linux/arch/powerpc/mm/pgtable.c

326 lines
8.5 KiB
C
Raw Normal View History

/*
* This file contains common routines for dealing with free of page tables
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
* Along with common page table handling code
*
* Derived from arch/powerpc/mm/tlb_64.c:
* Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
*
* Modifications by Paul Mackerras (PowerMac) (paulus@cs.anu.edu.au)
* and Cort Dougan (PReP) (cort@cs.nmt.edu)
* Copyright (C) 1996 Paul Mackerras
*
* Derived from "arch/i386/mm/init.c"
* Copyright (C) 1991, 1992, 1993, 1994 Linus Torvalds
*
* Dave Engebretsen <engebret@us.ibm.com>
* Rework for PPC64 port.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/init.h>
#include <linux/percpu.h>
#include <linux/hardirq.h>
#include <asm/pgalloc.h>
#include <asm/tlbflush.h>
#include <asm/tlb.h>
DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);
#ifdef CONFIG_SMP
/*
* Handle batching of page table freeing on SMP. Page tables are
* queued up and send to be freed later by RCU in order to avoid
* freeing a page table page that is being walked without locks
*/
static DEFINE_PER_CPU(struct pte_freelist_batch *, pte_freelist_cur);
static unsigned long pte_freelist_forced_free;
struct pte_freelist_batch
{
struct rcu_head rcu;
unsigned int index;
pgtable_free_t tables[0];
};
#define PTE_FREELIST_SIZE \
((PAGE_SIZE - sizeof(struct pte_freelist_batch)) \
/ sizeof(pgtable_free_t))
static void pte_free_smp_sync(void *arg)
{
/* Do nothing, just ensure we sync with all CPUs */
}
/* This is only called when we are critically out of memory
* (and fail to get a page in pte_free_tlb).
*/
static void pgtable_free_now(pgtable_free_t pgf)
{
pte_freelist_forced_free++;
smp_call_function(pte_free_smp_sync, NULL, 1);
pgtable_free(pgf);
}
static void pte_free_rcu_callback(struct rcu_head *head)
{
struct pte_freelist_batch *batch =
container_of(head, struct pte_freelist_batch, rcu);
unsigned int i;
for (i = 0; i < batch->index; i++)
pgtable_free(batch->tables[i]);
free_page((unsigned long)batch);
}
static void pte_free_submit(struct pte_freelist_batch *batch)
{
INIT_RCU_HEAD(&batch->rcu);
call_rcu(&batch->rcu, pte_free_rcu_callback);
}
void pgtable_free_tlb(struct mmu_gather *tlb, pgtable_free_t pgf)
{
/* This is safe since tlb_gather_mmu has disabled preemption */
struct pte_freelist_batch **batchp = &__get_cpu_var(pte_freelist_cur);
if (atomic_read(&tlb->mm->mm_users) < 2 ||
cpumask_equal(mm_cpumask(tlb->mm), cpumask_of(smp_processor_id()))){
pgtable_free(pgf);
return;
}
if (*batchp == NULL) {
*batchp = (struct pte_freelist_batch *)__get_free_page(GFP_ATOMIC);
if (*batchp == NULL) {
pgtable_free_now(pgf);
return;
}
(*batchp)->index = 0;
}
(*batchp)->tables[(*batchp)->index++] = pgf;
if ((*batchp)->index == PTE_FREELIST_SIZE) {
pte_free_submit(*batchp);
*batchp = NULL;
}
}
void pte_free_finish(void)
{
/* This is safe since tlb_gather_mmu has disabled preemption */
struct pte_freelist_batch **batchp = &__get_cpu_var(pte_freelist_cur);
if (*batchp == NULL)
return;
pte_free_submit(*batchp);
*batchp = NULL;
}
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
#endif /* CONFIG_SMP */
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
static inline int is_exec_fault(void)
{
return current->thread.regs && TRAP(current->thread.regs) == 0x400;
}
/* We only try to do i/d cache coherency on stuff that looks like
* reasonably "normal" PTEs. We currently require a PTE to be present
* and we avoid _PAGE_SPECIAL and _PAGE_NO_CACHE. We also only do that
* on userspace PTEs
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
*/
static inline int pte_looks_normal(pte_t pte)
{
return (pte_val(pte) &
(_PAGE_PRESENT | _PAGE_SPECIAL | _PAGE_NO_CACHE | _PAGE_USER)) ==
(_PAGE_PRESENT | _PAGE_USER);
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
}
struct page * maybe_pte_to_page(pte_t pte)
{
unsigned long pfn = pte_pfn(pte);
struct page *page;
if (unlikely(!pfn_valid(pfn)))
return NULL;
page = pfn_to_page(pfn);
if (PageReserved(page))
return NULL;
return page;
}
#if defined(CONFIG_PPC_STD_MMU) || _PAGE_EXEC == 0
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
/* Server-style MMU handles coherency when hashing if HW exec permission
* is supposed per page (currently 64-bit only). If not, then, we always
* flush the cache for valid PTEs in set_pte. Embedded CPU without HW exec
* support falls into the same category.
