Am 01.10.25 um 00:35 schrieb Jerry Delisle:
On Tue, Sep 30, 2025, 12:33 PM Harald Anlauf <[email protected]> wrote:
Dear All,
the attached patch fixes the problem that UBSAN did generate a check
of the passed stride of an optional array dummy argument even in the
case the actual was not present, as seen by the reporter when compiling
with -O0. Wrapping the setup of the descriptor by a test of presence
of the argument eliminates this path.
Regarding the testcase, as I could not find a simple failing
reproducer, I chose to verify the new test in the tree-dump
and the reduced UBSAN count in the optimized dump.
Regtested on x86_64-pc-linux-gnu. OK for mainline?
Thanks,
Harald
Yes, thanks Harald. OK
Pushed as r16-4171-gb12d5a6ba64f09 .
Thanks for the review!
Harald
Jerry
