Message ID | 1523975611-15978-7-git-send-email-ldufour@linux.vnet.ibm.com (mailing list archive) |
---|---|
State | Superseded, archived |
Headers | show |
Series | Speculative page faults | expand |
On Tue, Apr 17, 2018 at 04:33:12PM +0200, Laurent Dufour wrote: > pte_unmap_same() is making the assumption that the page table are still > around because the mmap_sem is held. > This is no more the case when running a speculative page fault and > additional check must be made to ensure that the final page table are still > there. > > This is now done by calling pte_spinlock() to check for the VMA's > consistency while locking for the page tables. > > This is requiring passing a vm_fault structure to pte_unmap_same() which is > containing all the needed parameters. > > As pte_spinlock() may fail in the case of a speculative page fault, if the > VMA has been touched in our back, pte_unmap_same() should now return 3 > cases : > 1. pte are the same (0) > 2. pte are different (VM_FAULT_PTNOTSAME) > 3. a VMA's changes has been detected (VM_FAULT_RETRY) > > The case 2 is handled by the introduction of a new VM_FAULT flag named > VM_FAULT_PTNOTSAME which is then trapped in cow_user_page(). I don't see such logic in this patch. Maybe you introduces it later? If so, please comment on it. Or just return 0 in case of 2 without introducing VM_FAULT_PTNOTSAME. > If VM_FAULT_RETRY is returned, it is passed up to the callers to retry the > page fault while holding the mmap_sem. > > Acked-by: David Rientjes <rientjes@google.com> > Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com> > --- > include/linux/mm.h | 1 + > mm/memory.c | 39 ++++++++++++++++++++++++++++----------- > 2 files changed, 29 insertions(+), 11 deletions(-) > > diff --git a/include/linux/mm.h b/include/linux/mm.h > index 4d1aff80669c..714da99d77a3 100644 > --- a/include/linux/mm.h > +++ b/include/linux/mm.h > @@ -1208,6 +1208,7 @@ static inline void clear_page_pfmemalloc(struct page *page) > #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables > * and needs fsync() to complete (for > * synchronous page faults in DAX) */ > +#define VM_FAULT_PTNOTSAME 0x4000 /* Page table entries have changed */ > > #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ > VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \ > diff --git a/mm/memory.c b/mm/memory.c > index 0b9a51f80e0e..f86efcb8e268 100644 > --- a/mm/memory.c > +++ b/mm/memory.c > @@ -2309,21 +2309,29 @@ static inline bool pte_map_lock(struct vm_fault *vmf) > * parts, do_swap_page must check under lock before unmapping the pte and > * proceeding (but do_wp_page is only called after already making such a check; > * and do_anonymous_page can safely check later on). > + * > + * pte_unmap_same() returns: > + * 0 if the PTE are the same > + * VM_FAULT_PTNOTSAME if the PTE are different > + * VM_FAULT_RETRY if the VMA has changed in our back during > + * a speculative page fault handling. > */ > -static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd, > - pte_t *page_table, pte_t orig_pte) > +static inline int pte_unmap_same(struct vm_fault *vmf) > { > - int same = 1; > + int ret = 0; > + > #if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT) > if (sizeof(pte_t) > sizeof(unsigned long)) { > - spinlock_t *ptl = pte_lockptr(mm, pmd); > - spin_lock(ptl); > - same = pte_same(*page_table, orig_pte); > - spin_unlock(ptl); > + if (pte_spinlock(vmf)) { > + if (!pte_same(*vmf->pte, vmf->orig_pte)) > + ret = VM_FAULT_PTNOTSAME; > + spin_unlock(vmf->ptl); > + } else > + ret = VM_FAULT_RETRY; > } > #endif > - pte_unmap(page_table); > - return same; > + pte_unmap(vmf->pte); > + return ret; > } > > static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma) > @@ -2912,10 +2920,19 @@ int do_swap_page(struct vm_fault *vmf) > pte_t pte; > int locked; > int exclusive = 0; > - int ret = 0; > + int ret; > > - if (!pte_unmap_same(vma->vm_mm, vmf->pmd, vmf->pte, vmf->orig_pte)) > + ret = pte_unmap_same(vmf); > + if (ret) { > + /* > + * If pte != orig_pte, this means another thread did the > + * swap operation in our back. > + * So nothing else to do. > + */ > + if (ret == VM_FAULT_PTNOTSAME) > + ret = 0; > goto out; > + } > > entry = pte_to_swp_entry(vmf->orig_pte); > if (unlikely(non_swap_entry(entry))) { > -- > 2.7.4 >
