find-debuginfo.sh: decompress DWARF compressed ELF sections

[ignatenkobrain]

debugedit and dwz do not support DWARF compressed ELF sections, let's
just decompress those before extracting debuginfo.

Tested-by: Igor Gnatenko i.gnatenko.brain@gmail.com

[pmatilai]

This is causing test-suite failures on the following tests:
142: rpmbuild implicit suid binary FAILED (rpmbuild.at:700)
163: rpmbuild buildid hardlink alldebug FAILED (rpmbuildid.at:699)
165: rpmbuild buildid hardlink separate FAILED (rpmbuildid.at:829)
167: rpmbuild buildid hardlink compat FAILED (rpmbuildid.at:968)

[ignatenkobrain]

@pmatilai update elfutils.

[ignatenkobrain]

https://bugzilla.redhat.com/show_bug.cgi?id=1607044

[pmatilai]

Okay that helps, but requiring a bleeding-edge elfutils for normal operation is not that great...

[ignatenkobrain]

@pmatilai that's kinda 2-sided problem. On one side if you have your binary with compressed ELF sections, then find-debuginfo would not find anything and just fail your build... On the other side, just tests fail if you have old elfutils.

[rpm-maint]

On Tue, 2018-07-31 at 02:48 -0700, Igor Gnatenko wrote: @pmatilai that's kinda 2-sided problem. On one side if you have your binary with compressed ELF sections, then find-debuginfo would not find anything and just fail your build... On the other side, just tests fail if you have old elfutils.

Some of those tests are pretty important though. It means that files will loose their suid bit. Also note that this bug fix isn't in any released elfutils yet. I wouldn't add this unconditionally. It tries to fix something that isn't very common (go binaries using the very latest go toolchain).

[pmatilai]

Yeah, breaking common and previously cases to fix an esoteric one is not a good deal.

We'll need something else here that doesn't wreak havoc with older, widely used elfutils versions.

[pmatilai]

This is also breaking our CI at the moment, which is no good either. Can we easily detect compressed sections and only call eu-elfcompress in that case? Otherwise we'll probably need to revert this for the time being at least.

Requiring an explicit %attr for suid/sgid might be a perfectly reasonable thing to do overall but a random elfutils bug can't really be the driver for such a change.

[pmatilai]

Apparently nothing is going to happen if I just let it be (not that it's unexpected in a case like this). Reverted as of commit fb8eb1c:

This is breaking suid bit handling in packaging unless an unreleased
bleeding edge version of elfutils is used. In other words it's not
an acceptable requirement and wont be for a long time. This needs
to be fixed in some other way.

[eclipseo]

@pmatilai With latest elfutils, is this commit still posing a problem?
It lives as a patch in Fedora's RPM SPEC, but is not present in EPEL8 which will cause issues with Golang packages.