[opensuse-factory] gcc-4.8 serious bug? gcc regression
Just tried to build (using configure) a package that failed in the gcc test. Seems like the graphite loop optimizer is missing. The manpage infers that the wrong build options were used.(?) (was present in earlier versions). The error was: conftest.c:1:0: sorry, unimplemented: Graphite loop optimizations cannot be used (-fgraphite, -fgraphite-identity, -floop-block, -floop-interchange, -floop-strip-mine, -floop-parallelize-all, and -ftree-loop-linear) /* confdefs.h */ The manpage says this about the loop optimizer: To use this code transformation, GCC has to be configured with --with-ppl and --with-cloog to enable the Graphite loop transformation infrastructure. gcc-4.8-1.3.x86_64.rpm dated Aug 5 rpm -V gcc verifies it as integrous. (yeah, it's a valid word) -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Tue, Oct 29, 2013 at 03:24:50PM -0700, Linda Walsh wrote:
Just tried to build (using configure) a package that failed in the gcc test.
Seems like the graphite loop optimizer is missing. The manpage infers that the wrong build options were used.(?)
(was present in earlier versions).
The error was: conftest.c:1:0: sorry, unimplemented: Graphite loop optimizations cannot be used (-fgraphite, -fgraphite-identity, -floop-block, -floop-interchange, -floop-strip-mine, -floop-parallelize-all, and -ftree-loop-linear) /* confdefs.h */
The manpage says this about the loop optimizer:
To use this code transformation, GCC has to be configured with --with-ppl and --with-cloog to enable the Graphite loop transformation infrastructure.
gcc-4.8-1.3.x86_64.rpm dated Aug 5
rpm -V gcc verifies it as integrous. (yeah, it's a valid word)
What is the version of the "gcc48"? And what compiler flags were tried by configure of those listed above? Ciao, Marcus -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Wed, 30 Oct 2013, Marcus Meissner wrote:
On Tue, Oct 29, 2013 at 03:24:50PM -0700, Linda Walsh wrote:
Just tried to build (using configure) a package that failed in the gcc test.
Seems like the graphite loop optimizer is missing. The manpage infers that the wrong build options were used.(?)
(was present in earlier versions).
The error was: conftest.c:1:0: sorry, unimplemented: Graphite loop optimizations cannot be used (-fgraphite, -fgraphite-identity, -floop-block, -floop-interchange, -floop-strip-mine, -floop-parallelize-all, and -ftree-loop-linear) /* confdefs.h */
Seriously the package shouldn't use any of these options.
The manpage says this about the loop optimizer:
To use this code transformation, GCC has to be configured with --with-ppl and --with-cloog to enable the Graphite loop transformation infrastructure.
gcc-4.8-1.3.x86_64.rpm dated Aug 5
rpm -V gcc verifies it as integrous. (yeah, it's a valid word)
What is the version of the "gcc48"?
And what compiler flags were tried by configure of those listed above?
I see from succeeded/obs-x86_64/gcc48
[ 111s] checking for version 0.10 of ISL... no
[ 111s] checking for version 0.11 of ISL... no
[ 111s] checking for version 0.12 of ISL... yes
[ 111s] checking for version 0.17.0 of CLooG... no
[ 111s] checking for version 0.18.0 of CLooG... yes
you may hit a time where the version check didn't support ISL 0.12
but factory had updated to that version.
Richard.
--
Richard Biener
On 10/30/2013 3:04 AM, Richard Biener wrote:
And what compiler flags were tried by configure of those listed above?
I ran a config script that set CFLAGS .. Attached it to bug Bug 848156 ... as well as the config log. These are all the packages w/gcc in them. gcc-c++-4.8-1.3.x86_64 gcc47-4.7.3_20130411-1.1.x86_64 gcc47-c++-4.7.3_20130411-1.1.x86_64 gcc48-32bit-4.8.1_20130806-1.2.x86_64 gcc48-4.8.1_20130806-1.2.x86_64 gcc48-ada-4.8.1_20130806-1.2.x86_64 gcc48-c++-4.8.1_20130806-1.2.x86_64 gcc48-fortran-4.8.1_20130806-1.2.x86_64 gcc48-gij-4.8.0_20130322-1.1.x86_64 gcc48-info-4.8.0_20130322-1.1.noarch gcc48-java-4.8.1_20130806-1.2.x86_64 gcc48-locale-4.8.1_20130806-1.2.x86_64 gcc48-objc-4.8.1_20130806-1.2.x86_64 gccmakedep-1.0.2-5.1.noarch libffi4-gcc48-4.8.0_20130322-1.1.x86_64 libgcc_s1-32bit-4.8.1_20130806-1.2.x86_64 libgcc_s1-4.8.1_20130806-1.2.x86_64 libstdc++6-gcc48-32bit-4.8.0_20130322-1.1.x86_64 libstdc++6-gcc48-4.8.0_20130322-1.1.x86_64 libstlport_gcc4-4.6.2-18.1.x86_64 Except for the gcc47 packages which apparently didn't get removed (maybe they coexist?) most look like they are from august: For cloog, I have: cloog-isl-devel-0.18.0-4.3.x86_64 libcloog0-0.15.10+ppl-14.2.x86_64 libcloog-isl4-0.18.0-4.3.x86_64 Shouldn't the gcc47 packages have been removed when I installed the 48 ones? or is there a reason both are in distro directory? (i.e. they work together?) It's not something I told it to keep, FWIW...
I see from succeeded/obs-x86_64/gcc48
[ 111s] checking for version 0.10 of ISL... no [ 111s] checking for version 0.11 of ISL... no [ 111s] checking for version 0.12 of ISL... yes [ 111s] checking for version 0.17.0 of CLooG... no [ 111s] checking for version 0.18.0 of CLooG... yes
you may hit a time where the version check didn't support ISL 0.12 but factory had updated to that version.
Richard.
-- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Wednesday 2013-10-30 13:41, Linda Walsh wrote:
Shouldn't the gcc47 packages have been removed when I installed the 48 ones?
No, there is no conflict.
or is there a reason both are in distro directory? (i.e. they work together?)
absolutely
It's not something I told it to keep, FWIW...
It's not something you told it to delete, FWIW. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On 10/30/2013 7:54 AM, Jan Engelhardt wrote:
On Wednesday 2013-10-30 13:41, Linda Walsh wrote:
Shouldn't the gcc47 packages have been removed when I installed the 48 ones?
No, there is no conflict.
Great!
or is there a reason both are in distro directory? (i.e. they work together?)
absolutely
--- Fantastic!... I was just concerned as I as cutting and pasting that something had gotten messed up and I'd be blamed (not that it belonging there means I won't, I know, but ... you know what I mean... ;-)
It's not something I told it to keep, FWIW...
It's not something you told it to delete, FWIW.
--- True.. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
participants (4)
-
Jan Engelhardt -
Linda Walsh -
Marcus Meissner -
Richard Biener