Eric Lee / linux-smarc-t335x-v3.2

Download as
Browse Code ยป

Commit d992895ba2b27cf5adf1ba0ad6d27662adc54c5e

Authored by Nick Piggin
Committed by Linus Torvalds
1 parent 40193713df
Exists in master and in 4 other branches v3.2_AM335xPSP_04.06.00.11, v3.2_NEWT335xPSP_04.06.00.11, v3.2_SBC_SMARTMEN, v3.2_SMARCT335xPSP_04.06.00.11

[PATCH] Lazy page table copies in fork() 

Defer copying of ptes until fault time when it is possible to reconstruct
the pte from backing store. Idea from Andi Kleen and Nick Piggin.

Thanks to input from Rik van Riel and Linus and to Hugh for correcting
my blundering.

Ray Fucillo <fucillo@intersystems.com> reports:

  "I applied this latest patch to a 2.6.12 kernel and found that it does
   resolve the problem.  Prior to the patch on this machine, I was
   seeing about 23ms spent in fork for ever 100MB of shared memory
   segment.

   After applying the patch, fork is taking about 1ms regardless of the
   shared memory size."

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Showing 1 changed file with 11 additions and 0 deletions Side-by-side Diff

... ... @@ -498,6 +498,17 @@
498 498 unsigned long addr = vma->vm_start;
499 499 unsigned long end = vma->vm_end;
500 500  
  501 + /*
  502 + * Don't copy ptes where a page fault will fill them correctly.
  503 + * Fork becomes much lighter when there are big shared or private
  504 + * readonly mappings. The tradeoff is that copy_page_range is more
  505 + * efficient than faulting.
  506 + */
  507 + if (!(vma->vm_flags & (VM_HUGETLB|VM_NONLINEAR|VM_RESERVED))) {
  508 + if (!vma->anon_vma)
  509 + return 0;
  510 + }
  511 +
501 512 if (is_vm_hugetlb_page(vma))
502 513 return copy_hugetlb_page_range(dst_mm, src_mm, vma);
503 514