[media_build] regression at 3a17e11 "update v4.10_sched_signal.patch"

[media_build] regression at 3a17e11 "update v4.10_sched_signal.patch"

[Vincent McIntyre]

Hi

I think the build was broken by this commit.

  3a17e11 "update v4.10_sched_signal.patch"

It's been fun learning about git bisect.

$ cat /etc/lsb-release 
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=16.04
DISTRIB_CODENAME=xenial
DISTRIB_DESCRIPTION="Ubuntu 16.04.2 LTS"

The build script falls over on these kernels, in the same way

 4.4.0-77-generic x86_64
 4.8.0-53-generic x86_64


$ ./build --main-git --depth 1 -v 1
...<elided>
******************
* Start building *
* ******************
* $ make allyesconfig
* make -C /home/me/git/clones/media_build/v4l allyesconfig
* make[1]: Entering directory '/home/me/git/clones/media_build/v4l'
* make[2]: Entering directory '/home/me/git/clones/media_build/linux'
* Syncing with dir ../media/
* Applying patches for kernel 4.4.0-77-generic
* patch -s -f -N -p1 -i ../backports/api_version.patch
* patch -s -f -N -p1 -i ../backports/pr_fmt.patch
* patch -s -f -N -p1 -i ../backports/debug.patch
* patch -s -f -N -p1 -i ../backports/drx39xxj.patch
* patch -s -f -N -p1 -i ../backports/v4.10_sched_signal.patch
* 1 out of 1 hunk FAILED
* Makefile:137: recipe for target 'apply_patches' failed
* make[2]: *** [apply_patches] Error 1
* make[2]: Leaving directory '/home/me/git/clones/media_build/linux'
* Makefile:383: recipe for target 'allyesconfig' failed
* make[1]: *** [allyesconfig] Error 2
* make[1]: Leaving directory '/home/me/git/clones/media_build/v4l'
* Makefile:26: recipe for target 'allyesconfig' failed
* make: *** [allyesconfig] Error 2
* can't select all drivers at ./build.new line 521
*

Hopefully this of some use to someone
Vince

Re: [media_build] regression at 3a17e11 "update v4.10_sched_signal.patch"

[Vincent McIntyre]

I managed to find the failing patch, not sure what the fix is.

$ cd linux/
$ patch -f -N -p1 -i ../backports/v4.10_sched_signal.patch
patching file drivers/media/dvb-core/dvb_ca_en50221.c
Hunk #1 succeeded at 35 (offset 1 line).
patching file drivers/media/dvb-core/dvb_demux.c
Hunk #1 succeeded at 20 with fuzz 1 (offset 1 line).
patching file drivers/media/dvb-core/dvb_frontend.c
Hunk #1 succeeded at 30 (offset 1 line).
patching file drivers/media/pci/cx18/cx18-driver.h
patching file drivers/media/pci/ivtv/ivtv-driver.c
patching file drivers/media/pci/ivtv/ivtv-driver.h
Hunk #1 succeeded at 39 (offset 1 line).
patching file drivers/media/pci/pt1/pt1.c
patching file drivers/media/pci/pt3/pt3.c
patching file drivers/media/pci/solo6x10/solo6x10-i2c.c
patching file drivers/media/pci/zoran/zoran_device.c
patching file drivers/media/platform/vivid/vivid-radio-rx.c
patching file drivers/media/platform/vivid/vivid-radio-tx.c
patching file drivers/media/rc/lirc_dev.c
Hunk #1 FAILED at 18.
1 out of 1 hunk FAILED -- saving rejects to file drivers/media/rc/lirc_dev.c.rej
patching file drivers/media/usb/cpia2/cpia2_core.c
patching file drivers/media/usb/gspca/cpia1.c
Hunk #1 succeeded at 28 (offset 1 line).
patching file drivers/media/v4l2-core/videobuf-dma-sg.c
patching file drivers/staging/media/lirc/lirc_zilog.c
patching file include/media/v4l2-ioctl.h

