<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
</head>
<body bgcolor="#ffffff" text="#000000">
<pre wrap="">Yes, well, I searched the user list, as I'm not a developer. But it does answer my question. I wish I was, but my programming skills were never at that high a level. And the kernel issue you mention is one I've been experiencing with DMA. I look forward to the patch making it into a future kernel. Keep me posted or send me a URL to the filed report so I can monitor its status. I probably should break down and break out the kernel source...but I admit I probably won't for now.
From: Bruce Markey <a class="moz-txt-link-rfc2396E" href="mailto:bjm@lvcm.com"><bjm@lvcm.com></a>
Subject: Re: [mythtv-users] Firewire Issues
To: Discussion about mythtv <a class="moz-txt-link-rfc2396E" href="mailto:mythtv-users@mythtv.org"><mythtv-users@mythtv.org></a>
Message-ID: <a class="moz-txt-link-rfc2396E" href="mailto:470D2039.4060902@lvcm.com"><470D2039.4060902@lvcm.com></a>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
DISAFAN wrote:
</pre>
<blockquote qctoggled="false" type="cite">
<pre wrap=""><span class="moz-txt-citetags">> </span>I confess, I am suffering from a lack of understanding of how to proceed
<span class="moz-txt-citetags">> </span>on my firewire issues. Positively, all of the analog channels...the
<span class="moz-txt-citetags">> </span>basic cable stuff...seems to come through on my SA3250HD. The HD
<span class="moz-txt-citetags">> </span>channels seem to come through. The digital versions of the analog feed
<span class="moz-txt-citetags">> </span>aren't tuning...but I figure if I can get rid of the PVR-150 hooked into
<span class="moz-txt-citetags">> </span>the box output, it will allow me to free up a PCI slot for another QAM
<span class="moz-txt-citetags">> </span>tuner.
<span class="moz-txt-citetags">> </span>
<span class="moz-txt-citetags">> </span>That said, my firewire recording, like many peoples, is inconsistent.
<span class="moz-txt-citetags">> </span>When it records, it records. When it doesn't, it doesn't, and I have to
<span class="moz-txt-citetags">> </span>stop MythTV, reset the firewire bus, and that seems to clear the problem.
<span class="moz-txt-citetags">> </span>
<span class="moz-txt-citetags">> </span>I have high hopes a future bugfix will stabilize the connection...it
</pre>
</blockquote>
<pre wrap=""><!---->
Wed, 18 Jul 2007 is actually in the past. If you are using a 64
bit kernel, there is a bug that causes a memory error which make
recording and firewire_tester appear to fail due to a connection
problem. However, it is a kernel memory error and the connection
is actually okay. "grep dma_region_alloc /var/log/syslog".
: root@bigboy \# ; zcat /var/log/syslog.4.gz | grep dma_region_alloc
Oct 3 21:54:54 bigboy kernel: [<ffffffff880652c7>] :ieee1394:dma_region_alloc+0x37/0x126
Oct 3 21:54:54 bigboy kernel: dma_region_alloc: vmalloc_32() failed
Oct 3 21:54:56 bigboy kernel: [<ffffffff880652c7>] :ieee1394:dma_region_alloc+0x37/0x126
...
I had to go back in my logs to find these as they haven't happened
since I applied this patch the evening of Oct 3:
<a class="moz-txt-link-freetext" href="http://lkml.org/lkml/2007/7/18/34">http://lkml.org/lkml/2007/7/18/34</a>
Since booting the patched kernel, I have never had one firewire
recording fail. I need to run "firewire_tester -p -n 0 -R" once
after rebooting then the connection stays good until the next
reboot (the memory bug would make firewire_tester give a false
negative but most likely the connection was always good).
...
</pre>
<blockquote type="cite">
<pre wrap=""><span class="moz-txt-citetags">> </span>I have searched through the archives and seem no closer to a best
<span class="moz-txt-citetags">> </span>solution, so I decided to seek commentary.
</pre>
</blockquote>
<pre wrap=""><!---->
You missed the dev list thread "LFireDev Error: Starting A/V
streaming: Cannot allocate memory". I'm lining up my ducks
to submit a complete and proper bug report for kernel.org so
this can be fixed without patching in future kernels.
-- bjm
</pre>
</body>
</html>