On 23/04/2018 08:31, Minchan Kim wrote: > On Tue, Apr 17, 2018 at 04:33:12PM +0200, Laurent Dufour wrote: >> pte_unmap_same() is making the assumption that the page table are still >> around because the mmap_sem is held. >> This is no more the case when running a speculative page fault and >> additional check must be made to ensure that the final page table are still >> there. >> >> This is now done by calling pte_spinlock() to check for the VMA's >> consistency while locking for the page tables. >> >> This is requiring passing a vm_fault structure to pte_unmap_same() which is >> containing all the needed parameters. >> >> As pte_spinlock() may fail in the case of a speculative page fault, if the >> VMA has been touched in our back, pte_unmap_same() should now return 3 >> cases : >> 1. pte are the same (0) >> 2. pte are different (VM_FAULT_PTNOTSAME) >> 3. a VMA's changes has been detected (VM_FAULT_RETRY) >> >> The case 2 is handled by the introduction of a new VM_FAULT flag named >> VM_FAULT_PTNOTSAME which is then trapped in cow_user_page(). > > I don't see such logic in this patch. > Maybe you introduces it later? If so, please comment on it. > Or just return 0 in case of 2 without introducing VM_FAULT_PTNOTSAME. Late in the series, pte_spinlock() will check for the VMA's changes and may return 1. This will be then required to handle the 3 cases presented above. I can move this handling later in the series, but I wondering if this will make it more easier to read. > >> If VM_FAULT_RETRY is returned, it is passed up to the callers to retry the >> page fault while holding the mmap_sem. >> >> Acked-by: David Rientjes <rientjes@google.com> >> Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com> >> --- >> include/linux/mm.h | 1 + >> mm/memory.c | 39 ++++++++++++++++++++++++++++----------- >> 2 files changed, 29 insertions(+), 11 deletions(-) >> >> diff --git a/include/linux/mm.h b/include/linux/mm.h >> index 4d1aff80669c..714da99d77a3 100644 >> --- a/include/linux/mm.h >> +++ b/include/linux/mm.h >> @@ -1208,6 +1208,7 @@ static inline void clear_page_pfmemalloc(struct page *page) >> #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables >> * and needs fsync() to complete (for >> * synchronous page faults in DAX) */ >> +#define VM_FAULT_PTNOTSAME 0x4000 /* Page table entries have changed */ >> >> #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ >> VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \ >> diff --git a/mm/memory.c b/mm/memory.c >> index 0b9a51f80e0e..f86efcb8e268 100644 >> --- a/mm/memory.c >> +++ b/mm/memory.c >> @@ -2309,21 +2309,29 @@ static inline bool pte_map_lock(struct vm_fault *vmf) >> * parts, do_swap_page must check under lock before unmapping the pte and >> * proceeding (but do_wp_page is only called after already making such a check; >> * and do_anonymous_page can safely check later on). >> + * >> + * pte_unmap_same() returns: >> + * 0 if the PTE are the same >> + * VM_FAULT_PTNOTSAME if the PTE are different >> + * VM_FAULT_RETRY if the VMA has changed in our back during >> + * a speculative page fault handling. >> */ >> -static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd, >> - pte_t *page_table, pte_t orig_pte) >> +static inline int pte_unmap_same(struct vm_fault *vmf) >> { >> - int same = 1; >> + int ret = 0; >> + >> #if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT) >> if (sizeof(pte_t) > sizeof(unsigned long)) { >> - spinlock_t *ptl = pte_lockptr(mm, pmd); >> - spin_lock(ptl); >> - same = pte_same(*page_table, orig_pte); >> - spin_unlock(ptl); >> + if (pte_spinlock(vmf)) { >> + if (!pte_same(*vmf->pte, vmf->orig_pte)) >> + ret = VM_FAULT_PTNOTSAME; >> + spin_unlock(vmf->ptl); >> + } else >> + ret = VM_FAULT_RETRY; >> } >> #endif >> - pte_unmap(page_table); >> - return same; >> + pte_unmap(vmf->pte); >> + return ret; >> } >> >> static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma) >> @@ -2912,10 +2920,19 @@ int do_swap_page(struct vm_fault *vmf) >> pte_t pte; >> int locked; >> int exclusive = 0; >> - int ret = 0; >> + int ret; >> >> - if (!pte_unmap_same(vma->vm_mm, vmf->pmd, vmf->pte, vmf->orig_pte)) >> + ret = pte_unmap_same(vmf); >> + if (ret) { >> + /* >> + * If pte != orig_pte, this means another thread did the >> + * swap operation in our back. >> + * So nothing else to do. >> + */ >> + if (ret == VM_FAULT_PTNOTSAME) >> + ret = 0; >> goto out; >> + } >> >> entry = pte_to_swp_entry(vmf->orig_pte); >> if (unlikely(non_swap_entry(entry))) { >> -- >> 2.7.4 >> >
