Re: make check (testing): how to re-test just 1 failed test src file?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, 19 Jul 2018 at 21:03, U.Mutlu <um@xxxxxxxxxxx> wrote:
>
> U.Mutlu wrote on 07/19/2018 09:02 PM:
> > Jonathan Wakely wrote on 07/19/2018 08:48 PM:
> >> On Thu, 19 Jul 2018 at 19:07, U.Mutlu wrote:
> >>> This is suspect: It seems running the above said single file test does
> >>> not produce any FAIL :-) (unlike the corrosponding FAILs in the big
> >>> list above) Ie. maybe some "false-positives" happening when doing the
> >>> big "make check"s. So, then there could be a bug in the testing
> >>> framework (?)
> >>
> >> Unlikely. Before assuming a bug check the test actually ran. Look in the
> >> .log or .sum file and see if it PASSed or simply wasn't run. If you're
> >> expecting a FAIL it's far more likely it just didn't run than you've
> >> found a bug in DejaGnu that nobody else has found.
> >
> > I definitely cannot reproduce the said FAIL (tried many times to do so as
> > said previously by testing just this specific case).
> >
> > There is no indication of any fail/FAIL in the output, nor in the new
> > report (after deleting *.sum's before running that single test); the report
> > is practically empty:
> >
> > ######## LAST_UPDATED: Mon Jul 16 19:30:10 UTC 2018 (revision 262747)
> >
> > Native configuration is x86_64-pc-linux-gnu
> >
> > === gcc tests ===
> >
> >
> > Running target unix
> >
> > === gcc Summary ===
> >
> > /data/sw/src/cross/my_cross_build/SAV_x86_64-linux/build-gcc/gcc/xgcc
> > version 9.0.0 20180716 (experimental) [trunk revision 262747] (GCC)
> >
> >
> > Compiler version: 9.0.0 20180716 (experimental) [trunk revision 262747]
> > (GCC) gcc Platform: x86_64-pc-linux-gnu configure flags:
> > --prefix=/sw/src/cross --target=x86_64-linux --enable-languages=c,c++
> > --disable-werror --disable-gdb --disable-bootstrap --disable-multilib
> > ########
> >
> >
> > I'll analyse (again) the output, the log and the sum to be sure. If all
> > fails, I'll also re-do the said full test again.
>
>
> Here's a full summary of what I did in my attempty to run the single test:
>
>    # go to my build dir:
>    cd /sw/src/cross/my_cross_build/SAV_x86_64-linux
>    cd build-gcc
>
>    # delete old *.sum files of prev test runs:
>    rm -fr *.sum
>
>    make check-gcc-c RUNTESTFLAGS="-a -v compile.exp=gcc.dg/torture/pr68264.c"
> 2>&1 | tee /tmp/test_one_output.txt
>
> And attached are the files
>    build-gcc/gcc/testsuite/gcc/gcc.log
>    build-gcc/gcc/testsuite/gcc/gcc.sum
>    /tmp/test_one_output.txt
>
> The output file says that it executed the test.

No it doesn't. It shows that no tests ran.


> But funny thing is: there is neither the expected FAILs nor any PASS.

Because no tests ran.

> Can someone analyse why there is neither a FAIL nor a PASS?

Because no tests ran :-)

The gcc.c-torture/compile/compile.exp driver might not support running
individual tests, I don't know.



[Index of Archives]     [Linux C Programming]     [Linux Kernel]     [eCos]     [Fedora Development]     [Fedora Announce]     [Autoconf]     [The DWARVES Debugging Tools]     [Yosemite Campsites]     [Yosemite News]     [Linux GCC]

  Powered by Linux