1 |
Hi, |
2 |
|
3 |
On Fri, 5 May 2006 17:28:06 +0200 "Hemmann, Volker Armin" |
4 |
<volker.armin.hemmann@××××××××××××.de> wrote: |
5 |
|
6 |
> This is an example: |
7 |
> |
8 |
> [ 1151.984763] oom-killer: gfp_mask=0x80d2, order=0 |
9 |
|
10 |
Huh? If I understand Linux' memory management correctly that says that |
11 |
the OOM condition was triggered by trying to reserve 1 page (order=0) |
12 |
of high memory (gfp_mask|0x0002). But: You don't have highmem (of |
13 |
course, because you're running on 64bit). |
14 |
|
15 |
> [ 1151.984770] DMA per-cpu: |
16 |
|
17 |
what about DMA32? Is this an older kernel? |
18 |
|
19 |
> [ 1151.984809] HighMem free:0kB min:128kB low:160kB high:192kB active:0kB inactive:0kB present:0kB pages_scanned:0 all_unreclaimable? no |
20 |
|
21 |
doesn't make me wonder on 64bit... |
22 |
|
23 |
> [ 1151.984829] Swap cache: add 71, delete 71, find 0/0, race 0+0 |
24 |
> [ 1151.984831] Free swap = 995736kB |
25 |
> [ 1151.984833] Total swap = 996020kB |
26 |
> [ 1151.984834] Free swap: 995736kB |
27 |
|
28 |
Errr, that basically says nearly full swap space is available, isn't it? |
29 |
|
30 |
I think you probably have some IO related driver that for whatever |
31 |
reason decides to claim highmem. This triggers the OOM (there's no |
32 |
highmem), and cc1plus just happens to be the most interesting task to |
33 |
kill for the kernel: |
34 |
- just started, |
35 |
- much memory recovered |
36 |
|
37 |
Further analysis would probably require patching the mm/oom_kill.c and |
38 |
inserting a few debug statements -- if the problem is still there for |
39 |
newest kernels (there's been some changes esp. reg. AMD64 in 2.6.14, |
40 |
IIRC). |
41 |
|
42 |
|
43 |
-hwh |
44 |
-- |
45 |
gentoo-user@g.o mailing list |