Can someone shed some light on the memory allocation system of ivtv? I've got 3 capture cards, 512MB RAM, driver 0.3.7b on kernel 2.6.12. I've been using the "stable recommended" firmware for the last couple of months and it has been working like a charm. Recording 3 shows and watching a 4th works great.
Last weekend I wanted to give the latest firmware another try; from what I've read, the latest firmware causes the ivtv-driver to switch to frame-based DMA in stead of block-based and to switch to dynamic buffer allocation. Unfortunatly I ran into "memory problems": Aug 1 17:19:55 [kernel] oom-killer: gfp_mask=0x1d2 Aug 1 17:19:56 [kernel] DMA per-cpu: Aug 1 17:19:56 [kernel] cpu 0 hot: low 2, high 6, batch 1 Aug 1 17:19:56 [kernel] cpu 0 cold: low 0, high 2, batch 1 Aug 1 17:19:56 [kernel] Normal per-cpu: Aug 1 17:19:56 [kernel] cpu 0 hot: low 62, high 186, batch 31 Aug 1 17:19:56 [kernel] cpu 0 cold: low 0, high 62, batch 31 Aug 1 17:19:56 [kernel] HighMem per-cpu: empty Aug 1 17:19:56 [kernel] Free pages: 67424kB (0kB HighMem) Aug 1 17:19:56 [kernel] Active:51080 inactive:50510 dirty:19 writeback:0 unstable:0 free:16856 slab:6985 mapped:1003 36 pagetables:1096 Aug 1 17:19:56 [kernel] DMA free:4028kB min:2048kB low:2560kB high:3072kB active:4768kB inactive:4136kB present:1638 4kB pages_scanned:10822 all_unreclaimable? yes Aug 1 17:19:56 [kernel] lowmem_reserve[]: 0 495 495 Aug 1 17:19:56 [kernel] Normal free:63396kB min:63484kB low:79352kB high:95224kB active:199552kB inactive:197904kB p resent:507840kB pages_scanned:477336 all_unreclaimable? yes Aug 1 17:19:56 [kernel] lowmem_reserve[]: 0 0 0 Aug 1 17:19:56 [kernel] HighMem free:0kB min:128kB low:160kB high:192kB active:0kB inactive:0kB present:0kB pages_sc anned:0 all_unreclaimable? no Aug 1 17:19:56 [kernel] lowmem_reserve[]: 0 0 0 Aug 1 17:19:56 [kernel] DMA: 1*4kB 1*8kB 1*16kB 1*32kB 0*64kB 1*128kB 1*256kB 1*512kB 1*1024kB 1*2048kB 0*4096kB = 4 028kB Aug 1 17:19:56 [kernel] Normal: 477*4kB 134*8kB 12*16kB 0*32kB 1*64kB 0*128kB 1*256kB 1*512kB 0*1024kB 1*2048kB 14*4 096kB = 63396kB Aug 1 17:19:56 [kernel] HighMem: empty Aug 1 17:19:56 [kernel] Swap cache: add 132927, delete 132888, find 1427/2189, race 0+0 Aug 1 17:19:56 [kernel] Free swap = 0kB Aug 1 17:19:56 [kernel] Total swap = 506036kB Aug 1 17:19:56 [kernel] Out of Memory: Killed process 3944 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11284 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11289 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11290 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11291 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11292 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11293 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11372 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11373 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11374 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11375 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 11376 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 24091 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 24092 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 24093 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 24098 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 3953 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 3954 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 4584 (mysqld). Aug 1 17:19:56 [kernel] Out of Memory: Killed process 4585 (mysqld). Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #0: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #1: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #2: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #3: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #4: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #5: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #6: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #7: Stealing a Buffer, 512 currently allocated Aug 1 17:20:03 [kernel] ivtv: ENC Stream 0 OVERFLOW #8: Stealing a Buffer, 512 currently allocated Aug 1 17:20:04 [kernel] ivtv: ENC Stream 0 OVERFLOW #9: Stealing a Buffer, 512 currently allocated Aug 1 17:20:04 [kernel] ivtv: ENC Stream 0 OVERFLOW #10: