* [PATCH 0/2] KVM: enable asynchronous page faults @ 2011-02-01 11:21 ` Gleb Natapov 0 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-01 11:21 UTC (permalink / raw) To: avi, mtosatti; +Cc: linux-mm, linux-kernel This patch series is on top of Huang Ying "Replace is_hwpoison_address with __get_user_pages" series: http://www.mail-archive.com/kvm@vger.kernel.org/msg48776.html Gleb Natapov (2): Allow GUP to fail instead of waiting on a page. KVM: Enable async page fault processing. include/linux/mm.h | 2 ++ mm/filemap.c | 6 ++++-- mm/memory.c | 5 ++++- virt/kvm/kvm_main.c | 23 +++++++++++++++++++++-- 4 files changed, 31 insertions(+), 5 deletions(-) ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH 0/2] KVM: enable asynchronous page faults @ 2011-02-01 11:21 ` Gleb Natapov 0 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-01 11:21 UTC (permalink / raw) To: avi, mtosatti; +Cc: linux-mm, linux-kernel This patch series is on top of Huang Ying "Replace is_hwpoison_address with __get_user_pages" series: http://www.mail-archive.com/kvm@vger.kernel.org/msg48776.html Gleb Natapov (2): Allow GUP to fail instead of waiting on a page. KVM: Enable async page fault processing. include/linux/mm.h | 2 ++ mm/filemap.c | 6 ++++-- mm/memory.c | 5 ++++- virt/kvm/kvm_main.c | 23 +++++++++++++++++++++-- 4 files changed, 31 insertions(+), 5 deletions(-) -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH 1/2] Allow GUP to fail instead of waiting on a page. 2011-02-01 11:21 ` Gleb Natapov @ 2011-02-01 11:21 ` Gleb Natapov -1 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-01 11:21 UTC (permalink / raw) To: avi, mtosatti Cc: linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins, Andrew Morton GUP user may want to try to acquire a reference to a page if it is already in memory, but not if IO, to bring it in, is needed. For example KVM may tell vcpu to schedule another guest process if current one is trying to access swapped out page. Meanwhile, the page will be swapped in and the guest process, that depends on it, will be able to run again. This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY instead. Signed-off-by: Gleb Natapov <gleb@redhat.com> CC: Linus Torvalds <torvalds@linux-foundation.org> CC: Rik van Riel <riel@redhat.com> CC: Hugh Dickins <hughd@google.com> CC: Andrew Morton <akpm@linux-foundation.org> --- include/linux/mm.h | 2 ++ mm/filemap.c | 6 ++++-- mm/memory.c | 5 ++++- 3 files changed, 10 insertions(+), 3 deletions(-) diff --git a/include/linux/mm.h b/include/linux/mm.h index 9c3a7c8..fa5e562 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -151,6 +151,7 @@ extern pgprot_t protection_map[16]; #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ +#define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ /* * This interface is used by x86 PAT code to identify a pfn mapping that is @@ -1532,6 +1533,7 @@ struct page *follow_page(struct vm_area_struct *, unsigned long address, #define FOLL_GET 0x04 /* do get_page on page */ #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ +#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ #define FOLL_MLOCK 0x40 /* mark page as mlocked */ #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ diff --git a/mm/filemap.c b/mm/filemap.c index 83a45d3..312b6eb 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -621,8 +621,10 @@ int __lock_page_or_retry(struct page *page, struct mm_struct *mm, __lock_page(page); return 1; } else { - up_read(&mm->mmap_sem); - wait_on_page_locked(page); + if (!(flags & FAULT_FLAG_RETRY_NOWAIT)) { + up_read(&mm->mmap_sem); + wait_on_page_locked(page); + } return 0; } } diff --git a/mm/memory.c b/mm/memory.c index e7c9a99..a52a3e6 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -1569,6 +1569,8 @@ int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, fault_flags |= FAULT_FLAG_WRITE; if (nonblocking) fault_flags |= FAULT_FLAG_ALLOW_RETRY; + if (foll_flags & FOLL_NOWAIT) + fault_flags |= (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT); ret = handle_mm_fault(mm, vma, start, fault_flags); @@ -1595,7 +1597,8 @@ int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, tsk->min_flt++; if (ret & VM_FAULT_RETRY) { - *nonblocking = 0; + if (nonblocking) + *nonblocking = 0; return i; } -- 1.7.1 ^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PATCH 1/2] Allow GUP to fail instead of waiting on a page. @ 2011-02-01 11:21 ` Gleb Natapov 0 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-01 11:21 UTC (permalink / raw) To: avi, mtosatti Cc: linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins, Andrew Morton GUP user may want to try to acquire a reference to a page if it is already in memory, but not if IO, to bring it in, is needed. For example KVM may tell vcpu to schedule another guest process if current one is trying to access swapped out page. Meanwhile, the page will be swapped in and the guest process, that depends on it, will be able to run again. This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY instead. Signed-off-by: Gleb Natapov <gleb@redhat.com> CC: Linus Torvalds <torvalds@linux-foundation.org> CC: Rik van Riel <riel@redhat.com> CC: Hugh Dickins <hughd@google.com> CC: Andrew Morton <akpm@linux-foundation.org> --- include/linux/mm.h | 2 ++ mm/filemap.c | 6 ++++-- mm/memory.c | 5 ++++- 3 files changed, 10 insertions(+), 3 deletions(-) diff --git a/include/linux/mm.h b/include/linux/mm.h index 9c3a7c8..fa5e562 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -151,6 +151,7 @@ extern pgprot_t protection_map[16]; #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ +#define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ /* * This interface is used by x86 PAT code to identify a pfn mapping that is @@ -1532,6 +1533,7 @@ struct page *follow_page(struct vm_area_struct *, unsigned long address, #define FOLL_GET 0x04 /* do get_page on page */ #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ +#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ #define FOLL_MLOCK 0x40 /* mark page as mlocked */ #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ diff --git a/mm/filemap.c b/mm/filemap.c index 83a45d3..312b6eb 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -621,8 +621,10 @@ int __lock_page_or_retry(struct page *page, struct mm_struct *mm, __lock_page(page); return 1; } else { - up_read(&mm->mmap_sem); - wait_on_page_locked(page); + if (!(flags & FAULT_FLAG_RETRY_NOWAIT)) { + up_read(&mm->mmap_sem); + wait_on_page_locked(page); + } return 0; } } diff --git a/mm/memory.c b/mm/memory.c index e7c9a99..a52a3e6 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -1569,6 +1569,8 @@ int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, fault_flags |= FAULT_FLAG_WRITE; if (nonblocking) fault_flags |= FAULT_FLAG_ALLOW_RETRY; + if (foll_flags & FOLL_NOWAIT) + fault_flags |= (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_RETRY_NOWAIT); ret = handle_mm_fault(mm, vma, start, fault_flags); @@ -1595,7 +1597,8 @@ int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, tsk->min_flt++; if (ret & VM_FAULT_RETRY) { - *nonblocking = 0; + if (nonblocking) + *nonblocking = 0; return i; } -- 1.7.1 -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. 2011-02-01 11:21 ` Gleb Natapov @ 2011-02-01 16:48 ` Rik van Riel -1 siblings, 0 replies; 20+ messages in thread From: Rik van Riel @ 2011-02-01 16:48 UTC (permalink / raw) To: Gleb Natapov Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Hugh Dickins, Andrew Morton On 02/01/2011 06:21 AM, Gleb Natapov wrote: > GUP user may want to try to acquire a reference to a page if it is already > in memory, but not if IO, to bring it in, is needed. For example KVM may > tell vcpu to schedule another guest process if current one is trying to > access swapped out page. Meanwhile, the page will be swapped in and the > guest process, that depends on it, will be able to run again. > > This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and > FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in > conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that > it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY > instead. > > Signed-off-by: Gleb Natapov<gleb@redhat.com> > CC: Linus Torvalds<torvalds@linux-foundation.org> > CC: Rik van Riel<riel@redhat.com> > CC: Hugh Dickins<hughd@google.com> > CC: Andrew Morton<akpm@linux-foundation.org> Acked-by: Rik van Riel <riel@redhat.com> -- All rights reversed ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. @ 2011-02-01 16:48 ` Rik van Riel 0 siblings, 0 replies; 20+ messages in thread From: Rik van Riel @ 2011-02-01 16:48 UTC (permalink / raw) To: Gleb Natapov Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Hugh Dickins, Andrew Morton On 02/01/2011 06:21 AM, Gleb Natapov wrote: > GUP user may want to try to acquire a reference to a page if it is already > in memory, but not if IO, to bring it in, is needed. For example KVM may > tell vcpu to schedule another guest process if current one is trying to > access swapped out page. Meanwhile, the page will be swapped in and the > guest process, that depends on it, will be able to run again. > > This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and > FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in > conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that > it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY > instead. > > Signed-off-by: Gleb Natapov<gleb@redhat.com> > CC: Linus Torvalds<torvalds@linux-foundation.org> > CC: Rik van Riel<riel@redhat.com> > CC: Hugh Dickins<hughd@google.com> > CC: Andrew Morton<akpm@linux-foundation.org> Acked-by: Rik van Riel <riel@redhat.com> -- All rights reversed -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. 2011-02-01 11:21 ` Gleb Natapov @ 2011-02-02 0:42 ` Andrew Morton -1 siblings, 0 replies; 20+ messages in thread From: Andrew Morton @ 2011-02-02 0:42 UTC (permalink / raw) To: Gleb Natapov Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins On Tue, 1 Feb 2011 13:21:46 +0200 Gleb Natapov <gleb@redhat.com> wrote: > GUP user may want to try to acquire a reference to a page if it is already > in memory, but not if IO, to bring it in, is needed. For example KVM may > tell vcpu to schedule another guest process if current one is trying to > access swapped out page. Meanwhile, the page will be swapped in and the > guest process, that depends on it, will be able to run again. > > This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and > FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in > conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that > it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY > instead. > > ... > > +#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ The comment is a little misleading. Or incomplete. For both swap-backed and file-backed pages, the code will initiate the disk transfer and will then return without waiting for it to complete. This (important!) information isn't really presented in either the changelog or the code itself. This? --- a/include/linux/mm.h~mm-allow-gup-to-fail-instead-of-waiting-on-a-page-fix +++ a/include/linux/mm.h @@ -1537,7 +1537,8 @@ struct page *follow_page(struct vm_area_ #define FOLL_GET 0x04 /* do get_page on page */ #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ -#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ +#define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO + * and return without waiting upon it */ #define FOLL_MLOCK 0x40 /* mark page as mlocked */ #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ _ ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. @ 2011-02-02 0:42 ` Andrew Morton 0 siblings, 0 replies; 20+ messages in thread From: Andrew Morton @ 2011-02-02 0:42 UTC (permalink / raw) To: Gleb Natapov Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins On Tue, 1 Feb 2011 13:21:46 +0200 Gleb Natapov <gleb@redhat.com> wrote: > GUP user may want to try to acquire a reference to a page if it is already > in memory, but not if IO, to bring it in, is needed. For example KVM may > tell vcpu to schedule another guest process if current one is trying to > access swapped out page. Meanwhile, the page will be swapped in and the > guest process, that depends on it, will be able to run again. > > This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and > FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in > conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that > it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY > instead. > > ... > > +#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ The comment is a little misleading. Or incomplete. For