For help:Unexpected fail about testsuite of GCC

U.Mutlu um@mutluit.com
Mon Jul 9 09:51:00 GMT 2018


陈龙 wrote on 07/09/2018 04:19 AM:
> /Sorry ,the site
> //https://docs.philips.com/:f:/r/personal/long_chen_philips_com/Documents/g++log?csf=1&e=L28vsv// have
> a wrong direction,please ignore it, refer to
> //https://gcc.gnu.org/ml/gcc-testresults/2018-05/msg00583.html//./
>

 >     Is that mean my test is right even if there are so many unexpected
 >     failures?And could I use the latest gcc version(8.1.0)  before I correct
 >     these fails in my environment?/

Hi,
I had not the time to look further on the failing tests.
I'm just a user like yourself :-)

If you need just a working compiler then take a precompiled stable version 
from the repository of your linux distribution. It will be an older version, 
not the current 8.1/9 version.

The 8.1/9 version is the "developer version", ie. unfinished work-in-progress
version. This is intented for testers and gcc-developers (patch submitters etc.).

FYI: yesterday an official snapshot of v9 has been released by gccadmin:
https://gcc.gnu.org/ml/gcc/2018-07/msg00119.html

I would say yes, take the latest version, but if your program behaves 
differently than expected, then remember that it could be a compiler issue.

