Date:	Sun, 3 Aug 2008 23:12:25 +0200
From:	Hans Verkuil <hverkuil@...all.nl>
To:	Kyuma Ohta <whatisthis.sowhat@...il.com>,
	Jean Delvare <khali@...ux-fr.org>
Cc:	ivtv-devel ML <ivtv-devel@...vdriver.org>,
	Video4Linux ML <video4linux-list@...hat.com>,
	Linux Kernel list <linux-kernel@...r.kernel.org>
Subject: Re: [ivtv-devel] [PATCH AVAIL.]ivtv:Crash 2.6.26 with KUROTOSIKOU CX23416-STVLP

Hi Ohta,

Well, I picked up my card this weekend and tested it. It turns out to be 
an i2c-core.c bug: chips with i2c addresses in the 0x5x range are 
probed differently than other chips and the probe command contains an 
error. The upd64083 has an address in that range and so was hit by this 
bug. The attached patch for linux/drivers/i2c/i2c-core.c will fix it.

As you can see, this mail also goes to Jean Delvare so that he can move 
this upstream (should also go to the 2.6.26-stable series, Jean!).

For the ivtv driver this bug will only hit cards where ivtv has to probe 
for an upd64083.

SoB for this patch:

Signed-off-by: Hans Verkuil <hverkuil@...all.nl>

I've verified that this is only an issue with kernels 2.6.26 and up. 
Older kernels are not affected unless the ivtv driver from the v4l-dvb 
repository is used. To be more precise: this bug has been in i2c-core.c 
since 2.6.22, but the ivtv driver in 2.6.26 was the first driver that 
used i2c_new_probed_device() with an i2c address in a range that caused 
the broken probe to be used.

Thanks for the report Ohta!

Regards,

	Hans

On Friday 18 July 2008 01:13:45 Kyuma Ohta wrote:
> Dear Hans,
> Thanx for reply.
>
> Hans Verkuil wrote:
> > On Thursday 17 July 2008 17:20:14 Kyuma Ohta wrote:
> > > Hi,
> > > I'm testing 2.6.26/amd64 with Athlon64 x2 Box with
> > > KUROTOSIKOU CX23416-STVLP,always crash ivtv driver
> > > when loading upd64083 driver.
> > > I checked crash dump,this issue cause of loading
> > > upd64083.ko with i2c_probed_new_device().
> > > So,I fixed ivtv-i2c.c of 2.6.26 vanilla,and
> > > fixed *pretty* differnce memory allocation,structure
> > > of upd64083.c.
> > > I'm running patched 2.6.26 vanilla with below attached
> > > patches over 24hrs,and over 10hrs recording from ivtv,
> > > not happend anything;-)
> > > Please apply below to 2.6.26.x..
> > >
> > > Best regards,
> > > Ohta.
> >
> > Hi Ohta,
> >
> > Thanks for the patches. If I'm not mistaken there are several
> > variants of this card: without upd* devices, only with upd64083 and
> > with both upd devices. Which one do you have?
> >
> > Can you also show the dmesg output when ivtv loads?
> >
> > Looking at the four patches, I would say that the only relevant
> > patch is the fix-probing patch. If you try it with only that one
> > applied, does it still work correct for you? Note that this patch
> > will not work with a KUROTOSIKOU card that has no upd* devices at
> > all.
> >
> > Can you also give me the kernel backtrace when you load ivtv with
> > the vanilla 2.6.26? I do not quite understand why it should crash.
> >
> > Regards,
> >
> > 	Hans
>
> I have a ivtv card with *both* upd64083 and upd64031a.
> I don't still try testing apply only one of patch,only
> apply all of...
>
> I attach compressed logs when loading ivtv at boottime,
> parallel probing saa7134 v4l2 device,
> both applied (successed) ,not applied (failed).
>
> Best regards,
> Ohta
>
>
>
> E-Mail: whatisthis.sowhat@...il.com (Public)
> Home Page: http://d.hatena.ne.jp/artane/
>   (Sorry,not maintaining,and written in Japanese only...)
> Twitter: Artanejp (Mainly Japanese)
> ICQ: 366538955
> KEYID: 6B79F95F
> FINGERPRINT:
> 9AB3 8569 6033 FDBE 352B  CB6D DBFA B9E2 6B79 F95F



View attachment "i2c-core.diff" of type "text/x-diff" (572 bytes)

Powered by blists - more mailing lists