both swap-backed and file-backed pages, the code will initiate the disk transfer and will then return without waiting for it to complete. This (important!) information isn't really presented in either the changelog or the code itself. This? --- a/include/linux/mm.h~mm-allow-gup-to-fail-instead-of-waiting-on-a-page-fix +++ a/include/linux/mm.h @@ -1537,7 +1537,8 @@ struct page *follow_page(struct vm_area_ #define FOLL_GET 0x04 /* do get_page on page */ #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ -#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ +#define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO + * and return without waiting upon it */ #define FOLL_MLOCK 0x40 /* mark page as mlocked */ #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ _ -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. 2011-02-02 0:42 ` Andrew Morton @ 2011-02-02 13:31 ` Gleb Natapov -1 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-02 13:31 UTC (permalink / raw) To: Andrew Morton Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins On Tue, Feb 01, 2011 at 04:42:40PM -0800, Andrew Morton wrote: > On Tue, 1 Feb 2011 13:21:46 +0200 > Gleb Natapov <gleb@redhat.com> wrote: > > > GUP user may want to try to acquire a reference to a page if it is already > > in memory, but not if IO, to bring it in, is needed. For example KVM may > > tell vcpu to schedule another guest process if current one is trying to > > access swapped out page. Meanwhile, the page will be swapped in and the > > guest process, that depends on it, will be able to run again. > > > > This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and > > FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in > > conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that > > it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY > > instead. > > > > ... > > > > +#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ > > The comment is a little misleading. Or incomplete. > > For both swap-backed and file-backed pages, the code will initiate the > disk transfer and will then return without waiting for it to complete. > This (important!) information isn't really presented in either the > changelog or the code itself. > > This? > Yes, this is better. Thanks you. I see that the patch below is in your queue already. Should I re-spin my patch with improved comment anyway? > --- a/include/linux/mm.h~mm-allow-gup-to-fail-instead-of-waiting-on-a-page-fix > +++ a/include/linux/mm.h > @@ -1537,7 +1537,8 @@ struct page *follow_page(struct vm_area_ > #define FOLL_GET 0x04 /* do get_page on page */ > #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ > #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ > -#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ > +#define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO > + * and return without waiting upon it */ > #define FOLL_MLOCK 0x40 /* mark page as mlocked */ > #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ > > _ -- Gleb. ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. @ 2011-02-02 13:31 ` Gleb Natapov 0 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-02 13:31 UTC (permalink / raw) To: Andrew Morton Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins On Tue, Feb 01, 2011 at 04:42:40PM -0800, Andrew Morton wrote: > On Tue, 1 Feb 2011 13:21:46 +0200 > Gleb Natapov <gleb@redhat.com> wrote: > > > GUP user may want to try to acquire a reference to a page if it is already > > in memory, but not if IO, to bring it in, is needed. For example KVM may > > tell vcpu to schedule another guest process if current one is trying to > > access swapped out page. Meanwhile, the page will be swapped in and the > > guest process, that depends on it, will be able to run again. > > > > This patch adds FAULT_FLAG_RETRY_NOWAIT (suggested by Linus) and > > FOLL_NOWAIT follow_page flags. FAULT_FLAG_RETRY_NOWAIT, when used in > > conjunction with VM_FAULT_ALLOW_RETRY, indicates to handle_mm_fault that > > it shouldn't drop mmap_sem and wait on a page, but return VM_FAULT_RETRY > > instead. > > > > ... > > > > +#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ > > The comment is a little misleading. Or incomplete. > > For