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]

[PATCH] Fix PR 18746, fab causing more problems


With checking disable we don't call get_stmt_operands on a statement
before generating memory tags which tree-ssa-alias will generate but
they are not renamed at this point so we get a seg fault.

This patch fixes the problem by instead of calling modify_stmt, we call
mark_new_vars_to_rename which also marks the vops for rename just in
case we have new vops to rename.  This might also fix some Ada testcases
I saw fail but I have not tested for that yet.

OK? Bootstrapped and tested on ppc-darwin with all languages except
for Ada, plus bootstrapped with disable-checking which is where the
bug was found.

Thanks,
Andrew Pinski

PS I have no testcase which fails with enable checking that I could
find.  If someone wants me to add a testcase for the case where
disable checking failed just ask.

ChangeLog:
	* tree-ssa-ccp.c (execute_fold_all_builtins): Instead of saying the
	statement was modified, call mark_new_vars_to_rename.


Attachment: fixPR18746.diff.txt
Description: Text document


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]