diff options
| author | James Cowgill <james.cowgill@mips.com> | 2018-03-03 15:49:21 +0000 |
|---|---|---|
| committer | Maciej W. Rozycki <macro@mips.com> | 2018-03-03 15:49:21 +0000 |
| commit | dfb93f11587ca08b820c7c785278366f2505cfd1 (patch) | |
| tree | 19952490ddd453f8fb532d608c92012da84b0c18 /ld/ChangeLog | |
| parent | b9671caf8fe1abd737846edf7dcd627870f986cc (diff) | |
| download | gdb-dfb93f11587ca08b820c7c785278366f2505cfd1.zip gdb-dfb93f11587ca08b820c7c785278366f2505cfd1.tar.gz gdb-dfb93f11587ca08b820c7c785278366f2505cfd1.tar.bz2 | |
PR ld/21900: MIPS: Fix relocation processing with undefined symbols
Currently, when `mips_elf_calculate_relocation' is asked to relocate an
undefined symbol, it reports an error or a warning and immediately
returns without performing the relocation. This is fine if the link
fails, but if unresolved_syms_in_objects == RM_GENERATE_WARNING, the
link will continue and output some unrelocated code, which is a
regression from commit e7e2196da3f0 ("MIPS/BFD: Correctly report
undefined relocations").
Fix this by continuing after calling the `undefined_symbol' hook unless
this is an error condition.
bfd/
PR ld/21900
* elfxx-mips.c (mips_elf_calculate_relocation): Only return
after calling `undefined_symbol' hook if this is an error
condition. Assume the value of 0 for the symbol requested
otherwise.
ld/
PR ld/21900
* testsuite/ld-mips-elf/undefined-warn.d: New test.
* testsuite/ld-mips-elf/undefined.s: Add padding at the end.
* testsuite/ld-mips-elf/mips-elf.exp: Run the new test.
Diffstat (limited to 'ld/ChangeLog')
| -rw-r--r-- | ld/ChangeLog | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/ld/ChangeLog b/ld/ChangeLog index 6f9abfa..c796125 100644 --- a/ld/ChangeLog +++ b/ld/ChangeLog @@ -1,3 +1,11 @@ +2018-03-03 James Cowgill <james.cowgill@mips.com> + Maciej W. Rozycki <macro@mips.com> + + PR ld/21900 + * testsuite/ld-mips-elf/undefined-warn.d: New test. + * testsuite/ld-mips-elf/undefined.s: Add padding at the end. + * testsuite/ld-mips-elf/mips-elf.exp: Run the new test. + 2018-03-02 Jim Wilson <jimw@sifive.com> PR 22903 |