On Mon, Apr 30, 2018 at 04:07:30PM +0200, Laurent Dufour wrote: > On 23/04/2018 08:31, Minchan Kim wrote: > > On Tue, Apr 17, 2018 at 04:33:12PM +0200, Laurent Dufour wrote: > >> pte_unmap_same() is making the assumption that the page table are still > >> around because the mmap_sem is held. > >> This is no more the case when running a speculative page fault and > >> additional check must be made to ensure that the final page table are still > >> there. > >> > >> This is now done by calling pte_spinlock() to check for the VMA's > >> consistency while locking for the page tables. > >> > >> This is requiring passing a vm_fault structure to pte_unmap_same() which is > >> containing all the needed parameters. > >> > >> As pte_spinlock() may fail in the case of a speculative page fault, if the > >> VMA has been touched in our back, pte_unmap_same() should now return 3 > >> cases : > >> 1. pte are the same (0) > >> 2. pte are different (VM_FAULT_PTNOTSAME) > >> 3. a VMA's changes has been detected (VM_FAULT_RETRY) > >> > >> The case 2 is handled by the introduction of a new VM_FAULT flag named > >> VM_FAULT_PTNOTSAME which is then trapped in cow_user_page(). > > > > I don't see such logic in this patch. > > Maybe you introduces it later? If so, please comment on it. > > Or just return 0 in case of 2 without introducing VM_FAULT_PTNOTSAME. > > Late in the series, pte_spinlock() will check for the VMA's changes and may > return 1. This will be then required to handle the 3 cases presented above. > > I can move this handling later in the series, but I wondering if this will make > it more easier to read. Just nit: During reviewing this patch, I was just curious you introduced new thing here but I couldn't find any site where use it. It makes review hard. :( That's why I said to you that please commet on it if you will use new thing late in this series. If you think as-is is better for review, it would be better to mention it explicitly. Thanks.
On Tue, Apr 17, 2018 at 8:03 PM, Laurent Dufour <ldufour@linux.vnet.ibm.com> wrote: > pte_unmap_same() is making the assumption that the page table are still > around because the mmap_sem is held. > This is no more the case when running a speculative page fault and > additional check must be made to ensure that the final page table are still > there. > > This is now done by calling pte_spinlock() to check for the VMA's > consistency while locking for the page tables. > > This is requiring passing a vm_fault structure to pte_unmap_same() which is > containing all the needed parameters. > > As pte_spinlock() may fail in the case of a speculative page fault, if the > VMA has been touched in our back, pte_unmap_same() should now return 3 > cases : > 1. pte are the same (0) > 2. pte are different (VM_FAULT_PTNOTSAME) > 3. a VMA's changes has been detected (VM_FAULT_RETRY) > > The case 2 is handled by the introduction of a new VM_FAULT flag named > VM_FAULT_PTNOTSAME which is then trapped in cow_user_page(). > If VM_FAULT_RETRY is returned, it is passed up to the callers to retry the > page fault while holding the mmap_sem. > > Acked-by: David Rientjes <rientjes@google.com> > Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com> > --- > include/linux/mm.h | 1 + > mm/memory.c | 39 ++++++++++++++++++++++++++++----------- > 2 files changed, 29 insertions(+), 11 deletions(-) > > diff --git a/include/linux/mm.h b/include/linux/mm.h > index 4d1aff80669c..714da99d77a3 100644 > --- a/include/linux/mm.h > +++ b/include/linux/mm.h > @@ -1208,6 +1208,7 @@ static inline void clear_page_pfmemalloc(struct page *page) > #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables > * and needs fsync() to complete (for > * synchronous page faults in DAX) */ > +#define VM_FAULT_PTNOTSAME 0x4000 /* Page table entries have changed */ This has to be added to VM_FAULT_RESULT_TRACE ?
