Skip to contentSkip to conversation

V4L1 compatibility broken for VIDIOCGTUNER with radio

DISCUSSION Herton Ronaldo Krzesinski 3 participants 5 messages Dec 3, 2009 – Dec 4, 2009 Media (V4L/DVB)

lore
Share & tools
lore
#0 lore raw

Hi,

After commit 9bedc7f ("V4L/DVB (12429): v4l2-ioctl: fix G_STD and G_PARM default handlers"), radio software using V4L1 stopped to work on a saa7134 card, a git bisect pointed to this commit introducing the regression. All VIDIOCGTUNER calls on a v4l1 application are returning -EINVAL after this commit.

Investigating the issue, it turns out that v4l1_compat_get_tuner calls VIDIOC_G_STD ioctl, but as it is a radio device (saa7134-radio) it now is returning -EINVAL to user space applications which are being confused about this.

May be VIDIOC_G_STD change in the commit above should be reverted, or v4l1_compat_get_tuner changed to not return error with G_STD, or not call G_STD ioctl for a radio device?

--
[]'s
Herton

#1 lore raw

Hi,

Am Donnerstag, den 03.12.2009, 16:56 -0200 schrieb Herton Ronaldo Krzesinski:

Quoted 13 lines

Hi,

After commit 9bedc7f ("V4L/DVB (12429): v4l2-ioctl: fix G_STD and G_PARM default handlers"), radio software using V4L1 stopped to work on a saa7134 card, a git bisect pointed to this commit introducing the regression. All VIDIOCGTUNER calls on a v4l1 application are returning -EINVAL after this commit.

Investigating the issue, it turns out that v4l1_compat_get_tuner calls VIDIOC_G_STD ioctl, but as it is a radio device (saa7134-radio) it now is returning -EINVAL to user space applications which are being confused about this.

Quoted 7 lines

May be VIDIOC_G_STD change in the commit above should be reverted, or v4l1_compat_get_tuner changed to not return error with G_STD, or not call G_STD ioctl for a radio device?

--
[]'s
Herton

it was fixed here.

http://linuxtv.org/hg/v4l-dvb/rev/58ecda742a70

Maybe it was not ported to stable?

Cheers,
Hermann

#2 lore raw

Em Qui 03 Dez 2009, às 19:54:29, hermann pitton escreveu:

Quoted 21 lines

Hi,

Am Donnerstag, den 03.12.2009, 16:56 -0200 schrieb Herton Ronaldo Krzesinski:

Quoted 17 lines

Hi,

After commit 9bedc7f ("V4L/DVB (12429): v4l2-ioctl: fix G_STD and G_PARM default handlers"), radio software using V4L1 stopped to work on a saa7134 card, a git bisect pointed to this commit introducing the regression. All VIDIOCGTUNER calls on a v4l1 application are returning -EINVAL after this commit.

Investigating the issue, it turns out that v4l1_compat_get_tuner calls VIDIOC_G_STD ioctl, but as it is a radio device (saa7134-radio) it now is returning -EINVAL to user space applications which are being confused about this.

May be VIDIOC_G_STD change in the commit above should be reverted, or v4l1_compat_get_tuner changed to not return error with G_STD, or not call G_STD ioctl for a radio device?

Quoted 7 lines
Quoted 3 lines

--
[]'s
Herton

it was fixed here.

http://linuxtv.org/hg/v4l-dvb/rev/58ecda742a70

Indeed, thanks for the pointer. I forgot to check latest v4l1-compat.c /o\

Quoted 2 lines

Maybe it was not ported to stable?

Not on latest stable (2.6.31.6), perhaps it should be forwarded.

