This is the mail archive of the
mailing list for the GCC project.
Re: CHECK_DEAD_NOTES abort in sched1 for sh-elf
Sanjiv Kumar Gupta, Noida wrote:
I am not sure whether this is a problem with distribute_notes in combiner,
or update_life_info in sched1.
It is a problem with distribute_notes in combine.
We are combining 3 instructions into two. Pre combine, t is used and
clobbered in i2. Post combine, t is only used in i2, and not mentioned
in i3. That means we need to convert the REG_UNUSED note into a
REG_DEAD note and put it on i2, but we have no code to do that.
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com