On 08 Jan 16:49, Jakub Jelinek wrote: > BTW, today when looking at the TARGET_OPTION_NODE streaming caused > regressions, I've discovered that it is very hard to debug issues in the > offloading compiler. Would be nice if > -save-temps -v > printed enough information that it is actually possible to reproduce it, > e.g. while mkoffload command is printed, one can't cut and paste it easily, > because some env vars are required and those aren't printed in the -v dump.
I agree, this should be improved. Unfortunately, I didn't have time so far. > Similarly, the lto1 offloading compiler invocation is not printed, and It can be printed by -foffload="-save-temps -v", or should we pass through these options from host to offload compiler by default? > wrapping offloading compiler's lto1 into a script that runs gdb on it > doesn't work, because stdout/stderr (and stdin) is redirected. > > This is something that can be solved during stage4, but would be really nice > if it wasn't terribly hard to debug stuff. -- Ilya