfn.
*
* The other constraint is that all pages which make up a pagetable
* must be mapped read-only in the guest. This prevents uncontrolled
* guest updates to the pagetable. Xen strictly enforces this, and
* will disallow any pagetable update which will end up mapping a
* pagetable page RW, and will disallow using any writable page as a
* pagetable.
*
* Naively, when loading %cr3 with the base of a new pagetable, Xen
* would need to validate the whole pagetable before going on.
* Naturally, this is quite slow. The solution is to "pin" a
* pagetable, which enforces all the constraints on the pagetable even
* when it is not actively in use. This menas that Xen can be assured
* that it is still valid when you do load it into %cr3, and doesn't
* need to revalidate it.
*
* Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
*/
#include <linux/sched.h>
#include <linux/highmem.h>
#include <linux/debugfs.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include <linux/bug.h>
#include <linux/vmalloc.h>
#include <linux/module.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 16:04:11 +08:00
#include <linux/gfp.h>
#include <linux/memblock.h>
#include <linux/seq_file.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include <trace/events/xen.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <asm/fixmap.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include <asm/mmu_context.h>
#include <asm/setup.h>
#include <asm/paravirt.h>
#include <asm/e820.h>
#include <asm/linkage.h>
#include <asm/page.h>
#include <asm/init.h>
#include <asm/pat.h>
#include <asm/smp.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include <asm/xen/hypercall.h>
#include <asm/xen/hypervisor.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include <xen/xen.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include <xen/page.h>
#include <xen/interface/xen.h>
#include <xen/interface/hvm/hvm_op.h>
#include <xen/interface/version.h>
#include <xen/interface/memory.h>
#include <xen/hvc-console.h>
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include "multicalls.h"
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#include "mmu.h"
#include "debugfs.h"
/*
* Protects atomic reservation decrease/increase against concurrent increases.
* Also protects non-atomic updates of current_pages and balloon lists.
*/
DEFINE_SPINLOCK(xen_reservation_lock);
/*
* Identity map, in addition to plain kernel map. This needs to be
* large enough to allocate page table pages to allocate the rest.
* Each page can map 2MB.
*/
#define LEVEL1_IDENT_ENTRIES (PTRS_PER_PTE * 4)
static RESERVE_BRK_ARRAY(pte_t, level1_ident_pgt, LEVEL1_IDENT_ENTRIES);
#ifdef CONFIG_X86_64
/* l3 pud for userspace vsyscall mapping */
static pud_t level3_user_vsyscall[PTRS_PER_PUD] __page_aligned_bss;
#endif /* CONFIG_X86_64 */
/*
* Note about cr3 (pagetable base) values:
*
* xen_cr3 contains the current logical cr3 value; it contains the
* last set cr3. This may not be the current effective cr3, because
* its update may be being lazily deferred. However, a vcpu looking
* at its own cr3 can use this value knowing that it everything will
* be self-consistent.
*
* xen_current_cr3 contains the actual vcpu cr3; it is set once the
* hypercall to set the vcpu cr3 is complete (so it may be a little
* out of date, but it will never be set early). If one vcpu is
* looking at another vcpu's cr3 value, it should use this variable.
*/
DEFINE_PER_CPU(unsigned long, xen_cr3); /* cr3 stored as physaddr */
DEFINE_PER_CPU(unsigned long, xen_current_cr3); /* actual vcpu cr3 */
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
/*
* Just beyond the highest usermode address. STACK_TOP_MAX has a
* redzone above it, so round it up to a PGD boundary.
*/
#define USER_LIMIT ((STACK_TOP_MAX + PGDIR_SIZE - 1) & PGDIR_MASK)
unsigned long arbitrary_virt_to_mfn(void *vaddr)
{
xmaddr_t maddr = arbitrary_virt_to_machine(vaddr);
return PFN_DOWN(maddr.maddr);
}
xmaddr_t arbitrary_virt_to_machine(void *vaddr)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
unsigned long address = (unsigned long)vaddr;
unsigned int level;
xen: fix Xen domU boot with batched mprotect Impact: fix guest kernel boot crash on certain configs Recent i686 2.6.27 kernels with a certain amount of memory (between 736 and 855MB) have a problem booting under a hypervisor that supports batched mprotect (this includes the RHEL-5 Xen hypervisor as well as any 3.3 or later Xen hypervisor). The problem ends up being that xen_ptep_modify_prot_commit() is using virt_to_machine to calculate which pfn to update. However, this only works for pages that are in the p2m list, and the pages coming from change_pte_range() in mm/mprotect.c are kmap_atomic pages. Because of this, we can run into the situation where the lookup in the p2m table returns an INVALID_MFN, which we then try to pass to the hypervisor, which then (correctly) denies the request to a totally bogus pfn. The right thing to do is to use arbitrary_virt_to_machine, so that we can be sure we are modifying the right pfn. This unfortunately introduces a performance penalty because of a full page-table-walk, but we can avoid that penalty for pages in the p2m list by checking if virt_addr_valid is true, and if so, just doing the lookup in the p2m table. The attached patch implements this, and allows my 2.6.27 i686 based guest with 768MB of memory to boot on a RHEL-5 hypervisor again. Thanks to Jeremy for the suggestions about how to fix this particular issue. Signed-off-by: Chris Lalancette <clalance@redhat.com> Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Chris Lalancette <clalance@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-10-24 08:40:25 +08:00
pte_t *pte;
unsigned offset;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
xen: fix Xen domU boot with batched mprotect Impact: fix guest kernel boot crash on certain configs Recent i686 2.6.27 kernels with a certain amount of memory (between 736 and 855MB) have a problem booting under a hypervisor that supports batched mprotect (this includes the RHEL-5 Xen hypervisor as well as any 3.3 or later Xen hypervisor). The problem ends up being that xen_ptep_modify_prot_commit() is using virt_to_machine to calculate which pfn to update. However, this only works for pages that are in the p2m list, and the pages coming from change_pte_range() in mm/mprotect.c are kmap_atomic pages. Because of this, we can run into the situation where the lookup in the p2m table returns an INVALID_MFN, which we then try to pass to the hypervisor, which then (correctly) denies the request to a totally bogus pfn. The right thing to do is to use arbitrary_virt_to_machine, so that we can be sure we are modifying the right pfn. This unfortunately introduces a performance penalty because of a full page-table-walk, but we can avoid that penalty for pages in the p2m list by checking if virt_addr_valid is true, and if so, just doing the lookup in the p2m table. The attached patch implements this, and allows my 2.6.27 i686 based guest with 768MB of memory to boot on a RHEL-5 hypervisor again. Thanks to Jeremy for the suggestions about how to fix this particular issue. Signed-off-by: Chris Lalancette <clalance@redhat.com> Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Chris Lalancette <clalance@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-10-24 08:40:25 +08:00
/*
* if the PFN is in the linear mapped vaddr range, we can just use
* the (quick) virt_to_machine() p2m lookup
*/
if (virt_addr_valid(vaddr))
return virt_to_machine(vaddr);
/* otherwise we have to do a (slower) full page-table walk */
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
xen: fix Xen domU boot with batched mprotect Impact: fix guest kernel boot crash on certain configs Recent i686 2.6.27 kernels with a certain amount of memory (between 736 and 855MB) have a problem booting under a hypervisor that supports batched mprotect (this includes the RHEL-5 Xen hypervisor as well as any 3.3 or later Xen hypervisor). The problem ends up being that xen_ptep_modify_prot_commit() is using virt_to_machine to calculate which pfn to update. However, this only works for pages that are in the p2m list, and the pages coming from change_pte_range() in mm/mprotect.c are kmap_atomic pages. Because of this, we can run into the situation where the lookup in the p2m table returns an INVALID_MFN, which we then try to pass to the hypervisor, which then (correctly) denies the request to a totally bogus pfn. The right thing to do is to use arbitrary_virt_to_machine, so that we can be sure we are modifying the right pfn. This unfortunately introduces a performance penalty because of a full page-table-walk, but we can avoid that penalty for pages in the p2m list by checking if virt_addr_valid is true, and if so, just doing the lookup in the p2m table. The attached patch implements this, and allows my 2.6.27 i686 based guest with 768MB of memory to boot on a RHEL-5 hypervisor again. Thanks to Jeremy for the suggestions about how to fix this particular issue. Signed-off-by: Chris Lalancette <clalance@redhat.com> Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Chris Lalancette <clalance@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-10-24 08:40:25 +08:00
pte = lookup_address(address, &level);
BUG_ON(pte == NULL);
offset = address & ~PAGE_MASK;
return XMADDR(((phys_addr_t)pte_mfn(*pte) << PAGE_SHIFT) + offset);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
EXPORT_SYMBOL_GPL(arbitrary_virt_to_machine);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
void make_lowmem_page_readonly(void *vaddr)
{
pte_t *pte, ptev;
unsigned long address = (unsigned long)vaddr;
unsigned int level;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
pte = lookup_address(address, &level);
if (pte == NULL)
return; /* vaddr missing */
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
ptev = pte_wrprotect(*pte);
if (HYPERVISOR_update_va_mapping(address, ptev, 0))
BUG();
}
void make_lowmem_page_readwrite(void *vaddr)
{
pte_t *pte, ptev;
unsigned long address = (unsigned long)vaddr;
unsigned int level;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
pte = lookup_address(address, &level);
if (pte == NULL)
return; /* vaddr missing */
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
ptev = pte_mkwrite(*pte);
if (HYPERVISOR_update_va_mapping(address, ptev, 0))
BUG();
}
static bool xen_page_pinned(void *ptr)
{
struct page *page = virt_to_page(ptr);
return PagePinned(page);
}
void xen_set_domain_pte(pte_t *ptep, pte_t pteval, unsigned domid)
{
struct multicall_space mcs;
struct mmu_update *u;
trace_xen_mmu_set_domain_pte(ptep, pteval, domid);
mcs = xen_mc_entry(sizeof(*u));
u = mcs.args;
/* ptep might be kmapped when using 32-bit HIGHPTE */
u->ptr = virt_to_machine(ptep).maddr;
u->val = pte_val_ma(pteval);
MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, domid);
xen_mc_issue(PARAVIRT_LAZY_MMU);
}
EXPORT_SYMBOL_GPL(xen_set_domain_pte);
static void xen_extend_mmu_update(const struct mmu_update *update)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
struct multicall_space mcs;
struct mmu_update *u;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
mcs = xen_mc_extend_args(__HYPERVISOR_mmu_update, sizeof(*u));
if (mcs.mc != NULL) {
mcs.mc->args[1]++;
} else {
mcs = __xen_mc_entry(sizeof(*u));
MULTI_mmu_update(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
}
u = mcs.args;
*u = *update;
}
static void xen_extend_mmuext_op(const struct mmuext_op *op)
{
struct multicall_space mcs;
struct mmuext_op *u;
mcs = xen_mc_extend_args(__HYPERVISOR_mmuext_op, sizeof(*u));
if (mcs.mc != NULL) {
mcs.mc->args[1]++;
} else {
mcs = __xen_mc_entry(sizeof(*u));
MULTI_mmuext_op(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
}
u = mcs.args;
*u = *op;
}
static void xen_set_pmd_hyper(pmd_t *ptr, pmd_t val)
{
struct mmu_update u;
preempt_disable();
xen_mc_batch();
/* ptr may be ioremapped for 64-bit pagetable setup */
u.ptr = arbitrary_virt_to_machine(ptr).maddr;
u.val = pmd_val_ma(val);
xen_extend_mmu_update(&u);
xen_mc_issue(PARAVIRT_LAZY_MMU);
preempt_enable();
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static void xen_set_pmd(pmd_t *ptr, pmd_t val)
{
trace_xen_mmu_set_pmd(ptr, val);
/* If page is not pinned, we can just update the entry
directly */
if (!xen_page_pinned(ptr)) {
*ptr = val;
return;
}
xen_set_pmd_hyper(ptr, val);
}
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
/*
* Associate a virtual page frame with a given physical page frame
* and protection flags for that frame.