Stealing a Buffer, 512 currently allocated Aug 1 17:20:04 [kernel] ivtv: ENC Stream 0 OVERFLOW #11: Stealing a Buffer, 512 currently allocated Aug 1 17:20:04 [kernel] ivtv: ENC Stream 0 OVERFLOW #12: Stealing a Buffer, 512 currently allocated Aug 1 17:20:04 [kernel] ivtv: ENC Stream 0 OVERFLOW #13: Stealing a Buffer, 512 currently allocated ... I've switched back to the old firmware after that and for the rest of the week I had no problems. What's the recommended configuration, e.g. what is the recommended setting vm.min_free_kbytes / card (or doesn't that matter)? Should I add more memory (go to 1GB)? N. On 8/4/05, Graeme Wilford <[EMAIL PROTECTED]> wrote: > This seems to have worked! I've not done extensive testing but using > mpg_buffers=8 while throwing in dec_yuv_buffers=4 and > max_dec_yuv_buffers=4 for good luck allowed me to play a DVD via > Xv/YUV while simultaneously capturing an uncorrupted MPEG stream. > > I'll need to play around to see if this is consistent and to optimise > the buffer sizes... > > Later, ivtv was unable to dynamically allocate YUV buffers due to a > lack of free memory. I already had vm.min_free_kbytes=16384 but this > would have needed an increase using the settings above. Instead, I've > turned off dynamic buffer allocation and removed the min_free setting > for now. > > Thanks, John. > > Cheets, > Wilf. > > On 02/08/05, John Harvey <[EMAIL PROTECTED]> wrote: > > I'm not using the 350 for capturing but did notice a while ago that the > > playback suffered when capturing when the capturing ran out of buffers. > > Have you tried setting mpg_buffers=8 and/or turning off dynamic buffer > > allocation. > > > > John > > > > > -----Original Message----- > > > From: [EMAIL PROTECTED] [mailto:ivtv-devel- > > > [EMAIL PROTECTED] On Behalf Of Graeme Wilford > > > Sent: 02 August 2005 22:46 > > > To: ivtv-devel@lists.sourceforge.net > > > Subject: Re: [ivtv-devel] [PATCH] yuv improvements > > > > > > John, Ian - I've done some testing with this new patch. > > > > > > With mplayer, I get real bad frame drop and/or audio sync issues with > > > both avi's and DVDs - worse than before the patch *but* with xine, > > > playback is almost perfect in both cases. This is a little strange as > > > both xine and mplayer yielded similar results before the patch. > > > > > > Anyhow, on a 2.53GHz Celeron with 2.6.12-1.1372_FC3 kernel, X driver > > > 0.10.5, ivtv-0.3.7a (dynamic buffers) + this patch, latest firmware > > > and xine-0.99.3, I get perfect imaging with no tearing. Just the odd > > > discarded frame. Nice job! > > > > > > However, using YUV output on the 350 definitely causes simultaneous > > > mpeg recordings to corrupt and this is still evident with this patch. > > > The mpeg stream has lots of frames missing. I get no errors reported > > > by the ivtv driver so the loss of frames seems to occur at the > > > firmware level on the card, before the ivtv driver gets near it... > > > > > > If you can't reproduce the problem, please let me know the driver > > > options and firmware you're using. > > > > > > Cheers, > > > Wilf. > > > > > > On 02/08/05, John Harvey <[EMAIL PROTECTED]> wrote: > > > > This patch includes the previous patch I sent from Ian for filtering. It > > > > adds some improvements from Ian to fix some color alignment issues. > > > > Also thanks to Ian we have worked out how to do multi-buffering without > > > > crashing the firmware so this patch should fix all the tearing issues as > > > > well. > > > > > > > > Any problems let me know. > > > > > > > > John > > > > > > -- > > > [EMAIL PROTECTED] > > > > -- > [EMAIL PROTECTED] > > > ------------------------------------------------------- > SF.Net email is Sponsored by the Better Software Conference & EXPO > September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices > Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA > Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf > _______________________________________________ > ivtv-devel mailing list > ivtv-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/ivtv-devel > ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ ivtv-devel mailing list ivtv-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ivtv-devel