both swap-backed and file-backed pages, the code will initiate the > disk transfer and will then return without waiting for it to complete. > This (important!) information isn't really presented in either the > changelog or the code itself. > > This? > Yes, this is better. Thanks you. I see that the patch below is in your queue already. Should I re-spin my patch with improved comment anyway? > --- a/include/linux/mm.h~mm-allow-gup-to-fail-instead-of-waiting-on-a-page-fix > +++ a/include/linux/mm.h > @@ -1537,7 +1537,8 @@ struct page *follow_page(struct vm_area_ > #define FOLL_GET 0x04 /* do get_page on page */ > #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ > #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ > -#define FOLL_NOWAIT 0x20 /* return if disk transfer is needed */ > +#define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO > + * and return without waiting upon it */ > #define FOLL_MLOCK 0x40 /* mark page as mlocked */ > #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ > > _ -- Gleb. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. 2011-02-02 13:31 ` Gleb Natapov @ 2011-02-02 19:37 ` Andrew Morton -1 siblings, 0 replies; 20+ messages in thread From: Andrew Morton @ 2011-02-02 19:37 UTC (permalink / raw) To: Gleb Natapov Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins On Wed, 2 Feb 2011 15:31:57 +0200 Gleb Natapov <gleb@redhat.com> wrote: > > This? > > > Yes, this is better. Thanks you. I see that the patch below is in your queue > already. Should I re-spin my patch with improved comment anyway? Nope, that's OK - I fold fixup patches into the base patch before sending them onwards. There's always a risk that someone will get a hold of an earlier version of the patch, but a) sending out a v2 doesn't eliminate that risk and b) it's not very important anyway (in this case) and c) because I separate the base patch from the fixup patches, I'll easily notice if someone merges an earlier patch, because I'm left holding stray fixup patches. ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] Allow GUP to fail instead of waiting on a page. @ 2011-02-02 19:37 ` Andrew Morton 0 siblings, 0 replies; 20+ messages in thread From: Andrew Morton @ 2011-02-02 19:37 UTC (permalink / raw) To: Gleb Natapov Cc: avi, mtosatti, linux-mm, linux-kernel, Linus Torvalds, Rik van Riel, Hugh Dickins On Wed, 2 Feb 2011 15:31:57 +0200 Gleb Natapov <gleb@redhat.com> wrote: > > This? > > > Yes, this is better. Thanks you. I see that the patch below is in your queue > already. Should I re-spin my patch with improved comment anyway? Nope, that's OK - I fold fixup patches into the base patch before sending them onwards. There's always a risk that someone will get a hold of an earlier version of the patch, but a) sending out a v2 doesn't eliminate that risk and b) it's not very important anyway (in this case) and c) because I separate the base patch from the fixup patches, I'll easily notice if someone merges an earlier patch, because I'm left holding stray fixup patches. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH 2/2] KVM: Enable async page fault processing. 2011-02-01 11:21 ` Gleb Natapov @ 2011-02-01 11:21 ` Gleb Natapov -1 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-01 11:21 UTC (permalink / raw) To: avi, mtosatti; +Cc: linux-mm, linux-kernel If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to avoid sleeping on IO. Check for hwpoison is done at the same time, otherwise check_user_page_hwpoison() will call GUP again and will put vcpu to sleep. Signed-off-by: Gleb Natapov <gleb@redhat.com> --- virt/kvm/kvm_main.c | 23 +++++++++++++++++++++-- 1 files changed, 21 insertions(+), 2 deletions(-) diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 74d032a..80f42ab 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -1029,6 +1029,17 @@ static pfn_t get_fault_pfn(void) return fault_pfn; } +int get_user_page_nowait(struct task_struct *tsk, struct mm_struct *mm, + unsigned long start, int write, struct page **page) +{ + int flags = FOLL_TOUCH | FOLL_NOWAIT | FOLL_HWPOISON | FOLL_GET; + + if (write) + flags |= FOLL_WRITE; + + return __get_user_pages(tsk, mm, start, 1, flags, page, NULL, NULL); +} + static