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
*/
static pte_t set_pte_filter(pte_t pte)
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
{
pte = __pte(pte_val(pte) & ~_PAGE_HPTEFLAGS);
if (pte_looks_normal(pte) && !(cpu_has_feature(CPU_FTR_COHERENT_ICACHE) ||
cpu_has_feature(CPU_FTR_NOEXECUTE))) {
struct page *pg = maybe_pte_to_page(pte);
if (!pg)
return pte;
if (!test_bit(PG_arch_1, &pg->flags)) {
flush_dcache_icache_page(pg);
set_bit(PG_arch_1, &pg->flags);
}
}
return pte;
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
}
static pte_t set_access_flags_filter(pte_t pte, struct vm_area_struct *vma,
int dirty)
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
{
return pte;
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
}
#else /* defined(CONFIG_PPC_STD_MMU) || _PAGE_EXEC == 0 */
/* Embedded type MMU with HW exec support. This is a bit more complicated
* as we don't have two bits to spare for _PAGE_EXEC and _PAGE_HWEXEC so
* instead we "filter out" the exec permission for non clean pages.
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
*/
static pte_t set_pte_filter(pte_t pte)
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
{
struct page *pg;
/* No exec permission in the first place, move on */
if (!(pte_val(pte) & _PAGE_EXEC) || !pte_looks_normal(pte))
return pte;
/* If you set _PAGE_EXEC on weird pages you're on your own */
pg = maybe_pte_to_page(pte);
if (unlikely(!pg))
return pte;
/* If the page clean, we move on */
if (test_bit(PG_arch_1, &pg->flags))
return pte;
/* If it's an exec fault, we flush the cache and make it clean */
if (is_exec_fault()) {
flush_dcache_icache_page(pg);
set_bit(PG_arch_1, &pg->flags);
return pte;
}
/* Else, we filter out _PAGE_EXEC */
return __pte(pte_val(pte) & ~_PAGE_EXEC);
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
}
static pte_t set_access_flags_filter(pte_t pte, struct vm_area_struct *vma,
int dirty)
{
struct page *pg;
/* So here, we only care about exec faults, as we use them
* to recover lost _PAGE_EXEC and perform I$/D$ coherency
* if necessary. Also if _PAGE_EXEC is already set, same deal,
* we just bail out
*/
if (dirty || (pte_val(pte) & _PAGE_EXEC) || !is_exec_fault())
return pte;
#ifdef CONFIG_DEBUG_VM
/* So this is an exec fault, _PAGE_EXEC is not set. If it was
* an error we would have bailed out earlier in do_page_fault()
* but let's make sure of it
*/
if (WARN_ON(!(vma->vm_flags & VM_EXEC)))
return pte;
#endif /* CONFIG_DEBUG_VM */
/* If you set _PAGE_EXEC on weird pages you're on your own */
pg = maybe_pte_to_page(pte);
if (unlikely(!pg))
goto bail;
/* If the page is already clean, we move on */
if (test_bit(PG_arch_1, &pg->flags))
goto bail;
/* Clean the page and set PG_arch_1 */
flush_dcache_icache_page(pg);
set_bit(PG_arch_1, &pg->flags);
bail:
return __pte(pte_val(pte) | _PAGE_EXEC);
}
#endif /* !(defined(CONFIG_PPC_STD_MMU) || _PAGE_EXEC == 0) */
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
/*
* set_pte stores a linux PTE into the linux page table.
