* regression test results for 970904 on alpha-dec-osf4.0
@ 1997-09-04 21:47 scott snyder
1997-09-04 22:05 ` Jeffrey A Law
0 siblings, 1 reply; 6+ messages in thread
From: scott snyder @ 1997-09-04 21:47 UTC (permalink / raw)
To: egcs
hi -
Here are results from running the test suite on the 970904
snapshot on a alpha-dec-osf4.0 target.
Here's the gcc summary:
=== gcc Summary ===
# of expected passes 4863
# of unexpected failures 16
# of expected failures 4
# of unsupported tests 7
Executing on host: /d0sgif/data1/snyder/egcs-build/ALPHA/gcc/xgcc -v
Here are the failures:
FAIL: gcc.c-torture/compile/961203-1.c, -O0
FAIL: gcc.c-torture/compile/961203-1.c, -O1
FAIL: gcc.c-torture/compile/961203-1.c, -O2
FAIL: gcc.c-torture/compile/961203-1.c, -O2 -fomit-frame-pointer -finline-functions
This is a known problem with this test.
FAIL: gcc.c-torture/execute/cbrt.c execution, -O0
FAIL: gcc.c-torture/execute/cbrt.c execution, -O1
FAIL: gcc.c-torture/execute/cbrt.c execution, -O2
FAIL: gcc.c-torture/execute/cbrt.c execution, -O2 -fomit-frame-pointer -finline-functions
As reported earlier, this is due to assumption in this test case
that 2*sizeof(long) == sizeof(double), which is false for the alpha.
If this is fixed, this test passes.
FAIL: gcc.c-torture/execute/complex-5.c execution, -O0
FAIL: gcc.c-torture/execute/complex-5.c execution, -O1
FAIL: gcc.c-torture/execute/complex-5.c execution, -O2
FAIL: gcc.c-torture/execute/complex-5.c execution, -O2 -fomit-frame-pointer -finline-functions
As discussed before, passing __complex__ arguments is broken.
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O0
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O1
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O2
FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O2 -fomit-frame-pointer -finline-functions
This is apparently a problem with the test harness; the program needs
to be compiled with -mieee.
The failures which i saw previously with 920411-1.c and m-un-1.c
are gone now.
And for g++:
=== g++ Summary ===
# of expected passes 3260
# of unexpected failures 3
# of unexpected successes 6
# of expected failures 82
# of untested testcases 6
The changes from the last test are the same as for other platforms.
FAIL: g++.brendan/crash19.C (test for excess errors)
FAIL: g++.ext/pretty.C (test for excess errors)
FAIL: g++.mike/p811.C (test for excess errors)
XPASS: g++.brendan/array1.C overflow in array dimension.* , (test for errors, line 6)
XPASS: g++.brendan/scope4.C (test for excess errors)
XPASS: g++.law/profile1.C (test for excess errors)
XPASS: g++.law/profile1.C Execution test
XPASS: g++.law/temps7.C (test for excess errors)
XPASS: g++.mike/p7325.C *-*-* Execution test
sss
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: regression test results for 970904 on alpha-dec-osf4.0
1997-09-04 21:47 regression test results for 970904 on alpha-dec-osf4.0 scott snyder
@ 1997-09-04 22:05 ` Jeffrey A Law
1997-09-04 22:46 ` scott snyder
1997-09-04 23:06 ` Richard Henderson
0 siblings, 2 replies; 6+ messages in thread
From: Jeffrey A Law @ 1997-09-04 22:05 UTC (permalink / raw)
To: scott snyder; +Cc: egcs
In message < 199709050447.EAA14961@d0sgif.fnal.gov >you write:
> As reported earlier, this is due to assumption in this test case
> that 2*sizeof(long) == sizeof(double), which is false for the alpha.
> If this is fixed, this test passes.
A patch would be welcome.
It'd be nice if it worked on targets with 16bit ints and 32bit longs, but
if we can't make it work on such targets, then we can skip the test
with something like
if (sizeof (type) != somethinguseful)
exit (0);
Some tests already do this.
> FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O0
> FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O1
> FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O2
> FAIL: gcc.c-torture/execute/ieee/fp-cmp-1.c execution, -O2 -fomit-frame-poi
> nter -finline-functions
>
> This is apparently a problem with the test harness; the program needs
> to be compiled with -mieee.
Yes. We just added this to the Cygnus testsuite for another target; once
another snapshot of dejagnu is made I'll import the latest Cygnus testsuite
into egcs and we'll be able to deal with this problem.
> The failures which i saw previously with 920411-1.c and m-un-1.c
> are gone now.
Great.
Jeff
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: regression test results for 970904 on alpha-dec-osf4.0
1997-09-04 22:05 ` Jeffrey A Law
@ 1997-09-04 22:46 ` scott snyder
1997-09-04 22:58 ` Jeffrey A Law
1997-09-04 23:06 ` Richard Henderson
1 sibling, 1 reply; 6+ messages in thread
From: scott snyder @ 1997-09-04 22:46 UTC (permalink / raw)
To: law; +Cc: egcs
> In message < 199709050447.EAA14961@d0sgif.fnal.gov >you write:
> > As reported earlier, this is due to assumption in this test case
> > that 2*sizeof(long) == sizeof(double), which is false for the alpha.
