This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
This fixes PR 18010, which is an IA-64 unwind problem turned up by the libunwind testsuite. When basic block reordering copies instructions around, it does not copy the RTX_FRAME_RELATED_P bit. This causes us to lose unwind info when epilogues are duplicated. This is primarily a problem for IA-64, as most DWARF2 unwinders don't bother to instrument epilogue instructions. This is required by the IA-64 unwind API though. This was tested with an x86_64-linux bootstrap and make check. There were no regressions. This was also tested with an ia64-linux bootstrap and make check. There were two extra libjava passes, though I am not sure why. The libunwind author confirms that this solves his problems. I have checked in the patch. -- Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com
Attachment:
patch.copy.frame_related
Description: Text document
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |