[Bug rtl-optimization/45136] -fcompare-debug failure with -Os -fschedule-insns

[aoliva at gcc dot gnu dot org]

--- Comment #6 from aoliva at gcc dot gnu dot org  2010-08-30 19:13 ---
The debug insns aren't at the end of the block in the beginning of sched, so we
don't disregard them at first, but they become end-of-block insns after the
trailing non-debug insns are pulled into other earlier blocks in the same
region.  I'm testing a patch that revamps the way we deal with boundary notes
and debug insns.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45136

[Bug rtl-optimization/45136] -fcompare-debug failure with -Os -fschedule-insns

[aoliva at gcc dot gnu dot org]

--- Comment #7 from aoliva at gcc dot gnu dot org  2010-08-30 19:14 ---
The debug insns aren't at the end of the block in the beginning of sched, so we
don't disregard them at first, but they become end-of-block insns after the
trailing non-debug insns are pulled into other earlier blocks in the same
region.  I'm testing a patch that revamps the way we deal with boundary notes
and debug insns.


-- 

aoliva at gcc dot gnu dot org changed:

   What|Removed |Added

 AssignedTo|unassigned at gcc dot gnu   |aoliva at gcc dot gnu dot
   |dot org |org
 Status|NEW |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45136

[Bug rtl-optimization/45136] -fcompare-debug failure with -Os -fschedule-insns

[jakub at gcc dot gnu dot org]

--- Comment #5 from jakub at gcc dot gnu dot org  2010-07-30 09:40 ---
Confirmed, this seems to be sched1 fault.
In bb5 we have 2 normal insns, followed by (-g only) a debug_insn, followed by
two NOTE_INSN_DELETED created by combine (do we ever remove these from the
IL?),
followed again (-g only) by debug_insn and then other insns.
sched1 moves those two NOTE_INSN_DELETED in -g case to the start of the block,
keeps them where they were otherwise.

Alex, can you please have a look?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45136

[Bug rtl-optimization/45136] -fcompare-debug failure with -Os -fschedule-insns

[zsojka at seznam dot cz]

--- Comment #1 from zsojka at seznam dot cz  2010-07-29 20:21 ---
Created an attachment (id=21356)
 -- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=21356action=view)
reduced testcase

Valgrind doesn't report any invalid/uninitialised read while compiling this
testcase.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45136

[Bug rtl-optimization/45136] -fcompare-debug failure with -Os -fschedule-insns

[hjl dot tools at gmail dot com]

--- Comment #2 from hjl dot tools at gmail dot com  2010-07-30 00:32 ---
It is caused by revision 161655:

http://gcc.gnu.org/ml/gcc-cvs/2010-07/msg6.html


-- 

hjl dot tools at gmail dot com changed:

   What|Removed |Added

 CC||rguenth at gcc dot gnu dot
   ||org
 Status|UNCONFIRMED |NEW
 Ever Confirmed|0   |1
   Last reconfirmed|-00-00 00:00:00 |2010-07-30 00:32:50
   date||


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45136

[Bug rtl-optimization/45136] -fcompare-debug failure with -Os -fschedule-insns

[zsojka at seznam dot cz]

--- Comment #3 from zsojka at seznam dot cz  2010-07-30 01:35 ---
Thank you for the reply! Then, the bug had to be fixed/hiden before r161655, I
can reproduce it with r161170, r153685 and gcc 4.5.0. (maybe you replied to a
wrong PR?)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45136

[Bug rtl-optimization/45136] -fcompare-debug failure with -Os -fschedule-insns

[hjl dot tools at gmail dot com]

--- Comment #4 from hjl dot tools at gmail dot com  2010-07-30 04:11 ---
I was wrong. It never worked.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45136