Cheers
U.Mutlu


> /Best regards,/
> /CL/
>
>
>
>
>
>
> At 2018-07-09 10:01:46, "陈龙" <18116491546@163.com> wrote:
>
>     /Hi,all/
>     /Thanks for the response./
>     /I know we need analyse the log if we want to deal with the failures. But
>     before that ,I want to confirm some things described in my first mail ,I
>     bring them here as below:/
>     //
>      >> I have run the  testsuite of GCC and compared with results from a
>     similar configuration in the  gcc-testresults mailing list,  the results
>     just have a little difference,and my test summary is in chapter Summary as
>     below,. Both of the results have many unexpected fails, I want to know why
>     they failed but the g++.log couldn’t affort enough fail case
>     information,they were all almost described like this:'FAIL:
>     g++.dg/pr80481.C  -std=gnu++14  scan-assembler-not vmovaps'.So was it
>     right with so many fails? And could you tell me how could I debug the fail
>     case please?Thanks.
>      >>
>      >>
>      >>
>      >> Test Environment
>      >>
>      >>
>      >>
>      >>
>      >> - x86_64-pc-mingw64 and msys2
>      >>
>      >> - gcc8.1.0
>      >>
>      >>
>      >>
>      >>
>      >> Summary
>      >>
>      >>                 === g++ Summary ===
>      >>
>      >> # of expected passes            115078
>      >> # of unexpected failures        628
>      >> # of unexpected successes       3
>      >> # of expected failures          493
>      >> # of unresolved testcases       5
>      >> # of unsupported tests          5122
>      >> /mingw64/bin/c++  version 8.1.0 (x86_64-posix-seh-rev0, Built by
>     MinGW-W64 project)
>
>     /please look at the sentence with yellow background , my test got 628
>     unexpected  failures,which is a little of difference from the result with
>     a similar configuration in the  gcc-testresults mailing list(refer to
>     //https://docs.philips.com/:f:/r/personal/long_chen_philips_com/Documents/g++log?csf=1&e=L28vsv/
>     <https://docs.philips.com/:f:/r/personal/long_chen_philips_com/Documents/g++log?csf=1&e=L28vsv%20>/).
>     Is that mean my test is right even if there are so many unexpected
>     failures?And could I use the latest gcc version(8.1.0)  before I correct
>     these fails in my environment?/
>
>     /Best regards,/
>     /CL/
>
>
>     At 2018-07-06 20:54:33, "U.Mutlu" <um@mutluit.com <mailto:um@mutluit.com>> wrote:
>     >Jonathan Wakely wrote on 07/06/2018 02:20 PM:
>     >> On Fri, 6 Jul 2018 at 11:46, U.Mutlu <um@mutluit.com <mailto:um@mutluit.com>> wrote:
>     >>>
>     >>> Hello CL,
>     >>>
>     >>> [as said, I'm new to the gcc testsuite, much like you, so take my answer with
>     >>> grain of salt, or so to say :-)]
>     >>>
>     >>>
>     >>> In your original posting you had this failing example:
>     >>> 'FAIL: g++.dg/pr80481.C  -std=gnu++14  scan-assembler-not vmovaps'
>     >>>
>     >>> The test source file pr80481.C is in the source directory under:
>     >>>     $ find ../gcc_trunk/ -iname "pr80481.C" -print
>     >>>     ../gcc_trunk/gcc/testsuite/g++.dg/pr80481.C
>     >>> and is handled by the make-target "check-c++" (see the log file of "make -k
>     >>> check"; you have to save that log file for reference & searching...).
>     >>
>     >> Nope, see my previous email:
>     >> https://gcc.gnu.org/ml/gcc-help/2018-07/msg00020.html
>     >>
>     >>> So, now in the next step you have to make the following test that will
>     >>> test just the "check-c++" category of the tests but with the additional
>     >>> flags "RUNTESTFLAGS="-v -v" as it will give more verbosity in the log file:
>     >>>     make -k check-c++ RUNTESTFLAGS="-v -v"
>     >>> And again redirect the output into a log file for analysis.
>     >>
>     >> Look in $builddir/gcc/testsuite/g++/g++.log which already has the info
>     >> you need, and why it failed.
>     >
>     >I see, thx, but there is only the following short log excerpt with
>     >no indication of any "error: " entries for the said FAIL case.
>     >Is this maybe caused by the fact that I used "make -j ...",
>     >ie. is the log file maybe "garbled" by the many threads, and
>     >the "error: " entries are somewhere outside of this range below?
>     >
>     >
>     >Testing g++.dg/pr80481.C,  -std=gnu++14
>     >replacement dg-process-target: `{ target { i?86-*-* x86_64-*-* }  && { !
>     >*-*-solaris* } }'
>     >dg-process-target-1: `{target { i?86-*-* x86_64-*-* }  && { ! *-*-solaris* }}'
>     >replacement dg-process-target: `{target i?86-*-* x86_64-*-*}'
>     >dg-process-target-1: `{target i?86-*-* x86_64-*-*}'
>     >selector_list: ` i?86-*-* x86_64-*-* ' 1
>     >selector_expression: ` i?86-*-* x86_64-*-* ' 1
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/x86_64-linux-gnu/./libio/_G_config.h
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/x86_64-linux-gnu/libio/_G_config.h
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/libio/_G_config.h
>     >/data/sw/src/gcc_dev/libio/_G_config.h
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/x86_64-linux-gnu/./libio/iostream.list
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/x86_64-linux-gnu/libio/iostream.list
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/libio/iostream.list
>     >/data/sw/src/gcc_dev/libio/iostream.list
>     >/sw/src/gcc_dev/gcc_trunk/gcc/testsuite/libio/Makefile.in
>     >/sw/src/gcc_dev/gcc_trunk/gcc/libio/Makefile.in
>     >/sw/src/gcc_dev/gcc_trunk/libio/Makefile.in
>     >/sw/src/gcc_dev/libio/Makefile.in
>     >doing compile
>     >Invoking the compiler as
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/gcc/testsuite/g++3/../../xg++
>     >-B/data/sw/src/gcc_dev/my_build_dir_for_gcc/gcc/testsuite/g++3/../../
>     >/sw/src/gcc_dev/gcc_trunk/gcc/testsuite/
>     >Setting timeout to 300
>     >Executing on host:
>     >/data/sw/src/gcc_dev/my_build_dir_for_gcc/gcc/testsuite/g++3/../../xg++
>     >-B/data/sw/src/gcc_dev/my_build_dir_for_gcc/gcc/testsuite/g++3/../../
>     >/sw/src/gcc_dev/gcc_trunk/gcc/testsuite/g++.dg
>     >spawn /data/sw/src/gcc_dev/my_build_dir_for_gcc/gcc/testsuite/g++3/../../xg++
>     >-B/data/sw/src/gcc_dev/my_build_dir_for_gcc/gcc/testsuite/g++3/../../
>     >/sw/src/gcc_dev/gcc_trunk/gcc/testsuite/g++.dg/pr80481.C -f
>     >pid is 23149 -23149
>     >waitres is 23149 exp8 0 0
>     >output is  status 0
>     >Checking pattern "sparc-*-sunos*" with x86_64-pc-linux-gnu
>     >Checking pattern "alpha*-*-*" with x86_64-pc-linux-gnu
>     >Checking pattern "hppa*-*-hpux*" with x86_64-pc-linux-gnu
>     >Checking pattern "sparc-*-sunos*" with x86_64-pc-linux-gnu
>     >Checking pattern "alpha*-*-*" with x86_64-pc-linux-gnu
>     >Checking pattern "hppa*-*-hpux*" with x86_64-pc-linux-gnu
>     >PASS: g++.dg/pr80481.C  -std=gnu++14 (test for excess errors)
>     >FAIL: g++.dg/pr80481.C  -std=gnu++14  scan-assembler-not vmovaps
>     >
>




More information about the Gcc-help mailing list