http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58759
Jeffrey A. Law <law at redhat dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |vmakarov at redhat dot com
--- Comment #5 from Jeffrey A. Law <law at redhat dot com> ---
Really looks like an LRA/IRA issue to me. The problem insn doesn't exist until
reloading. It gets created here:
#0 make_insn_raw (pattern=0x7ffff0308000) at
/home/gcc/GIT/gcc/gcc/emit-rtl.c:3654
#1 0x0000000000789ead in emit_insn (x=0x7ffff0308000) at
/home/gcc/GIT/gcc/gcc/emit-rtl.c:4737
#2 0x0000000000df5214 in ix86_expand_move (mode=SImode,
operands=0x7fffffffdb60) at /home/gcc/GIT/gcc/gcc/config/i386/i386.c:16419
#3 0x0000000000eb7484 in gen_movsi (operand0=0x7ffff0300fe0,
operand1=0x7ffff03072a0) at /home/gcc/GIT/gcc/gcc/config/i386/i386.md:1788
#4 0x00000000007d0609 in insn_gen_fn::operator() (this=0x1512cd0
<insn_data+99216>, a0=0x7ffff0300fe0, a1=0x7ffff03072a0) at
/home/gcc/GIT/gcc/gcc/recog.h:284
#5 0x00000000007b4d94 in emit_move_insn_1 (x=0x7ffff0300fe0, y=0x7ffff03072a0)
at /home/gcc/GIT/gcc/gcc/expr.c:3447
#6 0x00000000009dd1c3 in gen_move_insn (x=0x7ffff0300fe0, y=0x7ffff03072a0) at
/home/gcc/GIT/gcc/gcc/optabs.c:4812
#7 0x000000000097534a in emit_spill_move (to_p=false,
mem_pseudo=0x7ffff03072a0, val=0x7ffff0300fe0) at
/home/gcc/GIT/gcc/gcc/lra-constraints.c:880
#8 0x000000000097eb18 in split_reg (before_p=false, original_regno=115,
insn=0x7ffff02ea400, next_usage_insns=0x7ffff0304828) at
/home/gcc/GIT/gcc/gcc/lra-constraints.c:4628
#9 0x0000000000980eee in inherit_in_ebb (head=0x7ffff02fa870,
tail=0x7ffff0304828) at /home/gcc/GIT/gcc/gcc/lra-constraints.c:5244
#10 0x00000000009811e8 in lra_inheritance () at
/home/gcc/GIT/gcc/gcc/lra-constraints.c:5311
#11 0x000000000096b92d in lra (f=0x0) at /home/gcc/GIT/gcc/gcc/lra.c:2338
#12 0x000000000091d079 in do_reload () at /home/gcc/GIT/gcc/gcc/ira.c:4686
We've had numerous problems with reload in the past trying to store into
readonly memory due to reloading values with known equivalent memory locations
(as is the case here). I wouldn't be surprised to see some of the same
problems showing up in IRA.
Vlad, can you take a look?