*/
void set_pte_mfn(unsigned long vaddr, unsigned long mfn, pgprot_t flags)
{
set_pte_vaddr(vaddr, mfn_pte(mfn, flags));
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static bool xen_batched_set_pte(pte_t *ptep, pte_t pteval)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
struct mmu_update u;
if (paravirt_get_lazy_mode() != PARAVIRT_LAZY_MMU)
return false;
xen_mc_batch();
u.ptr = virt_to_machine(ptep).maddr | MMU_NORMAL_PT_UPDATE;
u.val = pte_val_ma(pteval);
xen_extend_mmu_update(&u);
xen_mc_issue(PARAVIRT_LAZY_MMU);
return true;
}
static inline void __xen_set_pte(pte_t *ptep, pte_t pteval)
{
if (!xen_batched_set_pte(ptep, pteval))
native_set_pte(ptep, pteval);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static void xen_set_pte(pte_t *ptep, pte_t pteval)
{
trace_xen_mmu_set_pte(ptep, pteval);
__xen_set_pte(ptep, pteval);
}
static void xen_set_pte_at(struct mm_struct *mm, unsigned long addr,
pte_t *ptep, pte_t pteval)
{
trace_xen_mmu_set_pte_at(mm, addr, ptep, pteval);
__xen_set_pte(ptep, pteval);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
pte_t xen_ptep_modify_prot_start(struct mm_struct *mm,
unsigned long addr, pte_t *ptep)
{
/* Just return the pte as-is. We preserve the bits on commit */
trace_xen_mmu_ptep_modify_prot_start(mm, addr, ptep, *ptep);
return *ptep;
}
void xen_ptep_modify_prot_commit(struct mm_struct *mm, unsigned long addr,
pte_t *ptep, pte_t pte)
{
struct mmu_update u;
trace_xen_mmu_ptep_modify_prot_commit(mm, addr, ptep, pte);
xen_mc_batch();
u.ptr = virt_to_machine(ptep).maddr | MMU_PT_UPDATE_PRESERVE_AD;
u.val = pte_val_ma(pte);
xen_extend_mmu_update(&u);
xen_mc_issue(PARAVIRT_LAZY_MMU);
}
/* Assume pteval_t is equivalent to all the other *val_t types. */
static pteval_t pte_mfn_to_pfn(pteval_t val)
{
if (val & _PAGE_PRESENT) {
unsigned long mfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
pteval_t flags = val & PTE_FLAGS_MASK;
val = ((pteval_t)mfn_to_pfn(mfn) << PAGE_SHIFT) | flags;
}
return val;
}
static pteval_t pte_pfn_to_mfn(pteval_t val)
{
if (val & _PAGE_PRESENT) {
unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
pteval_t flags = val & PTE_FLAGS_MASK;
unsigned long mfn;
if (!xen_feature(XENFEAT_auto_translated_physmap))
mfn = get_phys_to_machine(pfn);
else
mfn = pfn;
/*
* If there's no mfn for the pfn, then just create an
* empty non-present pte. Unfortunately this loses
* information about the original pfn, so
* pte_mfn_to_pfn is asymmetric.
*/
if (unlikely(mfn == INVALID_P2M_ENTRY)) {
mfn = 0;
flags = 0;
} else {
/*
* Paramount to do this test _after_ the
* INVALID_P2M_ENTRY as INVALID_P2M_ENTRY &
* IDENTITY_FRAME_BIT resolves to true.
*/
mfn &= ~FOREIGN_FRAME_BIT;
if (mfn & IDENTITY_FRAME_BIT) {
mfn &= ~IDENTITY_FRAME_BIT;
flags |= _PAGE_IOMAP;
}
}
val = ((pteval_t)mfn << PAGE_SHIFT) | flags;
}
return val;
}
static pteval_t iomap_pte(pteval_t val)
{
if (val & _PAGE_PRESENT) {
unsigned long pfn = (val & PTE_PFN_MASK) >> PAGE_SHIFT;
pteval_t flags = val & PTE_FLAGS_MASK;
/* We assume the pte frame number is a MFN, so
just use it as-is. */
val = ((pteval_t)pfn << PAGE_SHIFT) | flags;
}
return val;
}
static pteval_t xen_pte_val(pte_t pte)
{
pteval_t pteval = pte.pte;
/* If this is a WC pte, convert back from Xen WC to Linux WC */
if ((pteval & (_PAGE_PAT | _PAGE_PCD | _PAGE_PWT)) == _PAGE_PAT) {
WARN_ON(!pat_enabled);
pteval = (pteval & ~_PAGE_PAT) | _PAGE_PWT;
}
if (xen_initial_domain() && (pteval & _PAGE_IOMAP))
return pteval;
return pte_mfn_to_pfn(pteval);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pte_val);
static pgdval_t xen_pgd_val(pgd_t pgd)
{
return pte_mfn_to_pfn(pgd.pgd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pgd_val);
/*
* Xen's PAT setup is part of its ABI, though I assume entries 6 & 7
* are reserved for now, to correspond to the Intel-reserved PAT
* types.
*
* We expect Linux's PAT set as follows:
*
* Idx PTE flags Linux Xen Default
* 0 WB WB WB
* 1 PWT WC WT WT
* 2 PCD UC- UC- UC-
* 3 PCD PWT UC UC UC
* 4 PAT WB WC WB
* 5 PAT PWT WC WP WT
* 6 PAT PCD UC- UC UC-
* 7 PAT PCD PWT UC UC UC
*/
void xen_set_pat(u64 pat)
{
/* We expect Linux to use a PAT setting of
* UC UC- WC WB (ignoring the PAT flag) */
WARN_ON(pat != 0x0007010600070106ull);
}
static pte_t xen_make_pte(pteval_t pte)
{
phys_addr_t addr = (pte & PTE_PFN_MASK);
/* If Linux is trying to set a WC pte, then map to the Xen WC.
* If _PAGE_PAT is set, then it probably means it is really
* _PAGE_PSE, so avoid fiddling with the PAT mapping and hope
* things work out OK...
*
* (We should never see kernel mappings with _PAGE_PSE set,
* but we could see hugetlbfs mappings, I think.).
*/
if (pat_enabled && !WARN_ON(pte & _PAGE_PAT)) {
if ((pte & (_PAGE_PCD | _PAGE_PWT)) == _PAGE_PWT)
pte = (pte & ~(_PAGE_PCD | _PAGE_PWT)) | _PAGE_PAT;
}
/*
* Unprivileged domains are allowed to do IOMAPpings for
* PCI passthrough, but not map ISA space. The ISA
* mappings are just dummy local mappings to keep other
* parts of the kernel happy.
*/
if (unlikely(pte & _PAGE_IOMAP) &&
(xen_initial_domain() || addr >= ISA_END_ADDRESS)) {
pte = iomap_pte(pte);
} else {
pte &= ~_PAGE_IOMAP;
pte = pte_pfn_to_mfn(pte);
}
return native_make_pte(pte);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pte);
static pgd_t xen_make_pgd(pgdval_t pgd)
{
pgd = pte_pfn_to_mfn(pgd);
return native_make_pgd(pgd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pgd);
static pmdval_t xen_pmd_val(pmd_t pmd)
{
return pte_mfn_to_pfn(pmd.pmd);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pmd_val);
static void xen_set_pud_hyper(pud_t *ptr, pud_t val)
{
struct mmu_update u;
preempt_disable();
xen_mc_batch();
/* ptr may be ioremapped for 64-bit pagetable setup */
u.ptr = arbitrary_virt_to_machine(ptr).maddr;
u.val = pud_val_ma(val);
xen_extend_mmu_update(&u);
xen_mc_issue(PARAVIRT_LAZY_MMU);
preempt_enable();
}
static void xen_set_pud(pud_t *ptr, pud_t val)
{
trace_xen_mmu_set_pud(ptr, val);
/* If page is not pinned, we can just update the entry
directly */
if (!xen_page_pinned(ptr)) {
*ptr = val;
return;
}
xen_set_pud_hyper(ptr, val);
}
#ifdef CONFIG_X86_PAE
static void xen_set_pte_atomic(pte_t *ptep, pte_t pte)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
trace_xen_mmu_set_pte_atomic(ptep, pte);
set_64bit((u64 *)ptep, native_pte_val(pte));
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static void xen_pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
trace_xen_mmu_pte_clear(mm, addr, ptep);
if (!xen_batched_set_pte(ptep, native_make_pte(0)))
native_pte_clear(mm, addr, ptep);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static void xen_pmd_clear(pmd_t *pmdp)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
trace_xen_mmu_pmd_clear(pmdp);
set_pmd(pmdp, __pmd(0));
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
#endif /* CONFIG_X86_PAE */
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
static pmd_t xen_make_pmd(pmdval_t pmd)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
pmd = pte_pfn_to_mfn(pmd);
return native_make_pmd(pmd);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pmd);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
#if PAGETABLE_LEVELS == 4
static pudval_t xen_pud_val(pud_t pud)
{
return pte_mfn_to_pfn(pud.pud);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_pud_val);
static pud_t xen_make_pud(pudval_t pud)
{
pud = pte_pfn_to_mfn(pud);
return native_make_pud(pud);
}
PV_CALLEE_SAVE_REGS_THUNK(xen_make_pud);
static pgd_t *xen_get_user_pgd(pgd_t *pgd)
{
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
pgd_t *pgd_page = (pgd_t *)(((unsigned long)pgd) & PAGE_MASK);
unsigned offset = pgd - pgd_page;
pgd_t *user_ptr = NULL;
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
if (offset < pgd_index(USER_LIMIT)) {
struct page *page = virt_to_page(pgd_page);
user_ptr = (pgd_t *)page->private;
if (user_ptr)
user_ptr += offset;
}
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
return user_ptr;
}
static void __xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
{
struct mmu_update u;
u.ptr = virt_to_machine(ptr).maddr;
u.val = pgd_val_ma(val);
xen_extend_mmu_update(&u);
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
}
/*
* Raw hypercall-based set_pgd, intended for in early boot before
* there's a page structure. This implies:
* 1. The only existing pagetable is the kernel's
* 2. It is always pinned
* 3. It has no user pagetable attached to it
*/
static void __init xen_set_pgd_hyper(pgd_t *ptr, pgd_t val)
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
{
preempt_disable();
xen_mc_batch();
__xen_set_pgd_hyper(ptr, val);
xen_mc_issue(PARAVIRT_LAZY_MMU);
preempt_enable();
}
static void xen_set_pgd(pgd_t *ptr, pgd_t val)
{
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
pgd_t *user_ptr = xen_get_user_pgd(ptr);
trace_xen_mmu_set_pgd(ptr, user_ptr, val);
/* If page is not pinned, we can just update the entry
directly */
if (!xen_page_pinned(ptr)) {
*ptr = val;
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
if (user_ptr) {
WARN_ON(xen_page_pinned(user_ptr));
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
*user_ptr = val;
}
return;
}
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
/* If it's pinned, then we can at least batch the kernel and
user updates together. */
xen_mc_batch();
__xen_set_pgd_hyper(ptr, val);
if (user_ptr)
__xen_set_pgd_hyper(user_ptr, val);
xen_mc_issue(PARAVIRT_LAZY_MMU);
}
#endif /* PAGETABLE_LEVELS == 4 */
/*
* (Yet another) pagetable walker. This one is intended for pinning a
* pagetable. This means that it walks a pagetable and calls the
* callback function on each page it finds making up the page table,
* at every level. It walks the entire pagetable, but it only bothers
* pinning pte pages which are below limit. In the normal case this
* will be STACK_TOP_MAX, but at boot we need to pin up to
* FIXADDR_TOP.
*
* For 32-bit the important bit is that we don't pin beyond there,
* because then we start getting into Xen's ptes.
*
* For 64-bit, we must skip the Xen hole in the middle of the address
* space, just after the big x86-64 virtual hole.
