ALT Linux Bugzilla – Full Text Bug Listing
| Summary: | verfiy-elf no longer reports unresolved symbols as errors if an .so is made PIE | ||
|---|---|---|---|
| Product: | Sisyphus | Reporter: | Ivan Zakharyaschev <imz> |
| Component: | rpm-build | Assignee: | Ivan Zakharyaschev <imz> |
| Status: | NEW --- | QA Contact: | qa-sisyphus |
| Severity: | normal | ||
| Priority: | P3 | CC: | arseny, glebfm, imz, ldv, placeholder, vt |
| Version: | unstable | ||
| Hardware: | all | ||
| OS: | Linux | ||
|
Description
Ivan Zakharyaschev
2016-12-06 22:56:02 MSK
$ fgrep './usr/lib64/lmms/RemoteZynAddSubFx: undefined symbol: _ZN8OscilGen22getcurrentbasefunctionEPf' gear-hsh.verify-ERR.log.0 gear-hsh.verify+PIE-WARN.log.1 gear-hsh.verify-ERR.log.0:verify-elf: ERROR: ./usr/lib64/lmms/RemoteZynAddSubFx: undefined symbol: _ZN8OscilGen22getcurrentbasefunctionEPf gear-hsh.verify+PIE-WARN.log.1:verify-elf: WARNING: ./usr/lib64/lmms/RemoteZynAddSubFx: undefined symbol: _ZN8OscilGen22getcurrentbasefunctionEPf This is a side effect of "set -o pipefail" introduced by commit 4.0.4-alt100.96~1. Every time you think about introducing "set -o pipefail", check whether the script uses grep and what's going to happen with grep exit status. (In reply to comment #2) > This is a side effect of "set -o pipefail" introduced by commit > 4.0.4-alt100.96~1. > > Every time you think about introducing "set -o pipefail", check whether the > script uses grep and what's going to happen with grep exit status. Sorry, I must have confused this bug report with another one. Anyway, please revisit the use of "set -o pipefail" in verify-elf. (In reply to comment #3) > (In reply to comment #2) > > This is a side effect of "set -o pipefail" introduced by commit > > 4.0.4-alt100.96~1. > > > > Every time you think about introducing "set -o pipefail", check whether the > > script uses grep and what's going to happen with grep exit status. > > Sorry, I must have confused this bug report with another one. This bugreport is based on the behaviour which I noticed in June 2016. I'll check prior releases of rpm, too. > Anyway, please revisit the use of "set -o pipefail" in verify-elf. Yes, you have noticed an important thing which I have not paid attention, too, and which I must revisit. (In reply to comment #4) > (In reply to comment #3) > > (In reply to comment #2) > > > This is a side effect of "set -o pipefail" introduced by commit > > > 4.0.4-alt100.96~1. > > > > > > Every time you think about introducing "set -o pipefail", check whether the > > > script uses grep and what's going to happen with grep exit status. > > > > Sorry, I must have confused this bug report with another one. > > This bugreport is based on the behaviour which I noticed in June 2016. I'll > check prior releases of rpm, too. verify_unresolved was not aware of PIEs and treated them as other shared objects that are traditionally subject to a more relaxed checking than executables. You can try commit 4.0.4-alt100.96-3-ga4fb4d4 and tell whether it makes things better. |