*/
void set_pte_at(struct mm_struct *mm, unsigned long addr, pte_t *ptep,
pte_t pte)
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
{
#ifdef CONFIG_DEBUG_VM
WARN_ON(pte_present(*ptep));
#endif
/* Note: mm->context.id might not yet have been assigned as
* this context might not have been activated yet when this
* is called.
*/
pte = set_pte_filter(pte);
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
/* Perform the setting of the PTE */
__set_pte_at(mm, addr, ptep, pte, 0);
}
/*
* This is called when relaxing access to a PTE. It's also called in the page
* fault path when we don't hit any of the major fault cases, ie, a minor
* update of _PAGE_ACCESSED, _PAGE_DIRTY, etc... The generic code will have
* handled those two for us, we additionally deal with missing execute
* permission here on some processors
*/
int ptep_set_access_flags(struct vm_area_struct *vma, unsigned long address,
pte_t *ptep, pte_t entry, int dirty)
{
int changed;
entry = set_access_flags_filter(entry, vma, dirty);
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
changed = !pte_same(*(ptep), entry);
if (changed) {
powerpc: Do not assert pte_locked for hugepage PTE entries With CONFIG_DEBUG_VM, an assertion is made when changing the protection flags of a PTE that the PTE is locked. Huge pages use a different pagetable format and the assertion is bogus and will always trigger with a bug looking something like Unable to handle kernel paging request for data at address 0xf1a00235800006f8 Faulting instruction address: 0xc000000000034a80 Oops: Kernel access of bad area, sig: 11 [#1] SMP NR_CPUS=32 NUMA Maple Modules linked in: dm_snapshot dm_mirror dm_region_hash dm_log dm_mod loop evdev ext3 jbd mbcache sg sd_mod ide_pci_generic pata_amd ata_generic ipr libata tg3 libphy scsi_mod windfarm_pid windfarm_smu_sat windfarm_max6690_sensor windfarm_lm75_sensor windfarm_cpufreq_clamp windfarm_core i2c_powermac NIP: c000000000034a80 LR: c000000000034b18 CTR: 0000000000000003 REGS: c000000003037600 TRAP: 0300 Not tainted (2.6.30-rc3-autokern1) MSR: 9000000000009032 <EE,ME,IR,DR> CR: 28002484 XER: 200fffff DAR: f1a00235800006f8, DSISR: 0000000040010000 TASK = c0000002e54cc740[2960] 'map_high_trunca' THREAD: c000000003034000 CPU: 2 GPR00: 4000000000000000 c000000003037880 c000000000895d30 c0000002e5a2e500 GPR04: 00000000a0000000 c0000002edc40880 0000005700000393 0000000000000001 GPR08: f000000011ac0000 01a00235800006e8 00000000000000f5 f1a00235800006e8 GPR12: 0000000028000484 c0000000008dd780 0000000000001000 0000000000000000 GPR16: fffffffffffff000 0000000000000000 00000000a0000000 c000000003037a20 GPR20: c0000002e5f4ece8 0000000000001000 c0000002edc40880 0000000000000000 GPR24: c0000002e5f4ece8 0000000000000000 00000000a0000000 c0000002e5f4ece8 GPR28: 0000005700000393 c0000002e5a2e500 00000000a0000000 c000000003037880 NIP [c000000000034a80] .assert_pte_locked+0xa4/0xd0 LR [c000000000034b18] .ptep_set_access_flags+0x6c/0xb4 Call Trace: [c000000003037880] [c000000003037990] 0xc000000003037990 (unreliable) [c000000003037910] [c000000000034b18] .ptep_set_access_flags+0x6c/0xb4 [c0000000030379b0] [c00000000014bef8] .hugetlb_cow+0x124/0x674 [c000000003037b00] [c00000000014c930] .hugetlb_fault+0x4e8/0x6f8 [c000000003037c00] [c00000000013443c] .handle_mm_fault+0xac/0x828 [c000000003037cf0] [c0000000000340a8] .do_page_fault+0x39c/0x584 [c000000003037e30] [c0000000000057b0] handle_page_fault+0x20/0x5c Instruction dump: 7d29582a 7d200074 7800d182 0b000000 3c004000 3960ffff 780007c6 796b00c4 7d290214 7929a302 1d290068 7d6b4a14 <800b0010> 7c000074 7800d182 0b000000 This patch fixes the problem by not asseting the PTE is locked for VMAs backed by huge pages. Signed-off-by: Mel Gorman <mel@csn.ul.ie> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-04-30 10:59:19 +00:00
