Skip to contentSkip to conversation

2.6.27.40 build error

DISCUSSION Greg KH 4 participants 6 messages Dec 8, 2009 – Dec 9, 2009

lore
Share & tools
lore

Greg KH

author
#0 lore raw

It turns out that the patch below causes a build failure on the 2.6.27.40 kernel.

Michael, should I just drop it, or should I take out the line with the "capture_transfer_done" in it, as that member is not in the structure in this kernel version.

thanks,

greg k-h

Quoted 34 lines · Devin Heitmueller

From: Devin Heitmueller <[email protected]>

commit 96fbf771d86a90ff006bc62ca4d4de6474b3de31 upstream.

Because the counters were not reset when starting up streaming, they would be reused from the previous run. This can result in cases such that when the second instance of streaming starts up, the "cnt" variable in em28xx_audio_isocirq() can end up being negative, resulting in attempting to write to memory before the start of runtime->dma_area (as well as having a negative number of bytes to copy).

---
 drivers/media/video/em28xx/em28xx-audio.c |    5 +++++
 1 file changed, 5 insertions(+)
drivers/media/video/em28xx/em28xx-audio.c +5 -0
@@ -365,6 +365,11 @@ static int snd_em28xx_hw_capture_free(st
 
 static int snd_em28xx_prepare(struct snd_pcm_substream *substream)
 {
+	struct em28xx *dev = snd_pcm_substream_chip(substream);
+
+	dev->adev.hwptr_done_capture = 0;
+	dev->adev.capture_transfer_done = 0;
+
 	return 0;
 }
 
#1 lore raw
Quoted 10 lines · Greg KH

It turns out that the patch below causes a build failure on the 2.6.27.40 kernel.

Michael, should I just drop it, or should I take out the line with the "capture_transfer_done" in it, as that member is not in the structure in this kernel version.

thanks,

greg k-h

There were several different leaks and crash conditions addressed since 2.6.27, and it is entirely possible that just removing the line is not the correct course of action.

I would propose just dropping the patch for 2.6.27, as we would have to rework it taking into consideration the other fixes which were done since then.

Devin

Signature · 2 lines
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com
#2 lore raw
Quoted 20 lines · Devin Heitmueller
Quoted 10 lines · Greg KH

It turns out that the patch below causes a build failure on the 2.6.27.40 kernel.

Michael, should I just drop it, or should I take out the line with the "capture_transfer_done" in it, as that member is not in the structure in this kernel version.

thanks,

greg k-h

There were several different leaks and crash conditions addressed since 2.6.27, and it is entirely possible that just removing the line is not the correct course of action.

I would propose just dropping the patch for 2.6.27, as we would have to rework it taking into consideration the other fixes which were done since then.

Quoted 5 lines

Devin

--
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com

Greg,

Yes -- please drop it for now. I didn't actually send this for 2.6.27.y anyway -- it must have gotten to you some other way.

Devin, if 2.6.27.y needs this fix, can you please generate the appropriate backport and test it before we resubmit?

Cheers,

Mike Krufky

#3 lore raw
Quoted 5 lines · Michael Krufky

Yes -- please drop it for now.  I didn't actually send this for 2.6.27.y anyway -- it must have gotten to you some other way.

Devin, if 2.6.27.y needs this fix, can you please generate the appropriate backport and test it before we resubmit?

We would have to decide whether it's worth the effort to get a 2.6.27 environment setup to do the necessary regression testing. At this point, I would argue that driver was screwed up enough in 2.6.27 that I wouldn't be inclined to tell users it worked even with this one fix.

Devin

Signature · 2 lines
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com
#4 lore raw
Quoted 7 lines · Devin Heitmueller
Quoted 5 lines · Michael Krufky

Yes -- please drop it for now.  I didn't actually send this for 2.6.27.y anyway -- it must have gotten to you some other way.

Devin, if 2.6.27.y needs this fix, can you please generate the appropriate backport and test it before we resubmit?

Quoted 6 lines

We would have to decide whether it's worth the effort to get a 2.6.27 environment setup to do the necessary regression testing.  At this point, I would argue that driver was screwed up enough in 2.6.27 that I wouldn't be inclined to tell users it worked even with this one fix.

Devin

Fine with me -- I'm glad the v4l-dvb development repositories hosted on both linuxtv.org and kernellabs.com are backwards compatible with recent kernels... :-)

I think all the crucial fixes (except for the pvrusb2 16k firmware patch) are all merged into 2.6.31.y as of now... So, now we can focus on new driver support for 2.6.32 :-)

Cheers,

Mike

#5 lore raw
In reply to Devin Heitmueller
Quoted 12 lines · Devin Heitmueller
Quoted 10 lines · Greg KH

It turns out that the patch below causes a build failure on the 2.6.27.40 kernel.

Michael, should I just drop it, or should I take out the line with the "capture_transfer_done" in it, as that member is not in the structure in this kernel version.

thanks,

greg k-h

Quoted 7 lines

There were several different leaks and crash conditions addressed since 2.6.27, and it is entirely possible that just removing the line is not the correct course of action.

I would propose just dropping the patch for 2.6.27, as we would have to rework it taking into consideration the other fixes which were done since then.

Ok, I'll go do that.

thanks,

greg k-h