inline int check_user_page_hwpoison(unsigned long addr) { int rc, flags = FOLL_TOUCH | FOLL_HWPOISON | FOLL_WRITE; @@ -1062,7 +1073,14 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, if (writable) *writable = write_fault; - npages = get_user_pages_fast(addr, 1, write_fault, page); + if (async) { + down_read(¤t->mm->mmap_sem); + npages = get_user_page_nowait(current, current->mm, + addr, write_fault, page); + up_read(¤t->mm->mmap_sem); + } else + npages = get_user_pages_fast(addr, 1, write_fault, + page); /* map read fault as writable if possible */ if (unlikely(!write_fault) && npages == 1) { @@ -1085,7 +1103,8 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, return get_fault_pfn(); down_read(¤t->mm->mmap_sem); - if (check_user_page_hwpoison(addr)) { + if (npages == -EHWPOISON || + (!async && check_user_page_hwpoison(addr))) { up_read(¤t->mm->mmap_sem); get_page(hwpoison_page); return page_to_pfn(hwpoison_page); -- 1.7.1 ^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PATCH 2/2] KVM: Enable async page fault processing. @ 2011-02-01 11:21 ` Gleb Natapov 0 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-02-01 11:21 UTC (permalink / raw) To: avi, mtosatti; +Cc: linux-mm, linux-kernel If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to avoid sleeping on IO. Check for hwpoison is done at the same time, otherwise check_user_page_hwpoison() will call GUP again and will put vcpu to sleep. Signed-off-by: Gleb Natapov <gleb@redhat.com> --- virt/kvm/kvm_main.c | 23 +++++++++++++++++++++-- 1 files changed, 21 insertions(+), 2 deletions(-) diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 74d032a..80f42ab 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -1029,6 +1029,17 @@ static pfn_t get_fault_pfn(void) return fault_pfn; } +int get_user_page_nowait(struct task_struct *tsk, struct mm_struct *mm, + unsigned long start, int write, struct page **page) +{ + int flags = FOLL_TOUCH | FOLL_NOWAIT | FOLL_HWPOISON | FOLL_GET; + + if (write) + flags |= FOLL_WRITE; + + return __get_user_pages(tsk, mm, start, 1, flags, page, NULL, NULL); +} + static inline int check_user_page_hwpoison(unsigned long addr) { int rc, flags = FOLL_TOUCH | FOLL_HWPOISON | FOLL_WRITE; @@ -1062,7 +1073,14 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, if (writable) *writable = write_fault; - npages = get_user_pages_fast(addr, 1, write_fault, page); + if (async) { + down_read(¤t->mm->mmap_sem); + npages = get_user_page_nowait(current, current->mm, + addr, write_fault, page); + up_read(¤t->mm->mmap_sem); + } else + npages = get_user_pages_fast(addr, 1, write_fault, + page); /* map read fault as writable if possible */ if (unlikely(!write_fault) && npages == 1) { @@ -1085,7 +1103,8 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, return get_fault_pfn(); down_read(¤t->mm->mmap_sem); - if (check_user_page_hwpoison(addr)) { + if (npages == -EHWPOISON || + (!async && check_user_page_hwpoison(addr))) { up_read(¤t->mm->mmap_sem); get_page(hwpoison_page); return page_to_pfn(hwpoison_page); -- 1.7.1 -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] KVM: Enable async page fault processing. 2011-02-01 11:21 ` Gleb Natapov @ 2011-02-22 8:38 ` Lai Jiangshan -1 siblings, 0 replies; 20+ messages in thread From: Lai Jiangshan @ 2011-02-22 8:38 UTC (permalink / raw) To: Gleb Natapov, avi; +Cc: mtosatti, linux-mm, linux-kernel On 01/-9/-28163 03:59 AM, Gleb Natapov wrote: > If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to > avoid sleeping on IO. Check for hwpoison is done at the same time, > otherwise check_user_page_hwpoison() will call GUP again and will put > vcpu to sleep. > > Signed-off-by: Gleb Natapov <gleb@redhat.com> > --- Acked-by: Lai Jiangshan <laijs@cn.fujitsu.com> ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] KVM: Enable async page fault processing. @ 2011-02-22 8:38 ` Lai Jiangshan 0 siblings, 0 replies; 20+ messages in thread From: Lai Jiangshan @ 2011-02-22 8:38 UTC (permalink / raw) To: Gleb Natapov, avi; +Cc: mtosatti, linux-mm, linux-kernel On 01/-9/-28163 03:59 AM, Gleb Natapov wrote: > If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to > avoid sleeping on IO. Check for hwpoison is done at the same time, > otherwise check_user_page_hwpoison() will call GUP again and will put > vcpu to sleep. > > Signed-off-by: Gleb Natapov <gleb@redhat.com> > --- Acked-by: Lai Jiangshan <laijs@cn.fujitsu.com> -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] KVM: Enable async page fault processing. 