public inbox for gcc-prs@sourceware.org
help / color / mirror / Atom feed
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-03-13 14:13 dje
0 siblings, 0 replies; 10+ messages in thread
From: dje @ 2002-03-13 14:13 UTC (permalink / raw)
To: aj, dje, dje, gcc-bugs, gcc-prs, nobody, pthomas
Synopsis: bootstraping CVS HEAD on powerpc fails in checked build
Responsible-Changed-From-To: unassigned->dje
Responsible-Changed-By: dje
Responsible-Changed-When: Wed Mar 13 14:13:37 2002
Responsible-Changed-Why:
powerpc
State-Changed-From-To: open->closed
State-Changed-By: dje
State-Changed-When: Wed Mar 13 14:13:37 2002
State-Changed-Why:
Fixed
http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5736
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 13:40 Graham Stott
0 siblings, 0 replies; 10+ messages in thread
From: Graham Stott @ 2002-02-20 13:40 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: Graham Stott <grahams@redhat.com>
To: David Edelsohn <dje@watson.ibm.com>
Cc: pthomas@suse.de, gcc-gnats@gcc.gnu.org, aj@suse.de,
aldyh@redhat.com
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
Date: Wed, 20 Feb 2002 20:40:01 +0000
David,
David Edelsohn wrote:
>
> How can this patch have an effect on the problem if Altivec is not
> enabled?
>
> David
I don't think Altivec needs to be enabled to hit the enable checking failure
we just need to pass one of the modes V16QImode, V8HImode, V4SFmode or
V4SImode into the LEGITIMATE_OFFSET_ADDRESS_P macro.
I did build a powerpc-suse-linux-gnu cross compiler and hit the abort in _mulsi3
applying the patch fixed the abort, but didn't continue the build past building
libgcc.a
Graham
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 10:26 Alan Matsuoka
0 siblings, 0 replies; 10+ messages in thread
From: Alan Matsuoka @ 2002-02-20 10:26 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: Alan Matsuoka <alanm@redhat.com>
To: David Edelsohn <dje@watson.ibm.com>
Cc: pthomas@suse.de, gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, aj@suse.de
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
Date: Wed, 20 Feb 2002 13:02:15 -0500
On Wed, 20 Feb 2002, David Edelsohn wrote:
> >>>>> Alan Matsuoka writes:
>
> Alan> Pre-Approved by Geoff Keating.
>
> Does this mean that you have already checked it in?
>
No. There was a typo .
Alan Matsuoka
GCC Engineering
Red Hat Canada, Ltd
mailto:alanm@redhat.com Tel: (416) 482-2661 x250 / Fax: (416) 482-6299
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 10:06 Alan Matsuoka
0 siblings, 0 replies; 10+ messages in thread
From: Alan Matsuoka @ 2002-02-20 10:06 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: Alan Matsuoka <alanm@redhat.com>
To: pthomas@suse.de
Cc: gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, aj@suse.de, dje@watson.ibm.com
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
Date: Wed, 20 Feb 2002 12:46:12 -0500
## > >
## >
## > You might want to try this patch it's completly untested. I don't have enough time or resources to test it.
I sent that one in to gcc-patches last night.
I've bootstrapped in on ppc linux without any regressions.
I caught this one when working on the AltiVec stuff with
checking turned on.
Pre-Approved by Geoff Keating.
alanm@redhat.com (Alan Matsuoka) writes:
# > Hey guy,
# >
# >
# > I was running a build with all the checking flags turned on
# > while looking for a problem with my stuff when I ran into an
# > abort inside LEGITIMATE_OFFSET_ADDRESS_P
# >
# > *************** typedef struct rs6000_args
# > *** 1969,1973 ****
# > && INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
# > && LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
# > ! && (! ALTIVEC_VECTOR_MODE (MODE) || INTVAL (X) == 0) \
# > && (((MODE) != DFmode && (MODE) != DImode) \
# > || (TARGET_32BIT \
# > --- 2116,2121 ----
# > && INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
# > && LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
# > ! && (! ALTIVEC_VECTOR_MODE (mode) \
# > ! || (GET_CODE (XEXP (X,1)) == CONST_INT && INTVAL (XEXP (X,1)) == 0)) \
# > && (((MODE) != DFmode && (MODE) != DImode) \
# > || (TARGET_32BIT \
# >
# >
# > rtl checking aborts in the INTVAL (X) because it was looking at a const_int
# > I think that this is what fixes it.
#
# Good catch!
#
# Please send this to the FSF list, with a ChangeLog entry, and you can
# say I approved it in advance.
#
# --
# - Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>
Bootstrapped on ppc linux without any regressions.
