kernel 2.6.8.1: memory leak? cdrecord problem?

kernel 2.6.8.1: memory leak? cdrecord problem?

[John McGowan]

KERNEL 2.6.8.1: Memory leak? CDrecord problem?

-----------------------------------
KERNEL 2.6.7: compiled cdrecord.
 Works fine. Burns data disks fine.
 Burns audio disks fine.

UPGRADED TO KERNEL 2.6.8.1 (using the patches):
 Used version of cdrecord compiled under 2.7
 Burns data disks fine.
 I have 256 Meg RAM (and 500+ Meg swap).
 Burning a 400+Meg wav file to audio CD (40+minute
 radio show) causes, after 130-180Meg have been
 burned (depending on what I am running):

   OUT OF MEMORY: closing down atd

  and memory keeps being used up:
 
   OUT OF MEMORY: closing down cupsd
   OUT OF MEMORY: closing down crond
   (etc.)
  
  If I am lucky, cdrecord gets closed down before
  something critical gets killed.
  Often I am unlucky.

UNDER KERNEL 2.6.8.1:
 Recompiled cdrecord under 2.6.8.1.
 Burns data disks fine.
 Burning a 400+ Meg wav file to audio CD causes:

   OUT OF MEMORY: closing down atd
   OUT OF MEMORY: closing down cupsd
   OUT OF MEMORY: closing down crond
   (etc.)

(Fedora Core1; kernel upgraded to (now) 2.6.7
               (I went back to 2.6.7 so I could
                burn audio CDs)
               Using Pioneer DVD burner with ATAPI
               (dev=ATAPI:0,0,0 in cdrecord) interface.)

Re: kernel 2.6.8.1: memory leak? cdrecord problem?

[Diego Calleja]

El Sat, 21 Aug 2004 13:26:46 -0400 John McGowan <jmcgowan@inch.com> escribió:

> KERNEL 2.6.8.1: Memory leak? CDrecord problem?


Yes, there's a memory leak, you can try 2.6.8.1-mm3
or apply the fix yourself (I think it's this one:
http://kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.8.1/2.6.8.1-mm3/broken-out/bio_uncopy_user-mem-leak.patch )

Re: kernel 2.6.8.1: memory leak? cdrecord problem?

[Con Kolivas]

[-- Attachment #1: Type: text/plain, Size: 656 bytes --]

Diego Calleja wrote:
> El Sat, 21 Aug 2004 13:26:46 -0400 John McGowan <jmcgowan@inch.com> escribió:
> 
> 
>>KERNEL 2.6.8.1: Memory leak? CDrecord problem?
> 
> 
> 
> Yes, there's a memory leak, you can try 2.6.8.1-mm3
> or apply the fix yourself (I think it's this one:
> http://kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.8.1/2.6.8.1-mm3/broken-out/bio_uncopy_user-mem-leak.patch )

That wont do by itself. It only fixes the memory leak. You _also_ need 
this patch for audio cds to not be stuffed:

http://www.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.8.1/2.6.8.1-mm3/broken-out/bio_uncopy_user-mem-leak-fix.patch

Cheers,
Con

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: kernel 2.6.8.1: memory leak? cdrecord problem?

[Frank Steiner]

Hi,

Con Kolivas wrote:

>> Yes, there's a memory leak, you can try 2.6.8.1-mm3
>> or apply the fix yourself (I think it's this one:
>> http://kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.8.1/2.6.8.1-mm3/broken-out/bio_uncopy_user-mem-leak.patch 
>> )
> 
> 
> That wont do by itself. It only fixes the memory leak. You _also_ need 
> this patch for audio cds to not be stuffed:
> 
> http://www.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.8.1/2.6.8.1-mm3/broken-out/bio_uncopy_user-mem-leak-fix.patch 

it doesn't work, even with these two patches applied. I tried to burn
a DVD image (3.5GB) with kernel 2.6.8.1 including those patches and
again the memory run full until the kernel started killing processes.
Unfortunately the host always crashes before it writes the messages
to the log files, but I saw some messages about "cpu0..." and DMA.


I was running "cdrecord -sao dev=ATAPI:/dev/cdrecorder -v 2.4.img"
on a NEC ND1300A.

Can I debug/provide more info?

