Fix two testsuite failures on k6-pc-linux-gnu

Joern Rennecke joern.rennecke@superh.com
Tue Jul 2 12:39:00 GMT 2002


>For the problem with the loop instruction, I'm not entirely sure how it
>should best be fixed.  One way would be your approach.  Another would
>be to fix genattrtab to mark the loop insn as having a varying length,
>another would be to twiddle the md file so that genattrtab recognizes
>it as having a variable length.  Thoughts folks?

Fixing genattrtab would be the cleanest approach, of course.  But then
we need a volunteer to do that first.

Now, regarding a workaround in i386.md: Since the information for the
general case is in the length attribute itself, we can't use it in
a special-case length attribute for the loop instruction.  To do that,
the length attribute would first have to renamed to something else,
then a new length attribute could use that other attribute.
That seems more complexity that just using the original length
attribute definition for the loop insn.

OTOH, I don't see the motivation for setting the type of the long
variant back to "ibr" .  I suppose "multi" is probably better for
scheduling.

-- 
--------------------------
SuperH
2430 Aztec West / Almondsbury / BRISTOL / BS32 4AQ
T:+44 1454 462330



More information about the Gcc-patches mailing list