[mythtv] glibc problems - mythbackend crash

Sat Jan 27 17:16:55 UTC 2007

Hi Daniel. Thanks for responding.

On 1/27/07, Daniel Kristjansson <danielk at cuymedia.net> wrote:
> On Sat, 2007-01-27 at 05:39 -0800, Mark Knecht wrote:
> > Hi,
> >    I sent the following message to this list some time ago but didn't
> > receive any responses so I'm trying again. mythbackend crashes once a
> > week with this sort of error.
> >
> >    Please see the earlier emails for complete examples of the crash.
> > I've posted just a bit of it here.
>
> Hmm, it looks like you've run out of memory. What version of
> the ivtv drivers are you using?

Currently running 0.7.1 here.

ivtv:  ==================== START INIT IVTV ====================
ivtv:  version 0.7.1 (tagged release) loading
ivtv:  Linux version: 2.6.17-gentoo-r2 SMP preempt mod_unload PENTIUM4 gcc-4.1
ivtv:  In case of problems please include the debug info between
ivtv:  the START INIT IVTV and END INIT IVTV lines, along with
ivtv:  any module options, when mailing the ivtv-users mailinglist.
ivtv0: Autodetected Hauppauge WinTV PVR-150 card (cx23416 based)
ACPI: PCI Interrupt 0000:01:01.0[A] -> GSI 22 (level, low) -> IRQ 22
ivtv0: Unreasonably low latency timer, setting to 64 (was 32)
tveeprom 0-0050: Hauppauge model 26032, rev C199, serial# 8095579

Looking in dmesg I do see these messages, possibly indicating it's
unable to allocate pages of memory?

ivtv-enc: page allocation failure. order:4, mode:0xd0
 <c013866e> __alloc_pages+0x246/0x25a  <c014978c> cache_alloc_refill+0x285/0x44e
 <c01499ae> __kmalloc+0x59/0x62  <df9432d0> ivtv_init_buffer+0x51/0x180 [ivtv]
 <df944826> enc_gather_free_buffers+0x97/0x27b [ivtv]  <df94e995>
ivtv_sched_DMA+0x584/0x874 [ivtv]
 <c0112698> default_wake_function+0x0/0x15  <df953017>
ivtv_enc_thread+0x166/0x1cb [ivtv]
 <c0127937> autoremove_wake_function+0x0/0x3a  <df952eb1>
ivtv_enc_thread+0x0/0x1cb [ivtv]
 <c0100bf5> kernel_thread_helper+0x5/0xb

>
> You might want to try something 0.7.x or earlier, assuming
> you are using a more recent driver. Some later versions of
> the driver can allocate multi-gigabyte chunks of RAM if they
> see an application written for an earlier version of the
> driver.
>
> Alternatively try SVN head, which has a partial fix now.
> Or try SVN head in a week or two when it should detect
> the API version of the ivtv driver before to use it in
> any serious way.

So this would be SVN of MythTV? I'll look into that but I'm not a
developer and Myth is a pretty big program.

>
> Note: This may not be your problem, the video file you
> were playing looks like it may have been corrupt. This
> could also overwrite memory, you can confirm the ivtv
> issue by seeing if your backend memory usage jumps
> significantly when recordings start.

OK, but I'm not seeing much more memory usage, at least using top.
Maybe I need to look some other way? Granted, the machine has 512MB
and it looks like most of that is 'used' but I know I cannot really
trust these numbers from top.

Here is memory usage with the backend idle:

top - 09:09:05 up 9 days, 12:17,  2 users,  load average: 0.00, 0.00, 0.00
Tasks:  90 total,   1 running,  89 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.0% us,  0.0% sy,  0.0% ni, 100.0% id,  0.0% wa,  0.0% hi,  0.0% si
Mem:    499704k total,   492536k used,     7168k free,    65168k buffers
Swap:  1052216k total,    25856k used,  1026360k free,   105768k cached

and here it is after I start a recording:

top - 09:10:25 up 9 days, 12:18,  2 users,  load average: 0.20, 0.05, 0.01
Tasks:  90 total,   1 running,  89 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.5% us,  1.3% sy,  0.0% ni, 92.5% id,  4.3% wa,  0.7% hi,  0.7% si
Mem:    499704k total,   492648k used,     7056k free,    62680k buffers
Swap:  1052216k total,    25856k used,  1026360k free,   108392k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 4395 mythtv    15   0  218m  26m  13m S    2  5.5  12:35.97
mythbackend
 3516 root       5 -10     0    0    0 S    1  0.0  80:53.30 ivtv-enc
 7572 root      10  -5     0    0    0 S    0  0.0  15:37.07 rpciod/0

One comment I got on the Gentoo list is that Myth is not stable for
this individual on SMP kernels. He said a number of his problems went
away when he down converted to UP. Has there been anyone else here
talking about that sort of thing? This machine has an Intel HT
processor and the kernel is SMP.

Again, thanks very much for the ideas.

Cheers,
Mark