Out of memory with Java 1.5 and 2.6.11.6

Out of memory with Java 1.5 and 2.6.11.6

[Robin Rosenberg]

I see regular crashes with 2.6.11.6 (mandrake-patched) and Java 1.5.02 (01 too 
btw, but not 1.4.2). Gentoo people report the same problem sugesting that it
may have appeared between 2.6.11.4 and 2.6.11.5.

If I start eclipse and then, outside of eclipse, starts a java 1.5 process 
eclipse dies instantly. Not always, but usually when I stop watching.

Any clues?

-- robin

oh, here is an excerpt from /var/log/messages.

Apr  5 22:05:20 xine kernel: oom-killer: gfp_mask=0xd0
Apr  5 22:05:20 xine kernel: DMA per-cpu:
Apr  5 22:05:20 xine kernel: cpu 0 hot: low 2, high 6, batch 1
Apr  5 22:05:20 xine kernel: cpu 0 cold: low 0, high 2, batch 1
Apr  5 22:05:20 xine kernel: Normal per-cpu:
Apr  5 22:05:20 xine kernel: cpu 0 hot: low 32, high 96, batch 16
Apr  5 22:05:20 xine kernel: cpu 0 cold: low 0, high 32, batch 16
Apr  5 22:05:20 xine kernel: HighMem per-cpu: empty
Apr  5 22:05:20 xine kernel:
Apr  5 22:05:20 xine kernel: Free pages:        5684kB (0kB HighMem)
Apr  5 22:05:20 xine kernel: Active:107335 inactive:5929 dirty:0 writeback:4 
unstable:0 free:1421 slab:9726 mapped:113240 pagetables:1705
Apr  5 22:05:20 xine kernel: DMA free:2068kB min:88kB low:108kB high:132kB 
active:8392kB inactive:0kB present:16384kB pages_scanned:8802 
all_unreclaimable? yes
Apr  5 22:05:20 xine kernel: lowmem_reserve[]: 0 495 495
Apr  5 22:05:21 xine kernel: Normal free:3616kB min:2800kB low:3500kB 
high:4200kB active:420948kB inactive:23716kB present:507576kB pages_scanned:0 
all_unreclaimable? no
Apr  5 22:05:21 xine kernel: lowmem_reserve[]: 0 0 0
Apr  5 22:05:21 xine kernel: HighMem free:0kB min:128kB low:160kB high:192kB 
active:0kB inactive:0kB present:0kB pages_scanned:0 all_unreclaimable? no
Apr  5 22:05:21 xine kernel: lowmem_reserve[]: 0 0 0
Apr  5 22:05:21 xine kernel: DMA: 1*4kB 0*8kB 1*16kB 0*32kB 2*64kB 1*128kB 
1*256kB 1*512kB 1*1024kB 0*2048kB 0*4096kB = 2068kB
Apr  5 22:05:21 xine kernel: Normal: 208*4kB 8*8kB 2*16kB 6*32kB 7*64kB 
0*128kB 0*256kB 0*512kB 0*1024kB 1*2048kB 0*4096kB = 3616kB
Apr  5 22:05:21 xine kernel: HighMem: empty
Apr  5 22:05:21 xine kernel: Swap cache: add 1267612, delete 1246233, find 
7825455/7951372, race 0+6
Apr  5 22:05:21 xine kernel: Free swap  = 760276kB
Apr  5 22:05:21 xine kernel: Total swap = 1060248kB
Apr  5 22:05:21 xine kernel: Out of Memory: Killed process 23770 (java).

Re: Out of memory with Java 1.5 and 2.6.11.6

[Ivan Yosifov]

Apr  5 22:05:20 xine kernel: oom-killer: gfp_mask=0xd0

The oom-killer a piece of code which starts killing processes when you
run out of memory. OOM-killer = Out Of Memory Killer. My guess is that
the second jvm consumed all the memory left ( this may me be a memory
leak, if so report it to SUN, not lkml ) and eclipse , being a
memory-heavyweight got slaughtered. The first thing that came to mind :)

