[hsflinux] Fun with Ubuntu Gutsy and kernel 2.6.22 et al
Linuxant support (Jonathan)
support at linuxant.com
Mon Oct 22 18:51:47 EDT 2007
the problem you have with the driver is likely related to the HDA bus
driver which has a problem to handle your hardware. The HSF driver ships
with a HDA bus driver provided by ALSA which is close to the current
code in ALSA mercurial. Hopefully newer code from the ALSA project will
provide better support for your HDA bus and this code could be
integrated in a future release of the driver.
I highly doubt that the kernel stack size is the issue, as far as I
know, x86 Ubuntu kernels have 8K of stacks which is more than enough for
the HSF driver.
Technical specialist / Linuxant
support at linuxant.com
Martin Espinoza wrote:
> Good morning, attached is my hsfdiag. The serial number's been stripped
> out of it (by me) because license status is not my problem.
> My problem is that the driver is quite unreliable. Interestingly, the
> source-code driver is less reliable than the prepackaged one. I hope
> that this can provide some insight into the problem.
> The following is a kernel message I got today:
> Oct 22 06:25:41 agamemnon kernel: [49775.292000] hda_intel:
> azx_get_response timeout, switching to polling mode...
> This is interesting because I watch network traffic with gkrellm and
> this message came after a break in the download. Naturally, this is not
> very scientific, because there are so many things which can affect such
> a download. However, being on a modem means that you are rarely waiting
> on the remote site; the opposite is generally true. In any case, then
> that error occurred, followed by a second or so of no transfer, and then
> the download picked back up again at the "full" rate - on my connection,
> about 3.1kB/sec :(
> I realize that I'm on the bleeding edge. Incidentally, I upgraded to
> 7.68.00.04 <http://7.68.00.04> (from 7.68.00.01 <http://7.68.00.01>)
> because .01 didn't work on 2.6.22 . I haven't read the release notes or
> anything, so this is probably well-known and the reason for the new
> release, but I thought I'd mention it anyway.
> When on Feisty I had built myself a kernel with the 16k stacks patch.
> But right now I need to do a lot of downloading before I can do any
> kernel building. I hope to get that done today while I'm someplace with
> a superior connection, and I'll see if it helps. I very much want to run
> the source release of the driver, so that I can upgrade my kernel and
> have it maybe work (it has in the past, though not going from 2.6.20 to
> 2.6.22. Still, no plan is perfect.)
> I sent a message in about my problems (I have and since feisty always
> have had the "cnxthsf_OsEventWaitTime(ea6136e0/HDA
> CodecUnsolicitedmessage, 40): returning OSEVENT_WAIT_TIMEOUT" problem)
> before upgrading to gutsy-rc, but the first time I sent it in without
> gzipping the hsfdiag, so I cancelled that one, then resent after
> gzipping it, and it was eaten by the listserver, or at some intermediate
> point. So I am now sending a more complete report, on a newer kernel.
> Is this driver going to live in kernel space forever, or is it going to
> go userspace? It seems like the whole linux world is heading towards a
> smaller stack size for performance reasons.
> hsflinux mailing list
> hsflinux at lists.linuxant.com
More information about the hsflinux