$ cat drivers/media/rc/lirc_dev.c.rej
--- drivers/media/rc/lirc_dev.c
+++ drivers/media/rc/lirc_dev.c
@@ -18,7 +18,7 @@
 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
 
 #include <linux/module.h>
-#include <linux/sched/signal.h>
+#include <linux/sched.h>
 #include <linux/ioctl.h>
 #include <linux/poll.h>
 #include <linux/mutex.h>

Re: [media_build] regression at 3a17e11 "update v4.10_sched_signal.patch"

[Vincent McIntyre]

> 
> $ cat drivers/media/rc/lirc_dev.c.rej
> --- drivers/media/rc/lirc_dev.c
> +++ drivers/media/rc/lirc_dev.c
> @@ -18,7 +18,7 @@
>  #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
>  
>  #include <linux/module.h>
> -#include <linux/sched/signal.h>
> +#include <linux/sched.h>
>  #include <linux/ioctl.h>
>  #include <linux/poll.h>
>  #include <linux/mutex.h>
> 

A bit of staring brings this to light:

The file that is being patched has extra lines relative to the patch

     18 #undef pr_fmt
     19 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
     20 
     21 #include <linux/module.h>
**   22 #include <linux/kernel.h>
     23 #include <linux/sched/signal.h>
**   24 #include <linux/errno.h>
     25 #include <linux/ioctl.h>
     26 #include <linux/fs.h>
     27 #include <linux/poll.h>
     28 #include <linux/completion.h>
     29 #include <linux/mutex.h>
     30 #include <linux/wait.h>

This hunk applies cleanly, and seems to match up with recent kernel
code (eg
   174cd4b1e5fbd0d74c68cf3a74f5bd4923485512
   sched/headers: Prepare to move signal wakeup & sigpending methods
    from <linux/sched.h> into <linux/sched/signal.h>)

@@ -20,7 +20,7 @@
 
 #include <linux/module.h>
 #include <linux/kernel.h>
-#include <linux/sched/signal.h>
+#include <linux/sched.h>
 #include <linux/errno.h>
 #include <linux/ioctl.h>
 #include <linux/fs.h>

HTH
Vince

Re: [media_build] regression at 3a17e11 "update v4.10_sched_signal.patch"

[Hans Verkuil]

On 08/06/17 15:28, Vincent McIntyre wrote:
> I managed to find the failing patch, not sure what the fix is.
> 
> $ cd linux/
> $ patch -f -N -p1 -i ../backports/v4.10_sched_signal.patch
> patching file drivers/media/dvb-core/dvb_ca_en50221.c
> Hunk #1 succeeded at 35 (offset 1 line).
> patching file drivers/media/dvb-core/dvb_demux.c
> Hunk #1 succeeded at 20 with fuzz 1 (offset 1 line).
> patching file drivers/media/dvb-core/dvb_frontend.c
> Hunk #1 succeeded at 30 (offset 1 line).
> patching file drivers/media/pci/cx18/cx18-driver.h
> patching file drivers/media/pci/ivtv/ivtv-driver.c
> patching file drivers/media/pci/ivtv/ivtv-driver.h
> Hunk #1 succeeded at 39 (offset 1 line).
> patching file drivers/media/pci/pt1/pt1.c
> patching file drivers/media/pci/pt3/pt3.c
> patching file drivers/media/pci/solo6x10/solo6x10-i2c.c
> patching file drivers/media/pci/zoran/zoran_device.c
> patching file drivers/media/platform/vivid/vivid-radio-rx.c
> patching file drivers/media/platform/vivid/vivid-radio-tx.c
> patching file drivers/media/rc/lirc_dev.c
> Hunk #1 FAILED at 18.
> 1 out of 1 hunk FAILED -- saving rejects to file drivers/media/rc/lirc_dev.c.rej
> patching file drivers/media/usb/cpia2/cpia2_core.c
> patching file drivers/media/usb/gspca/cpia1.c
> Hunk #1 succeeded at 28 (offset 1 line).
> patching file drivers/media/v4l2-core/videobuf-dma-sg.c
> patching file drivers/staging/media/lirc/lirc_zilog.c
> patching file include/media/v4l2-ioctl.h
> 
> $ cat drivers/media/rc/lirc_dev.c.rej
> --- drivers/media/rc/lirc_dev.c
> +++ drivers/media/rc/lirc_dev.c
> @@ -18,7 +18,7 @@
>  #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
>  
>  #include <linux/module.h>
> -#include <linux/sched/signal.h>
> +#include <linux/sched.h>
>  #include <linux/ioctl.h>
>  #include <linux/poll.h>
>  #include <linux/mutex.h>
> 

Odd, it applies cleanly here.