2011-02-01 11:21 ` Gleb Natapov @ 2011-03-24 12:22 ` Gleb Natapov -1 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-03-24 12:22 UTC (permalink / raw) To: avi, mtosatti; +Cc: linux-mm, linux-kernel On Tue, Feb 01, 2011 at 01:21:47PM +0200, Gleb Natapov wrote: > If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to > avoid sleeping on IO. Check for hwpoison is done at the same time, > otherwise check_user_page_hwpoison() will call GUP again and will put > vcpu to sleep. > FOLL_NOWAIT is now in Linus tree, so this patch can be applied now. I verified that it still applies and works. > Signed-off-by: Gleb Natapov <gleb@redhat.com> > --- > virt/kvm/kvm_main.c | 23 +++++++++++++++++++++-- > 1 files changed, 21 insertions(+), 2 deletions(-) > > diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c > index 74d032a..80f42ab 100644 > --- a/virt/kvm/kvm_main.c > +++ b/virt/kvm/kvm_main.c > @@ -1029,6 +1029,17 @@ static pfn_t get_fault_pfn(void) > return fault_pfn; > } > > +int get_user_page_nowait(struct task_struct *tsk, struct mm_struct *mm, > + unsigned long start, int write, struct page **page) > +{ > + int flags = FOLL_TOUCH | FOLL_NOWAIT | FOLL_HWPOISON | FOLL_GET; > + > + if (write) > + flags |= FOLL_WRITE; > + > + return __get_user_pages(tsk, mm, start, 1, flags, page, NULL, NULL); > +} > + > static inline int check_user_page_hwpoison(unsigned long addr) > { > int rc, flags = FOLL_TOUCH | FOLL_HWPOISON | FOLL_WRITE; > @@ -1062,7 +1073,14 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, > if (writable) > *writable = write_fault; > > - npages = get_user_pages_fast(addr, 1, write_fault, page); > + if (async) { > + down_read(¤t->mm->mmap_sem); > + npages = get_user_page_nowait(current, current->mm, > + addr, write_fault, page); > + up_read(¤t->mm->mmap_sem); > + } else > + npages = get_user_pages_fast(addr, 1, write_fault, > + page); > > /* map read fault as writable if possible */ > if (unlikely(!write_fault) && npages == 1) { > @@ -1085,7 +1103,8 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, > return get_fault_pfn(); > > down_read(¤t->mm->mmap_sem); > - if (check_user_page_hwpoison(addr)) { > + if (npages == -EHWPOISON || > + (!async && check_user_page_hwpoison(addr))) { > up_read(¤t->mm->mmap_sem); > get_page(hwpoison_page); > return page_to_pfn(hwpoison_page); > -- > 1.7.1 > > -- > To unsubscribe, send a message with 'unsubscribe linux-mm' in > the body to majordomo@kvack.org. For more info on Linux MM, > see: http://www.linux-mm.org/ . > Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ > Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> -- Gleb. ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] KVM: Enable async page fault processing. @ 2011-03-24 12:22 ` Gleb Natapov 0 siblings, 0 replies; 20+ messages in thread From: Gleb Natapov @ 2011-03-24 12:22 UTC (permalink / raw) To: avi, mtosatti; +Cc: linux-mm, linux-kernel On Tue, Feb 01, 2011 at 01:21:47PM +0200, Gleb Natapov wrote: > If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to > avoid sleeping on IO. Check for hwpoison is done at the same time, > otherwise check_user_page_hwpoison() will call GUP again and will put > vcpu to sleep. > FOLL_NOWAIT is now in Linus tree, so this patch can be applied now. I verified that it still applies and works. > Signed-off-by: Gleb Natapov <gleb@redhat.com> > --- > virt/kvm/kvm_main.c | 23 +++++++++++++++++++++-- > 1 files changed, 21 insertions(+), 2 deletions(-) > > diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c > index 74d032a..80f42ab 100644 > --- a/virt/kvm/kvm_main.c > +++ b/virt/kvm/kvm_main.c > @@ -1029,6 +1029,17 @@ static pfn_t get_fault_pfn(void) > return fault_pfn; > } > > +int get_user_page_nowait(struct task_struct *tsk, struct mm_struct *mm, > + unsigned long start, int write, struct page **page) > +{ > + int flags = FOLL_TOUCH | FOLL_NOWAIT | FOLL_HWPOISON | FOLL_GET; > + > + if (write) > + flags |= FOLL_WRITE; > + > + return __get_user_pages(tsk, mm, start, 1, flags, page, NULL, NULL); > +} > + > static inline int check_user_page_hwpoison(unsigned long addr) > { > int rc, flags = FOLL_TOUCH | FOLL_HWPOISON | FOLL_WRITE; > @@ -1062,7 +1073,14 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, > if (writable) > *writable = write_fault; > > - npages = get_user_pages_fast(addr, 1, write_fault, page); > + if (async) { > + down_read(¤t->mm->mmap_sem); > + npages = get_user_page_nowait(current, current->mm, > + addr, write_fault, page); > + up_read(¤t->mm->mmap_sem); > + } else > + npages = get_user_pages_fast(addr, 1, write_fault, > + page); > > /* map read fault as writable if possible */ > if (unlikely(!write_fault) && npages == 1) { > @@ -1085,7 +1103,8 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic, > return get_fault_pfn(); > > down_read(¤t->mm->mmap_sem); > - if (check_user_page_hwpoison(addr)) { > + if (npages == -EHWPOISON || > + (!async && check_user_page_hwpoison(addr))) { > up_read(¤t->mm->mmap_sem); > get_page(hwpoison_page); > return page_to_pfn(hwpoison_page); > -- > 1.7.1 > > -- > To unsubscribe, send a message with 'unsubscribe linux-mm' in > the body to majordomo@kvack.org. For more info on Linux MM, > see: http://www.linux-mm.org/ . > Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ > Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> -- Gleb. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] KVM: Enable async page fault processing. 2011-03-24 12:22 ` Gleb Natapov @ 2011-03-24 13:23 ` Avi Kivity -1 siblings, 0 replies; 20+ messages in thread From: Avi Kivity @ 2011-03-24 13:23 UTC (permalink / raw) To: Gleb Natapov; +Cc: mtosatti, linux-mm, linux-kernel On 03/24/2011 02:22 PM, Gleb Natapov wrote: > On Tue, Feb 01, 2011 at 01:21:47PM +0200, Gleb Natapov wrote: > > If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to > > avoid sleeping on IO. Check for hwpoison is done at the same time, > > otherwise check_user_page_hwpoison() will call GUP again and will put > > vcpu to sleep. > > > FOLL_NOWAIT is now in Linus tree, so this patch can be applied now. I > verified that it still applies and works. Thanks, applied and queued for 2.6.39. -- error compiling committee.c: too many arguments to function ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] KVM: Enable async page fault processing. @ 2011-03-24 13:23 ` Avi Kivity 0 siblings, 0 replies; 20+ messages in thread From: Avi Kivity @ 2011-03-24 13:23 UTC (permalink / raw) To: Gleb Natapov; +Cc: mtosatti, linux-mm, linux-kernel On 03/24/2011 02:22 PM, Gleb Natapov wrote: > On Tue, Feb 01, 2011 at 01:21:47PM +0200, Gleb Natapov wrote: > > If asynchronous hva_to_pfn() is requested call GUP with FOLL_NOWAIT to > > avoid sleeping on IO. Check for hwpoison is done at the same time, > > otherwise check_user_page_hwpoison() will call GUP again and will put > > vcpu to sleep. > > > FOLL_NOWAIT is now in Linus tree, so this patch can be applied now. I > verified that it still applies and works. Thanks, applied and queued for 2.6.39. -- error compiling committee.c: too many arguments to function -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a> ^ permalink raw reply [flat|nested] 20+ messages in thread
end of thread, other threads:[~2011-03-24 13:23 UTC | newest] Thread overview: 20+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2011-02-01 11:21 [PATCH 0/2] KVM: enable asynchronous page faults Gleb Natapov 2011-02-01 11:21 ` Gleb Natapov 2011-02-01 11:21 ` [PATCH 1/2] Allow GUP to fail instead of waiting on a page Gleb Natapov 2011-02-01 11:21 ` Gleb Natapov 2011-02-01 16:48 ` Rik van Riel 2011-02-01 16:48 ` Rik van Riel 2011-02-02 0:42 ` Andrew Morton 2011-02-02 0:42 ` Andrew Morton 2011-02-02 13:31 ` Gleb Natapov 2011-02-02 13:31 ` Gleb Natapov 2011-02-02 19:37 ` Andrew Morton 2011-02-02 19:37 ` Andrew Morton 2011-02-01 11:21 ` [PATCH 2/2] KVM: Enable async page fault processing Gleb Natapov 2011-02-01 11:21 ` Gleb Natapov 2011-02-22 8:38 ` Lai Jiangshan 2011-02-22 8:38 ` Lai Jiangshan 2011-03-24 12:22 ` Gleb Natapov 2011-03-24 12:22 ` Gleb Natapov 2011-03-24 13:23 ` Avi Kivity 2011-03-24 13:23 ` Avi Kivity
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.