diff options
author | Vladimir Makarov <vmakarov@redhat.com> | 2014-01-23 20:06:28 +0000 |
---|---|---|
committer | Vladimir Makarov <vmakarov@gcc.gnu.org> | 2014-01-23 20:06:28 +0000 |
commit | 6e23f2967868d307d367385411a5e61e6de2c02b (patch) | |
tree | 3c4587bb80623ea093acd96ddf83fb8e6f30c554 /gcc/lra-constraints.c | |
parent | f04dda3093541436d9ebd3194761ff8199c5ac79 (diff) | |
download | gcc-6e23f2967868d307d367385411a5e61e6de2c02b.zip gcc-6e23f2967868d307d367385411a5e61e6de2c02b.tar.gz gcc-6e23f2967868d307d367385411a5e61e6de2c02b.tar.bz2 |
re PR rtl-optimization/59915 (LRA ICE - Repeated looping over subreg reloads (gcc.c-torture/compile/simd-3.c))
2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
PR regression/59915
* lra-constraints.c (simplify_operand_subreg): Spill pseudo if
there is a danger of looping.
From-SVN: r207007
Diffstat (limited to 'gcc/lra-constraints.c')
-rw-r--r-- | gcc/lra-constraints.c | 17 |
1 files changed, 14 insertions, 3 deletions
diff --git a/gcc/lra-constraints.c b/gcc/lra-constraints.c index 7454229..34159f7 100644 --- a/gcc/lra-constraints.c +++ b/gcc/lra-constraints.c @@ -1291,9 +1291,20 @@ simplify_operand_subreg (int nop, enum machine_mode reg_mode) && ! LRA_SUBREG_P (operand)) || CONSTANT_P (reg) || GET_CODE (reg) == PLUS || MEM_P (reg)) { - /* The class will be defined later in curr_insn_transform. */ - enum reg_class rclass - = (enum reg_class) targetm.preferred_reload_class (reg, ALL_REGS); + enum reg_class rclass; + + if (REG_P (reg) + && curr_insn_set != NULL_RTX + && (REG_P (SET_SRC (curr_insn_set)) + || GET_CODE (SET_SRC (curr_insn_set)) == SUBREG)) + /* There is big probability that we will get the same class + for the new pseudo and we will get the same insn which + means infinite looping. So spill the new pseudo. */ + rclass = NO_REGS; + else + /* The class will be defined later in curr_insn_transform. */ + rclass + = (enum reg_class) targetm.preferred_reload_class (reg, ALL_REGS); if (get_reload_reg (curr_static_id->operand[nop].type, reg_mode, reg, rclass, "subreg reg", &new_reg)) |