Re: [AVR] Fix PR49881

[Georg-Johann Lay <avr at gjlay dot de>]

Richard Henderson wrote:
> Dang, forgot to add gcc-patches...
> 
> -------- Original Message --------
> Subject: [AVR] Fix PR49881
> 
> The following iteration fixes the two regressions reported
> in comment 7 of the PR.
> 
> These were ICEs due to emit_push_insn_single being "helpful"
> with pushing complex numbers.  Instead of recursing for the
> components of a complex number, it simply generated raw
> pre_dec patterns.  This is arguably a middle-end bug, but 
> it's easier to fix in the backend by listing complex modes
> in the macro for the push expander.
> 
> Ok?
> 
> 
> r~

After updating to r177084 (your patch is r177071) I still get *many*
test fails or untested (programs hangs?) fails.

Just few examples:

PASS: gcc.c-torture/execute/920726-1.c compilation,  -O1
FAIL: gcc.c-torture/execute/920726-1.c execution,  -O1

PASS: gcc.c-torture/execute/930513-1.c compilation,  -Os
UNTESTED: gcc.c-torture/execute/930513-1.c execution,  -Os

FAIL: gcc.c-torture/execute/920726-1.c execution,  -O0
FAIL: gcc.c-torture/execute/920726-1.c execution,  -O1
FAIL: gcc.c-torture/execute/920726-1.c execution,  -O2
FAIL: gcc.c-torture/execute/920726-1.c execution,  -O3 -fomit-frame-pointer
FAIL: gcc.c-torture/execute/920726-1.c execution,  -O3 -fomit-frame-pointer -funroll-loops
FAIL: gcc.c-torture/execute/920726-1.c execution,  -O3 -fomit-frame-pointer -funroll-all-loops
-finline-functions
FAIL: gcc.c-torture/execute/920726-1.c execution,  -O3 -g

There are still unrecognizables:

gcc.c-torture/execute/complex-7.c:56:1: error: unrecognizable insn:
(insn 17 14 18 3 (set (mem:SF (post_dec:HI (reg/f:HI 32 __SP_L__)) [0 S4 A8])
        (reg:SF 43 [ f5.0+4 ]))
/mnt/nfs/home/georg/gnu/gcc.gnu.org/trunk/gcc/testsuite/gcc.c-torture/execute/complex-7.c:52 -1
     (nil))

FAIL: gcc.c-torture/execute/complex-7.c compilation,  -O0

Is it possible that you run avr regressions?

Johann