*/
xen: pin correct PGD on suspend Impact: fix Xen guest boot failure commit eefb47f6a1e855653d275cb90592a3587ea93a09 ("xen: use spin_lock_nest_lock when pinning a pagetable") changed xen_pgd_walk to walk over mm->pgd rather than taking pgd as an argument. This breaks xen_mm_(un)pin_all() because it makes init_mm.pgd readonly instead of the pgd we are interested in and therefore the pin subsequently fails. (XEN) mm.c:2280:d15 Bad type (saw 00000000e8000001 != exp 0000000060000000) for mfn bc464 (pfn 21ca7) (XEN) mm.c:2665:d15 Error while pinning mfn bc464 [ 14.586913] 1 multicall(s) failed: cpu 0 [ 14.586926] Pid: 14, comm: kstop/0 Not tainted 2.6.28-rc5-x86_32p-xenU-00172-gee2f6cc #200 [ 14.586940] Call Trace: [ 14.586955] [<c030c17a>] ? printk+0x18/0x1e [ 14.586972] [<c0103df3>] xen_mc_flush+0x163/0x1d0 [ 14.586986] [<c0104bc1>] __xen_pgd_pin+0xa1/0x110 [ 14.587000] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587015] [<c0104d7b>] xen_mm_pin_all+0x4b/0x70 [ 14.587029] [<c022bcb9>] xen_suspend+0x39/0xe0 [ 14.587042] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587054] [<c015a3cd>] stop_cpu+0x9d/0xf0 [ 14.587067] [<c01417cd>] run_workqueue+0x8d/0x150 [ 14.587080] [<c030e4b3>] ? _spin_unlock_irqrestore+0x23/0x40 [ 14.587094] [<c014558a>] ? prepare_to_wait+0x3a/0x70 [ 14.587107] [<c0141918>] worker_thread+0x88/0xf0 [ 14.587120] [<c01453c0>] ? autoremove_wake_function+0x0/0x50 [ 14.587133] [<c0141890>] ? worker_thread+0x0/0xf0 [ 14.587146] [<c014509c>] kthread+0x3c/0x70 [ 14.587157] [<c0145060>] ? kthread+0x0/0x70 [ 14.587170] [<c0109d1b>] kernel_thread_helper+0x7/0x10 [ 14.587181] call 1/3: op=14 arg=[c0415000] result=0 [ 14.587192] call 2/3: op=14 arg=[e1ca2000] result=0 [ 14.587204] call 3/3: op=26 arg=[c1808860] result=-22 Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-11-21 18:21:33 +08:00
static int __xen_pgd_walk(struct mm_struct *mm, pgd_t *pgd,
int (*func)(struct mm_struct *mm, struct page *,
enum pt_level),
unsigned long limit)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
int flush = 0;
unsigned hole_low, hole_high;
unsigned pgdidx_limit, pudidx_limit, pmdidx_limit;
unsigned pgdidx, pudidx, pmdidx;
/* The limit is the last byte to be touched */
limit--;
BUG_ON(limit >= FIXADDR_TOP);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
if (xen_feature(XENFEAT_auto_translated_physmap))
return 0;
/*
* 64-bit has a great big hole in the middle of the address
* space, which contains the Xen mappings. On 32-bit these
* will end up making a zero-sized hole and so is a no-op.
*/
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
hole_low = pgd_index(USER_LIMIT);
hole_high = pgd_index(PAGE_OFFSET);
pgdidx_limit = pgd_index(limit);
#if PTRS_PER_PUD > 1
pudidx_limit = pud_index(limit);
#else
pudidx_limit = 0;
#endif
#if PTRS_PER_PMD > 1
pmdidx_limit = pmd_index(limit);
#else
pmdidx_limit = 0;
#endif
for (pgdidx = 0; pgdidx <= pgdidx_limit; pgdidx++) {
pud_t *pud;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
if (pgdidx >= hole_low && pgdidx < hole_high)
continue;
if (!pgd_val(pgd[pgdidx]))
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
continue;
pud = pud_offset(&pgd[pgdidx], 0);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
if (PTRS_PER_PUD > 1) /* not folded */
flush |= (*func)(mm, virt_to_page(pud), PT_PUD);
for (pudidx = 0; pudidx < PTRS_PER_PUD; pudidx++) {
pmd_t *pmd;
if (pgdidx == pgdidx_limit &&
pudidx > pudidx_limit)
goto out;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
if (pud_none(pud[pudidx]))
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
continue;
pmd = pmd_offset(&pud[pudidx], 0);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
if (PTRS_PER_PMD > 1) /* not folded */
flush |= (*func)(mm, virt_to_page(pmd), PT_PMD);
for (pmdidx = 0; pmdidx < PTRS_PER_PMD; pmdidx++) {
struct page *pte;
if (pgdidx == pgdidx_limit &&
pudidx == pudidx_limit &&
pmdidx > pmdidx_limit)
goto out;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
if (pmd_none(pmd[pmdidx]))
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
continue;
pte = pmd_page(pmd[pmdidx]);
flush |= (*func)(mm, pte, PT_PTE);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
}
}
out:
/* Do the top level last, so that the callbacks can use it as
a cue to do final things like tlb flushes. */
flush |= (*func)(mm, virt_to_page(pgd), PT_PGD);
return flush;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
xen: pin correct PGD on suspend Impact: fix Xen guest boot failure commit eefb47f6a1e855653d275cb90592a3587ea93a09 ("xen: use spin_lock_nest_lock when pinning a pagetable") changed xen_pgd_walk to walk over mm->pgd rather than taking pgd as an argument. This breaks xen_mm_(un)pin_all() because it makes init_mm.pgd readonly instead of the pgd we are interested in and therefore the pin subsequently fails. (XEN) mm.c:2280:d15 Bad type (saw 00000000e8000001 != exp 0000000060000000) for mfn bc464 (pfn 21ca7) (XEN) mm.c:2665:d15 Error while pinning mfn bc464 [ 14.586913] 1 multicall(s) failed: cpu 0 [ 14.586926] Pid: 14, comm: kstop/0 Not tainted 2.6.28-rc5-x86_32p-xenU-00172-gee2f6cc #200 [ 14.586940] Call Trace: [ 14.586955] [<c030c17a>] ? printk+0x18/0x1e [ 14.586972] [<c0103df3>] xen_mc_flush+0x163/0x1d0 [ 14.586986] [<c0104bc1>] __xen_pgd_pin+0xa1/0x110 [ 14.587000] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587015] [<c0104d7b>] xen_mm_pin_all+0x4b/0x70 [ 14.587029] [<c022bcb9>] xen_suspend+0x39/0xe0 [ 14.587042] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587054] [<c015a3cd>] stop_cpu+0x9d/0xf0 [ 14.587067] [<c01417cd>] run_workqueue+0x8d/0x150 [ 14.587080] [<c030e4b3>] ? _spin_unlock_irqrestore+0x23/0x40 [ 14.587094] [<c014558a>] ? prepare_to_wait+0x3a/0x70 [ 14.587107] [<c0141918>] worker_thread+0x88/0xf0 [ 14.587120] [<c01453c0>] ? autoremove_wake_function+0x0/0x50 [ 14.587133] [<c0141890>] ? worker_thread+0x0/0xf0 [ 14.587146] [<c014509c>] kthread+0x3c/0x70 [ 14.587157] [<c0145060>] ? kthread+0x0/0x70 [ 14.587170] [<c0109d1b>] kernel_thread_helper+0x7/0x10 [ 14.587181] call 1/3: op=14 arg=[c0415000] result=0 [ 14.587192] call 2/3: op=14 arg=[e1ca2000] result=0 [ 14.587204] call 3/3: op=26 arg=[c1808860] result=-22 Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-11-21 18:21:33 +08:00
static int xen_pgd_walk(struct mm_struct *mm,
int (*func)(struct mm_struct *mm, struct page *,
enum pt_level),
unsigned long limit)
{
return __xen_pgd_walk(mm, mm->pgd, func, limit);
}
/* If we're using split pte locks, then take the page's lock and
return a pointer to it. Otherwise return NULL. */
static spinlock_t *xen_pte_lock(struct page *page, struct mm_struct *mm)
{
spinlock_t *ptl = NULL;
#if USE_SPLIT_PTLOCKS
ptl = __pte_lockptr(page);
spin_lock_nest_lock(ptl, &mm->page_table_lock);
#endif
return ptl;
}
static void xen_pte_unlock(void *v)
{
spinlock_t *ptl = v;
spin_unlock(ptl);
}
static void xen_do_pin(unsigned level, unsigned long pfn)
{
struct mmuext_op op;
op.cmd = level;
op.arg1.mfn = pfn_to_mfn(pfn);
xen_extend_mmuext_op(&op);
}
static int xen_pin_page(struct mm_struct *mm, struct page *page,
enum pt_level level)
{
unsigned pgfl = TestSetPagePinned(page);
int flush;
if (pgfl)
flush = 0; /* already pinned */
else if (PageHighMem(page))
/* kmaps need flushing if we found an unpinned
highpage */
flush = 1;
else {
void *pt = lowmem_page_address(page);
unsigned long pfn = page_to_pfn(page);
struct multicall_space mcs = __xen_mc_entry(0);
spinlock_t *ptl;
flush = 0;
/*
* We need to hold the pagetable lock between the time
* we make the pagetable RO and when we actually pin
* it. If we don't, then other users may come in and
* attempt to update the pagetable by writing it,
* which will fail because the memory is RO but not
* pinned, so Xen won't do the trap'n'emulate.
*
* If we're using split pte locks, we can't hold the
* entire pagetable's worth of locks during the
* traverse, because we may wrap the preempt count (8
* bits). The solution is to mark RO and pin each PTE
* page while holding the lock. This means the number
* of locks we end up holding is never more than a
* batch size (~32 entries, at present).
*
* If we're not using split pte locks, we needn't pin
* the PTE pages independently, because we're
* protected by the overall pagetable lock.
*/
ptl = NULL;
if (level == PT_PTE)
ptl = xen_pte_lock(page, mm);
MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
pfn_pte(pfn, PAGE_KERNEL_RO),
level == PT_PGD ? UVMF_TLB_FLUSH : 0);
if (ptl) {
xen_do_pin(MMUEXT_PIN_L1_TABLE, pfn);
/* Queue a deferred unlock for when this batch
is completed. */
xen_mc_callback(xen_pte_unlock, ptl);
}
}
return flush;
}
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
/* This is called just after a mm has been created, but it has not
been used yet. We need to make sure that its pagetable is all
read-only, and can be pinned. */
static void __xen_pgd_pin(struct mm_struct *mm, pgd_t *pgd)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
trace_xen_mmu_pgd_pin(mm, pgd);
xen_mc_batch();
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
xen: pin correct PGD on suspend Impact: fix Xen guest boot failure commit eefb47f6a1e855653d275cb90592a3587ea93a09 ("xen: use spin_lock_nest_lock when pinning a pagetable") changed xen_pgd_walk to walk over mm->pgd rather than taking pgd as an argument. This breaks xen_mm_(un)pin_all() because it makes init_mm.pgd readonly instead of the pgd we are interested in and therefore the pin subsequently fails. (XEN) mm.c:2280:d15 Bad type (saw 00000000e8000001 != exp 0000000060000000) for mfn bc464 (pfn 21ca7) (XEN) mm.c:2665:d15 Error while pinning mfn bc464 [ 14.586913] 1 multicall(s) failed: cpu 0 [ 14.586926] Pid: 14, comm: kstop/0 Not tainted 2.6.28-rc5-x86_32p-xenU-00172-gee2f6cc #200 [ 14.586940] Call Trace: [ 14.586955] [<c030c17a>] ? printk+0x18/0x1e [ 14.586972] [<c0103df3>] xen_mc_flush+0x163/0x1d0 [ 14.586986] [<c0104bc1>] __xen_pgd_pin+0xa1/0x110 [ 14.587000] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587015] [<c0104d7b>] xen_mm_pin_all+0x4b/0x70 [ 14.587029] [<c022bcb9>] xen_suspend+0x39/0xe0 [ 14.587042] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587054] [<c015a3cd>] stop_cpu+0x9d/0xf0 [ 14.587067] [<c01417cd>] run_workqueue+0x8d/0x150 [ 14.587080] [<c030e4b3>] ? _spin_unlock_irqrestore+0x23/0x40 [ 14.587094] [<c014558a>] ? prepare_to_wait+0x3a/0x70 [ 14.587107] [<c0141918>] worker_thread+0x88/0xf0 [ 14.587120] [<c01453c0>] ? autoremove_wake_function+0x0/0x50 [ 14.587133] [<c0141890>] ? worker_thread+0x0/0xf0 [ 14.587146] [<c014509c>] kthread+0x3c/0x70 [ 14.587157] [<c0145060>] ? kthread+0x0/0x70 [ 14.587170] [<c0109d1b>] kernel_thread_helper+0x7/0x10 [ 14.587181] call 1/3: op=14 arg=[c0415000] result=0 [ 14.587192] call 2/3: op=14 arg=[e1ca2000] result=0 [ 14.587204] call 3/3: op=26 arg=[c1808860] result=-22 Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-11-21 18:21:33 +08:00
if (__xen_pgd_walk(mm, pgd, xen_pin_page, USER_LIMIT)) {
/* re-enable interrupts for flushing */
xen_mc_issue(0);
kmap_flush_unused();
xen_mc_batch();
}
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
#ifdef CONFIG_X86_64
{
pgd_t *user_pgd = xen_get_user_pgd(pgd);
xen_do_pin(MMUEXT_PIN_L4_TABLE, PFN_DOWN(__pa(pgd)));
if (user_pgd) {
xen_pin_page(mm, virt_to_page(user_pgd), PT_PGD);
xen_do_pin(MMUEXT_PIN_L4_TABLE,
PFN_DOWN(__pa(user_pgd)));
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
}
}
#else /* CONFIG_X86_32 */
#ifdef CONFIG_X86_PAE
/* Need to make sure unshared kernel PMD is pinnable */
xen_pin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]),
PT_PMD);
#endif
xen_do_pin(MMUEXT_PIN_L3_TABLE, PFN_DOWN(__pa(pgd)));
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
#endif /* CONFIG_X86_64 */
xen_mc_issue(0);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static void xen_pgd_pin(struct mm_struct *mm)
{
__xen_pgd_pin(mm, mm->pgd);
}
/*
* On save, we need to pin all pagetables to make sure they get their
* mfns turned into pfns. Search the list for any unpinned pgds and pin
* them (unpinned pgds are not currently in use, probably because the
* process is under construction or destruction).