Wed Feb 20 00:03:25 EST 2002 Alan Matsuoka <alanm@redhat.com>
* config/rs6000/rs6000.h (LEGITIMATE_OFFSET_ADDRESS_P): Look
for (const_int 0) in X not just INTVAL.
Index: rs6000.h
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/rs6000/rs6000.h,v
retrieving revision 1.181
diff -d -c -p -r1.181 rs6000.h
*** rs6000.h 2002/02/19 02:53:36 1.181
--- rs6000.h 2002/02/20 04:52:08
*************** typedef struct rs6000_args
*** 1971,1977 ****
&& GET_CODE (XEXP (X, 0)) == REG \
&& INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
&& LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
! && (! ALTIVEC_VECTOR_MODE (MODE) || INTVAL (X) == 0) \
&& (((MODE) != DFmode && (MODE) != DImode) \
|| (TARGET_32BIT \
? LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 4) \
--- 1971,1978 ----
&& GET_CODE (XEXP (X, 0)) == REG \
&& INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
&& LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
! && (! ALTIVEC_VECTOR_MODE (mode) \
! || (GET_CODE (XEXP (X,1)) == CONST_INT && INTVAL (XEXP (X,1)) == 0)) \
&& (((MODE) != DFmode && (MODE) != DImode) \
|| (TARGET_32BIT \
? LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 4) \
## Graham.
##
## ChangeLog
##
## * config/rs6000/rs6000.h (LEGITIMATE_OFFSET_ADDRESS_P): Fix typo.
##
## ------------------------------------------------------------------------
## Index: gcc/config/rs6000/rs6000.h
## ===================================================================
## RCS file: /cvs/gcc/gcc/gcc/config/rs6000/rs6000.h,v
## retrieving revision 1.182
## diff -c -p -r1.182 rs6000.h
## *** rs6000.h 2002/02/20 04:56:34 1.182
## --- rs6000.h 2002/02/20 17:24:00
## *************** typedef struct rs6000_args
## *** 1974,1980 ****
## && GET_CODE (XEXP (X, 0)) == REG \
## && INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
## && LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
## ! && (! ALTIVEC_VECTOR_MODE (MODE) || INTVAL (X) == 0) \
## && (((MODE) != DFmode && (MODE) != DImode) \
## || (TARGET_32BIT \
## ? LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 4) \
## --- 1974,1981 ----
## && GET_CODE (XEXP (X, 0)) == REG \
## && INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
## && LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
## ! && (! ALTIVEC_VECTOR_MODE (MODE) \
## ! || INTVAL (XEXP (X, 1)) == 0) \
## && (((MODE) != DFmode && (MODE) != DImode) \
## || (TARGET_32BIT \
## ? LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 4) \
## ------------------------------------------------------------------------
Alan Matsuoka
GCC Engineering
Red Hat Canada, Ltd
mailto:alanm@redhat.com Tel: (416) 482-2661 x250 / Fax: (416) 482-6299
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 10:06 David Edelsohn
0 siblings, 0 replies; 10+ messages in thread
From: David Edelsohn @ 2002-02-20 10:06 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: David Edelsohn <dje@watson.ibm.com>
To: Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
Cc: pthomas@suse.de, Aldy Hernandez <aldyh@redhat.com>, gcc-gnats@gcc.gnu.org,
aj@suse.de
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
Date: Wed, 20 Feb 2002 12:57:47 -0500
This already has been superceded. by Alan's patch. This patch
from Franz and others will be ignored.
David
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 10:06 David Edelsohn
0 siblings, 0 replies; 10+ messages in thread
From: David Edelsohn @ 2002-02-20 10:06 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: David Edelsohn <dje@watson.ibm.com>
To: Alan Matsuoka <alanm@redhat.com>
Cc: pthomas@suse.de, gcc-prs@gcc.gnu.org, gcc-gnats@gcc.gnu.org, aj@suse.de
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
Date: Wed, 20 Feb 2002 12:58:27 -0500
>>>>> Alan Matsuoka writes:
Alan> Pre-Approved by Geoff Keating.
Does this mean that you have already checked it in?
