* [PATCH 0/2] hugetlb minor improvements
@ 2014-09-26 20:44 Naoya Horiguchi
2014-09-26 20:44 ` [mmotm][PATCH 1/2] mm/hugetlb: improve suboptimal migration/hwpoisoned entry check Naoya Horiguchi
2014-09-26 20:44 ` [mmotm][PATCH 2/2] mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)() Naoya Horiguchi
0 siblings, 2 replies; 3+ messages in thread
From: Naoya Horiguchi @ 2014-09-26 20:44 UTC (permalink / raw)
To: Andrew Morton
Cc: Hugh Dickins, David Rientjes, linux-mm, linux-kernel, Naoya Horiguchi
This patchset does minor improvements and cleanups for mm/hugetlb.c.
It's based on mmotm-2014-09-25-16-28 and shows no regression in libhugetlbfs test.
Tree: git@github.com:Naoya-Horiguchi/linux.git
Branch: mmotm-2014-09-25-16-28/hugetlb_minor_improvements
---
Summary:
Naoya Horiguchi (2):
mm/hugetlb: improve suboptimal migration/hwpoisoned entry check
mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)()
mm/hugetlb.c | 60 ++++++++++++++++++++++--------------------------------------
1 file changed, 22 insertions(+), 38 deletions(-)
^ permalink raw reply [flat|nested] 3+ messages in thread
* [mmotm][PATCH 1/2] mm/hugetlb: improve suboptimal migration/hwpoisoned entry check
2014-09-26 20:44 [PATCH 0/2] hugetlb minor improvements Naoya Horiguchi
@ 2014-09-26 20:44 ` Naoya Horiguchi
2014-09-26 20:44 ` [mmotm][PATCH 2/2] mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)() Naoya Horiguchi
1 sibling, 0 replies; 3+ messages in thread
From: Naoya Horiguchi @ 2014-09-26 20:44 UTC (permalink / raw)
To: Andrew Morton
Cc: Hugh Dickins, David Rientjes, linux-mm, linux-kernel, Naoya Horiguchi
Currently hugetlb_fault() checks at first whether pte of the faulted
address is a migration or hwpoisoned entry. The reason of this approach
is that without the checks, the BUG_ON() in huge_pte_alloc() is triggered,
because it assumes that when pte is not none, it always points to a
normal hugepage, which was correct originally but not after hugetlb
supports page migration or hwpoison.
In order to iron out this weird workaround, this patch changes the
wrongly assumed BUG_ON() in huge_pte_alloc(). This allows us to check
pte_present() case only in proper place, which makes code simpler.
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
mm/hugetlb.c | 28 ++++++++++++----------------
1 file changed, 12 insertions(+), 16 deletions(-)
diff --git mmotm-2014-09-25-16-28.orig/mm/hugetlb.c mmotm-2014-09-25-16-28/mm/hugetlb.c
index 1ecb625bc498..e6543359be4d 100644
--- mmotm-2014-09-25-16-28.orig/mm/hugetlb.c
+++ mmotm-2014-09-25-16-28/mm/hugetlb.c
@@ -3130,20 +3130,10 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
struct hstate *h = hstate_vma(vma);
struct address_space *mapping;
int need_wait_lock = 0;
+ int need_wait_migration = 0;
address &= huge_page_mask(h);
- ptep = huge_pte_offset(mm, address);
- if (ptep) {
- entry = huge_ptep_get(ptep);
- if (unlikely(is_hugetlb_entry_migration(entry))) {
- migration_entry_wait_huge(vma, mm, ptep);
- return 0;
- } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry)))
- return VM_FAULT_HWPOISON_LARGE |
- VM_FAULT_SET_HINDEX(hstate_index(h));
- }
-
ptep = huge_pte_alloc(mm, address, huge_page_size(h));
if (!ptep)
return VM_FAULT_OOM;
@@ -3169,12 +3159,16 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
/*
* entry could be a migration/hwpoison entry at this point, so this
* check prevents the kernel from going below assuming that we have
- * a active hugepage in pagecache. This goto expects the 2nd page fault,
- * and is_hugetlb_entry_(migration|hwpoisoned) check will properly
- * handle it.
+ * a active hugepage in pagecache.
*/
- if (!pte_present(entry))
+ if (!pte_present(entry)) {
+ if (is_hugetlb_entry_migration(entry))
+ need_wait_migration = 1;
+ else if (is_hugetlb_entry_hwpoisoned(entry))
+ ret = VM_FAULT_HWPOISON_LARGE |
+ VM_FAULT_SET_HINDEX(hstate_index(h));
goto out_mutex;
+ }
/*
* If we are going to COW the mapping later, we examine the pending
@@ -3242,6 +3236,8 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
}
out_mutex:
mutex_unlock(&htlb_fault_mutex_table[hash]);
+ if (need_wait_migration)
+ migration_entry_wait_huge(vma, mm, ptep);
if (need_wait_lock)
wait_on_page_locked(page);
return ret;
@@ -3664,7 +3660,7 @@ pte_t *huge_pte_alloc(struct mm_struct *mm,
pte = (pte_t *)pmd_alloc(mm, pud, addr);
}
}
- BUG_ON(pte && !pte_none(*pte) && !pte_huge(*pte));
+ BUG_ON(pte && !pte_none(*pte) && pte_present(*pte) && !pte_huge(*pte));
return pte;
}
--
1.9.3
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [mmotm][PATCH 2/2] mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)()
2014-09-26 20:44 [PATCH 0/2] hugetlb minor improvements Naoya Horiguchi
2014-09-26 20:44 ` [mmotm][PATCH 1/2] mm/hugetlb: improve suboptimal migration/hwpoisoned entry check Naoya Horiguchi
@ 2014-09-26 20:44 ` Naoya Horiguchi
1 sibling, 0 replies; 3+ messages in thread
From: Naoya Horiguchi @ 2014-09-26 20:44 UTC (permalink / raw)
To: Andrew Morton
Cc: Hugh Dickins, David Rientjes, linux-mm, linux-kernel, Naoya Horiguchi
non_swap_entry() returns true if a given swp_entry_t is a migration
entry or hwpoisoned entry. So non_swap_entry() && is_migration_entry() is
identical with just is_migration_entry(). By removing non_swap_entry(),
we can write is_hugetlb_entry_(migration|hwpoisoned)() more simply.
