Hide Forgot
Description of problem: [ben@ulna2:~]$ readelf --unwind /usr/lib64/libm.so The decoding of unwind sections for machine type AArch64 is not currently supported. Version-Release number of selected component (if applicable): binutils-2.27-28.base.el7_5.1.aarch64 How reproducible: always Steps to Reproduce: see above. Actual results: The decoding of unwind sections for machine type AArch64 is not currently supported. Expected results: a dump of the .eh_frame section. Additional info: http://post-office.corp.redhat.com/archives/tools/2016-August/msg00054.html May be slightly related to: https://bugzilla.redhat.com/enter_bug.cgi?product=Red%20Hat%20Enterprise%20Linux%207
oops may be related to: https://bugzilla.redhat.com/show_bug.cgi?id=1626614
Whoa this isn't just ARM evidently we can't process x86_64 either [ben@Mustang ~]$ readelf -u /usr/lib64/libm.so.6 The decoding of unwind sections for machine type Advanced Micro Devices X86-64 is not currently supported. I thought it was just an ARM thing. [ben@Mustang ~]$ rpm -q binutils binutils-2.29.1-23.fc28.x86_64
Useful information on how to read the .eh_frame section: https://www.airs.com/blog/archives/460 Maybe the best plan is to extend readelf to process the .eh_frame sections for unwind info.
A workaround is readelf -wF or -wf however, it is unclear to me why -u isn't wired up to do the same thing as -wf. Is that something worth considering?
Hi Ben, Would you be terribly upset if I delayed this bug to say RHEL 7.7 ? It is a new feature request and I would prefer to have the feature implemented upstream first and then backported once it has been tested a bit... Cheers Nick
(In reply to Nick Clifton from comment #6) > Hi Ben, > > Would you be terribly upset if I delayed this bug to say RHEL 7.7 ? For this one — not a problem. > > It is a new feature request and I would prefer to have the feature > implemented upstream first and then backported once it has been > tested a bit... > > Cheers > Nick
The reason for this bug is that there is evidently a problem with the AARCH64 unwind information in some cases and the information doesn't match the .eh_frame debug data. I was hoping to use this feature to help debug the problem. I was unable to spot the problem when I was doing some testing and so the customer promised me a reproducer but got tied up with other problems. So the original problem report is on hold for the time being. Even having a prototype decoder would be helpful when this problem report pops back up again.
*** This bug has been marked as a duplicate of bug 1640745 ***