On 10/05/2018 18:15, vinayak menon wrote: > On Tue, Apr 17, 2018 at 8:03 PM, Laurent Dufour > <ldufour@linux.vnet.ibm.com> wrote: >> pte_unmap_same() is making the assumption that the page table are still >> around because the mmap_sem is held. >> This is no more the case when running a speculative page fault and >> additional check must be made to ensure that the final page table are still >> there. >> >> This is now done by calling pte_spinlock() to check for the VMA's >> consistency while locking for the page tables. >> >> This is requiring passing a vm_fault structure to pte_unmap_same() which is >> containing all the needed parameters. >> >> As pte_spinlock() may fail in the case of a speculative page fault, if the >> VMA has been touched in our back, pte_unmap_same() should now return 3 >> cases : >> 1. pte are the same (0) >> 2. pte are different (VM_FAULT_PTNOTSAME) >> 3. a VMA's changes has been detected (VM_FAULT_RETRY) >> >> The case 2 is handled by the introduction of a new VM_FAULT flag named >> VM_FAULT_PTNOTSAME which is then trapped in cow_user_page(). >> If VM_FAULT_RETRY is returned, it is passed up to the callers to retry the >> page fault while holding the mmap_sem. >> >> Acked-by: David Rientjes <rientjes@google.com> >> Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com> >> --- >> include/linux/mm.h | 1 + >> mm/memory.c | 39 ++++++++++++++++++++++++++++----------- >> 2 files changed, 29 insertions(+), 11 deletions(-) >> >> diff --git a/include/linux/mm.h b/include/linux/mm.h >> index 4d1aff80669c..714da99d77a3 100644 >> --- a/include/linux/mm.h >> +++ b/include/linux/mm.h >> @@ -1208,6 +1208,7 @@ static inline void clear_page_pfmemalloc(struct page *page) >> #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables >> * and needs fsync() to complete (for >> * synchronous page faults in DAX) */ >> +#define VM_FAULT_PTNOTSAME 0x4000 /* Page table entries have changed */ > > > This has to be added to VM_FAULT_RESULT_TRACE ? Indeed there is no chance that the macro VM_FAULT_RESULT_TRACE would have to translate that code to a string since VM_FAULT_PTNOTSAME is currently only returned by pte_unmap_same() and then converted by its only caller do_swap_page() to return 0. So VM_FAULT_PTNOTSAME is not expected to be seen outside of these services which are never using VM_FAULT_RESULT_TRACE(). This being said, this may be a good idea to add it in the case of future potential usage.
diff --git a/include/linux/mm.h b/include/linux/mm.h index 4d1aff80669c..714da99d77a3 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -1208,6 +1208,7 @@ static inline void clear_page_pfmemalloc(struct page *page) #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables * and needs fsync() to complete (for * synchronous page faults in DAX) */ +#define VM_FAULT_PTNOTSAME 0x4000 /* Page table entries have changed */ #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \ diff --git a/mm/memory.c b/mm/memory.c index 0b9a51f80e0e..f86efcb8e268 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -2309,21 +2309,29 @@ static inline bool pte_map_lock(struct vm_fault *vmf) * parts, do_swap_page must check under lock before unmapping the pte and * proceeding (but do_wp_page is only called after already making such a check; * and do_anonymous_page can safely check later on). + * + * pte_unmap_same() returns: + * 0 if the PTE are the same + * VM_FAULT_PTNOTSAME if the PTE are different + * VM_FAULT_RETRY if the VMA has changed in our back during + * a speculative page fault handling. */ -static inline int pte_unmap_same(struct mm_struct *mm, pmd_t *pmd, - pte_t *page_table, pte_t orig_pte) +static inline int pte_unmap_same(struct vm_fault *vmf) { - int same = 1; + int ret = 0; + #if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT) if (sizeof(pte_t) > sizeof(unsigned long)) { - spinlock_t *ptl = pte_lockptr(mm, pmd); - spin_lock(ptl); - same = pte_same(*page_table, orig_pte); - spin_unlock(ptl); + if (pte_spinlock(vmf)) { + if (!pte_same(*vmf->pte, vmf->orig_pte)) + ret = VM_FAULT_PTNOTSAME; + spin_unlock(vmf->ptl); + } else + ret = VM_FAULT_RETRY; } #endif - pte_unmap(page_table); - return same; + pte_unmap(vmf->pte); + return ret; } static inline void cow_user_page(struct page *dst, struct page *src, unsigned long va, struct vm_area_struct *vma) @@ -2912,10 +2920,19 @@ int do_swap_page(struct vm_fault *vmf) pte_t pte; int locked; int exclusive = 0; - int ret = 0; + int ret; - if (!pte_unmap_same(vma->vm_mm, vmf->pmd, vmf->pte, vmf->orig_pte)) + ret = pte_unmap_same(vmf); + if (ret) { + /* + * If pte != orig_pte, this means another thread did the + * swap operation in our back. + * So nothing else to do. + */ + if (ret == VM_FAULT_PTNOTSAME) + ret = 0; goto out; + } entry = pte_to_swp_entry(vmf->orig_pte); if (unlikely(non_swap_entry(entry))) {