SOme bugfixes in expr.c

[guerby@acm.org]

This indeed fixes 9 ACATS expr.c ICEs:

<<
gcc -c -I/home/guerby/work/gcc/acats/acats4gnat-0.3/work/support c37213d.adb
+===========================GNAT BUG DETECTED==============================+
| 5.00w (20010924) (i686-pc-linux-gnu) GCC error:                          |
| Internal compiler error in find_placeholder, at expr.c:6086              |

=> 9 times
>>

But it probably added a few failures to the C testsuite. Yesterday the FAIL count
was at 1, today it's 19. However I can't say for sure this was caused by your patch,
since there were other commited in the meantime.

<<
FAIL: gcc.c-torture/compile/20000224-1.c,  -O1  
FAIL: gcc.c-torture/compile/20000224-1.c,  -O2  
FAIL: gcc.c-torture/compile/20000224-1.c,  -O3 -fomit-frame-pointer  
FAIL: gcc.c-torture/compile/20000224-1.c,  -O3 -fomit-frame-pointer -funroll-loops  
FAIL: gcc.c-torture/compile/20000224-1.c,  -O3 -fomit-frame-pointer -funroll-all-loops -finline-functions  
FAIL: gcc.c-torture/compile/20000224-1.c,  -O3 -g  
FAIL: gcc.c-torture/compile/20000224-1.c,  -Os  
FAIL: gcc.c-torture/execute/20000412-5.c compilation,  -O1 
FAIL: gcc.c-torture/execute/20000412-5.c compilation,  -O2 
FAIL: gcc.c-torture/execute/20000412-5.c compilation,  -O3 -fomit-frame-pointer 
FAIL: gcc.c-torture/execute/20000412-5.c compilation,  -O3 -g 
FAIL: gcc.c-torture/execute/20000412-5.c compilation,  -Os 
FAIL: gcc.c-torture/execute/20010114-1.c compilation,  -O1 
FAIL: gcc.c-torture/execute/20010114-1.c compilation,  -O2 
FAIL: gcc.c-torture/execute/20010114-1.c compilation,  -O3 -fomit-frame-pointer 
FAIL: gcc.c-torture/execute/20010114-1.c compilation,  -O3 -g 
FAIL: gcc.c-torture/execute/20010114-1.c compilation,  -Os 
FAIL: gcc.dg/format/ext-2.c bad use of I flag (test for warnings, line 72)
FAIL: gcc.dg/format/ext-2.c (test for excess errors)

		=== gcc Summary ===

# of expected passes		16083
# of unexpected failures	19
# of expected failures		71
# of unresolved testcases	10
# of unsupported tests		24
/home/guerby/work/gcc/build-alt2/gcc/xgcc version 3.1 20011028 (experimental)
>>

Is there any policy about the status of the C testsuite on a major
platform like i686-pc-linux-gnu versus commits? Since the wall clock
time needed is 24 minutes on a 1GHz P3 (vs 45 minutes for bootstrap),
this doesn't seem unreasonable to require the testsuite to stay in
good shape, or at least that new FAIL are announced by people
commiting patches?

-- 
Laurent Guerby <guerby@acm.org>