[linux-dvb] Hauppauge WinTV regreession from 2.6.24 to 2.6.25

Mauro Carvalho Chehab mchehab at infradead.org
Sat Apr 26 13:59:18 CEST 2008


On Fri, 25 Apr 2008 11:06:57 -0400
mkrufky at linuxtv.org wrote:

> Mauro Carvalho Chehab wrote:
> > On Fri, 25 Apr 2008 10:40:14 -0400
> > "Michael Krufky" <mkrufky at linuxtv.org> wrote:
> >
> >   
> >> On Fri, Apr 25, 2008 at 9:56 AM, Mauro Carvalho Chehab
> >> <mchehab at infradead.org> wrote:
> >>     
> >>> On Thu, 24 Apr 2008 05:55:28 +0200
> >>>  hermann pitton <hermann-pitton at arcor.de> wrote:
> >>>
> >>>  > > > >>>> I am testing a kernel upgrade from 2.6.24.to 2.6.25, and the
> drivers
> >>>  > > > >>>> for   the Hauppauge WinTV appear to have suffered some
> regression
> >>>  > > > >>>> between the two kernel versions.
> >>>
> >>>
> >>>       
> >>>> do you see the auto detection issue?
> >>>>         
> >>>  >
> >>>  > Either tell it is just nothing, what I very seriously doubt, or
> please
> >>>  > comment.
> >>>  >
> >>>  > I don't like to end up on LKML again getting told that written rules
> >>>  > don't exist ;)
> >>>
> >>>  Sorry for now answer earlier. Too busy here, due to the merge window.
> >>>
> >>>  This seems to be an old bug. On several cases, tuner_type information
> came from
> >>>  some sort of autodetection schema, but the proper setup is not sent to
> tuner.
> >>>
> >>>  Please test the enclosed patch. It warrants that TUNER_SET_TYPE_ADDR is
> called
> >>>  at saa7134_board_init2() for all those boards:
> >>>
> >>>  SAA7134_BOARD_ADS_DUO_CARDBUS_PTV331
> >>>  SAA7134_BOARD_ASUS_EUROPA2_HYBRID
> >>>  SAA7134_BOARD_ASUSTeK_P7131_DUAL
> >>>  SAA7134_BOARD_ASUSTeK_P7131_HYBRID_LNA
> >>>  SAA7134_BOARD_AVERMEDIA_SUPER_007
> >>>  SAA7134_BOARD_BEHOLD_COLUMBUS_TVFM
> >>>  SAA7134_BOARD_BMK_MPEX_NOTUNER
> >>>  SAA7134_BOARD_BMK_MPEX_TUNER
> >>>  SAA7134_BOARD_CINERGY_HT_PCI
> >>>  SAA7134_BOARD_CINERGY_HT_PCMCIA
> >>>  SAA7134_BOARD_CREATIX_CTX953
> >>>  SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS
> >>>  SAA7134_BOARD_FLYDVB_TRIO
> >>>  SAA7134_BOARD_HAUPPAUGE_HVR1110
> >>>  SAA7134_BOARD_KWORLD_ATSC110
> >>>  SAA7134_BOARD_KWORLD_DVBT_210
> >>>  SAA7134_BOARD_MD7134
> >>>  SAA7134_BOARD_MEDION_MD8800_QUADRO
> >>>  SAA7134_BOARD_PHILIPS_EUROPA
> >>>  SAA7134_BOARD_PHILIPS_TIGER
> >>>  SAA7134_BOARD_PHILIPS_TIGER_S
> >>>  SAA7134_BOARD_PINNACLE_PCTV_310i
> >>>  SAA7134_BOARD_TEVION_DVBT_220RF
> >>>  SAA7134_BOARD_TWINHAN_DTV_DVB_3056
> >>>  SAA7134_BOARD_VIDEOMATE_DVBT_200
> >>>  SAA7134_BOARD_VIDEOMATE_DVBT_200A
> >>>  SAA7134_BOARD_VIDEOMATE_DVBT_300
> >>>
> >>>  It is important to test the above boards, to be sure that no regression
> is
> >>>  caused.
> >>>
> >>>  Signed-off-by: Mauro Carvalho Chehab <mchehab at infradead.org>
> >>>
> >>>  diff -r 60110897e86a linux/drivers/media/video/saa7134/saa7134-cards.c
> >>>  --- a/linux/drivers/media/video/saa7134/saa7134-cards.c Fri Apr 25
> 08:04:54 2008 -0300
> >>>  +++ b/linux/drivers/media/video/saa7134/saa7134-cards.c Fri Apr 25
> 10:44:16 2008 -0300
> >>>       
> >> Mauro,
> >>
> >> I didn't review your patch yet, and it needs to be tested, however,
> >> the bug reported in this thread deals with the same regression that
> >> you are attempting to repair, but on the cx88 driver -- not the
> >> saa7134 driver.
> >>     
> >
> > Hmm... it seems that people merged two similar issues together, on
> different
> > drivers. At least, part of the reports at the thread were with saa7134
> driver.
> >
> > I'll investigate if this solution will also work for cx88.
> 
> Mauro,
> 
> "...people merged two similar issues together, on different drivers..."  
> It was you -- did you forget?
> 
> cx88: http://linuxtv.org/hg/v4l-dvb/rev/2eb392c86745
> 
> saa7134: http://linuxtv.org/hg/v4l-dvb/rev/e7668fc3666c
> 
> I'm surprised that you don't remember this -- you pushed this to Linus 
> late in the 2.6.25-rcX, after I had strongly advised against this -- I 
> warned you that this may create regressions, needed thorough testing, 
> and was too risky a change to push into the middle of 2.6.25-rc
> 
> I hate to say, "I told you so" .... but.............
> 
> ;-)
> 
> Lets get your fixes tested ASAP so we can fix 2.6.25-stable.

The fix weren't caused by the changesets you've pointed, but are much older. If
you try to revert both changesets and load tuner before cx88/saa7134, or
compile tuner statically, you'll notice exact the same issue.

The issue is that set_type_addr were called at the wrong place.

Anyway, I've just committed a patch that should fix this for cx88. I'll soon
use the same logic to fix also saa7134.

I've also added a patch for tuner-core, to improve debug (of course, this
doesn't need to go to -stable). This helps to see the bug, if tuner debug is
enabled.

Cheers,
Mauro



More information about the linux-dvb mailing list