PATCH: PR libfortran/25305: [4.0 regression]: libfortran failed fma3d in SPEC CPU 2K

[H. J. Lu]

On Thu, Dec 15, 2005 at 07:21:28PM +0100, Paul Thomas wrote:
> H. J. Lu wrote:
> 
> >Could please please take a look at this 4.0 regression
> >
> >http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25305
> >
> >and approve the patch?
> >
> >Thanks.
> > 
> >
> This is proving to be a bit mysterious for us without access to the SPEC 
> tests! The patch can be examined, if you describe what is broken and 
> provide a test for the testsuite that demonstrates thatit is fixed.
> 


Here is a testcase from Grigory. I got

		=== gfortran tests ===


Running target unix/-m32
FAIL: gfortran.dg/end_of_record.f90  -O0  execution test
FAIL: gfortran.dg/end_of_record.f90  -O1  execution test
FAIL: gfortran.dg/end_of_record.f90  -O2  execution test
FAIL: gfortran.dg/end_of_record.f90  -O3 -fomit-frame-pointer
execution test
FAIL: gfortran.dg/end_of_record.f90  -O3 -fomit-frame-pointer
-funroll-loops  execution test
FAIL: gfortran.dg/end_of_record.f90  -O3 -fomit-frame-pointer
-funroll-all-loops -finline-functions  execution test
FAIL: gfortran.dg/end_of_record.f90  -O3 -g  execution test
FAIL: gfortran.dg/end_of_record.f90  -Os  execution test

in

http://gcc.gnu.org/ml/gcc-testresults/2005-12/msg01043.html

Without my patch, it passes:

http://gcc.gnu.org/ml/gcc-testresults/2005-12/msg01045.html

Gcc 4.2 also passes.


H.J.
----
2005-12-19  Grigory Zagorodnev <grigory_zagorodnev@linux.intel.com>

         PR libfortran/25305
         * gfortran.dg/end_of_record.f90: New.

--- gcc/testsuite/gfortran.dg/end_of_record.f90.xxx	2005-12-19 09:31:41.000000000 -0800
+++ gcc/testsuite/gfortran.dg/end_of_record.f90	2005-12-19 09:33:51.000000000 -0800
@@ -0,0 +1,13 @@
+! { dg-do run }
+! PR25305: 4.0 regression caused by a libfortran patch 
+program test
+  CHARACTER RD*8
+  CHARACTER WR*8
+  CHARACTER(1) C1
+
+  RD='N 1'
+  READ (RD(1:2),*) C1
+
+  WRITE (WR,*) 'passed'  !! Unexpected 'End of record' error
+  WRITE (*,*) WR
+end program test