> > If this is fixed, this test passes.
>A patch would be welcome.
hi -
I included a little patch with the previous regression report i sent;
however, that was a little cheesy in that it tested for just __alpha.
Here's one which might be a little more generally applicable.
This still assumes that a double is 64 bits, but i'm not sure how
to reliably test for that using the preprocessor.
sss
*** /d0sgif/data1/snyder/egcs/gcc/testsuite/gcc.c-torture/execute/cbrt.c Mon Aug 25 11:37:23 1997
--- cbrt.c Fri Sep 5 01:37:58 1997
***************
*** 9,14 ****
--- 9,16 ----
* ====================================================
*/
+ #include <limits.h>
+
#ifndef __vax__
static const unsigned long
B1 = 715094163, /* B1 = (682-0.03306235651)*2**20 */
***************
*** 30,36 ****
--- 32,46 ----
unsigned sign;
union {
double t;
+ #ifndef WORD_BIT
+ unsigned long pt[2];
+ #elif (WORD_BIT == 32)
+ unsigned int pt[2];
+ #elif (LONG_BIT == 32)
unsigned long pt[2];
+ #else
+ #error No suitable integer type found.
+ #endif
} ut, ux;
int n0;
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: regression test results for 970904 on alpha-dec-osf4.0
1997-09-04 22:46 ` scott snyder
@ 1997-09-04 22:58 ` Jeffrey A Law
0 siblings, 0 replies; 6+ messages in thread
From: Jeffrey A Law @ 1997-09-04 22:58 UTC (permalink / raw)
To: scott snyder; +Cc: egcs
In message < 199709050546.FAA15859@d0sgif.fnal.gov >you write:
>
> > In message < 199709050447.EAA14961@d0sgif.fnal.gov >you write:
> > > As reported earlier, this is due to assumption in this test case
> > > that 2*sizeof(long) == sizeof(double), which is false for the alpha.
> > > If this is fixed, this test passes.
> >A patch would be welcome.
>
> hi -
>
> I included a little patch with the previous regression report i sent;
> however, that was a little cheesy in that it tested for just __alpha.
> Here's one which might be a little more generally applicable.
> This still assumes that a double is 64 bits, but i'm not sure how
> to reliably test for that using the preprocessor.
Not exactly what I was looking for :)
I think the way to go is make "pt" an array of ints.
Then in main do something like this
if (sizeof (double) != 2 * sizeof (int)
|| sizeof (double) != 8)
exit (0);
Jeff
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: regression test results for 970904 on alpha-dec-osf4.0
1997-09-04 22:05 ` Jeffrey A Law
1997-09-04 22:46 ` scott snyder
@ 1997-09-04 23:06 ` Richard Henderson
1997-09-04 23:20 ` Jeffrey A Law
1 sibling, 1 reply; 6+ messages in thread
From: Richard Henderson @ 1997-09-04 23:06 UTC (permalink / raw)
To: law; +Cc: snyder, egcs
> > As reported earlier, this is due to assumption in this test case
> > that 2*sizeof(long) == sizeof(double), which is false for the alpha.
> > If this is fixed, this test passes.
> A patch would be welcome.
How about this.
r~
Index: cbrt.c
===================================================================
RCS file: /cvs/cvsfiles/egcs/gcc/testsuite/gcc.c-torture/execute/cbrt.c,v
retrieving revision 1.1.1.1
diff -u -p -r1.1.1.1 cbrt.c
--- cbrt.c 1997/08/19 07:35:59 1.1.1.1
+++ cbrt.c 1997/09/05 06:04:25
@@ -28,9 +28,10 @@ cbrtl (double x)
double r,s,w;
double lt;
unsigned sign;
+ typedef unsigned unsigned32 __attribute__((mode(SI)));
union {
double t;
- unsigned long pt[2];
+ unsigned32 pt[2];
} ut, ux;
int n0;
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: regression test results for 970904 on alpha-dec-osf4.0
1997-09-04 23:06 ` Richard Henderson
@ 1997-09-04 23:20 ` Jeffrey A Law
0 siblings, 0 replies; 6+ messages in thread
From: Jeffrey A Law @ 1997-09-04 23:20 UTC (permalink / raw)
To: Richard Henderson; +Cc: snyder, egcs
In message < 199709050605.XAA09203@rtl.cygnus.com >you write:
> > > As reported earlier, this is due to assumption in this test case
> > > that 2*sizeof(long) == sizeof(double), which is false for the alpha.
> > > If this is fixed, this test passes.
> > A patch would be welcome.
>
> How about this.
Thanks. That's even better than what I was proposing...
jeff
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~1997-09-04 23:20 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
1997-09-04 21:47 regression test results for 970904 on alpha-dec-osf4.0 scott snyder
1997-09-04 22:05 ` Jeffrey A Law
1997-09-04 22:46 ` scott snyder
1997-09-04 22:58 ` Jeffrey A Law
1997-09-04 23:06 ` Richard Henderson
1997-09-04 23:20 ` Jeffrey A Law
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).