sparc/mm/fault_32.c: Port OOM changes to do_sparc_fault
PATCH v3 19/20 sparc/mm/fault_32.c: Port OOM changes to do_sparc_fault
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 32-bit sparc.
--- arch/sparc/mm/fault_32.c | 37 ++++++++++++++++++++++++++++++------- 1 files changed, 30 insertions(+), 7 deletions(-)
arch/sparc/mm/fault_32.c +30 -7
@@ -226,6 +226,8 @@ asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,unsigned long g2;int from_user = !(regs->psr &PSR_PS );int fault, code; +unsigned int flags = (FAULT_FLAG_ALLOW_RETRY |FAULT_FLAG_KILLABLE | + (write ?FAULT_FLAG_WRITE :0 ));if (text_fault) address = regs->pc;
@@ -252,6 +254,7 @@ asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS ,1 , regs, address); +retry:down_read (&mm->mmap_sem);/*
@@ -290,7 +293,11 @@ good_area:* make sure we exit gracefully rather than endlessly redo * the fault. */ - fault =handle_mm_fault (mm, vma, address, write ?FAULT_FLAG_WRITE :0 ); + 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;
@@ -298,13 +305,29 @@ good_area: goto do_sigbus;BUG (); } -if (fault &VM_FAULT_MAJOR ) { - current->maj_flt++; -perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MAJ ,1 , regs, address); - }else { - current->min_flt++; -perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MIN ,1 , regs, address); + +if (flags &FAULT_FLAG_ALLOW_RETRY ) { +if (fault &VM_FAULT_MAJOR ) { + current->maj_flt++; +perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MAJ , +1 , regs, address); + }else { + current->min_flt++; +perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MIN , +1 , regs, address); + } +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; + } } +up_read (&mm->mmap_sem);return ;
Signature · 1 line
1.7.5.4
On Mon, 2012-03-26 at 11:32 -0400, Kautuk Consul wrote:
[]
arch/sparc/mm/fault_32.c asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write, · line 230
Quoted 10 lines
arch/sparc/mm/fault_32.c +2
-0@@ -226,6 +226,8 @@ asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,unsigned long g2;int from_user = !(regs->psr &PSR_PS );int fault, code; +unsigned int flags = (FAULT_FLAG_ALLOW_RETRY |FAULT_FLAG_KILLABLE | + (write ?FAULT_FLAG_WRITE :0 ));
David meant that the indentation should be aligned immediately after the open parenthesis:
+ unsigned int flags = (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE |
+ (write ? FAULT_FLAG_WRITE : 0));
though perhaps this is clearer as:
unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE; if (write) flags |= FAULT_FLAG_WRITE;
or even separate sets.
if (write) flags = foo; else flags = bar;
Quoted 13 lines
David meant that the indentation should be aligned immediately after the open parenthesis:
+ unsigned int flags = (
FAULT_FLAG_ALLOW_RETRY|FAULT_FLAG_KILLABLE|
+ (write ?FAULT_FLAG_WRITE: 0));though perhaps this is clearer as:
unsigned int flags =
FAULT_FLAG_ALLOW_RETRY|FAULT_FLAG_KILLABLE;if (write)
flags |=FAULT_FLAG_WRITE;
Quoted 6 lines
or even separate sets.
if (write)
flags = foo;
else
flags = bar;
This is how it is in the patch file(immediately after the paramthesis). I have now attached both sparc patches to this email.
you must be having the same problem in viewing the patch file
as me due to the email client itself inserting invalid spaces here and there.
by the way, what tab-spaces are you using when you open it in vim ?
From a95d00b2293d41a5afa193e2045c3d8bb81bbc4a Mon Sep 17 00:00:00 2001
From: Kautuk Consul <[email protected]>
Date: Sun, 25 Mar 2012 13:50:30 -0400
Subject: [PATCH 19/20 v3] sparc/mm/fault_32.c: Port OOM changes to do_sparc_fault
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 32-bit sparc.