David
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 10:06 Franz Sirl
0 siblings, 0 replies; 10+ messages in thread
From: Franz Sirl @ 2002-02-20 10:06 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: Franz Sirl <Franz.Sirl-kernel@lauterbach.com>
To: pthomas@suse.de
Cc: Aldy Hernandez <aldyh@redhat.com>,gcc-gnats@gcc.gnu.org,aj@suse.de,
dje@watson.ibm.com
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in
checked build
Date: Wed, 20 Feb 2002 18:56:29 +0100
At 18:14 20.02.2002, pthomas@suse.de wrote:
> >Number: 5736
> >Category: target
> >Synopsis: bootstraping CVS HEAD on powerpc fails in checked build
> >Confidential: no
> >Severity: serious
> >Priority: medium
> >Responsible: unassigned
> >State: open
> >Class: sw-bug
> >Submitter-Id: net
> >Arrival-Date: Wed Feb 20 09:16:01 PST 2002
> >Closed-Date:
> >Last-Modified:
> >Originator: pthomas@suse.de
> >Release: 3.1 20020220
> >Organization:
> >Environment:
>powerpc-suse-linux-gnu
>
>gcc configured with:
>--enable-shared --with-system-zlib --enable-threads
>--host=powerpc-suse-linux-gnu --target=powerpc-suse-linux-gnu
>--build=powerpc-suse-linux-gnu --enable-checking=misc,tree,rtl,gc,gcac
>Thread model: posix
>
> >Description:
>Bootstrapping CVS HEAD with -O2 fails when compiling libgcc2, specifically
>_negdi2 with
>cc1: RTL check: expected code `const_int', have `plus' in
>rs6000_legitimate_address, at config/rs6000/rs6000.c:2014
Hmm, looking at the code I would think this line in LEGITIMATE_OFFSET_ADDRESS_P
&& (! ALTIVEC_VECTOR_MODE (MODE) || INTVAL (X) == 0)
should actually read
&& (! ALTIVEC_VECTOR_MODE (MODE) || INTVAL (XEXP (X, 1)) == 0)
? Aldy?
Maybe this even causes PR 5693?
Franz.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 9:56 David Edelsohn
0 siblings, 0 replies; 10+ messages in thread
From: David Edelsohn @ 2002-02-20 9:56 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: David Edelsohn <dje@watson.ibm.com>
To: Graham Stott <grahams@redhat.com>
Cc: pthomas@suse.de, gcc-gnats@gcc.gnu.org, aj@suse.de
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
Date: Wed, 20 Feb 2002 12:45:59 -0500
How can this patch have an effect on the problem if Altivec is not
enabled?
David
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 9:56 Graham Stott
0 siblings, 0 replies; 10+ messages in thread
From: Graham Stott @ 2002-02-20 9:56 UTC (permalink / raw)
To: nobody; +Cc: gcc-prs
The following reply was made to PR target/5736; it has been noted by GNATS.
From: Graham Stott <grahams@redhat.com>
To: pthomas@suse.de
Cc: gcc-gnats@gcc.gnu.org, aj@suse.de, dje@watson.ibm.com
Subject: Re: target/5736: bootstraping CVS HEAD on powerpc fails in checked build
Date: Wed, 20 Feb 2002 17:33:57 +0000
All,
pthomas@suse.de wrote:
>
> >Number: 5736
> >Category: target
> >Synopsis: bootstraping CVS HEAD on powerpc fails in checked build
> >Confidential: no
> >Severity: serious
> >Priority: medium
> >Responsible: unassigned
> >State: open
> >Class: sw-bug
> >Submitter-Id: net
> >Arrival-Date: Wed Feb 20 09:16:01 PST 2002
> >Closed-Date:
> >Last-Modified:
> >Originator: pthomas@suse.de
> >Release: 3.1 20020220
> >Organization:
> >Environment:
> powerpc-suse-linux-gnu
>
> gcc configured with:
> --enable-shared --with-system-zlib --enable-threads --host=powerpc-suse-linux-gnu --target=powerpc-suse-linux-gnu --build=powerpc-suse-linux-gnu --enable-checking=misc,tree,rtl,gc,gcac
> Thread model: posix
>
> >Description:
> Bootstrapping CVS HEAD with -O2 fails when compiling libgcc2, specifically _negdi2 with
> cc1: RTL check: expected code `const_int', have `plus' in rs6000_legitimate_address, at config/rs6000/rs6000.c:2014
>
Looks to be a simple typo in LEGITIMATE_OFFSET_ADDRESS_P.
> Compiling at lower optimisation succeeds.
>
> >How-To-Repeat:
> Compile attatched _negdi2.i with:
> ./xgcc -B./ -O2 -DIN_GCC -fPIC -g1 -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED -fPIC -mstrict-align -DL_negdi2 -c _negdi2.i
> >Fix:
>
You might want to try this patch it's completly untested. I don't have enough time or resources to test it.
Graham.
ChangeLog
* config/rs6000/rs6000.h (LEGITIMATE_OFFSET_ADDRESS_P): Fix typo.