And the name is_hugetlb_entry_(migration|hwpoisoned) is lengthy and
it's not predictable from naming convention around pte_* family.
Just pte_migration() looks better, but these function contains hugetlb
specific (so architecture dependent) huge_pte_none() check, so let's
rename them as huge_pte_(migration|hwpoisoned).
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
mm/hugetlb.c | 36 ++++++++++++------------------------
1 file changed, 12 insertions(+), 24 deletions(-)
diff --git mmotm-2014-09-25-16-28.orig/mm/hugetlb.c mmotm-2014-09-25-16-28/mm/hugetlb.c
index e6543359be4d..e70da7ae36ed 100644
--- mmotm-2014-09-25-16-28.orig/mm/hugetlb.c
+++ mmotm-2014-09-25-16-28/mm/hugetlb.c
@@ -2516,30 +2516,18 @@ static void set_huge_ptep_writable(struct vm_area_struct *vma,
update_mmu_cache(vma, address, ptep);
}
-static int is_hugetlb_entry_migration(pte_t pte)
+static inline int huge_pte_migration(pte_t pte)
{
- swp_entry_t swp;
-
if (huge_pte_none(pte) || pte_present(pte))
return 0;
- swp = pte_to_swp_entry(pte);
- if (non_swap_entry(swp) && is_migration_entry(swp))
- return 1;
- else
- return 0;
+ return is_migration_entry(pte_to_swp_entry(pte));
}
-static int is_hugetlb_entry_hwpoisoned(pte_t pte)
+static inline int huge_pte_hwpoisoned(pte_t pte)
{
- swp_entry_t swp;
-
if (huge_pte_none(pte) || pte_present(pte))
return 0;
- swp = pte_to_swp_entry(pte);
- if (non_swap_entry(swp) && is_hwpoison_entry(swp))
- return 1;
- else
- return 0;
+ return is_hwpoison_entry(pte_to_swp_entry(pte));
}
int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
@@ -2583,8 +2571,8 @@ int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
entry = huge_ptep_get(src_pte);
if (huge_pte_none(entry)) { /* skip none entry */
;
- } else if (unlikely(is_hugetlb_entry_migration(entry) ||
- is_hugetlb_entry_hwpoisoned(entry))) {
+ } else if (unlikely(huge_pte_migration(entry) ||
+ huge_pte_hwpoisoned(entry))) {
swp_entry_t swp_entry = pte_to_swp_entry(entry);
if (is_write_migration_entry(swp_entry) && cow) {
@@ -3162,9 +3150,9 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
* a active hugepage in pagecache.
*/
if (!pte_present(entry)) {
- if (is_hugetlb_entry_migration(entry))
+ if (huge_pte_migration(entry))
need_wait_migration = 1;
- else if (is_hugetlb_entry_hwpoisoned(entry))
+ else if (huge_pte_hwpoisoned(entry))
ret = VM_FAULT_HWPOISON_LARGE |
VM_FAULT_SET_HINDEX(hstate_index(h));
goto out_mutex;
@@ -3291,8 +3279,8 @@ long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
* (in which case hugetlb_fault waits for the migration,) and
* hwpoisoned hugepages (in which case we need to prevent the
* caller from accessing to them.) In order to do this, we use
- * here is_swap_pte instead of is_hugetlb_entry_migration and
- * is_hugetlb_entry_hwpoisoned. This is because it simply covers
+ * here is_swap_pte instead of huge_pte_migration and
+ * huge_pte_hwpoisoned. This is because it simply covers
* both cases, and because we can't follow correct pages
* directly from any kind of swap entries.
*/
@@ -3370,11 +3358,11 @@ unsigned long hugetlb_change_protection(struct vm_area_struct *vma,
continue;
}
pte = huge_ptep_get(ptep);
- if (unlikely(is_hugetlb_entry_hwpoisoned(pte))) {
+ if (unlikely(huge_pte_hwpoisoned(pte))) {
spin_unlock(ptl);
continue;
}
- if (unlikely(is_hugetlb_entry_migration(pte))) {
+ if (unlikely(huge_pte_migration(pte))) {
swp_entry_t entry = pte_to_swp_entry(pte);
if (is_write_migration_entry(entry)) {
--
1.9.3
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2014-09-26 20:44 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-09-26 20:44 [PATCH 0/2] hugetlb minor improvements Naoya Horiguchi
2014-09-26 20:44 ` [mmotm][PATCH 1/2] mm/hugetlb: improve suboptimal migration/hwpoisoned entry check Naoya Horiguchi
2014-09-26 20:44 ` [mmotm][PATCH 2/2] mm/hugetlb: cleanup and rename is_hugetlb_entry_(migration|hwpoisoned)() Naoya Horiguchi
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).