https://gcc.gnu.org/g:956bc995a904df30b510801de68b71965e747cc5
commit r14-12715-g956bc995a904df30b510801de68b71965e747cc5 Author: Stefan Schulze Frielinghaus <[email protected]> Date: Fri Jun 12 11:15:14 2026 +0200 ira: Allow multiple uses within an insn [PR125173] For the attached test we end up during IRA with 6: r99:DI=`s' REG_EQUIV `s' 7: {[`b']=asm_operands; [const(`s'+0x4)]=asm_operands; [`a']=asm_operands;clobber flags:CC;} REG_DEAD r99:DI REG_UNUSED flags:CC where r99 is used in each asm_operands (mem/c:SI (plus:DI (reg/f:DI 99) (const_int 4 [0x4])) [2 s.d+0 S4 A32]) This in turn means that we hit multiple times the very same insn while iterating over all uses of r99 which in the end triggers the assert. This patch relaxes the assert since what we want to ensure here is that a register is not used by multiple insns since otherwise we cannot trivially move or delete the definition. PR rtl-optimization/125173 gcc/ChangeLog: * ira.cc (combine_and_move_insns): Allow multiple uses within an insn. gcc/testsuite/ChangeLog: * gcc.dg/pr125173-1.c: New test. (cherry picked from commit 107d2ad166d09883b43ec1a5cf1e37c68b134a93) Diff: --- gcc/ira.cc | 12 +++++++++++- gcc/testsuite/gcc.dg/pr125173-1.c | 17 +++++++++++++++++ 2 files changed, 28 insertions(+), 1 deletion(-) diff --git a/gcc/ira.cc b/gcc/ira.cc index 5642aea3caae..1619259ec042 100644 --- a/gcc/ira.cc +++ b/gcc/ira.cc @@ -3908,6 +3908,7 @@ combine_and_move_insns (void) continue; rtx_insn *use_insn = 0; + bool multiple_insns = false; for (df_ref use = DF_REG_USE_CHAIN (regno); use; use = DF_REF_NEXT_REG (use)) @@ -3915,11 +3916,20 @@ combine_and_move_insns (void) { if (DEBUG_INSN_P (DF_REF_INSN (use))) continue; - gcc_assert (!use_insn); + if (use_insn && DF_REF_INSN (use) != use_insn) + { + multiple_insns = true; + break; + } use_insn = DF_REF_INSN (use); } gcc_assert (use_insn); + /* If a register is used by more than one insn, we cannot trivially move + or delete the definition anymore. */ + if (multiple_insns) + continue; + /* Don't substitute into jumps. indirect_jump_optimize does this for anything we are prepared to handle. */ if (JUMP_P (use_insn)) diff --git a/gcc/testsuite/gcc.dg/pr125173-1.c b/gcc/testsuite/gcc.dg/pr125173-1.c new file mode 100644 index 000000000000..5f66b24fcb00 --- /dev/null +++ b/gcc/testsuite/gcc.dg/pr125173-1.c @@ -0,0 +1,17 @@ +/* { dg-do compile } */ +/* { dg-options "-O1 -fno-fuse-ops-with-volatile-access -fno-forward-propagate --param=max-cse-insns=0" } */ + +int a; +volatile short b; + +struct +{ + char c; + int d; +} s; + +void +foo () +{ + __asm__ ("":"+m" (b), "+m" (s.d), "+m" (a)); +}