------------------------------------------------------------------------
Index: gcc/config/rs6000/rs6000.h
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/rs6000/rs6000.h,v
retrieving revision 1.182
diff -c -p -r1.182 rs6000.h
*** rs6000.h 2002/02/20 04:56:34 1.182
--- rs6000.h 2002/02/20 17:24:00
*************** typedef struct rs6000_args
*** 1974,1980 ****
&& GET_CODE (XEXP (X, 0)) == REG \
&& INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
&& LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
! && (! ALTIVEC_VECTOR_MODE (MODE) || INTVAL (X) == 0) \
&& (((MODE) != DFmode && (MODE) != DImode) \
|| (TARGET_32BIT \
? LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 4) \
--- 1974,1981 ----
&& GET_CODE (XEXP (X, 0)) == REG \
&& INT_REG_OK_FOR_BASE_P (XEXP (X, 0), (STRICT)) \
&& LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 0) \
! && (! ALTIVEC_VECTOR_MODE (MODE) \
! || INTVAL (XEXP (X, 1)) == 0) \
&& (((MODE) != DFmode && (MODE) != DImode) \
|| (TARGET_32BIT \
? LEGITIMATE_ADDRESS_INTEGER_P (XEXP (X, 1), 4) \
------------------------------------------------------------------------
^ permalink raw reply [flat|nested] 10+ messages in thread
* target/5736: bootstraping CVS HEAD on powerpc fails in checked build
@ 2002-02-20 9:26 pthomas
0 siblings, 0 replies; 10+ messages in thread
From: pthomas @ 2002-02-20 9:26 UTC (permalink / raw)
To: gcc-gnats; +Cc: aj, dje
>Number: 5736
>Category: target
>Synopsis: bootstraping CVS HEAD on powerpc fails in checked build
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: unassigned
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Feb 20 09:16:01 PST 2002
>Closed-Date:
>Last-Modified:
>Originator: pthomas@suse.de
>Release: 3.1 20020220
>Organization:
>Environment:
powerpc-suse-linux-gnu
gcc configured with:
--enable-shared --with-system-zlib --enable-threads --host=powerpc-suse-linux-gnu --target=powerpc-suse-linux-gnu --build=powerpc-suse-linux-gnu --enable-checking=misc,tree,rtl,gc,gcac
Thread model: posix
>Description:
Bootstrapping CVS HEAD with -O2 fails when compiling libgcc2, specifically _negdi2 with
cc1: RTL check: expected code `const_int', have `plus' in rs6000_legitimate_address, at config/rs6000/rs6000.c:2014
Compiling at lower optimisation succeeds.
>How-To-Repeat:
Compile attatched _negdi2.i with:
./xgcc -B./ -O2 -DIN_GCC -fPIC -g1 -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED -fPIC -mstrict-align -DL_negdi2 -c _negdi2.i
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="_negdi2.i"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="_negdi2.i"
dHlwZWRlZiBpbnQgU0l0eXBlIF9fYXR0cmlidXRlX18gKChtb2RlIChTSSkpKTsKdHlwZWRlZiB1
bnNpZ25lZCBpbnQgVVNJdHlwZSBfX2F0dHJpYnV0ZV9fICgobW9kZSAoU0kpKSk7CgpzdHJ1Y3Qg
RFdzdHJ1Y3Qge1NJdHlwZSBoaWdoLCBsb3c7fTsKCnR5cGVkZWYgdW5pb24KewogIHN0cnVjdCBE
V3N0cnVjdCBzOwogIERJdHlwZSBsbDsKfSBEV3VuaW9uOwoKREl0eXBlCl9fbmVnZGkyIChESXR5
cGUgdSkKewogIERXdW5pb24gdzsKICBEV3VuaW9uIHV1OwoKICB1dS5sbCA9IHU7CgogIHcucy5s
b3cgPSAtdXUucy5sb3c7CiAgdy5zLmhpZ2ggPSAtdXUucy5oaWdoIC0gKChVU0l0eXBlKSB3LnMu
bG93ID4gMCk7CgogIHJldHVybiB3LmxsOwp9Cg==
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2002-03-13 22:13 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-03-13 14:13 target/5736: bootstraping CVS HEAD on powerpc fails in checked build dje
-- strict thread matches above, loose matches on Subject: below --
2002-02-20 13:40 Graham Stott
2002-02-20 10:26 Alan Matsuoka
2002-02-20 10:06 Alan Matsuoka
2002-02-20 10:06 Franz Sirl
2002-02-20 10:06 David Edelsohn
2002-02-20 10:06 David Edelsohn
2002-02-20 9:56 Graham Stott
2002-02-20 9:56 David Edelsohn
2002-02-20 9:26 pthomas
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).