http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59290
Bug ID: 59290
Summary: [ARM] regression on negdi-2.c (big-endian)
Product: gcc
Version: unknown
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: christophe.lyon at st dot com
Since commit 203828 (new cortexa9_extra_costs table), I have noticed a
regression on:
gcc.target/arm/negdi-2.c scan-assembler-times mov 1
in big-endian targets:
armeb-none-linux-gnueabihf
mode: arm or thumb
cpu=cortex-a9
fpu=neon-fp16
The preceding test in the same testcase (scan-assembler-times "negs\\tr0, r0"
1) FAILs too, but it was already the case before that commit.
The testcase might need to be adapted to cope with big-endian targets.
