Skip to contentSkip to conversation

Swap delay accounting, include lock_page() delays

PATCH In review Balbir Singh 2 participants 4 messages Oct 31, 2007

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Swap delay accounting, include lock_page() delays

The delay incurred in lock_page() should also be accounted in swap delay accounting

---

 mm/memory.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff -puN mm/swapfile.c~fix-delay-accounting-swap-accounting mm/swapfile.c

mm/memory.c renamed from mm/memory.c~fix-delay-accounting-swap-accounting +1 -1
diff -puN mm/memory.c~fix-delay-accounting-swap-accounting mm/memory.c
@@ -2084,9 +2084,9 @@ static int do_swap_page(struct mm_struct
 		count_vm_event(PGMAJFAULT);
 	}
 
-	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
 	mark_page_accessed(page);
 	lock_page(page);
+	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
 
 	/*
 	 * Back out if somebody else already faulted in this pte.

_

Signature · 10 lines
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
#1 lore raw
Quoted 6 lines · Balbir Singh

The delay incurred in lock_page() should also be accounted in swap delay accounting

Ah right, I forgot to resend this one, sorry. Thanks for remembering.

Quoted 22 lines

---

 mm/memory.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff -puN mm/swapfile.c~fix-delay-accounting-swap-accounting mm/swapfile.c diff -puN mm/memory.c~fix-delay-accounting-swap-accounting mm/memory.c
---
linux-2.6-latest/mm/memory.c~fix-delay-accounting-swap-accounting 2007-10-3
1 12:58:05.000000000 +0530 +++
linux-2.6-latest-balbir/mm/memory.c 2007-10-31 13:02:50.000000000 +0530 @@
-2084,9 +2084,9 @@ static int do_swap_page(struct mm_struct

mm/memory.c renamed from mm/memory.c~fix-delay-accounting-swap-accounting +1 -1
 		count_vm_event(PGMAJFAULT);
 	}
 
-	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
 	mark_page_accessed(page);
 	lock_page(page);
+	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
 
 	/*
 	 * Back out if somebody else already faulted in this pte.
Quoted 1 line

_

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#2 lore raw
Quoted 9 lines · Nick Piggin
Quoted 6 lines · Balbir Singh

The delay incurred in lock_page() should also be accounted in swap delay accounting

Ah right, I forgot to resend this one, sorry. Thanks for remembering.

Although, I think I had a bit more detail in the changelog which I think should be kept.

Basically, swap delay accounting seems quite broken as of now, because what it is counting is the time required to allocate a new page and submit the IO, but not actually the time to perform the IO at all (which I'd expect will be significant, although possibly in some workloads the actual page allocation will dominate).

Quoted 16 lines
Quoted 16 lines

---

 mm/memory.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff -puN mm/swapfile.c~fix-delay-accounting-swap-accounting mm/swapfile.c diff -puN mm/memory.c~fix-delay-accounting-swap-accounting mm/memory.c ---
linux-2.6-latest/mm/memory.c~fix-delay-accounting-swap-accounting 2007-10
-3 1 12:58:05.000000000 +0530 +++
linux-2.6-latest-balbir/mm/memory.c 2007-10-31 13:02:50.000000000 +0530

mm/memory.c renamed from mm/memory.c~fix-delay-accounting-swap-accounting +0 -0
@@ -2084,9 +2084,9 @@ static int do_swap_page(struct mm_struct
 		count_vm_event(PGMAJFAULT);
 	}
 
Quoted 8 lines
Quoted 8 lines
mm/memory.c renamed from mm/memory.c~fix-delay-accounting-swap-accounting +1 -1
-	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
 	mark_page_accessed(page);
 	lock_page(page);
+	delayacct_clear_flag(DELAYACCT_PF_SWAPIN);
 
 	/*
 	 * Back out if somebody else already faulted in this pte.

_

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

#3 lore raw
Quoted 10 lines · Nick Piggin
Quoted 8 lines · Nick Piggin
Quoted 6 lines · Balbir Singh

The delay incurred in lock_page() should also be accounted in swap delay accounting

Ah right, I forgot to resend this one, sorry. Thanks for remembering.

Quoted 9 lines

Although, I think I had a bit more detail in the changelog which I think should be kept.

Basically, swap delay accounting seems quite broken as of now, because what it is counting is the time required to allocate a new page and submit the IO, but not actually the time to perform the IO at all (which I'd expect will be significant, although possibly in some workloads the actual page allocation will dominate).

This looks quite good to me. I'm off attending a wedding, I'll resend the patch when I am back.

Signature · 10 lines
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>