*
* Expected to be called in stop_machine() ("equivalent to taking
* every spinlock in the system"), so the locking doesn't really
* matter all that much.
*/
void xen_mm_pin_all(void)
{
struct page *page;
spin_lock(&pgd_lock);
list_for_each_entry(page, &pgd_list, lru) {
if (!PagePinned(page)) {
__xen_pgd_pin(&init_mm, (pgd_t *)page_address(page));
SetPageSavePinned(page);
}
}
spin_unlock(&pgd_lock);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
/*
* The init_mm pagetable is really pinned as soon as its created, but
* that's before we have page structures to store the bits. So do all
* the book-keeping now.
*/
static int __init xen_mark_pinned(struct mm_struct *mm, struct page *page,
enum pt_level level)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
SetPagePinned(page);
return 0;
}
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
static void __init xen_mark_init_mm_pinned(void)
{
xen_pgd_walk(&init_mm, xen_mark_pinned, FIXADDR_TOP);
}
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
static int xen_unpin_page(struct mm_struct *mm, struct page *page,
enum pt_level level)
{
unsigned pgfl = TestClearPagePinned(page);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
if (pgfl && !PageHighMem(page)) {
void *pt = lowmem_page_address(page);
unsigned long pfn = page_to_pfn(page);
spinlock_t *ptl = NULL;
struct multicall_space mcs;
/*
* Do the converse to pin_page. If we're using split
* pte locks, we must be holding the lock for while
* the pte page is unpinned but still RO to prevent
* concurrent updates from seeing it in this
* partially-pinned state.
*/
if (level == PT_PTE) {
ptl = xen_pte_lock(page, mm);
if (ptl)
xen_do_pin(MMUEXT_UNPIN_TABLE, pfn);
}
mcs = __xen_mc_entry(0);
MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
pfn_pte(pfn, PAGE_KERNEL),
level == PT_PGD ? UVMF_TLB_FLUSH : 0);
if (ptl) {
/* unlock when batch completed */
xen_mc_callback(xen_pte_unlock, ptl);
}
}
return 0; /* never need to flush on unpin */
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
/* Release a pagetables pages back as normal RW */
static void __xen_pgd_unpin(struct mm_struct *mm, pgd_t *pgd)
{
trace_xen_mmu_pgd_unpin(mm, pgd);
xen_mc_batch();
xen_do_pin(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
#ifdef CONFIG_X86_64
{
pgd_t *user_pgd = xen_get_user_pgd(pgd);
if (user_pgd) {
xen_do_pin(MMUEXT_UNPIN_TABLE,
PFN_DOWN(__pa(user_pgd)));
xen_unpin_page(mm, virt_to_page(user_pgd), PT_PGD);
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
}
}
#endif
#ifdef CONFIG_X86_PAE
/* Need to make sure unshared kernel PMD is unpinned */
xen_unpin_page(mm, pgd_page(pgd[pgd_index(TASK_SIZE)]),
PT_PMD);
#endif
xen64: allocate and manage user pagetables Because the x86_64 architecture does not enforce segment limits, Xen cannot protect itself with them as it does in 32-bit mode. Therefore, to protect itself, it runs the guest kernel in ring 3. Since it also runs the guest userspace in ring3, the guest kernel must maintain a second pagetable for its userspace, which does not map kernel space. Naturally, the guest kernel pagetables map both kernel and userspace. The userspace pagetable is attached to the corresponding kernel pagetable via the pgd's page->private field. It is allocated and freed at the same time as the kernel pgd via the paravirt_pgd_alloc/free hooks. Fortunately, the user pagetable is almost entirely shared with the kernel pagetable; the only difference is the pgd page itself. set_pgd will populate all entries in the kernel pagetable, and also set the corresponding user pgd entry if the address is less than STACK_TOP_MAX. The user pagetable must be pinned and unpinned with the kernel one, but because the pagetables are aliased, pgd_walk() only needs to be called on the kernel pagetable. The user pgd page is then pinned/unpinned along with the kernel pgd page. xen_write_cr3 must write both the kernel and user cr3s. The init_mm.pgd pagetable never has a user pagetable allocated for it, because it can never be used while running usermode. One awkward area is that early in boot the page structures are not available. No user pagetable can exist at that point, but it complicates the logic to avoid looking at the page structure. Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Cc: Stephen Tweedie <sct@redhat.com> Cc: Eduardo Habkost <ehabkost@redhat.com> Cc: Mark McLoughlin <markmc@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-09 06:07:13 +08:00
xen: pin correct PGD on suspend Impact: fix Xen guest boot failure commit eefb47f6a1e855653d275cb90592a3587ea93a09 ("xen: use spin_lock_nest_lock when pinning a pagetable") changed xen_pgd_walk to walk over mm->pgd rather than taking pgd as an argument. This breaks xen_mm_(un)pin_all() because it makes init_mm.pgd readonly instead of the pgd we are interested in and therefore the pin subsequently fails. (XEN) mm.c:2280:d15 Bad type (saw 00000000e8000001 != exp 0000000060000000) for mfn bc464 (pfn 21ca7) (XEN) mm.c:2665:d15 Error while pinning mfn bc464 [ 14.586913] 1 multicall(s) failed: cpu 0 [ 14.586926] Pid: 14, comm: kstop/0 Not tainted 2.6.28-rc5-x86_32p-xenU-00172-gee2f6cc #200 [ 14.586940] Call Trace: [ 14.586955] [<c030c17a>] ? printk+0x18/0x1e [ 14.586972] [<c0103df3>] xen_mc_flush+0x163/0x1d0 [ 14.586986] [<c0104bc1>] __xen_pgd_pin+0xa1/0x110 [ 14.587000] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587015] [<c0104d7b>] xen_mm_pin_all+0x4b/0x70 [ 14.587029] [<c022bcb9>] xen_suspend+0x39/0xe0 [ 14.587042] [<c015a330>] ? stop_cpu+0x0/0xf0 [ 14.587054] [<c015a3cd>] stop_cpu+0x9d/0xf0 [ 14.587067] [<c01417cd>] run_workqueue+0x8d/0x150 [ 14.587080] [<c030e4b3>] ? _spin_unlock_irqrestore+0x23/0x40 [ 14.587094] [<c014558a>] ? prepare_to_wait+0x3a/0x70 [ 14.587107] [<c0141918>] worker_thread+0x88/0xf0 [ 14.587120] [<c01453c0>] ? autoremove_wake_function+0x0/0x50 [ 14.587133] [<c0141890>] ? worker_thread+0x0/0xf0 [ 14.587146] [<c014509c>] kthread+0x3c/0x70 [ 14.587157] [<c0145060>] ? kthread+0x0/0x70 [ 14.587170] [<c0109d1b>] kernel_thread_helper+0x7/0x10 [ 14.587181] call 1/3: op=14 arg=[c0415000] result=0 [ 14.587192] call 2/3: op=14 arg=[e1ca2000] result=0 [ 14.587204] call 3/3: op=26 arg=[c1808860] result=-22 Signed-off-by: Ian Campbell <ian.campbell@citrix.com> Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-11-21 18:21:33 +08:00
__xen_pgd_walk(mm, pgd, xen_unpin_page, USER_LIMIT);
xen_mc_issue(0);
}
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
static void xen_pgd_unpin(struct mm_struct *mm)
{
__xen_pgd_unpin(mm, mm->pgd);
}
/*
* On resume, undo any pinning done at save, so that the rest of the
* kernel doesn't see any unexpected pinned pagetables.
*/
void xen_mm_unpin_all(void)
{
struct page *page;
spin_lock(&pgd_lock);
list_for_each_entry(page, &pgd_list, lru) {
if (PageSavePinned(page)) {
BUG_ON(!PagePinned(page));
__xen_pgd_unpin(&init_mm, (pgd_t *)page_address(page));
ClearPageSavePinned(page);
}
}
spin_unlock(&pgd_lock);
}
static void xen_activate_mm(struct mm_struct *prev, struct mm_struct *next)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
spin_lock(&next->page_table_lock);
xen_pgd_pin(next);
spin_unlock(&next->page_table_lock);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static void xen_dup_mmap(struct mm_struct *oldmm, struct mm_struct *mm)
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
{
spin_lock(&mm->page_table_lock);
xen_pgd_pin(mm);
spin_unlock(&mm->page_table_lock);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
#ifdef CONFIG_SMP
/* Another cpu may still have their %cr3 pointing at the pagetable, so
we need to repoint it somewhere else before we can unpin it. */
static void drop_other_mm_ref(void *info)
{
struct mm_struct *mm = info;
struct mm_struct *active_mm;
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
active_mm = percpu_read(cpu_tlbstate.active_mm);
xen mmu: fix a race window causing leave_mm BUG() There's a race window in xen_drop_mm_ref, where remote cpu may exit dirty bitmap between the check on this cpu and the point where remote cpu handles drop request. So in drop_other_mm_ref we need check whether TLB state is still lazy before calling into leave_mm. This bug is rarely observed in earlier kernel, but exaggerated by the commit 831d52bc153971b70e64eccfbed2b232394f22f8 ("x86, mm: avoid possible bogus tlb entries by clearing prev mm_cpumask after switching mm") which clears bitmap after changing the TLB state. the call trace is as below: --------------------------------- kernel BUG at arch/x86/mm/tlb.c:61! invalid opcode: 0000 [#1] SMP last sysfs file: /sys/devices/system/xen_memory/xen_memory0/info/current_kb CPU 1 Modules linked in: 8021q garp xen_netback xen_blkback blktap blkback_pagemap nbd bridge stp llc autofs4 ipmi_devintf ipmi_si ipmi_msghandler lockd sunrpc bonding ipv6 xenfs dm_multipath video output sbs sbshc parport_pc lp parport ses enclosure snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device serio_raw bnx2 snd_pcm_oss snd_mixer_oss snd_pcm snd_timer iTCO_wdt snd soundcore snd_page_alloc i2c_i801 iTCO_vendor_support i2c_core pcs pkr pata_acpi ata_generic ata_piix shpchp mptsas mptscsih mptbase [last unloaded: freq_table] Pid: 25581, comm: khelper Not tainted 2.6.32.36fixxen #1 Tecal RH2285 RIP: e030:[<ffffffff8103a3cb>] [<ffffffff8103a3cb>] leave_mm+0x15/0x46 RSP: e02b:ffff88002805be48 EFLAGS: 00010046 RAX: 0000000000000000 RBX: 0000000000000001 RCX: ffff88015f8e2da0 RDX: ffff88002805be78 RSI: 0000000000000000 RDI: 0000000000000001 RBP: ffff88002805be48 R08: ffff88009d662000 R09: dead000000200200 R10: dead000000100100 R11: ffffffff814472b2 R12: ffff88009bfc1880 R13: ffff880028063020 R14: 00000000000004f6 R15: 0000000000000000 FS: 00007f62362d66e0(0000) GS:ffff880028058000(0000) knlGS:0000000000000000 CS: e033 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000003aabc11909 CR3: 000000009b8ca000 CR4: 0000000000002660 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 00000000000000 00 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process khelper (pid: 25581, threadinfo ffff88007691e000, task ffff88009b92db40) Stack: ffff88002805be68 ffffffff8100e4ae 0000000000000001 ffff88009d733b88 <0> ffff88002805be98 ffffffff81087224 ffff88002805be78 ffff88002805be78 <0> ffff88015f808360 00000000000004f6 ffff88002805bea8 ffffffff81010108 Call Trace: <IRQ> [<ffffffff8100e4ae>] drop_other_mm_ref+0x2a/0x53 [<ffffffff81087224>] generic_smp_call_function_single_interrupt+0xd8/0xfc [<ffffffff81010108>] xen_call_function_single_interrupt+0x13/0x28 [<ffffffff810a936a>] handle_IRQ_event+0x66/0x120 [<ffffffff810aac5b>] handle_percpu_irq+0x41/0x6e [<ffffffff8128c1c0>] __xen_evtchn_do_upcall+0x1ab/0x27d [<ffffffff8128dd11>] xen_evtchn_do_upcall+0x33/0x46 [<ffffffff81013efe>] xen_do_hyper visor_callback+0x1e/0x30 <EOI> [<ffffffff814472b2>] ? _spin_unlock_irqrestore+0x15/0x17 [<ffffffff8100f8cf>] ? xen_restore_fl_direct_end+0x0/0x1 [<ffffffff81113f71>] ? flush_old_exec+0x3ac/0x500 [<ffffffff81150dc5>] ? load_elf_binary+0x0/0x17ef [<ffffffff81150dc5>] ? load_elf_binary+0x0/0x17ef [<ffffffff8115115d>] ? load_elf_binary+0x398/0x17ef [<ffffffff81042fcf>] ? need_resched+0x23/0x2d [<ffffffff811f4648>] ? process_measurement+0xc0/0xd7 [<ffffffff81150dc5>] ? load_elf_binary+0x0/0x17ef [<ffffffff81113094>] ? search_binary_handler+0xc8/0x255 [<ffffffff81114362>] ? do_execve+0x1c3/0x29e [<ffffffff8101155d>] ? sys_execve+0x43/0x5d [<ffffffff8106fc45>] ? __call_usermodehelper+0x0/0x6f [<ffffffff81013e28>] ? kernel_execve+0x68/0xd0 [<ffffffff 8106fc45>] ? __call_usermodehelper+0x0/0x6f [<ffffffff8100f8cf>] ? xen_restore_fl_direct_end+0x0/0x1 [<ffffffff8106fb64>] ? ____call_usermodehelper+0x113/0x11e [<ffffffff81013daa>] ? child_rip+0xa/0x20 [<ffffffff8106fc45>] ? __call_usermodehelper+0x0/0x6f [<ffffffff81012f91>] ? int_ret_from_sys_call+0x7/0x1b [<ffffffff8101371d>] ? retint_restore_args+0x5/0x6 [<ffffffff81013da0>] ? child_rip+0x0/0x20 Code: 41 5e 41 5f c9 c3 55 48 89 e5 0f 1f 44 00 00 e8 17 ff ff ff c9 c3 55 48 89 e5 0f 1f 44 00 00 65 8b 04 25 c8 55 01 00 ff c8 75 04 <0f> 0b eb fe 65 48 8b 34 25 c0 55 01 00 48 81 c6 b8 02 00 00 e8 RIP [<ffffffff8103a3cb>] leave_mm+0x15/0x46 RSP <ffff88002805be48> ---[ end trace ce9cee6832a9c503 ]--- Tested-by: Maoxiaoyun<tinnycloud@hotmail.com> Signed-off-by: Kevin Tian <kevin.tian@intel.com> [v1: Fleshed out the git description a bit] Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
2011-05-12 10:56:08 +08:00
if (active_mm == mm && percpu_read(cpu_tlbstate.state) != TLBSTATE_OK)
leave_mm(smp_processor_id());
/* If this cpu still has a stale cr3 reference, then make sure
it has been flushed. */
if (percpu_read(xen_current_cr3) == __pa(mm->pgd))
load_cr3(swapper_pg_dir);
}
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
static void xen_drop_mm_ref(struct mm_struct *mm)
{
cpumask_var_t mask;
unsigned cpu;
if (current->active_mm == mm) {
if (current->mm == mm)
load_cr3(swapper_pg_dir);
else
leave_mm(smp_processor_id());
}
/* Get the "official" set of cpus referring to our pagetable. */
if (!alloc_cpumask_var(&mask, GFP_ATOMIC)) {
for_each_online_cpu(cpu) {
if (!cpumask_test_cpu(cpu, mm_cpumask(mm))
&& per_cpu(xen_current_cr3, cpu) != __pa(mm->pgd))
continue;
smp_call_function_single(cpu, drop_other_mm_ref, mm, 1);
}
return;
}
cpumask_copy(mask, mm_cpumask(mm));
/* It's possible that a vcpu may have a stale reference to our
cr3, because its in lazy mode, and it hasn't yet flushed
its set of pending hypercalls yet. In this case, we can
look at its actual current cr3 value, and force it to flush
if needed. */
for_each_online_cpu(cpu) {
if (per_cpu(xen_current_cr3, cpu) == __pa(mm->pgd))
cpumask_set_cpu(cpu, mask);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
if (!cpumask_empty(mask))
smp_call_function_many(mask, drop_other_mm_ref, mm, 1);
free_cpumask_var(mask);
}
#else
static void xen_drop_mm_ref(struct mm_struct *mm)
{
if (current->active_mm == mm)
load_cr3(swapper_pg_dir);
}
#endif
/*
* While a process runs, Xen pins its pagetables, which means that the
* hypervisor forces it to be read-only, and it controls all updates
* to it. This means that all pagetable updates have to go via the
* hypervisor, which is moderately expensive.
*
* Since we're pulling the pagetable down, we switch to use init_mm,
* unpin old process pagetable and mark it all read-write, which
* allows further operations on it to be simple memory accesses.
*
* The only subtle point is that another CPU may be still using the
* pagetable because of lazy tlb flushing. This means we need need to
* switch all CPUs off this pagetable before we can unpin it.
*/
static void xen_exit_mmap(struct mm_struct *mm)
{
get_cpu(); /* make sure we don't move around */
xen_drop_mm_ref(mm);
put_cpu();
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
spin_lock(&mm->page_table_lock);
/* pgd may not be pinned in the error exit path of execve */
if (xen_page_pinned(mm->pgd))
xen_pgd_unpin(mm);
spin_unlock(&mm->page_table_lock);
xen: virtual mmu Xen pagetable handling, including the machinery to implement direct pagetables. Xen presents the real CPU's pagetables directly to guests, with no added shadowing or other layer of abstraction. Naturally this means the hypervisor must maintain close control over what the guest can put into the pagetable. When the guest modifies the pte/pmd/pgd, it must convert its domain-specific notion of a "physical" pfn into a global machine frame number (mfn) before inserting the entry into the pagetable. Xen will check to make sure the domain is allowed to create a mapping of the given mfn. Xen also requires that all mappings the guest has of its own active pagetable are read-only. This is relatively easy to implement in Linux because all pagetables share the same pte pages for kernel mappings, so updating the pte in one pagetable will implicitly update the mapping in all pagetables. Normally a pagetable becomes active when you point to it with cr3 (or the Xen equivalent), but when you do so, Xen must check the whole pagetable for correctness, which is clearly a performance problem. Xen solves this with pinning which keeps a pagetable effectively active even if its currently unused, which means that all the normal update rules are enforced. This means that it need not revalidate the pagetable when loading cr3. This patch has a first-cut implementation of pinning, but it is more fully implemented in a later patch. Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com> Signed-off-by: Chris Wright <chrisw@sous-sol.org>
2007-07-18 09:37:04 +08:00
}
static void __init xen_pagetable_setup_start(pgd_t *base)
{
}
x86,xen: introduce x86_init.mapping.pagetable_reserve Introduce a new x86_init hook called pagetable_reserve that at the end of init_memory_mapping is used to reserve a range of memory addresses for the kernel pagetable pages we used and free the other ones. On native it just calls memblock_x86_reserve_range while on xen it also takes care of setting the spare memory previously allocated for kernel pagetable pages from RO to RW, so that it can be used for other purposes. A detailed explanation of the reason why this hook is needed follows. As a consequence of the commit: commit 4b239f458c229de044d6905c2b0f9fe16ed9e01e Author: Yinghai Lu <yinghai@kernel.org> Date: Fri Dec 17 16:58:28 2010 -0800 x86-64, mm: Put early page table high at some point init_memory_mapping is going to reach the pagetable pages area and map those pages too (mapping them as normal memory that falls in the range of addresses passed to init_memory_mapping as argument). Some of those pages are already pagetable pages (they are in the range pgt_buf_start-pgt_buf_end) therefore they are going to be mapped RO and everything is fine. Some of these pages are not pagetable pages yet (they fall in the range pgt_buf_end-pgt_buf_top; for example the page at pgt_buf_end) so they are going to be mapped RW. When these pages become pagetable pages and are hooked into the pagetable, xen will find that the guest has already a RW mapping of them somewhere and fail the operation. The reason Xen requires pagetables to be RO is that the hypervisor needs to verify that the pagetables are valid before using them. The validation operations are called "pinning" (more details in arch/x86/xen/mmu.c). In order to fix the issue we mark all the pages in the entire range pgt_buf_start-pgt_buf_top as RO, however when the pagetable allocation is completed only the range pgt_buf_start-pgt_buf_end is reserved by init_memory_mapping. Hence the kernel is going to crash as soon as one of the pages in the range pgt_buf_end-pgt_buf_top is reused (b/c those ranges are RO). For this reason we need a hook to reserve the kernel pagetable pages we used and free the other ones so that they can be reused for other purposes. On native it just means calling memblock_x86_reserve_range, on Xen it also means marking RW the pagetable pages that we allocated before but that haven't been used before. Another way to fix this is without using the hook is by adding a 'if (xen_pv_domain)' in the 'init_memory_mapping' code and calling the Xen counterpart, but that is just nasty. Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Yinghai Lu <yinghai@kernel.org> Acked-by: H. Peter Anvin <hpa@zytor.com> Cc: Ingo Molnar <mingo@elte.hu> Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
2011-04-14 22:49:41 +08:00
static __init void xen_mapping_pagetable_reserve(u64 start, u64 end)
{
/* reserve the range used */
native_pagetable_reserve(start, end);
/* set as RW the rest */
printk(KERN_DEBUG "xen: setting RW the range %llx - %llx\n", end,
PFN_PHYS(pgt_buf_top));
while (end < PFN_PHYS(pgt_buf_top)) {
make_lowmem_page_readwrite(__va(end));
end += PAGE_SIZE;
}
}
static void xen_post_allocator_init(void);
static void __init xen_pagetable_setup_done(pgd_t *base)
{
xen_setup_shared_info();
xen_post_allocator_init();
}
static void xen_write_cr2(unsigned long cr2)
{
percpu_read(xen_vcpu)->arch.cr2 = cr2;
}
static unsigned long xen_read_cr2(void)
{
return percpu_read(xen_vcpu)->arch.cr2;
}
unsigned long xen_read_cr2_direct(void)
{
return percpu_read(xen_vcpu_info.arch.cr2);
}
static void xen_flush_tlb(void)
{
struct mmuext_op *op;
struct multicall_space mcs;
trace_xen_mmu_flush_tlb(0);
preempt_disable();
mcs = xen_mc_entry(sizeof(*op));
op = mcs.args;
op->cmd = MMUEXT_TLB_FLUSH_LOCAL;
MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
xen_mc_issue(PARAVIRT_LAZY_MMU);
preempt_enable();
}
static void xen_flush_tlb_single(unsigned long addr)
{
struct mmuext_op *op;
struct multicall_space mcs;
trace_xen_mmu_flush_tlb_single(addr);
preempt_disable();
mcs = xen_mc_entry(sizeof(*op));
op = mcs.args;
op->cmd = MMUEXT_INVLPG_LOCAL;
op->arg1.linear_addr = addr & PAGE_MASK;
MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
xen_mc_issue(PARAVIRT_LAZY_MMU);
preempt_enable();
}
static void xen_flush_tlb_others(const struct cpumask *cpus,
struct mm_struct *mm, unsigned long va)
{
struct {
struct mmuext_op op;
#ifdef CONFIG_SMP
DECLARE_BITMAP(mask, num_processors);
#else
DECLARE_BITMAP(mask, NR_CPUS);
#endif
} *args;
struct multicall_space mcs;
trace_xen_mmu_flush_tlb_others(cpus, mm, va);
if (cpumask_empty(cpus))
return; /* nothing to do */
mcs = xen_mc_entry(sizeof(*args));
args = mcs.args;
args->op.arg2.vcpumask = to_cpumask(args->mask);
/* Remove us, and any offline CPUS. */
cpumask_and(to_cpumask(args->mask), cpus, cpu_online_mask);
cpumask_clear_cpu(smp_processor_id(), to_cpumask(args->mask));
if (va == TLB_FLUSH_ALL) {
args->op.cmd = MMUEXT_TLB_FLUSH_MULTI;
} else {
args->op.cmd = MMUEXT_INVLPG_MULTI;
args->op.arg1.linear_addr = va;
}
MULTI_mmuext_op(mcs.mc, &args->op, 1, NULL, DOMID_SELF);
xen_mc_issue(PARAVIRT_LAZY_MMU);
}
static unsigned long xen_read_cr3(void)
{
return percpu_read(xen_cr3);
}
static void set_current_cr3(void *v)
{
percpu_write(xen_current_cr3, (unsigned long)v);
}
static void __xen_write_cr3(bool kernel, unsigned long cr3)
{
struct mmuext_op op;
unsigned long mfn;
trace_xen_mmu_write_cr3(kernel, cr3);
if (cr3)
mfn = pfn_to_mfn(PFN_DOWN(cr3));
else
mfn = 0;
WARN_ON(mfn == 0 && kernel);
op.cmd = kernel ? MMUEXT_NEW_BASEPTR : MMUEXT_NEW_USER_BASEPTR;
op.arg1.mfn = mfn;
xen_extend_mmuext_op(&op);
if (kernel) {
percpu_write(xen_cr3, cr3);
/* Update xen_current_cr3 once the batch has actually
been submitted. */
xen_mc_callback(set_current_cr3, (void *)cr3);
}
}
static void xen_write_cr3(unsigned long cr3)
{
BUG_ON(preemptible());
xen_mc_batch(); /* disables interrupts */
/* Update while interrupts are disabled, so its atomic with
respect to ipis */
percpu_write(xen_cr3, cr3);
__xen_write_cr3(true, cr3);
#ifdef CONFIG_X86_64
{
pgd_t *user_pgd = xen_get_user_pgd(__va(cr3));
if (user_pgd)
__xen_write_cr3(false, __pa(user_pgd));
else
__xen_write_cr3(false, 0);
}
#endif
xen_mc_issue(PARAVIRT_LAZY_CPU); /* interrupts restored */
}
static int xen_pgd_alloc(struct mm_struct *mm)
{
pgd_t *pgd = mm->pgd;
int ret = 0;
BUG_ON(PagePinned(virt_to_page(pgd)));
#ifdef CONFIG_X86_64
{
struct page *page = virt_to_page(pgd);
pgd_t *user_pgd;
BUG_ON(page->private != 0);
ret = -ENOMEM;
user_pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO);
page->private = (unsigned long)user_pgd;
if (user_pgd != NULL) {
user_pgd[pgd_index(VSYSCALL_START)] =
__pgd(__pa(level3_user_vsyscall) | _PAGE_TABLE);
ret = 0;
}
BUG_ON(PagePinned(virt_to_page(xen_get_user_pgd(pgd))));
}
#endif
return ret;
}
static void xen_pgd_free(struct mm_struct *mm, pgd_t *pgd)
{
#ifdef CONFIG_X86_64
pgd_t *user_pgd = xen_get_user_pgd(pgd);
if (user_pgd)
free_page((unsigned long)user_pgd);
#endif
}
#ifdef CONFIG_X86_32
static pte_t __init mask_rw_pte(pte_t *ptep, pte_t pte)
{
/* If there's an existing pte, then don't allow _PAGE_RW to be set */
if (pte_val_ma(*ptep) & _PAGE_PRESENT)
pte = __pte_ma(((pte_val_ma(*ptep) & _PAGE_RW) | ~_PAGE_RW) &
pte_val_ma(pte));
return pte;
}
#else /* CONFIG_X86_64 */
static pte_t __init mask_rw_pte(pte_t *ptep, pte_t pte)
{
unsigned long pfn = pte_pfn(pte);
/*
* If the new pfn is within the range of the newly allocated
* kernel pagetable, and it isn't being mapped into an
* early_ioremap fixmap slot as a freshly allocated page, make sure
* it is RO.
