Hi! The following testcase ICEs, because we decide to fold a comparison into just one of its operands, we call recalculate_side_effects on that and ICE on the assertion that it isn't called on anything unexpected (here PARM_DECL). Already some time ago we had to add an exception for SSA_NAME for the same reason. The tcc_declaration case is slightly different, TREE_SIDE_EFFECTS is sometimes present on those if they are TREE_THIS_VOLATILE, but it is something the FE should take care of when creating those decls, not a business of the gimplifier.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk/16.2? 2026-07-30 Jakub Jelinek <[email protected]> PR middle-end/126497 * gimplify.cc (recalculate_side_effects): Return for tcc_declaration. * gcc.dg/bitint-141.c: New test. --- gcc/gimplify.cc.jj 2026-07-29 23:02:59.826913052 +0200 +++ gcc/gimplify.cc 2026-07-30 16:36:45.710098749 +0200 @@ -3361,6 +3361,13 @@ recalculate_side_effects (tree t) /* No side-effects. */ return; + case tcc_declaration: + /* These can have side-effects if TREE_THIS_VOLATILE, + but those should be set elsewhere, not in + recalculate_side_effects. Can be triggered e.g. if + a comparison is folded into one of its operands. */ + return; + default: if (code == SSA_NAME) /* No side-effects. */ --- gcc/testsuite/gcc.dg/bitint-141.c.jj 2026-07-30 16:40:32.362324944 +0200 +++ gcc/testsuite/gcc.dg/bitint-141.c 2026-07-30 16:41:04.970925875 +0200 @@ -0,0 +1,19 @@ +/* PR middle-end/126497 */ +/* { dg-do compile { target bitint } } */ +/* { dg-options "-std=c23" } */ + +typedef unsigned _BitInt (1) U; + +U +foo (U a) +{ + U t = a >= 1uwb; + return t; +} + +U +bar (U a) +{ + U t = a == 1uwb; + return t; +} Jakub