But don't bother, the media_build totally broke after the latest round of
commits to the master. We're looking into that, but it might take a bit
of time before it is resolved.

Regards,

	Hans

Re: [media_build] regression at 3a17e11 "update v4.10_sched_signal.patch"

[Vincent McIntyre]

On Thu, Jun 08, 2017 at 04:42:30PM +0200, Hans Verkuil wrote:
> On 08/06/17 15:28, Vincent McIntyre wrote:
> > I managed to find the failing patch, not sure what the fix is.
> > 
> > $ cd linux/
> > $ patch -f -N -p1 -i ../backports/v4.10_sched_signal.patch
> > patching file drivers/media/dvb-core/dvb_ca_en50221.c
> > Hunk #1 succeeded at 35 (offset 1 line).
> > patching file drivers/media/dvb-core/dvb_demux.c
> > Hunk #1 succeeded at 20 with fuzz 1 (offset 1 line).
> > patching file drivers/media/dvb-core/dvb_frontend.c
> > Hunk #1 succeeded at 30 (offset 1 line).
> > patching file drivers/media/pci/cx18/cx18-driver.h
> > patching file drivers/media/pci/ivtv/ivtv-driver.c
> > patching file drivers/media/pci/ivtv/ivtv-driver.h
> > Hunk #1 succeeded at 39 (offset 1 line).
> > patching file drivers/media/pci/pt1/pt1.c
> > patching file drivers/media/pci/pt3/pt3.c
> > patching file drivers/media/pci/solo6x10/solo6x10-i2c.c
> > patching file drivers/media/pci/zoran/zoran_device.c
> > patching file drivers/media/platform/vivid/vivid-radio-rx.c
> > patching file drivers/media/platform/vivid/vivid-radio-tx.c
> > patching file drivers/media/rc/lirc_dev.c
> > Hunk #1 FAILED at 18.
> > 1 out of 1 hunk FAILED -- saving rejects to file drivers/media/rc/lirc_dev.c.rej
> > patching file drivers/media/usb/cpia2/cpia2_core.c
> > patching file drivers/media/usb/gspca/cpia1.c
> > Hunk #1 succeeded at 28 (offset 1 line).
> > patching file drivers/media/v4l2-core/videobuf-dma-sg.c
> > patching file drivers/staging/media/lirc/lirc_zilog.c
> > patching file include/media/v4l2-ioctl.h
> > 
> > $ cat drivers/media/rc/lirc_dev.c.rej
> > --- drivers/media/rc/lirc_dev.c
> > +++ drivers/media/rc/lirc_dev.c
> > @@ -18,7 +18,7 @@
> >  #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
> >  
> >  #include <linux/module.h>
> > -#include <linux/sched/signal.h>
> > +#include <linux/sched.h>
> >  #include <linux/ioctl.h>
> >  #include <linux/poll.h>
> >  #include <linux/mutex.h>
> > 
> 
> Odd, it applies cleanly here.
> 
> But don't bother, the media_build totally broke after the latest round of
> commits to the master. We're looking into that, but it might take a bit
> of time before it is resolved.
> 

Just to follow up on this.
The build still fell over during patching if I ran
   build --main-git --depth 1 -v 1
but did not if I ran
   build -v 1
(which causes the tarball to download instead)

This suggests an inconsistency between the two,
which is surprising and likely unintended.

I poked around a bit and found that the media subdir
was out of date and the build script was not updating it;
--depth 1 appears to suppress that.

Once I did a manual update, the patching succeeded.
I'll try to work out a patch so 'build' avoids this issue in future.

Vince