--- arch/sparc/mm/fault_32.c | 37 ++++++++++++++++++++++++++++++------- 1 files changed, 30 insertions(+), 7 deletions(-)
arch/sparc/mm/fault_32.c +30 -7
@@ -226,6 +226,8 @@ asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,unsigned long g2;int from_user = !(regs->psr &PSR_PS );int fault, code; +unsigned int flags = (FAULT_FLAG_ALLOW_RETRY |FAULT_FLAG_KILLABLE | + (write ?FAULT_FLAG_WRITE :0 ));if (text_fault) address = regs->pc;
@@ -252,6 +254,7 @@ asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS ,1 , regs, address); +retry:down_read (&mm->mmap_sem);/*
@@ -290,7 +293,11 @@ good_area:* make sure we exit gracefully rather than endlessly redo * the fault. */ - fault =handle_mm_fault (mm, vma, address, write ?FAULT_FLAG_WRITE :0 ); + 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;
@@ -298,13 +305,29 @@ good_area: goto do_sigbus;BUG (); } -if (fault &VM_FAULT_MAJOR ) { - current->maj_flt++; -perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MAJ ,1 , regs, address); - }else { - current->min_flt++; -perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MIN ,1 , regs, address); + +if (flags &FAULT_FLAG_ALLOW_RETRY ) { +if (fault &VM_FAULT_MAJOR ) { + current->maj_flt++; +perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MAJ , +1 , regs, address); + }else { + current->min_flt++; +perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MIN , +1 , regs, address); + } +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; + } } +up_read (&mm->mmap_sem);return ;
Signature · 1 line
1.7.5.4
From 3f4a3166b5baff9d043fd39f0f0c2c8044c85cce Mon Sep 17 00:00:00 2001
From: Kautuk Consul <[email protected]>
Date: Mon, 26 Mar 2012 11:28:50 -0400
Subject: [PATCH 20/20 v3] sparc/mm/fault_64.c: Port OOM changes to
do_sparc64_fault
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 64-bit sparc.
--- arch/sparc/mm/fault_64.c | 38 +++++++++++++++++++++++++++++++------- 1 files changed, 31 insertions(+), 7 deletions(-)
arch/sparc/mm/fault_64.c +31 -7
@@ -279,6 +279,7 @@ asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs)unsigned int insn =0 ;int si_code, fault_code, fault;unsigned long address, mm_rss; +unsigned int flags =FAULT_FLAG_ALLOW_RETRY |FAULT_FLAG_KILLABLE ; fault_code =get_thread_fault_code ();
@@ -333,6 +334,8 @@ asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs) insn =get_fault_insn (regs, insn); goto handle_kernel_fault; } + +retry:down_read (&mm->mmap_sem); }
@@ -423,7 +426,13 @@ good_area: goto bad_area; } - fault =handle_mm_fault (mm, vma, address, (fault_code &FAULT_CODE_WRITE ) ?FAULT_FLAG_WRITE :0 ); + fault =handle_mm_fault (mm, vma, address, + flags | ((fault_code &FAULT_CODE_WRITE ) ? + FAULT_FLAG_WRITE : 0); + +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;
@@ -431,12 +440,27 @@ good_area: goto do_sigbus;BUG (); } -if (fault &VM_FAULT_MAJOR ) { - current->maj_flt++; -perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MAJ ,1 , regs, address); - }else { - current->min_flt++; -perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MIN ,1 , regs, address); + +if (flags &FAULT_FLAG_ALLOW_RETRY ) { +if (fault &VM_FAULT_MAJOR ) { + current->maj_flt++; +perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MAJ , +1 , regs, address); + }else { + current->min_flt++; +perf_sw_event (PERF_COUNT_SW_PAGE_FAULTS_MIN , +1 , regs, address); + } +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; + } }up_read (&mm->mmap_sem);
Signature · 1 line
1.7.5.4
Quoted 20 lines · Kautuk Consul
Quoted 13 lines
David meant that the indentation should be aligned immediately after the open parenthesis:
arch/sparc/mm/fault_32.c +2
-0+unsigned int flags = (FAULT_FLAG_ALLOW_RETRY |FAULT_FLAG_KILLABLE | + (write ?FAULT_FLAG_WRITE :0 ));though perhaps this is clearer as:
unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE; if (write) flags |= FAULT_FLAG_WRITE;Quoted 6 lines
or even separate sets.
if (write) flags = foo; else flags = bar;
Quoted 5 lines
This is how it is in the patch file(immediately after the paramthesis). I have now attached both sparc patches to this email.
you must be having the same problem in viewing the patch file
as me due to the email client itself inserting invalid spaces here and there.
