* [PATCH 16/20] tile/mm/fault.c: Port OOM changes to handle_page_fault
@ 2012-03-20 13:26 Kautuk Consul
2012-03-20 16:09 ` Kautuk Consul
0 siblings, 1 reply; 2+ messages in thread
From: Kautuk Consul @ 2012-03-20 13:26 UTC (permalink / raw)
To: Chris Metcalf, paul.mckenney, Lucas De Marchi, Josh Triplett
Cc: linux-kernel, Kautuk Consul
Commit d065bd810b6deb67d4897a14bfe21f8eb526ba99
(mm: retry page fault when blocking on disk transfer) and
commit 37b23e0525d393d48a7d59f870b3bc061a30ccdb
(x86,mm: make pagefault killable)
The above commits introduced changes into the x86 pagefault handler
for making the page fault handler retryable as well as killable.
These changes reduce the mmap_sem hold time, which is crucial
during OOM killer invocation.
Port these changes to tile.
Signed-off-by: Kautuk Consul <consul.kautuk@gmail.com>
---
arch/tile/mm/fault.c | 31 ++++++++++++++++++++++++++-----
1 files changed, 26 insertions(+), 5 deletions(-)
diff --git a/arch/tile/mm/fault.c b/arch/tile/mm/fault.c
index c1eaaa1..a097631 100644
--- a/arch/tile/mm/fault.c
+++ b/arch/tile/mm/fault.c
@@ -267,6 +267,8 @@ static int handle_page_fault(struct pt_regs *regs,
int si_code;
int is_kernel_mode;
pgd_t *pgd;
+ unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE |
+ (write ? FAULT_FLAG_WRITE : 0);
/* on TILE, protection faults are always writes */
if (!is_page_fault)
@@ -373,6 +375,8 @@ static int handle_page_fault(struct pt_regs *regs,
vma = NULL; /* happy compiler */
goto bad_area_nosemaphore;
}
+
+retry:
down_read(&mm->mmap_sem);
}
@@ -420,7 +424,11 @@ good_area:
* make sure we exit gracefully rather than endlessly redo
* the fault.
*/
- fault = handle_mm_fault(mm, vma, address, write);
+ fault = handle_mm_fault(mm, vma, address, flags);
+
+ if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
+ return;
+
if (unlikely(fault & VM_FAULT_ERROR)) {
if (fault & VM_FAULT_OOM)
goto out_of_memory;
@@ -428,10 +436,23 @@ good_area:
goto do_sigbus;
BUG();
}
- if (fault & VM_FAULT_MAJOR)
- tsk->maj_flt++;
- else
- tsk->min_flt++;
+ if (flags & FAULT_FLAG_ALLOW_RETRY) {
+ if (fault & VM_FAULT_MAJOR)
+ tsk->maj_flt++;
+ else
+ tsk->min_flt++;
+ if (fault & VM_FAULT_RETRY) {
+ flags &= ~FAULT_FLAG_ALLOW_RETRY;
+
+ /*
+ * No need to up_read(&mm->mmap_sem) as we would
+ * have already released it in __lock_page_or_retry
+ * in mm/filemap.c.
+ */
+
+ goto retry;
+ }
+ }
#if CHIP_HAS_TILE_DMA() || CHIP_HAS_SN_PROC()
/*
--
1.7.5.4
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH 16/20] tile/mm/fault.c: Port OOM changes to handle_page_fault
2012-03-20 13:26 [PATCH 16/20] tile/mm/fault.c: Port OOM changes to handle_page_fault Kautuk Consul
@ 2012-03-20 16:09 ` Kautuk Consul
0 siblings, 0 replies; 2+ messages in thread
From: Kautuk Consul @ 2012-03-20 16:09 UTC (permalink / raw)
To: Chris Metcalf, paul.mckenney, Lucas De Marchi, Josh Triplett
Cc: linux-kernel, Kautuk Consul
hi,
There is a defect in this patch.
So, Ill post a v2 to this right away.
On Tue, Mar 20, 2012 at 9:26 AM, Kautuk Consul <consul.kautuk@gmail.com> wrote:
> Commit d065bd810b6deb67d4897a14bfe21f8eb526ba99
> (mm: retry page fault when blocking on disk transfer) and
> commit 37b23e0525d393d48a7d59f870b3bc061a30ccdb
> (x86,mm: make pagefault killable)
>
> The above commits introduced changes into the x86 pagefault handler
> for making the page fault handler retryable as well as killable.
>
> These changes reduce the mmap_sem hold time, which is crucial
> during OOM killer invocation.
>
> Port these changes to tile.
>
> Signed-off-by: Kautuk Consul <consul.kautuk@gmail.com>
> ---
> arch/tile/mm/fault.c | 31 ++++++++++++++++++++++++++-----
> 1 files changed, 26 insertions(+), 5 deletions(-)
>
> diff --git a/arch/tile/mm/fault.c b/arch/tile/mm/fault.c
> index c1eaaa1..a097631 100644
> --- a/arch/tile/mm/fault.c
> +++ b/arch/tile/mm/fault.c
> @@ -267,6 +267,8 @@ static int handle_page_fault(struct pt_regs *regs,
> int si_code;
> int is_kernel_mode;
> pgd_t *pgd;
> + unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE |
> + (write ? FAULT_FLAG_WRITE : 0);
>
> /* on TILE, protection faults are always writes */
> if (!is_page_fault)
> @@ -373,6 +375,8 @@ static int handle_page_fault(struct pt_regs *regs,
> vma = NULL; /* happy compiler */
> goto bad_area_nosemaphore;
> }
> +
> +retry:
> down_read(&mm->mmap_sem);
> }
>
> @@ -420,7 +424,11 @@ good_area:
> * make sure we exit gracefully rather than endlessly redo
> * the fault.
> */
> - fault = handle_mm_fault(mm, vma, address, write);
> + fault = handle_mm_fault(mm, vma, address, flags);
> +
> + if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
> + return;
Incorrect!
This function returns an int.
> +
> if (unlikely(fault & VM_FAULT_ERROR)) {
> if (fault & VM_FAULT_OOM)
> goto out_of_memory;
> @@ -428,10 +436,23 @@ good_area:
> goto do_sigbus;
> BUG();
> }
> - if (fault & VM_FAULT_MAJOR)
> - tsk->maj_flt++;
> - else
> - tsk->min_flt++;
> + if (flags & FAULT_FLAG_ALLOW_RETRY) {
> + if (fault & VM_FAULT_MAJOR)
> + tsk->maj_flt++;
> + else
> + tsk->min_flt++;
> + if (fault & VM_FAULT_RETRY) {
> + flags &= ~FAULT_FLAG_ALLOW_RETRY;
> +
> + /*
> + * No need to up_read(&mm->mmap_sem) as we would
> + * have already released it in __lock_page_or_retry
> + * in mm/filemap.c.
> + */
> +
> + goto retry;
> + }
> + }
>
> #if CHIP_HAS_TILE_DMA() || CHIP_HAS_SN_PROC()
> /*
> --
> 1.7.5.4
>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2012-03-20 16:09 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-03-20 13:26 [PATCH 16/20] tile/mm/fault.c: Port OOM changes to handle_page_fault Kautuk Consul
2012-03-20 16:09 ` Kautuk Consul
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).