http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45932
--- Comment #2 from jules at gcc dot gnu.org 2010-10-08 16:17:58 UTC --- Could you be more specific about testsuite options, etc. needed to observe this apparent regression? I haven't been able to reproduce it using default options, nor by specifying [-mthumb] -march=armv7-a -mfpu=neon -mfloat-abi=softfp to the test run. I do see a failure like so: .../gcc/testsuite/gcc.c-torture/execute/pr37573.c:64:1: sorry, unimplemented: gimple bytecode streams do not support machine specific builtin functions on this target for -flto and -fwhopr, but those happen irrespective of whether the misalignment patch is applied or not. Thanks!