*/
if (((!is_early_ioremap_ptep(ptep) &&
pfn >= pgt_buf_start && pfn < pgt_buf_top)) ||
(is_early_ioremap_ptep(ptep) && pfn != (pgt_buf_end - 1)))
pte = pte_wrprotect(pte);
return pte;
}
#endif /* CONFIG_X86_64 */
/* Init-time set_pte while constructing initial pagetables, which
doesn't allow RO pagetable pages to be remapped RW */
static void __init xen_set_pte_init(pte_t *ptep, pte_t pte)
{
pte = mask_rw_pte(ptep, pte);
xen_set_pte(ptep, pte);
}
static void pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
{
struct mmuext_op op;
op.cmd = cmd;
op.arg1.mfn = pfn_to_mfn(pfn);
if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
BUG();
}
/* Early in boot, while setting up the initial pagetable, assume
everything is pinned. */
static void __init xen_alloc_pte_init(struct mm_struct *mm, unsigned long pfn)
{
#ifdef CONFIG_FLATMEM
BUG_ON(mem_map); /* should only be used early */
#endif
make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);
}
/* Used for pmd and pud */
static void __init xen_alloc_pmd_init(struct mm_struct *mm, unsigned long pfn)
{
#ifdef CONFIG_FLATMEM
BUG_ON(mem_map); /* should only be used early */
#endif
make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
}
/* Early release_pte assumes that all pts are pinned, since there's
only init_mm and anything attached to that is pinned. */
static void __init xen_release_pte_init(unsigned long pfn)
{
pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);
make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
}
static void __init xen_release_pmd_init(unsigned long pfn)
{
make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
}
static inline void __pin_pagetable_pfn(unsigned cmd, unsigned long pfn)
{
struct multicall_space mcs;
struct mmuext_op *op;
mcs = __xen_mc_entry(sizeof(*op));
op = mcs.args;
op->cmd = cmd;
op->arg1.mfn = pfn_to_mfn(pfn);
MULTI_mmuext_op(mcs.mc, mcs.args, 1, NULL, DOMID_SELF);
}
static inline void __set_pfn_prot(unsigned long pfn, pgprot_t prot)
{
struct multicall_space mcs;
unsigned long addr = (unsigned long)__va(pfn << PAGE_SHIFT);
mcs = __xen_mc_entry(0);
MULTI_update_va_mapping(mcs.mc, (unsigned long)addr,
pfn_pte(pfn, prot), 0);
}
/* This needs to make sure the new pte page is pinned iff its being
attached to a pinned pagetable. */
static inline void xen_alloc_ptpage(struct mm_struct *mm, unsigned long pfn,
unsigned level)
{
bool pinned = PagePinned(virt_to_page(mm->pgd));
trace_xen_mmu_alloc_ptpage(mm, pfn, level, pinned);
if (pinned) {
struct page *page = pfn_to_page(pfn);
SetPagePinned(page);
if (!PageHighMem(page)) {
xen_mc_batch();
__set_pfn_prot(pfn, PAGE_KERNEL_RO);
if (level == PT_PTE && USE_SPLIT_PTLOCKS)
__pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);
xen_mc_issue(PARAVIRT_LAZY_MMU);
} else {
/* make sure there are no stray mappings of
this page */
kmap_flush_unused();
}
}
}
static void xen_alloc_pte(struct mm_struct *mm, unsigned long pfn)
{
xen_alloc_ptpage(mm, pfn, PT_PTE);
}
static void xen_alloc_pmd(struct mm_struct *mm, unsigned long pfn)
{
xen_alloc_ptpage(mm, pfn, PT_PMD);
}
/* This should never happen until we're OK to use struct page */
static inline void xen_release_ptpage(unsigned long pfn, unsigned level)
{
struct page *page = pfn_to_page(pfn);
bool pinned = PagePinned(page);
trace_xen_mmu_release_ptpage(pfn, level, pinned);
if (pinned) {
if (!PageHighMem(page)) {
xen_mc_batch();
if (level == PT_PTE && USE_SPLIT_PTLOCKS)
__pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);
__set_pfn_prot(pfn, PAGE_KERNEL);
xen_mc_issue(PARAVIRT_LAZY_MMU);
}
ClearPagePinned(page);
}
}
static void xen_release_pte(unsigned long pfn)
{
xen_release_ptpage(pfn, PT_PTE);
}
static void xen_release_pmd(unsigned long pfn)
{
xen_release_ptpage(pfn, PT_PMD);
}
#if PAGETABLE_LEVELS == 4
static void xen_alloc_pud(struct mm_struct *mm, unsigned long pfn)
{
xen_alloc_ptpage(mm, pfn, PT_PUD);
}
static void xen_release_pud(unsigned long pfn)
{
xen_release_ptpage(pfn, PT_PUD);
}
#endif
void __init xen_reserve_top(void)
{
#ifdef CONFIG_X86_32
unsigned long top = HYPERVISOR_VIRT_START;
struct xen_platform_parameters pp;
if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0)
top = pp.virt_start;
reserve_top_address(-top);
#endif /* CONFIG_X86_32 */
}
/*
* Like __va(), but returns address in the kernel mapping (which is
* all we have until the physical memory mapping has been set up.
*/
static void *__ka(phys_addr_t paddr)
{
#ifdef CONFIG_X86_64
return (void *)(paddr + __START_KERNEL_map);
#else
return __va(paddr);
#endif
}
/* Convert a machine address to physical address */
static unsigned long m2p(phys_addr_t maddr)
{
phys_addr_t paddr;
maddr &= PTE_PFN_MASK;
paddr = mfn_to_pfn(maddr >> PAGE_SHIFT) << PAGE_SHIFT;
return paddr;
}
/* Convert a machine address to kernel virtual */
static void *m2v(phys_addr_t maddr)
{
return __ka(m2p(maddr));
}
/* Set the page permissions on an identity-mapped pages */
static void set_page_prot(void *addr, pgprot_t prot)
{
unsigned long pfn = __pa(addr) >> PAGE_SHIFT;
pte_t pte = pfn_pte(pfn, prot);
if (HYPERVISOR_update_va_mapping((unsigned long)addr, pte, 0))
BUG();
}
static void __init xen_map_identity_early(pmd_t *pmd, unsigned long max_pfn)
{
unsigned pmdidx, pteidx;
unsigned ident_pte;
unsigned long pfn;
level1_ident_pgt = extend_brk(sizeof(pte_t) * LEVEL1_IDENT_ENTRIES,
PAGE_SIZE);
ident_pte = 0;
pfn = 0;
for (pmdidx = 0; pmdidx < PTRS_PER_PMD && pfn < max_pfn; pmdidx++) {
pte_t *pte_page;
/* Reuse or allocate a page of ptes */
if (pmd_present(pmd[pmdidx]))
pte_page = m2v(pmd[pmdidx].pmd);
else {
/* Check for free pte pages */
if (ident_pte == LEVEL1_IDENT_ENTRIES)
break;
pte_page = &level1_ident_pgt[ident_pte];
ident_pte += PTRS_PER_PTE;
pmd[pmdidx] = __pmd(__pa(pte_page) | _PAGE_TABLE);
}
/* Install mappings */
for (pteidx = 0; pteidx < PTRS_PER_PTE; pteidx++, pfn++) {
pte_t pte;
#ifdef CONFIG_X86_32
if (pfn > max_pfn_mapped)
max_pfn_mapped = pfn;
#endif
if (!pte_none(pte_page[pteidx]))
continue;
pte = pfn_pte(pfn, PAGE_KERNEL_EXEC);
pte_page[pteidx] = pte;
}
}
for (pteidx = 0; pteidx < ident_pte; pteidx += PTRS_PER_PTE)
set_page_prot(&level1_ident_pgt[pteidx], PAGE_KERNEL_RO);
set_page_prot(pmd, PAGE_KERNEL_RO);
}
void __init xen_setup_machphys_mapping(void)
{
struct xen_machphys_mapping mapping;
if (HYPERVISOR_memory_op(XENMEM_machphys_mapping, &mapping) == 0) {
machine_to_phys_mapping = (unsigned long *)mapping.v_start;
machine_to_phys_nr = mapping.max_mfn + 1;
} else {
machine_to_phys_nr = MACH2PHYS_NR_ENTRIES;
}
#ifdef CONFIG_X86_32
WARN_ON((machine_to_phys_mapping + (machine_to_phys_nr - 1))
< machine_to_phys_mapping);
#endif
}
#ifdef CONFIG_X86_64
static void convert_pfn_mfn(void *v)
{
pte_t *pte = v;
int i;
/* All levels are converted the same way, so just treat them
as ptes. */
for (i = 0; i < PTRS_PER_PTE; i++)
pte[i] = xen_make_pte(pte[i].pte);
}
/*
* Set up the initial kernel pagetable.
