| Message ID | alpine.LNX.2.00.1012141853520.5687@monoid.intra.ispras.ru |
|---|---|
| State | New |
| Headers | show |
On 12/14/2010 12:29 PM, Alexander Monakov wrote: > Hi, > > Our BB removal machinery can not delete the first BB in the region. This > patch restores the correct behaviour for purge_empty_blocks: we should skip > the first BB unconditionally, not only if it's a preheader. It was > accidentally broken a while ago while fixing a different bug because I was not > aware of the limitation. > > Bootstrapped& tested on x86_64-linux, OK for trunk? > Ok, thanks. > 2010-12-14 Alexander Monakov<amonakov@ispras.ru> > > PR rtl-optimization/46649 > * sel-sched-ir.c (purge_empty_blocks): Unconditionally skip the first > basic block in the region. > > testsuite: > g++.dg/opt/pr46649.C: New.
diff --git a/gcc/sel-sched-ir.c b/gcc/sel-sched-ir.c index 468dfd7..43619d9 100644 --- a/gcc/sel-sched-ir.c +++ b/gcc/sel-sched-ir.c @@ -3766,10 +3766,10 @@ tidy_control_flow (basic_block xbb, bool full_tidying) void purge_empty_blocks (void) { - /* Do not attempt to delete preheader. */ - int i = sel_is_loop_preheader_p (BASIC_BLOCK (BB_TO_BLOCK (0))) ? 1 : 0; + int i; - while (i < current_nr_blocks) + /* Do not attempt to delete the first basic block in the region. */ + for (i = 1; i < current_nr_blocks; ) { basic_block b = BASIC_BLOCK (BB_TO_BLOCK (i)); diff --git a/gcc/testsuite/g++.dg/opt/pr46649.C b/gcc/testsuite/g++.dg/opt/pr46649.C new file mode 100644 index 0000000..4dfaea4 --- /dev/null +++ b/gcc/testsuite/g++.dg/opt/pr46649.C @@ -0,0 +1,9 @@ +// { dg-do compile { target powerpc*-*-* ia64-*-* x86_64-*-* } } +// { dg-options "-fschedule-insns -fselective-scheduling" } + +void foo () +{ + for (;;) + for (;;({break;})) + ; +}