OMAP and arm-soc's for-next branch

OMAP and arm-soc's for-next branch

[Russell King - ARM Linux]

I've just re-merged the build tree for the nightly builds so it's now
based on v3.3, and pulled in the latest for-next from arm-soc.

I notice that the problem I reported earlier (see the "arm-soc + rmk's
tree boot failure on OMAP4430SDP" thread), which results in OMAP
being totally unable to boot, is still present.

As far as I'm concerned, as long as this remains unfixed in arm-soc,
arm-soc is not to push any branch containing the broken commit
(3ec2decb) upstream.

Re: OMAP and arm-soc's for-next branch

[Russell King - ARM Linux]

On Sat, Mar 24, 2012 at 12:22:56PM +0000, Russell King - ARM Linux wrote:
> I've just re-merged the build tree for the nightly builds so it's now
> based on v3.3, and pulled in the latest for-next from arm-soc.
> 
> I notice that the problem I reported earlier (see the "arm-soc + rmk's
> tree boot failure on OMAP4430SDP" thread), which results in OMAP
> being totally unable to boot, is still present.
> 
> As far as I'm concerned, as long as this remains unfixed in arm-soc,
> arm-soc is not to push any branch containing the broken commit
> (3ec2decb) upstream.

Just noticed that the fix for this has already been pushed into mainline.
So, what this means is that anyone trying to bisect across this merge
window with an OMAP platform will hit a range of commits which just won't
boot.  That's really great stuff when the merge window contains the most
probable set of commits for causing issues which would need to be bisected.

Re: OMAP and arm-soc's for-next branch

[Tony Lindgren]

* Russell King - ARM Linux <linux@arm.linux.org.uk> [120324 05:53]:
> On Sat, Mar 24, 2012 at 12:22:56PM +0000, Russell King - ARM Linux wrote:
> > I've just re-merged the build tree for the nightly builds so it's now
> > based on v3.3, and pulled in the latest for-next from arm-soc.
> > 
> > I notice that the problem I reported earlier (see the "arm-soc + rmk's
> > tree boot failure on OMAP4430SDP" thread), which results in OMAP
> > being totally unable to boot, is still present.
> > 
> > As far as I'm concerned, as long as this remains unfixed in arm-soc,
> > arm-soc is not to push any branch containing the broken commit
> > (3ec2decb) upstream.
> 
> Just noticed that the fix for this has already been pushed into mainline.
> So, what this means is that anyone trying to bisect across this merge
> window with an OMAP platform will hit a range of commits which just won't
> boot.  That's really great stuff when the merge window contains the most
> probable set of commits for causing issues which would need to be bisected.

Yes the boot issue got fixed by commit dc7e57fa (OMAPDSS: register dss drivers
in module init) as discussed in the "arm-soc + rmk's tree boot failure on
OMAP4430SDP" thread.

It's too bad that the problem causing commit and the fix for a driver come
from separate development trees as that causes bisect issues like you mention
above. Better coordination and testing is needed between drivers and core
omap changes to avoid issues like this in the future.

Regards,

Tony

Re: OMAP and arm-soc's for-next branch

[Arnd Bergmann]

On Saturday 24 March 2012, Russell King - ARM Linux wrote:
> On Sat, Mar 24, 2012 at 12:22:56PM +0000, Russell King - ARM Linux wrote:
> > I've just re-merged the build tree for the nightly builds so it's now
> > based on v3.3, and pulled in the latest for-next from arm-soc.
> > 
> > I notice that the problem I reported earlier (see the "arm-soc + rmk's
> > tree boot failure on OMAP4430SDP" thread), which results in OMAP
> > being totally unable to boot, is still present.
> > 
> > As far as I'm concerned, as long as this remains unfixed in arm-soc,
> > arm-soc is not to push any branch containing the broken commit
> > (3ec2decb) upstream.
> 
> Just noticed that the fix for this has already been pushed into mainline.
> So, what this means is that anyone trying to bisect across this merge
> window with an OMAP platform will hit a range of commits which just won't
> boot.  That's really great stuff when the merge window contains the most
> probable set of commits for causing issues which would need to be bisected.

I had already pulled the omap_dss2 branch from Tomi into the next/cleanup
branch of arm-soc, in order to resolve a bunch of merge conflicts that we
discussed earlier.

This should limit the number of broken commits in the history to five,
and I could reduce that further to two commits if I rebuild the
next/cleanup branch in a different order, but I think it's not worth
it for that.

I definitely agree that the cleanup branches need a little more care.
The idea is really that large bug harmless changes go in there, also
to help simplify the bisection process. If stuff breaks in the cleanups,
I consider that worse than bugs that come in through new code.

	Arnd

Re: OMAP and arm-soc's for-next branch

[Russell King - ARM Linux]

On Sat, Mar 24, 2012 at 06:14:42PM +0000, Arnd Bergmann wrote:
> On Saturday 24 March 2012, Russell King - ARM Linux wrote:
> > On Sat, Mar 24, 2012 at 12:22:56PM +0000, Russell King - ARM Linux wrote:
> > > I've just re-merged the build tree for the nightly builds so it's now
> > > based on v3.3, and pulled in the latest for-next from arm-soc.
> > > 
> > > I notice that the problem I reported earlier (see the "arm-soc + rmk's
> > > tree boot failure on OMAP4430SDP" thread), which results in OMAP
> > > being totally unable to boot, is still present.
> > > 
> > > As far as I'm concerned, as long as this remains unfixed in arm-soc,
> > > arm-soc is not to push any branch containing the broken commit
> > > (3ec2decb) upstream.
> > 
> > Just noticed that the fix for this has already been pushed into mainline.
> > So, what this means is that anyone trying to bisect across this merge
> > window with an OMAP platform will hit a range of commits which just won't
> > boot.  That's really great stuff when the merge window contains the most
> > probable set of commits for causing issues which would need to be bisected.
> 
> I had already pulled the omap_dss2 branch from Tomi into the next/cleanup
> branch of arm-soc, in order to resolve a bunch of merge conflicts that we
> discussed earlier.

I don't see the commit which Tony talked about - dc7e57fa (OMAPDSS:
register dss drivers in module init) in last night's arm-soc for-next
branch.  So, the only reason the autobuilds are able to boot is because
I'm manually reverting 3ec2decb.

Re: OMAP and arm-soc's for-next branch

[Arnd Bergmann]

On Sunday 25 March 2012, Russell King - ARM Linux wrote:
> On Sat, Mar 24, 2012 at 06:14:42PM +0000, Arnd Bergmann wrote:
> > On Saturday 24 March 2012, Russell King - ARM Linux wrote:
> > I had already pulled the omap_dss2 branch from Tomi into the next/cleanup
> > branch of arm-soc, in order to resolve a bunch of merge conflicts that we
> > discussed earlier.
> 
> I don't see the commit which Tony talked about - dc7e57fa (OMAPDSS:
> register dss drivers in module init) in last night's arm-soc for-next
> branch.  So, the only reason the autobuilds are able to boot is because
> I'm manually reverting 3ec2decb.
> 

Sorry, it seems I pulled the wrong subset. I've updated the omap/cleanup branch
with all of the fbdev updates now, which includes the commit you mentioned here.

Hope I got it right this time.

Sorry for the repeated screwups on this.

	Arnd