*
* We can construct this by grafting the Xen provided pagetable into
* head_64.S's preconstructed pagetables. We copy the Xen L2's into
* level2_ident_pgt, level2_kernel_pgt and level2_fixmap_pgt. This
* means that only the kernel has a physical mapping to start with -
* but that's enough to get __va working. We need to fill in the rest
* of the physical mapping once some sort of allocator has been set
* up.
*/
pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
unsigned long max_pfn)
{
pud_t *l3;
pmd_t *l2;
/* max_pfn_mapped is the last pfn mapped in the initial memory
* mappings. Considering that on Xen after the kernel mappings we
* have the mappings of some pages that don't exist in pfn space, we
* set max_pfn_mapped to the last real pfn mapped. */
max_pfn_mapped = PFN_DOWN(__pa(xen_start_info->mfn_list));
/* Zap identity mapping */
init_level4_pgt[0] = __pgd(0);
/* Pre-constructed entries are in pfn, so convert to mfn */
convert_pfn_mfn(init_level4_pgt);
convert_pfn_mfn(level3_ident_pgt);
convert_pfn_mfn(level3_kernel_pgt);
l3 = m2v(pgd[pgd_index(__START_KERNEL_map)].pgd);
l2 = m2v(l3[pud_index(__START_KERNEL_map)].pud);
memcpy(level2_ident_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
memcpy(level2_kernel_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
l3 = m2v(pgd[pgd_index(__START_KERNEL_map + PMD_SIZE)].pgd);
l2 = m2v(l3[pud_index(__START_KERNEL_map + PMD_SIZE)].pud);
memcpy(level2_fixmap_pgt, l2, sizeof(pmd_t) * PTRS_PER_PMD);
/* Set up identity map */
xen_map_identity_early(level2_ident_pgt, max_pfn);
/* Make pagetable pieces RO */
set_page_prot(init_level4_pgt, PAGE_KERNEL_RO);
set_page_prot(level3_ident_pgt, PAGE_KERNEL_RO);
set_page_prot(level3_kernel_pgt, PAGE_KERNEL_RO);
set_page_prot(level3_user_vsyscall, PAGE_KERNEL_RO);
set_page_prot(level2_kernel_pgt, PAGE_KERNEL_RO);
set_page_prot(level2_fixmap_pgt, PAGE_KERNEL_RO);
/* Pin down new L4 */
pin_pagetable_pfn(MMUEXT_PIN_L4_TABLE,
PFN_DOWN(__pa_symbol(init_level4_pgt)));
/* Unpin Xen-provided one */
pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));
/* Switch over */
pgd = init_level4_pgt;
/*
* At this stage there can be no user pgd, and no page
* structure to attach it to, so make sure we just set kernel
* pgd.
*/
xen_mc_batch();
__xen_write_cr3(true, __pa(pgd));
xen_mc_issue(PARAVIRT_LAZY_CPU);
memblock_reserve(__pa(xen_start_info->pt_base),
xen_start_info->nr_pt_frames * PAGE_SIZE);
return pgd;
}
#else /* !CONFIG_X86_64 */
static RESERVE_BRK_ARRAY(pmd_t, initial_kernel_pmd, PTRS_PER_PMD);
static RESERVE_BRK_ARRAY(pmd_t, swapper_kernel_pmd, PTRS_PER_PMD);
static void __init xen_write_cr3_init(unsigned long cr3)
{
unsigned long pfn = PFN_DOWN(__pa(swapper_pg_dir));
BUG_ON(read_cr3() != __pa(initial_page_table));
BUG_ON(cr3 != __pa(swapper_pg_dir));
/*
* We are switching to swapper_pg_dir for the first time (from
* initial_page_table) and therefore need to mark that page
* read-only and then pin it.
*
* Xen disallows sharing of kernel PMDs for PAE
* guests. Therefore we must copy the kernel PMD from
* initial_page_table into a new kernel PMD to be used in
* swapper_pg_dir.
*/
swapper_kernel_pmd =
extend_brk(sizeof(pmd_t) * PTRS_PER_PMD, PAGE_SIZE);
memcpy(swapper_kernel_pmd, initial_kernel_pmd,
sizeof(pmd_t) * PTRS_PER_PMD);
swapper_pg_dir[KERNEL_PGD_BOUNDARY] =
__pgd(__pa(swapper_kernel_pmd) | _PAGE_PRESENT);
set_page_prot(swapper_kernel_pmd, PAGE_KERNEL_RO);
set_page_prot(swapper_pg_dir, PAGE_KERNEL_RO);
xen_write_cr3(cr3);
pin_pagetable_pfn(MMUEXT_PIN_L3_TABLE, pfn);
pin_pagetable_pfn(MMUEXT_UNPIN_TABLE,
PFN_DOWN(__pa(initial_page_table)));
set_page_prot(initial_page_table, PAGE_KERNEL);
set_page_prot(initial_kernel_pmd, PAGE_KERNEL);
pv_mmu_ops.write_cr3 = &xen_write_cr3;
}
pgd_t * __init xen_setup_kernel_pagetable(pgd_t *pgd,
unsigned long max_pfn)
{
pmd_t *kernel_pmd;
initial_kernel_pmd =
extend_brk(sizeof(pmd_t) * PTRS_PER_PMD, PAGE_SIZE);
max_pfn_mapped = PFN_DOWN(__pa(xen_start_info->pt_base) +
xen_start_info->nr_pt_frames * PAGE_SIZE +
512*1024);
kernel_pmd = m2v(pgd[KERNEL_PGD_BOUNDARY].pgd);
memcpy(initial_kernel_pmd, kernel_pmd, sizeof(pmd_t) * PTRS_PER_PMD);
xen_map_identity_early(initial_kernel_pmd, max_pfn);
memcpy(initial_page_table, pgd, sizeof(pgd_t) * PTRS_PER_PGD);
initial_page_table[KERNEL_PGD_BOUNDARY] =
__pgd(__pa(initial_kernel_pmd) | _PAGE_PRESENT);
set_page_prot(initial_kernel_pmd, PAGE_KERNEL_RO);
set_page_prot(initial_page_table, PAGE_KERNEL_RO);
set_page_prot(empty_zero_page, PAGE_KERNEL_RO);
pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));
pin_pagetable_pfn(MMUEXT_PIN_L3_TABLE,
PFN_DOWN(__pa(initial_page_table)));
xen_write_cr3(__pa(initial_page_table));
memblock_reserve(__pa(xen_start_info->pt_base),
xen_start_info->nr_pt_frames * PAGE_SIZE));
return initial_page_table;
}
#endif /* CONFIG_X86_64 */
static unsigned char dummy_mapping[PAGE_SIZE] __page_aligned_bss;
static void xen_set_fixmap(unsigned idx, phys_addr_t phys, pgprot_t prot)
{
pte_t pte;
phys >>= PAGE_SHIFT;
switch (idx) {
case FIX_BTMAP_END ... FIX_BTMAP_BEGIN:
#ifdef CONFIG_X86_F00F_BUG
case FIX_F00F_IDT:
#endif
#ifdef CONFIG_X86_32
case FIX_WP_TEST:
case FIX_VDSO:
# ifdef CONFIG_HIGHMEM
case FIX_KMAP_BEGIN ... FIX_KMAP_END:
# endif
#else
case VSYSCALL_LAST_PAGE ... VSYSCALL_FIRST_PAGE:
case VVAR_PAGE:
#endif
case FIX_TEXT_POKE0:
case FIX_TEXT_POKE1:
/* All local page mappings */
pte = pfn_pte(phys, prot);
break;
#ifdef CONFIG_X86_LOCAL_APIC
case FIX_APIC_BASE: /* maps dummy local APIC */
pte = pfn_pte(PFN_DOWN(__pa(dummy_mapping)), PAGE_KERNEL);
break;
#endif
#ifdef CONFIG_X86_IO_APIC
case FIX_IO_APIC_BASE_0 ... FIX_IO_APIC_BASE_END:
/*
* We just don't map the IO APIC - all access is via
* hypercalls. Keep the address in the pte for reference.
*/
pte = pfn_pte(PFN_DOWN(__pa(dummy_mapping)), PAGE_KERNEL);
break;
#endif
case FIX_PARAVIRT_BOOTMAP:
/* This is an MFN, but it isn't an IO mapping from the
IO domain */
pte = mfn_pte(phys, prot);
break;
default:
/* By default, set_fixmap is used for hardware mappings */
pte = mfn_pte(phys, __pgprot(pgprot_val(prot) | _PAGE_IOMAP));
break;
}
__native_set_fixmap(idx, pte);
#ifdef CONFIG_X86_64
/* Replicate changes to map the vsyscall page into the user
pagetable vsyscall mapping. */
if ((idx >= VSYSCALL_LAST_PAGE && idx <= VSYSCALL_FIRST_PAGE) ||
idx == VVAR_PAGE) {
unsigned long vaddr = __fix_to_virt(idx);
set_pte_vaddr_pud(level3_user_vsyscall, vaddr, pte);
}
#endif
}
void __init xen_ident_map_ISA(void)
{
unsigned long pa;
/*
* If we're dom0, then linear map the ISA machine addresses into
* the kernel's address space.
