lkml.org 
[lkml]   [2009]   [Mar]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Subject[RFC] x86: gup_fast() batch limit (was: DRM lock ordering fix series)
From
Date
On Sat, 2009-03-28 at 11:48 +0100, Peter Zijlstra wrote:
> On Sat, 2009-03-28 at 09:46 +0100, Brice Goglin wrote:
> > Peter Zijlstra wrote:
> > > Also note that doing large gup() with gup_fast() will be undesirable due
> > > to it disabling IRQs. So iterating say several MB worth of pages will
> > > hurt like crazy. Currently all gup_fast() users do a single page lookup.
> > >
> >
> > In 2.6.29, fs/bio.c:955, fs/direct-io.c:153 and fs/splice.c:1222 do
> > multiple-pages lookup at once. The latter might be limited to 16 pages
> > because of the pipe-depth, I don't know about the formers.
> >
> > Is there some sort of reasonable limit? A couple dozens pages at once maybe?
>
> Depends on your latency requirements, looking at the code I'd say we'd
> have to add that batch limit the comment talks about. I'd see preempt-rt
> wanting to lower that significantly.
>
> Regular mainline could do with 32-64 I guess, max irq latency is well
> over 10ms on mainline anyway.

I'm not really trusting my brain today, but something like the below
should work I think.

Nick, any thoughts?

Not-Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
---
arch/x86/mm/gup.c | 24 +++++++++++++++++++++---
1 files changed, 21 insertions(+), 3 deletions(-)
diff --git a/arch/x86/mm/gup.c b/arch/x86/mm/gup.c
index be54176..4ded5c3 100644
--- a/arch/x86/mm/gup.c
+++ b/arch/x86/mm/gup.c
@@ -11,6 +11,8 @@

#include <asm/pgtable.h>

+#define GUP_BATCH 32
+
static inline pte_t gup_get_pte(pte_t *ptep)
{
#ifndef CONFIG_X86_PAE
@@ -91,7 +93,8 @@ static noinline int gup_pte_range(pmd_t pmd, unsigned long addr,
get_page(page);
pages[*nr] = page;
(*nr)++;
-
+ if (*nr > GUP_BATCH)
+ break;
} while (ptep++, addr += PAGE_SIZE, addr != end);
pte_unmap(ptep - 1);

@@ -157,6 +160,8 @@ static int gup_pmd_range(pud_t pud, unsigned long addr, unsigned long end,
if (!gup_pte_range(pmd, addr, next, write, pages, nr))
return 0;
}
+ if (*nr > GUP_BATCH)
+ break;
} while (pmdp++, addr = next, addr != end);

return 1;
@@ -214,6 +219,8 @@ static int gup_pud_range(pgd_t pgd, unsigned long addr, unsigned long end,
if (!gup_pmd_range(pud, addr, next, write, pages, nr))
return 0;
}
+ if (*nr > GUP_BATCH)
+ break;
} while (pudp++, addr = next, addr != end);

return 1;
@@ -226,7 +233,7 @@ int get_user_pages_fast(unsigned long start, int nr_pages, int write,
unsigned long addr, len, end;
unsigned long next;
pgd_t *pgdp;
- int nr = 0;
+ int batch = 0, nr = 0;

start &= PAGE_MASK;
addr = start;
@@ -254,6 +261,7 @@ int get_user_pages_fast(unsigned long start, int nr_pages, int write,
* (which we do on x86, with the above PAE exception), we can follow the
* address down to the the page and take a ref on it.
*/
+again:
local_irq_disable();
pgdp = pgd_offset(mm, addr);
do {
@@ -262,11 +270,21 @@ int get_user_pages_fast(unsigned long start, int nr_pages, int write,
next = pgd_addr_end(addr, end);
if (pgd_none(pgd))
goto slow;
- if (!gup_pud_range(pgd, addr, next, write, pages, &nr))
+ if (!gup_pud_range(pgd, addr, next, write, pages, &batch))
goto slow;
+ if (batch > GUP_BATCH) {
+ local_irq_enable();
+ addr += batch << PAGE_SHIFT;
+ nr += batch;
+ batch = 0;
+ if (addr != end)
+ goto again;
+ }
} while (pgdp++, addr = next, addr != end);
local_irq_enable();

+ nr += batch;
+
VM_BUG_ON(nr != (end - start) >> PAGE_SHIFT);
return nr;




\
 
 \ /
  Last update: 2009-03-28 13:25    [from the cache]
©2003-2011 Jasper Spaans