Nope. That's not it.
Quoted 1 line
by the way, what tab-spaces are you using when you open it in vim ?
I use vim as an editor as infrequently as possible.
Your original patch used 6 tabs and 2 spaces.
Tab size in kernel style is always 8.
Again, make sure your tab indentation is set to 8.
Quoted 6 lines
Quoted 3 lines
you must be having the same problem in viewing the patch file
as me due to the email client itself inserting invalid spaces here and there.Nope. That's not it.
Quoted 9 lines
Quoted 1 line
by the way, what tab-spaces are you using when you open it in vim ?
I use vim as an editor as infrequently as possible.
Your original patch used 6 tabs and 2 spaces.
Tab size in kernel style is always 8.Again, make sure your tab indentation is set to 8.
I see what you are saying.
One question: Is it possible to include this kinda checking (multiline statements with paranthesis checking assuming tab indentation 8) in the
checkpatch.pl script ?
It would be easier to not make such mistakes in future with that kind of tool.
Quoted 5 lines · Kautuk Consul
Is it possible to include this kinda checking
(multiline statements with paranthesis checking assuming tab indentation 8) in the
checkpatch.pl script ?
It would be easier to not make such mistakes in future with that kind of tool.
Funny you should ask...
Use the latest version of checkpatch and
add the --strict option to the command line.
https://lkml.org/lkml/2012/3/2/15
It works for most cases.
Quoted 7 lines · Joe Perches
Quoted 5 lines · Kautuk Consul
Is it possible to include this kinda checking
(multiline statements with paranthesis checking assuming tab indentation 8) in the
checkpatch.pl script ?
It would be easier to not make such mistakes in future with that kind of tool.
Quoted 9 lines
Funny you should ask...
Use the latest version of checkpatch and
add the --strict option to the command line.https://lkml.org/lkml/2012/3/2/15
It works for most cases.
Ah yes, I am not in the habit of using the --strict option. I will use that in future, sorry for my ignorance.
And thanks to all you guys for helping me out here. :)
Quoted 23 lines · Kautuk Consul
Commit
d065bd810b6deb67d4897a14bfe21f8eb526ba99(mm: retry page fault when blocking on disk transfer) and commit37b23e0525d393d48a7d59f870b3bc061a30ccdb(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_semhold time, which is crucial during OOM killer invocation.Port these changes to 32-bit sparc.
Signed-off-by Kautuk Consul <[email protected]>--- arch/sparc/mm/fault_32.c | 37 ++++++++++++++++++++++++++++++------- 1 files changed, 30 insertions(+), 7 deletions(-)arch/sparc/mm/fault_32.c +0
-0
arch/sparc/mm/fault_32.c asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write, · line 230
Quoted 5 lines
arch/sparc/mm/fault_32.c +2
-0@@ -226,6 +226,8 @@ asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write,unsigned long g2;int from_user = !(regs->psr &PSR_PS );int fault, code; +unsigned int flags = (FAULT_FLAG_ALLOW_RETRY |FAULT_FLAG_KILLABLE | + (write ?FAULT_FLAG_WRITE :0 ));
Hi Kautuk.
The indention still look like shit.
It looks like this (tabs converted to spaces):
unsigned int flags = (FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE |
(write ? FAULT_FLAG_WRITE : 0));
Sam
David Miller
maintainerFrom: Kautuk Consul <[email protected]>
Date: Mon, 26 Mar 2012 11:32:26 -0400
arch/sparc/mm/fault_32.c asmlinkage void do_sparc_fault(struct pt_regs *regs, int text_fault, int write, · line 230
Quoted 2 lines
arch/sparc/mm/fault_32.c +2
-0+unsigned int flags = (FAULT_FLAG_ALLOW_RETRY |FAULT_FLAG_KILLABLE | + (write ?FAULT_FLAG_WRITE :0 ));
Indentation is still wrong. The "(write ..." on the second line must line up with the column of the first "F" in "(FAULT_FLAG..." on the first line.
+ perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ,
+ 1, regs, address);
+ perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN,
+ 1, regs, address);
These are still wrong too, for the same reason.
You didn't fix anything.