On Wed, 2005-04-06 at 21:31 +0200, Robin Rosenberg wrote:
> I see regular crashes with 2.6.11.6 (mandrake-patched) and Java 1.5.02 (01 too 
> btw, but not 1.4.2). Gentoo people report the same problem sugesting that it
> may have appeared between 2.6.11.4 and 2.6.11.5.
> 
> If I start eclipse and then, outside of eclipse, starts a java 1.5 process 
> eclipse dies instantly. Not always, but usually when I stop watching.
> 
> Any clues?
> 
> -- robin
> 
> oh, here is an excerpt from /var/log/messages.
> 
> Apr  5 22:05:20 xine kernel: oom-killer: gfp_mask=0xd0
> Apr  5 22:05:20 xine kernel: DMA per-cpu:
> Apr  5 22:05:20 xine kernel: cpu 0 hot: low 2, high 6, batch 1
> Apr  5 22:05:20 xine kernel: cpu 0 cold: low 0, high 2, batch 1
> Apr  5 22:05:20 xine kernel: Normal per-cpu:
> Apr  5 22:05:20 xine kernel: cpu 0 hot: low 32, high 96, batch 16
> Apr  5 22:05:20 xine kernel: cpu 0 cold: low 0, high 32, batch 16
> Apr  5 22:05:20 xine kernel: HighMem per-cpu: empty
> Apr  5 22:05:20 xine kernel:
> Apr  5 22:05:20 xine kernel: Free pages:        5684kB (0kB HighMem)
> Apr  5 22:05:20 xine kernel: Active:107335 inactive:5929 dirty:0 writeback:4 
> unstable:0 free:1421 slab:9726 mapped:113240 pagetables:1705
> Apr  5 22:05:20 xine kernel: DMA free:2068kB min:88kB low:108kB high:132kB 
> active:8392kB inactive:0kB present:16384kB pages_scanned:8802 
> all_unreclaimable? yes
> Apr  5 22:05:20 xine kernel: lowmem_reserve[]: 0 495 495
> Apr  5 22:05:21 xine kernel: Normal free:3616kB min:2800kB low:3500kB 
> high:4200kB active:420948kB inactive:23716kB present:507576kB pages_scanned:0 
> all_unreclaimable? no
> Apr  5 22:05:21 xine kernel: lowmem_reserve[]: 0 0 0
> Apr  5 22:05:21 xine kernel: HighMem free:0kB min:128kB low:160kB high:192kB 
> active:0kB inactive:0kB present:0kB pages_scanned:0 all_unreclaimable? no
> Apr  5 22:05:21 xine kernel: lowmem_reserve[]: 0 0 0
> Apr  5 22:05:21 xine kernel: DMA: 1*4kB 0*8kB 1*16kB 0*32kB 2*64kB 1*128kB 
> 1*256kB 1*512kB 1*1024kB 0*2048kB 0*4096kB = 2068kB
> Apr  5 22:05:21 xine kernel: Normal: 208*4kB 8*8kB 2*16kB 6*32kB 7*64kB 
> 0*128kB 0*256kB 0*512kB 0*1024kB 1*2048kB 0*4096kB = 3616kB
> Apr  5 22:05:21 xine kernel: HighMem: empty
> Apr  5 22:05:21 xine kernel: Swap cache: add 1267612, delete 1246233, find 
> 7825455/7951372, race 0+6
> Apr  5 22:05:21 xine kernel: Free swap  = 760276kB
> Apr  5 22:05:21 xine kernel: Total swap = 1060248kB
> Apr  5 22:05:21 xine kernel: Out of Memory: Killed process 23770 (java).
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: Out of memory with Java 1.5 and 2.6.11.6

[Chris Wright]

* Robin Rosenberg (robin.rosenberg.lists@dewire.com) wrote:
> 
> I see regular crashes with 2.6.11.6 (mandrake-patched) and Java 1.5.02 (01 too
> btw, but not 1.4.2). Gentoo people report the same problem sugesting that it
> may have appeared between 2.6.11.4 and 2.6.11.5.

Sounds very unlikely, we didn't change anything in -stable that would
trigger.

Any chance 

# echo 1 >  /proc/sys/vm/legacy_va_layout

makes any difference?  I've seen this once before (apparently with old
glibc threading library as well).

thanks,
-chris