BTW, the bio.c from 2.6.9rc1 and 2.6.8.1 don't differ, so I assume it's
still unfixed in 2.6.9-rc1, too...?

cu,
Frank

-- 
Dipl.-Inform. Frank Steiner   Web:  http://www.bio.ifi.lmu.de/~steiner/
Lehrstuhl f. Bioinformatik    Mail: http://www.bio.ifi.lmu.de/~steiner/m/
LMU, Amalienstr. 17           Phone: +49 89 2180-4049
80333 Muenchen, Germany       Fax:   +49 89 2180-99-4049

Re: kernel 2.6.8.1: memory leak? cdrecord problem?

[Frank Steiner]

Frank Steiner wrote:

> Unfortunately the host always crashes before it writes the messages
> to the log files, but I saw some messages about "cpu0..." and DMA.

Now I got it to do an emergency sync :-)

Aug 26 14:13:42 wirth kernel: oom-killer: gfp_mask=0xd0
Aug 26 14:13:42 wirth kernel: DMA per-cpu:
Aug 26 14:13:42 wirth kernel: cpu 0 hot: low 2, high 6, batch 1
Aug 26 14:13:42 wirth kernel: cpu 0 cold: low 0, high 2, batch 1
Aug 26 14:13:42 wirth kernel: Normal per-cpu:
Aug 26 14:13:42 wirth kernel: cpu 0 hot: low 32, high 96, batch 16
Aug 26 14:13:42 wirth kernel: cpu 0 cold: low 0, high 32, batch 16
Aug 26 14:13:42 wirth kernel: HighMem per-cpu:
Aug 26 14:13:42 wirth kernel: cpu 0 hot: low 32, high 96, batch 16
Aug 26 14:13:42 wirth kernel: cpu 0 cold: low 0, high 32, batch 16
Aug 26 14:13:42 wirth kernel:
Aug 26 14:13:42 wirth kernel: Free pages:      107700kB (105792kB HighMem)
Aug 26 14:13:42 wirth kernel: Active:13356 inactive:253643 dirty:0 writeback:0 unstable:0 free:26925 slab:2333 mapped:12041 pagetables:151
Aug 26 14:13:42 wirth kernel: DMA free:44kB min:16kB low:32kB high:48kB active:4kB inactive:0kB present:16384kB
Aug 26 14:13:42 wirth kernel: protections[]: 8 476 732
Aug 26 14:13:42 wirth kernel: Normal free:1864kB min:936kB low:1872kB high:2808kB active:1236kB inactive:1060kB present:901120kB
Aug 26 14:13:42 wirth kernel: protections[]: 0 468 724
Aug 26 14:13:42 wirth kernel: HighMem free:105792kB min:512kB low:1024kB high:1536kB active:52184kB inactive:1013512kB present:1179632kB
Aug 26 14:13:42 wirth kernel: protections[]: 0 0 256
Aug 26 14:13:42 wirth kernel: DMA: 1*4kB 1*8kB 0*16kB 1*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 44kB
Aug 26 14:13:42 wirth kernel: Normal: 0*4kB 1*8kB 0*16kB 0*32kB 1*64kB 0*128kB 1*256kB 1*512kB 1*1024kB 0*2048kB 0*4096kB = 1864kB
Aug 26 14:13:42 wirth kernel: HighMem: 222*4kB 417*8kB 216*16kB 96*32kB 121*64kB 42*128kB 8*256kB 6*512kB 1*1024kB 1*2048kB 18*4096kB = 105792kB
Aug 26 14:13:42 wirth kernel: Swap cache: add 0, delete 0, find 0/0, race 0+0
Aug 26 14:13:42 wirth kdm[6478]: Greeter exited unexpectedly
Aug 26 14:13:42 wirth kernel: Out of Memory: Killed process 6906 (kdm_greet).
...
(continues to kill processes)

If hope that helps...

-- 
Dipl.-Inform. Frank Steiner   Web:  http://www.bio.ifi.lmu.de/~steiner/
Lehrstuhl f. Bioinformatik    Mail: http://www.bio.ifi.lmu.de/~steiner/m/
LMU, Amalienstr. 17           Phone: +49 89 2180-4049
80333 Muenchen, Germany       Fax:   +49 89 2180-99-4049