[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: {firewire,modem,superdrive}problems with 2.4.21-ben2

On Wed, 2003-07-09 at 15:20, Soeren Sonnenburg wrote:
> Hi...
> I still have trouble with the stuff in the subject on a 15" 1Ghz G4
> Powerbook. The neat thing about it is that it is all reproducable.
> Here we go:
> 1. firewire
>  do:
> 	modprobe ieee1394
> 	modprobe ohci1394 (<- bang leaves me in xmon with a nice message
> 	Jul  9 12:26:05 no kernel: kernel BUG at sched.c:720!
> 	Jul  9 12:26:05 no kernel: kernel BUG at
> 	/usr/src/linux/include/asm/semaphore.h:84!)
> leaving xmon twice returns me to the console and everything seems to be
> fine again.

I have approx. the same machine, I've had success with earlier 2.4.21
version, but it may be something broken in the latest. I'll check that
out and eventually merge a more recent 2.4 branch from the ieee1394
svn tree.

> modprobing sbp2 then finds the attached drives and accessing it seems to
> be ok, except that I my logs fill with stuff like:
> ...
> Jul  9 15:08:52 no kernel: ieee1394: unsolicited response packet received - np
> Jul  9 15:08:52 no kernel: ieee1394: contents: ffc09d20 ffc10000 00000000 0e945194
> Jul  9 15:09:52 no kernel: ieee1394: unsolicited response packet received - np
> Jul  9 15:09:52 no kernel: ieee1394: contents: ffc02520 ffc10000 00000000 db025194
> ...
> 2.modem
> I use the latest hcfusb driver as from their website. The driver loads
> ok and I can even use minicom dial up to some remote machine and the
> like. However when using wvdial I get a successful connection again and
> an IP assigned, but as soon as I ask about routing tables (route) or
> ping to the gateway I land in xmon.

Hrm... I did a quick test a couple of monthes ago when I helped fix
the wrong parameters issue (that caused the driver to work only once)
and it worked, but I try to avoid this modem as much as possible ;)
There might be something still wrong. I have spotted a couple of
possible races in the USB code in there, I plan to work with Marc
Boucher during OLS on these.

> 3.superdrive
> when using dvdrecord or the patched cdrecord, it seems to dvd-burn ok,
> but from time to time I get 
> Jul  9 12:21:51 no kernel: hdc: timeout waiting                        
> for dbdma command stop
> Jul  9 12:21:51 no kernel: hdc: bad status at DMA end, dstat=8400
> and the whole dvd-r is just garbage...
> I will happily help tracing bugs when someone with in detail knowlege
> tells me what to do.

The latest is interesting... Was it followed by an IDE or ATAPI error ?
If not, then we have a "short transfer" (that is the driver transfering
less than what was asked). That is nasty and is not dealt properly by
the current driver. I need to work with Jens Axboe on this issue, the
workaround of just terminating the transfer on the disk interrupt
isn't very satisfying and may expose some races.


Reply to: