Executing on host: /home/dave/gnu/gcc/objdir/gcc/xgcc -B/home/dave/gnu/gcc/objdi r/gcc/ /home/dave/gnu/gcc/gcc/gcc/testsuite/gcc.c-torture/execute/pr42248.c -w -O1 -lm -o /home/dave/gnu/gcc/objdir/gcc/testsuite/gcc/pr42248.x1 (timeo ut = 300) /home/dave/gnu/gcc/gcc/gcc/testsuite/gcc.c-torture/execute/pr42248.c: In functio n 'check': /home/dave/gnu/gcc/gcc/gcc/testsuite/gcc.c-torture/execute/pr42248.c:9:1: intern al compiler error: in make_decl_rtl, at varasm.c:1316 Breakpoint 2, make_decl_rtl (decl=0x4008f7e0) at ../../gcc/gcc/varasm.c:1312 1312 gcc_assert (TREE_CODE (decl) != VAR_DECL (gdb) p debug_tree (decl) <var_decl 0x4008f7e0 x.2 type <record_type 0x4008f000 Scf10 type_0 BLK size <integer_cst 0x400107f8 constant 256> unit size <integer_cst 0x400100d8 constant 32> align 64 symtab 0 alias set -1 canonical type 0x40077f60 fields <field_decl 0x400190b0 a type <complex_type 0x40027060 complex double> DC file /home/dave/gnu/gcc/gcc/gcc/testsuite/gcc.c-torture/execute/pr42248.c line 2 col 19 size <integer_cst 0x40010468 constant 128> unit size <integer_cst 0x40010480 constant 16> align 64 offset_align 64 offset <integer_cst 0x40010108 constant 0> bit offset <integer_cst 0x40010570 constant 0> context <record_type 0x40077f60> chain <field_decl 0x40019108 b>> context <translation_unit_decl 0x40023dd0 D.1206> pointer_to_this <pointer_type 0x4008f240> chain <type_decl 0x40023d00 D.1191>> used BLK file /home/dave/gnu/gcc/gcc/gcc/testsuite/gcc.c-torture/execute/pr42248.c line 9 col 1 size <integer_cst 0x400107f8 256> unit size <integer_cst 0x400100d8 32> align 64 context <function_decl 0x40083600 check> chain <var_decl 0x4008f840 y.3>> (gdb) bt #0 make_decl_rtl (decl=0x4008f7e0) at ../../gcc/gcc/varasm.c:1312 #1 0x00892cf8 in expand_expr_real_1 (exp=0x4008f7e0, target=0x0, tmode=VOIDmode, modifier=EXPAND_NORMAL, alt_rtl=0x0) at ../../gcc/gcc/expr.c:8438 #2 0x0088a0ac in expand_expr_real (exp=0x4008f7e0, target=0x0, tmode=VOIDmode, modifier=EXPAND_NORMAL, alt_rtl=0x0) at ../../gcc/gcc/expr.c:7188 ...
Can you check if the patch that introduced the testcase is responsible for the failure?
Subject: Re: FAIL: gcc.c-torture/execute/pr42248.c compilation at -O1 and above > Can you check if the patch that introduced the testcase is responsible for the > failure? Will do. Dave
Subject: Re: FAIL: gcc.c-torture/execute/pr42248.c compilation at -O1 and above > Can you check if the patch that introduced the testcase is responsible for the > failure? It's not. Dave
Subject: Re: FAIL: gcc.c-torture/execute/pr42248.c compilation at -O1 and above > > Can you check if the patch that introduced the testcase is responsible for the > > failure? > > It's not. Testcase doesn't fail with Debian 4.3.4-6 and Debian 4.4.2-8. Dave
This bug was introduced in revision 147980: 2009-05-29 Martin Jambor <mjambor@suse.cz> * tree-sra.c: New implementation of SRA.
Confirmed with a cross to hppa-linux. I believe this is again related to DECL_VALUE_EXPR. SRA does: @@ -122,8 +60,9 @@ <bb 2>: x.0 = x; + SR.3_7 = x.a; y.1_6 = y_1(D); - D.1959_2 = x.0.a; + D.1959_2 = SR.3_7; y.2_3 = y.1_6; if (D.1959_2 != y.2_3) goto <bb 3>; but x is replaced with x.0 (callee copy). SRA shouldn't undo this (thus, never create replacements involving a DECL with DECL_HAS_VALUE_EXPR_P set). Reduced testcase: typedef struct { _Complex double a; _Complex double b; } Scf10; void check (Scf10 x, _Complex double y) { if (x.a != y) __builtin_abort (); }
I have a patch.
*** This bug has been marked as a duplicate of 41250 ***