if (!(vma->vm_flags & VM_HUGETLB))
assert_pte_locked(vma->vm_mm, address);
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
__ptep_set_access_flags(ptep, entry);
flush_tlb_page_nohash(vma, address);
}
return changed;
}
#ifdef CONFIG_DEBUG_VM
void assert_pte_locked(struct mm_struct *mm, unsigned long addr)
{
pgd_t *pgd;
pud_t *pud;
pmd_t *pmd;
if (mm == &init_mm)
return;
pgd = mm->pgd + pgd_index(addr);
BUG_ON(pgd_none(*pgd));
pud = pud_offset(pgd, addr);
BUG_ON(pud_none(*pud));
pmd = pmd_offset(pud, addr);
BUG_ON(!pmd_present(*pmd));
assert_spin_locked(pte_lockptr(mm, pmd));
powerpc/mm: Rework I$/D$ coherency (v3) This patch reworks the way we do I and D cache coherency on PowerPC. The "old" way was split in 3 different parts depending on the processor type: - Hash with per-page exec support (64-bit and >= POWER4 only) does it at hashing time, by preventing exec on unclean pages and cleaning pages on exec faults. - Everything without per-page exec support (32-bit hash, 8xx, and 64-bit < POWER4) does it for all page going to user space in update_mmu_cache(). - Embedded with per-page exec support does it from do_page_fault() on exec faults, in a way similar to what the hash code does. That leads to confusion, and bugs. For example, the method using update_mmu_cache() is racy on SMP where another processor can see the new PTE and hash it in before we have cleaned the cache, and then blow trying to execute. This is hard to hit but I think it has bitten us in the past. Also, it's inefficient for embedded where we always end up having to do at least one more page fault. This reworks the whole thing by moving the cache sync into two main call sites, though we keep different behaviours depending on the HW capability. The call sites are set_pte_at() which is now made out of line, and ptep_set_access_flags() which joins the former in pgtable.c The base idea for Embedded with per-page exec support, is that we now do the flush at set_pte_at() time when coming from an exec fault, which allows us to avoid the double fault problem completely (we can even improve the situation more by implementing TLB preload in update_mmu_cache() but that's for later). If for some reason we didn't do it there and we try to execute, we'll hit the page fault, which will do a minor fault, which will hit ptep_set_access_flags() to do things like update _PAGE_ACCESSED or _PAGE_DIRTY if needed, we just make this guys also perform the I/D cache sync for exec faults now. This second path is the catch all for things that weren't cleaned at set_pte_at() time. For cpus without per-pag exec support, we always do the sync at set_pte_at(), thus guaranteeing that when the PTE is visible to other processors, the cache is clean. For the 64-bit hash with per-page exec support case, we keep the old mechanism for now. I'll look into changing it later, once I've reworked a bit how we use _PAGE_EXEC. This is also a first step for adding _PAGE_EXEC support for embedded platforms Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2009-02-10 16:02:37 +00:00
}
#endif /* CONFIG_DEBUG_VM */