Re: Kernel memory requirements and BK

Re: Kernel memory requirements and BK

[Chuck Ebbert]

Art Haas wrote:

> With the 2.6.9 and 2.6.10-rc kernels, BK bombs out with
> out-of-memory errors once the repository checking begins.

There's nothing wrong with BK's performance that can't be
solved by a terahertz processor and a terabyte of RAM. ;)

But these patches might help:

===============================================================================
# vm_pages_scanned_active_list.patch
#
#       Stop kswapd from looping.
#
#       Patch by Nick Piggin 24 Oct 2004
#       Signed-off-by: Andrew Morton <akpm@osdl.org>
#       Signed-off-by: Linus Torvalds <torvalds@osdl.org>
#       Status: in 2.6.10
#
--- 2.6.9/mm/vmscan.c
+++ 2.6.9.1/mm/vmscan.c
@@ -574,7 +574,6 @@ static void shrink_cache(struct zone *zo
                        nr_taken++;
                }
                zone->nr_inactive -= nr_taken;
-               zone->pages_scanned += nr_taken;
                spin_unlock_irq(&zone->lru_lock);
 
                if (nr_taken == 0)
@@ -675,6 +674,7 @@ refill_inactive_zone(struct zone *zone, 
                }
                pgscanned++;
        }
+       zone->pages_scanned += pgscanned;
        zone->nr_active -= pgmoved;
        spin_unlock_irq(&zone->lru_lock);
 
===============================================================================
# spurious_oomkill.patch
#
#       Prevent spurious out of memory process kills.
#       Reported to work by testers on lkml.
#
#       Patch by Rik van Riel <riel@redhat.com>
#       Status: NOT in 2.6.10
#
--- 2.6.9/mm/vmscan.c
+++ 2.6.9.1/mm/vmscan.c
@@ -379,7 +379,7 @@
 
                referenced = page_referenced(page, 1);
                /* In active use or really unfreeable?  Activate it. */
-               if (referenced && page_mapping_inuse(page))
+               if (referenced && sc->priority && page_mapping_inuse(page))
                        goto activate_locked;
 
 #ifdef CONFIG_SWAP
@@ -715,7 +715,7 @@
                if (page_mapped(page)) {
                        if (!reclaim_mapped ||
                            (total_swap_pages == 0 && PageAnon(page)) ||
-                           page_referenced(page, 0)) {
+                           (page_referenced(page, 0) && sc->priority)) {
                                list_add(&page->lru, &l_active);
                                continue;
                        }
===============================================================================

--Chuck Ebbert  06-Nov-04  23:55:23

Re: Kernel memory requirements and BK [SOLVED]

[Art Haas]

On Sun, Nov 07, 2004 at 12:02:07AM -0500, Chuck Ebbert wrote:
> Art Haas wrote:
> 
> > With the 2.6.9 and 2.6.10-rc kernels, BK bombs out with
> > out-of-memory errors once the repository checking begins.
> 
> There's nothing wrong with BK's performance that can't be
> solved by a terahertz processor and a terabyte of RAM. ;)
> 
> But these patches might help:
> 

[ snip first patch added into Linus's BK tree]

> ============================================================================
> # spurious_oomkill.patch
> #
> #       Prevent spurious out of memory process kills.
> #       Reported to work by testers on lkml.
> #
> #       Patch by Rik van Riel <riel@redhat.com>
> #       Status: NOT in 2.6.10
> #
> --- 2.6.9/mm/vmscan.c
> +++ 2.6.9.1/mm/vmscan.c
> @@ -379,7 +379,7 @@
>  
>                 referenced = page_referenced(page, 1);
>                 /* In active use or really unfreeable?  Activate it. */
> -               if (referenced && page_mapping_inuse(page))
> +               if (referenced && sc->priority && page_mapping_inuse(page))
>                         goto activate_locked;
>  
>  #ifdef CONFIG_SWAP
> @@ -715,7 +715,7 @@
>                 if (page_mapped(page)) {
>                         if (!reclaim_mapped ||
>                             (total_swap_pages == 0 && PageAnon(page)) ||
> -                           page_referenced(page, 0)) {
> +                           (page_referenced(page, 0) && sc->priority)) {
>                                 list_add(&page->lru, &l_active);
>                                 continue;
>                         }

This patch did the trick for me. I booted up my 2.6.8.1 kernel, did a
'bk pull' to get the code (Sunday Morning), and added in this patch. After
building and installing the patched kernel, I did another 'bk pull'
later and things worked fine. Consider me one more person to vouch for
this patch solving the OOM problem!

Art Haas
-- 
Man once surrendering his reason, has no remaining guard against absurdities
the most monstrous, and like a ship without rudder, is the sport of every wind.

-Thomas Jefferson to James Smith, 1822