Quoted 3 lines

Cheers,
Hermann

--
[]'s
Herton

#3 lore raw

Am Donnerstag, den 03.12.2009, 21:04 -0200 schrieb Herton Ronaldo Krzesinski:

Quoted 28 lines

Em Qui 03 Dez 2009, às 19:54:29, hermann pitton escreveu:

Quoted 27 lines

Hi,

Am Donnerstag, den 03.12.2009, 16:56 -0200 schrieb Herton Ronaldo Krzesinski:

Quoted 13 lines

Hi,

After commit 9bedc7f ("V4L/DVB (12429): v4l2-ioctl: fix G_STD and G_PARM default handlers"), radio software using V4L1 stopped to work on a saa7134 card, a git bisect pointed to this commit introducing the regression. All VIDIOCGTUNER calls on a v4l1 application are returning -EINVAL after this commit.

Investigating the issue, it turns out that v4l1_compat_get_tuner calls VIDIOC_G_STD ioctl, but as it is a radio device (saa7134-radio) it now is returning -EINVAL to user space applications which are being confused about this.

Quoted 7 lines

May be VIDIOC_G_STD change in the commit above should be reverted, or v4l1_compat_get_tuner changed to not return error with G_STD, or not call G_STD ioctl for a radio device?

--
[]'s
Herton

it was fixed here.

Quoted 9 lines
Quoted 1 line

http://linuxtv.org/hg/v4l-dvb/rev/58ecda742a70

Indeed, thanks for the pointer. I forgot to check latest v4l1-compat.c /o\

Quoted 2 lines

Maybe it was not ported to stable?

Not on latest stable (2.6.31.6), perhaps it should be forwarded.

Yes, for sure. It's our fault.

Seems we had an "internal server error" :(

I came across it by accident.

Quoted 3 lines

The only other issue I'm aware of is that radio is broken since guessed 8 weeks on my tuners, only realized when testing on enabling external active antenna voltage for DVB-T on a/some 310i.

I did the bisect with some delay and Hans marked the fix with priority "high", but we missed to point Mike at it for stable explicitly.

Mike, please review and forward.

Sorry,
Hermann

#4 lore raw
Quoted 51 lines · hermann pitton

Am Donnerstag, den 03.12.2009, 21:04 -0200 schrieb Herton Ronaldo Krzesinski:

Quoted 28 lines

Em Qui 03 Dez 2009, às 19:54:29, hermann pitton escreveu:

Quoted 27 lines

Hi,

Am Donnerstag, den 03.12.2009, 16:56 -0200 schrieb Herton Ronaldo Krzesinski:

Quoted 13 lines

Hi,

After commit 9bedc7f ("V4L/DVB (12429): v4l2-ioctl: fix G_STD and G_PARM default handlers"), radio software using V4L1 stopped to work on a saa7134 card, a git bisect pointed to this commit introducing the regression. All VIDIOCGTUNER calls on a v4l1 application are returning -EINVAL after this commit.

Investigating the issue, it turns out that v4l1_compat_get_tuner calls VIDIOC_G_STD ioctl, but as it is a radio device (saa7134-radio) it now is returning -EINVAL to user space applications which are being confused about this.

Quoted 7 lines

May be VIDIOC_G_STD change in the commit above should be reverted, or v4l1_compat_get_tuner changed to not return error with G_STD, or not call G_STD ioctl for a radio device?

--
[]'s
Herton

it was fixed here.

Quoted 9 lines
Quoted 1 line

http://linuxtv.org/hg/v4l-dvb/rev/58ecda742a70

Indeed, thanks for the pointer. I forgot to check latest v4l1-compat.c /o\

Quoted 2 lines

Maybe it was not ported to stable?

Not on latest stable (2.6.31.6), perhaps it should be forwarded.

Yes, for sure. It's our fault.

Seems we had an "internal server error" :(

I came across it by accident.

Quoted 3 lines

The only other issue I'm aware of is that radio is broken since guessed 8 weeks on my tuners, only realized when testing on enabling external active antenna voltage for DVB-T on a/some 310i.

Quoted 7 lines

I did the bisect with some delay and Hans marked the fix with priority "high", but we missed to point Mike at it for stable explicitly.

Mike, please review and forward.

Sorry,
Hermann

Already done. It's queued for 2.6.31.7 The delay was due to the standard bureaucracy ... nothing we haven't seen before -- hopefully the distros will take in in once it's merged to stable, which should probably be within the next week or two.

Thanks,

Mike