[syslinux] memdisk from 3.36 crashing on Dell PowerEdge 1955

[Andreas Kotes]

Hello again,

* H. Peter Anvin <hpa at zytor.com> [20070320 22:19]:
> Andreas Kotes wrote:
> >* Andreas Kotes <count-linux at flatline.de> [20070315 19:03]:
> >>... and then nothing ... I've no idea what to try now .. unless you've
> >>got any suggestions, I'll drop a mail to Dell Support tomorrow ...
> >
> >more findings:
> >
> >memdisk 3.31 works
> >memdisk 3.35 doesn't
> >memdisk 3.50 (git) says 'Error reading PBR'
> >
> >didn't dig deep enough into git to find out how to get to a specific
> >revision (if there is something like a revision at all with git), so I
> >didn't find out which patch broke it .. I'll be using memdisk from 3.31
> >for now, and will be happy to try code if this helps you.
> >
> 
> There is; there is also a tool called "git bisect" to help find the image.

damn cool tool, once you learn how to use it .. it seems to have it's
problems with multi-parent merges ..

I suspect the problem crept in with

[6a6ccb3746477ae339bd7e86f38d54d173ca3dc7] Merge with
git+ssh://master.kernel.org/pub/scm/boot/syslinux/syslinux.git#syslinux-3.3x

but can't pin it exactly so far ... :(

my trail:

testsys32:~/s2g-serv-bs-count/bom-work/syslinux# git bisect good
Bisecting: 8 revisions left to test after this
[e75b8e02220acac6314e98df0709580bc20f9ca2] spec: handle variants of "i386" like "i686"
(...ok...)
testsys32:~/s2g-serv-bs-count/bom-work/syslinux# git bisect good
Bisecting: 4 revisions left to test after this
[3048448410ff316881314dbd29740f67b57c04d2] Patch: Parse append arguments properly in mboot module
(...bad...)
testsys32:~/s2g-serv-bs-count/bom-work/syslinux# git bisect bad
Bisecting: 3 revisions left to test after this
[6a6ccb3746477ae339bd7e86f38d54d173ca3dc7] Merge with git+ssh://master.kernel.org/pub/scm/boot/syslinux/syslinux.git#syslinux-3.3x
(...bad, and stays bad...)
testsys32:~/s2g-serv-bs-count/bom-work/syslinux# git bisect bad
672e1a7b5133e49d29c924732419296a03ebe39c is first bad commit
(...)
testsys32:~/s2g-serv-bs-count/bom-work/syslinux# git checkout syslinux-3.3m
(...I used bad=syslinux-3.3m and good=syslinux-3.2x...)
testsys32:~/s2g-serv-bs-count/bom-work/syslinux# git revert 672e1a7b5133e49d29c924732419296a03ebe39c
(...still a bad version :(...)

I'd suspect I could only start dissecting the merge (ugh!), albeit the 'Error reading PBR' seemed to (re)appear some time during the 'bad' part of the bisection, and is present in the current master - may this be a pointer which could help with suggestions how to proceed?

Best regards,

   Andreas

-- 
"God is a comedian playing to an audience too afraid to laugh." -- H.L.Mencken