*/
if (!xen_initial_domain())
return;
xen_raw_printk("Xen: setup ISA identity maps\n");
for (pa = ISA_START_ADDRESS; pa < ISA_END_ADDRESS; pa += PAGE_SIZE) {
pte_t pte = mfn_pte(PFN_DOWN(pa), PAGE_KERNEL_IO);
if (HYPERVISOR_update_va_mapping(PAGE_OFFSET + pa, pte, 0))
BUG();
}
xen_flush_tlb();
}
static void __init xen_post_allocator_init(void)
{
pv_mmu_ops.set_pte = xen_set_pte;
pv_mmu_ops.set_pmd = xen_set_pmd;
pv_mmu_ops.set_pud = xen_set_pud;
#if PAGETABLE_LEVELS == 4
pv_mmu_ops.set_pgd = xen_set_pgd;
#endif
/* This will work as long as patching hasn't happened yet
(which it hasn't) */
pv_mmu_ops.alloc_pte = xen_alloc_pte;
pv_mmu_ops.alloc_pmd = xen_alloc_pmd;
pv_mmu_ops.release_pte = xen_release_pte;
pv_mmu_ops.release_pmd = xen_release_pmd;
#if PAGETABLE_LEVELS == 4
pv_mmu_ops.alloc_pud = xen_alloc_pud;
pv_mmu_ops.release_pud = xen_release_pud;
#endif
#ifdef CONFIG_X86_64
SetPagePinned(virt_to_page(level3_user_vsyscall));
#endif
xen_mark_init_mm_pinned();
}
static void xen_leave_lazy_mmu(void)
{
preempt_disable();
xen_mc_flush();
paravirt_leave_lazy_mmu();
preempt_enable();
}
static const struct pv_mmu_ops xen_mmu_ops __initconst = {
.read_cr2 = xen_read_cr2,
.write_cr2 = xen_write_cr2,
.read_cr3 = xen_read_cr3,
#ifdef CONFIG_X86_32
.write_cr3 = xen_write_cr3_init,
#else
.write_cr3 = xen_write_cr3,
#endif
.flush_tlb_user = xen_flush_tlb,
.flush_tlb_kernel = xen_flush_tlb,
.flush_tlb_single = xen_flush_tlb_single,
.flush_tlb_others = xen_flush_tlb_others,
.pte_update = paravirt_nop,
.pte_update_defer = paravirt_nop,
.pgd_alloc = xen_pgd_alloc,
.pgd_free = xen_pgd_free,
.alloc_pte = xen_alloc_pte_init,
.release_pte = xen_release_pte_init,
.alloc_pmd = xen_alloc_pmd_init,
.release_pmd = xen_release_pmd_init,
.set_pte = xen_set_pte_init,
.set_pte_at = xen_set_pte_at,
.set_pmd = xen_set_pmd_hyper,
.ptep_modify_prot_start = __ptep_modify_prot_start,
.ptep_modify_prot_commit = __ptep_modify_prot_commit,
.pte_val = PV_CALLEE_SAVE(xen_pte_val),
.pgd_val = PV_CALLEE_SAVE(xen_pgd_val),
.make_pte = PV_CALLEE_SAVE(xen_make_pte),
.make_pgd = PV_CALLEE_SAVE(xen_make_pgd),
#ifdef CONFIG_X86_PAE
.set_pte_atomic = xen_set_pte_atomic,
.pte_clear = xen_pte_clear,
.pmd_clear = xen_pmd_clear,
#endif /* CONFIG_X86_PAE */
.set_pud = xen_set_pud_hyper,
.make_pmd = PV_CALLEE_SAVE(xen_make_pmd),
.pmd_val = PV_CALLEE_SAVE(xen_pmd_val),
#if PAGETABLE_LEVELS == 4
.pud_val = PV_CALLEE_SAVE(xen_pud_val),
.make_pud = PV_CALLEE_SAVE(xen_make_pud),
.set_pgd = xen_set_pgd_hyper,
.alloc_pud = xen_alloc_pmd_init,
.release_pud = xen_release_pmd_init,
#endif /* PAGETABLE_LEVELS == 4 */
.activate_mm = xen_activate_mm,
.dup_mmap = xen_dup_mmap,
.exit_mmap = xen_exit_mmap,
.lazy_mode = {
.enter = paravirt_enter_lazy_mmu,
.leave = xen_leave_lazy_mmu,
},
.set_fixmap = xen_set_fixmap,
};
void __init xen_init_mmu_ops(void)
{
x86,xen: introduce x86_init.mapping.pagetable_reserve Introduce a new x86_init hook called pagetable_reserve that at the end of init_memory_mapping is used to reserve a range of memory addresses for the kernel pagetable pages we used and free the other ones. On native it just calls memblock_x86_reserve_range while on xen it also takes care of setting the spare memory previously allocated for kernel pagetable pages from RO to RW, so that it can be used for other purposes. A detailed explanation of the reason why this hook is needed follows. As a consequence of the commit: commit 4b239f458c229de044d6905c2b0f9fe16ed9e01e Author: Yinghai Lu <yinghai@kernel.org> Date: Fri Dec 17 16:58:28 2010 -0800 x86-64, mm: Put early page table high at some point init_memory_mapping is going to reach the pagetable pages area and map those pages too (mapping them as normal memory that falls in the range of addresses passed to init_memory_mapping as argument). Some of those pages are already pagetable pages (they are in the range pgt_buf_start-pgt_buf_end) therefore they are going to be mapped RO and everything is fine. Some of these pages are not pagetable pages yet (they fall in the range pgt_buf_end-pgt_buf_top; for example the page at pgt_buf_end) so they are going to be mapped RW. When these pages become pagetable pages and are hooked into the pagetable, xen will find that the guest has already a RW mapping of them somewhere and fail the operation. The reason Xen requires pagetables to be RO is that the hypervisor needs to verify that the pagetables are valid before using them. The validation operations are called "pinning" (more details in arch/x86/xen/mmu.c). In order to fix the issue we mark all the pages in the entire range pgt_buf_start-pgt_buf_top as RO, however when the pagetable allocation is completed only the range pgt_buf_start-pgt_buf_end is reserved by init_memory_mapping. Hence the kernel is going to crash as soon as one of the pages in the range pgt_buf_end-pgt_buf_top is reused (b/c those ranges are RO). For this reason we need a hook to reserve the kernel pagetable pages we used and free the other ones so that they can be reused for other purposes. On native it just means calling memblock_x86_reserve_range, on Xen it also means marking RW the pagetable pages that we allocated before but that haven't been used before. Another way to fix this is without using the hook is by adding a 'if (xen_pv_domain)' in the 'init_memory_mapping' code and calling the Xen counterpart, but that is just nasty. Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Acked-by: Yinghai Lu <yinghai@kernel.org> Acked-by: H. Peter Anvin <hpa@zytor.com> Cc: Ingo Molnar <mingo@elte.hu> Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
2011-04-14 22:49:41 +08:00
x86_init.mapping.pagetable_reserve = xen_mapping_pagetable_reserve;
x86_init.paging.pagetable_setup_start = xen_pagetable_setup_start;
x86_init.paging.pagetable_setup_done = xen_pagetable_setup_done;
pv_mmu_ops = xen_mmu_ops;
memset(dummy_mapping, 0xff, PAGE_SIZE);
}
/* Protected by xen_reservation_lock. */
#define MAX_CONTIG_ORDER 9 /* 2MB */
static unsigned long discontig_frames[1<<MAX_CONTIG_ORDER];
#define VOID_PTE (mfn_pte(0, __pgprot(0)))
static void xen_zap_pfn_range(unsigned long vaddr, unsigned int order,
unsigned long *in_frames,
unsigned long *out_frames)
{
int i;
struct multicall_space mcs;
xen_mc_batch();
for (i = 0; i < (1UL<<order); i++, vaddr += PAGE_SIZE) {
mcs = __xen_mc_entry(0);
if (in_frames)
in_frames[i] = virt_to_mfn(vaddr);
MULTI_update_va_mapping(mcs.mc, vaddr, VOID_PTE, 0);
__set_phys_to_machine(virt_to_pfn(vaddr), INVALID_P2M_ENTRY);
if (out_frames)
out_frames[i] = virt_to_pfn(vaddr);
}
xen_mc_issue(0);
}
/*
* Update the pfn-to-mfn mappings for a virtual address range, either to
* point to an array of mfns, or contiguously from a single starting
* mfn.
*/
static void xen_remap_exchanged_ptes(unsigned long vaddr, int order,
unsigned long *mfns,
unsigned long first_mfn)
{
unsigned i, limit;
unsigned long mfn;
xen_mc_batch();
limit = 1u << order;
for (i = 0; i < limit; i++, vaddr += PAGE_SIZE) {
struct multicall_space mcs;
unsigned flags;
mcs = __xen_mc_entry(0);
if (mfns)
mfn = mfns[i];
else
mfn = first_mfn + i;
if (i < (limit - 1))
flags = 0;
else {
if (order == 0)
flags = UVMF_INVLPG | UVMF_ALL;
else
flags = UVMF_TLB_FLUSH | UVMF_ALL;
}
MULTI_update_va_mapping(mcs.mc, vaddr,
mfn_pte(mfn, PAGE_KERNEL), flags);
set_phys_to_machine(virt_to_pfn(vaddr), mfn);
}
xen_mc_issue(0);
}
/*
* Perform the hypercall to exchange a region of our pfns to point to
* memory with the required contiguous alignment. Takes the pfns as
* input, and populates mfns as output.
*
* Returns a success code indicating whether the hypervisor was able to
* satisfy the request or not.
*/
static int xen_exchange_memory(unsigned long extents_in, unsigned int order_in,
unsigned long *pfns_in,
unsigned long extents_out,
unsigned int order_out,
unsigned long *mfns_out,
unsigned int address_bits)
{
long rc;
int success;
struct xen_memory_exchange exchange = {
.in = {
.nr_extents = extents_in,
.extent_order = order_in,
.extent_start = pfns_in,
.domid = DOMID_SELF
},
.out = {
.nr_extents = extents_out,
.extent_order = order_out,
.extent_start = mfns_out,
.address_bits = address_bits,
.domid = DOMID_SELF
}
};
BUG_ON(extents_in << order_in != extents_out << order_out);
rc = HYPERVISOR_memory_op(XENMEM_exchange, &exchange);
success = (exchange.nr_exchanged == extents_in);
BUG_ON(!success && ((exchange.nr_exchanged != 0) || (rc == 0)));
BUG_ON(success && (rc != 0));
return success;
}
int xen_create_contiguous_region(unsigned long vstart, unsigned int order,
unsigned int address_bits)
{
unsigned long *in_frames = discontig_frames, out_frame;
unsigned long flags;
int success;
/*
* Currently an auto-translated guest will not perform I/O, nor will
* it require PAE page directories below 4GB. Therefore any calls to
* this function are redundant and can be ignored.
*/
if (xen_feature(XENFEAT_auto_translated_physmap))
return 0;
if (unlikely(order > MAX_CONTIG_ORDER))
return -ENOMEM;
memset((void *) vstart, 0, PAGE_SIZE << order);
spin_lock_irqsave(&xen_reservation_lock, flags);
/* 1. Zap current PTEs, remembering MFNs. */
xen_zap_pfn_range(vstart, order, in_frames, NULL);
/* 2. Get a new contiguous memory extent. */
out_frame = virt_to_pfn(vstart);
success = xen_exchange_memory(1UL << order, 0, in_frames,
1, order, &out_frame,
address_bits);
/* 3. Map the new extent in place of old pages. */
if (success)
xen_remap_exchanged_ptes(vstart, order, NULL, out_frame);
else
xen_remap_exchanged_ptes(vstart, order, in_frames, 0);
spin_unlock_irqrestore(&xen_reservation_lock, flags);
return success ? 0 : -ENOMEM;
}
EXPORT_SYMBOL_GPL(xen_create_contiguous_region);
void xen_destroy_contiguous_region(unsigned long vstart, unsigned int order)
{
unsigned long *out_frames = discontig_frames, in_frame;
unsigned long flags;
int success;
if (xen_feature(XENFEAT_auto_translated_physmap))
return;
if (unlikely(order > MAX_CONTIG_ORDER))
return;
memset((void *) vstart, 0, PAGE_SIZE << order);
spin_lock_irqsave(&xen_reservation_lock, flags);
/* 1. Find start MFN of contiguous extent. */
in_frame = virt_to_mfn(vstart);
/* 2. Zap current PTEs. */
xen_zap_pfn_range(vstart, order, NULL, out_frames);
/* 3. Do the exchange for non-contiguous MFNs. */
success = xen_exchange_memory(1, order, &in_frame, 1UL << order,
0, out_frames, 0);
/* 4. Map new pages in place of old pages. */
if (success)
xen_remap_exchanged_ptes(vstart, order, out_frames, 0);
else
xen_remap_exchanged_ptes(vstart, order, NULL, in_frame);
spin_unlock_irqrestore(&xen_reservation_lock, flags);
}
EXPORT_SYMBOL_GPL(xen_destroy_contiguous_region);
#ifdef CONFIG_XEN_PVHVM
static void xen_hvm_exit_mmap(struct mm_struct *mm)
{
struct xen_hvm_pagetable_dying a;
int rc;
a.domid = DOMID_SELF;
a.gpa = __pa(mm->pgd);
rc = HYPERVISOR_hvm_op(HVMOP_pagetable_dying, &a);
WARN_ON_ONCE(rc < 0);
}
static int is_pagetable_dying_supported(void)
{
struct xen_hvm_pagetable_dying a;
int rc = 0;
a.domid = DOMID_SELF;
a.gpa = 0x00;
rc = HYPERVISOR_hvm_op(HVMOP_pagetable_dying, &a);
if (rc < 0) {
printk(KERN_DEBUG "HVMOP_pagetable_dying not supported\n");
return 0;
}
return 1;
}
void __init xen_hvm_init_mmu_ops(void)
{
if (is_pagetable_dying_supported())
pv_mmu_ops.exit_mmap = xen_hvm_exit_mmap;
}
#endif
#define REMAP_BATCH_SIZE 16
struct remap_data {
unsigned long mfn;
pgprot_t prot;
struct mmu_update *mmu_update;
};
static int remap_area_mfn_pte_fn(pte_t *ptep, pgtable_t token,
unsigned long addr, void *data)
{
struct remap_data *rmd = data;
pte_t pte = pte_mkspecial(pfn_pte(rmd->mfn++, rmd->prot));
rmd->mmu_update->ptr = virt_to_machine(ptep).maddr;
rmd->mmu_update->val = pte_val_ma(pte);
rmd->mmu_update++;
return 0;
}
int xen_remap_domain_mfn_range(struct vm_area_struct *vma,
unsigned long addr,
unsigned long mfn, int nr,
pgprot_t prot, unsigned domid)
{
struct remap_data rmd;
struct mmu_update mmu_update[REMAP_BATCH_SIZE];
int batch;
unsigned long range;
int err = 0;
prot = __pgprot(pgprot_val(prot) | _PAGE_IOMAP);
BUG_ON(!((vma->vm_flags & (VM_PFNMAP | VM_RESERVED | VM_IO)) ==
(VM_PFNMAP | VM_RESERVED | VM_IO)));
rmd.mfn = mfn;
rmd.prot = prot;
while (nr) {
batch = min(REMAP_BATCH_SIZE, nr);
range = (unsigned long)batch << PAGE_SHIFT;
rmd.mmu_update = mmu_update;
err = apply_to_page_range(vma->vm_mm, addr, range,
remap_area_mfn_pte_fn, &rmd);
if (err)
goto out;
err = -EFAULT;
if (HYPERVISOR_mmu_update(mmu_update, batch, NULL, domid) < 0)
goto out;
nr -= batch;
addr += range;
}
err = 0;
out:
flush_tlb_all();
return err;
}
EXPORT_SYMBOL_GPL(xen_remap_domain_mfn_range);