Bug 1412348

Hi Orion,

  Please could you upload the aarch64 rootn.exe file so that I can examine it and find out what readelf is complaining about ?

Cheers
  Nick

Created attachment 1240381 [details]
rootn.exe

Here you go.

Hi Orion,

> Here you go.

Thanks.  Unfortunately I cannot reproduce the problem. :-(  Running "readelf -W -S rootn.exe" produces the expected output and no error messages about the dynamic segment.  Plus when I check the dynamic segment manually, it does indeed fit in the file.

I am at a loss as how to proceed with this bug.  The code that checks the dynamic section size in readelf looks good to me.  The rootn.exe file looks OK.  I suppose that it is possible that the find-debuginfo.sh script (which is invoking readelf and then failing) is picking up an old version of readelf.  But I have tried versions from F23 (based upon binutils 2.25) and RHEL 6.9 (based upon binutils 2.20) and they both work as well.

Possibly the problem is output synchronization - ie the error message about the dynamic segment being too big is not caused by the rootn.exe file at all, but some other file that has been processed by readelf.  The error message is printed to stderr, so it is quite possible that the build log is not synchronising stdout and stderr output, or that multiple processes are running at the same time.  So, with that in mind, would you mind trying out a rawhide rebuild which should pick up a new binutils rpm (binutils-2.27-14.fc26) which includes the file name in all error messages.

Cheers
  Nick

I see that there is a new version of the rootn package (6.08.04) as compared to the version whose builds were failing in koji (6.08.02).  Does this new version also experience these build problems ?

Seems to be building now - https://koji.fedoraproject.org/koji/taskinfo?taskID=17392791

Thanks for taking a look.

I'm seeing the same issue while building HandBrake for RPMFusion (reproducible in mock locally on x86_64) and the result is an empty -debuginfo package:
[...]
+ /usr/lib/rpm/find-debuginfo.sh --strict-build-id -m --run-dwz --dwz-low-mem-die-limit 10000000 --dwz-max-die-limit 110000000 /builddir/build/BUILD/HandBrake-1.0.2
extracting debug info from /builddir/build/BUILDROOT/HandBrake-1.0.2-1.fc26.x86_64/usr/bin/ghb
readelf: Error: the dynamic segment offset + size exceeds the size of the file
extracting debug info from /builddir/build/BUILDROOT/HandBrake-1.0.2-1.fc26.x86_64/usr/bin/HandBrakeCLI
readelf: Error: the dynamic segment offset + size exceeds the size of the file
dwz: Too few files for multifile optimization
/usr/lib/rpm/sepdebugcrcfix: Updated 0 CRC32s, 2 CRC32s did match.
[...]

Here's the affected package:
https://www.greysector.net/~rathann/review/HandBrake/HandBrake-1.0.2-1.fc25.src.rpm

Created attachment 1245468 [details]
one of HandBrake's binaries about which readelf is complaining

Hi Dominik,

  This is very strange.  I cannot reproduce this failure.  I downloaded the binary you provided in comment #6 and tried to get readelf to complain about it, but to no avail.  All the versions that I tried, including rawhide, F25, F23 and even RHEL 6.9, all think that the binary is OK and that there are no problems with the dynamic segment.

  Can I ask a few questions:

  * Which version of the binutils rpm do you have installed ?

  * Are you running the build on a 32-bit host ?

  * If you run "readelf -W -S HandBrakeCLI" are any errors reported ?

Cheers
  Nick

This bug appears to have been reported against 'rawhide' during the Fedora 26 development cycle.
Changing version to '26'.

Another case, this time fedora-26-i386's systemd:

e.g.: https://kojipkgs.fedoraproject.org//packages/systemd/233/3.fc26/data/logs/i686/build.log

...
extracting debug info from /builddir/build/BUILDROOT/systemd-233-3.fc26.i386/usr/lib/systemd/tests/test-arphrd-list
readelf: /builddir/build/BUILDROOT/systemd-233-3.fc26.i386/usr/lib/debug/usr/lib/systemd/tests/test-arphrd-list.debug: Error: the dynamic segment offset + size exceeds the size of the file
extracting debug info from /builddir/build/BUILDROOT/systemd-233-3.fc26.i386/usr/lib/systemd/tests/test-condition
...

ATM, this case can be reproduced by rebuilding systemd-233-3.fc26.src.rpm from fedora's git (fc26 branch) in local mocks (on x86_64, in case this should matter)

Hi Ralf,

  Could this be related to:

https://bugzilla.redhat.com/show_bug.cgi?id=1434050

  If so, does the problem still exist with the binutils-2.27-21.fc26 rpm that was created today ?

Cheers
  Nick

Actually the empty debuginfo package was not readelf's fault. Sorry for the false positive.

Also sorry, but I didn't have time to dig into this. The bug you referenced in comment 10 looks like it could be the same. I'll try to keep an eye out for these error messages in my next builds.

(In reply to Nick Clifton from comment #10)
>   If so, does the problem still exist with the binutils-2.27-21.fc26 rpm
> that was created today ?

I can confirm, this error-message is gone when building 
systemd-233-3.fc26.i386 w/ binutils-2.27-21.fc26.

*** This bug has been marked as a duplicate of bug 1434050 ***

binutils-2.27-21.fc26 has been submitted as an update to Fedora 26. https://bodhi.fedoraproject.org/updates/FEDORA-2017-519de08d5c

binutils-2.27-21.fc26 has been pushed to the Fedora 26 testing repository. If problems still persist, please make note of it in this bug report.
See https://fedoraproject.org/wiki/QA:Updates_Testing for
instructions on how to install test updates.
You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2017-519de08d5c

Experienced the same issue on Fedora 25, binutils-2.27-21.fc26.x86_64 fixed the issue.

binutils-2.27-21.fc26 has been pushed to the Fedora 26 stable repository. If problems still persist, please make note of it in this bug report.