http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58234
Andrew Pinski pinskia at gcc dot gnu.org changed:
What|Removed |Added
Status|UNCONFIRMED |RESOLVED
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58234
--- Comment #2 from Larry Baker baker at usgs dot gov ---
Andrew,
Thank you for your prompt reply. Fair enough.
Can you direct me to where glibc bugs are reported?
I have already filed a bug report with Intel.
The in-line asm is not quite
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58234
--- Comment #3 from Andrew Pinski pinskia at gcc dot gnu.org ---
(In reply to Larry Baker from comment #2)
Andrew,
Thank you for your prompt reply. Fair enough.
Can you direct me to where glibc bugs are reported?
Except there is no bug in
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58234
--- Comment #4 from Larry Baker baker at usgs dot gov ---
Actually, there is: the useless movl instead of a movq of the updated address
pointer into __d1 on x86_64. But, that is a benign flaw.
Can you answer either of my questions?
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58234
--- Comment #5 from Andrew Pinski pinskia at gcc dot gnu.org ---
Can you tell me how it is possible to specify the clobber side effects without
requiring output constraints?
It is too hard if you have inputs in those registers too. You could
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58234
--- Comment #6 from Larry Baker baker at usgs dot gov ---
Thank you.
The example I found (mov_blk) that does not use output constraints, but
specifies that the input registers are clobbered, is from a 2003 document. It
too fails using today's