Re: Broken "make install" in 2.6.14-git1

["Larry W. Finger" <Larry.Finger@lwfinger.net>]

At 04:19 PM 10/30/2005, Andrew Morton wrote:
> larry.finger@att.net (Larry.Finger@lwfinger.net) wrote:
> >
> >  -------------- Original message ----------------------
> > From: Andrew Morton <akpm@osdl.org>
> > > larry.finger@att.net (Larry.Finger@lwfinger.net) wrote:
> > > >
> > > > The changes introduced the commit 
> 596c96ba05e5d56e72451e02f93f4e15e17458df
> > > break the initrd building step of the "make install" process. 
> The console output
> > > is as follows:
> > >
> > > I'm unable to locate that commit, perhaps due to a local lack 
> of gittiness.
> > >
> > > Can you describe the patch less cryptically?
> > >
> >
> > I'm not very good with git. All that bisect visualize shows for 
> the patch is the following:
> >
> > Author: Linus Torvalds <torvalds@g5.osdl.org>  2005-10-29 16:02:16
> > Committer: Linus Torvalds <torvalds@g5.osdl.org>  2005-10-29 16:02:16
> > Parent: e9d52234e35b27ea4ea5f2ab64ca47b1a0c740ab (Merge branch 
> 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus)
> >
> > Obviously, it consists of a number of patches, but I don't know 
> how to elaborate. Is there some git command to get the full description?
>
> Yes, that's a sort of empty marker which indicates the point at which Linus
> merged the MIPS git tree.  It's rather bad of git-bisect if it told you
> that this was offending patch.
>
> > > > >sudo make install
> > > >   CHK     include/linux/version.h
> > > >   CHK     include/linux/compile.h
> > > >   SKIPPED include/linux/compile.h
> > > >   CHK     usr/initramfs_list
> > > > Kernel: arch/i386/boot/bzImage is ready  (#97)
> > > > sh /home/finger/kernel/linux/arch/i386/boot/install.sh 
> 2.6.14-g596c96ba
> > > arch/i386/boot/bzImage System.map "/boot"
> > > > Root device:    /dev/hda6 (mounted on / as reiserfs)
> > > > Module list:    via82cxxx processor thermal fan reiserfs
> > > >
> > > > Kernel image:   /boot/vmlinuz-2.6.14-g596c96ba
> > > > Initrd image:   /boot/initrd-2.6.14-g596c96ba
> > > > Shared libs:    lib/ld-2.3.5.so lib/libblkid.so.1.0 lib/libc-2.3.5.so
> > > lib/libselinux.so.1 lib/libuuid.so.1.2
> > > > Driver modules: via82cxxx processor thermal fan reiserfs
> > > > Filesystem modules:
> > > > Including:      klibc initramfs udev fsck.reiserfs
> > > > Bootsplash:     SuSE (1024x768)
> > > > 8358 blocks
> > > > no record for '/block/hdc/uevent' in database
> > > > Use of uninitialized value in scalar chomp at
> > > > /usr/lib/perl5/vendor_perl/5.8.7/Bootloader/Tools.pm line 139.
> > > > Use of uninitialized value in concatenation (.) or string at
> > > /usr/lib/perl5/vendor_perl/5.8.7/Bootloader/Tools.pm line 140.
> > > > ......
> > > >
> > > > I used git bisect to localize the bad commit. I also observed 
> that if the
> > > kernel created /sys/block/hdc/uevent, it failed. If this "file" 
> does not exist,
> > > the install worked.
> > >
> > > What does "it failed" mean?   Is this the same bug, or a different one?
> >
> > Same bug. _make install_ fails if the uevent files are present. 
> Sorry for the imprecision.
> >
>
> I don't know what'a happening here.  What program is saying "no record for
> '/block/hdc/uevent' in database"?

The problem turned out to be in Bootloader/Tools.pm where the code 
did not know how to handle the uevent files. Once I modified the perl 
script to skip over uevent items, all is well.

Thanks for your attention to this matter, and I'm sorry to have 
bothered the kernel group with a bug in a peripheral system.

Larry



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/