Tests Failing on x86_64_gnu

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Tests Failing on x86_64_gnu

Nicholas Krause
Greetings,

I just build,configuring and ran the tests from the upstream gcc git repo like this:
mdkir odbjir
cd obdjir
$PWD/../gcc/configure  --prefix=$HOME/GCC-deve
make -j8
make bootstrap
make -k check

Those are the steps from the offical docs unless I am missing a step i.e. does
make install need to be done or not? Don't know if this a reported issue or not.

Cheers,

Nick
Reply | Threaded
Open this post in threaded view
|

Re: Tests Failing on x86_64_gnu

Jonathan Wakely-4
On Mon, 29 Oct 2018 at 18:05, nick <[hidden email]> wrote:
>
> Greetings,
>
> I just build,configuring and ran the tests from the upstream gcc git repo like this:
> mdkir odbjir
> cd obdjir
> $PWD/../gcc/configure  --prefix=$HOME/GCC-deve
> make -j8
> make bootstrapq

This is redundant, just saying "make" does the same thing.

> make -k check
>
> Those are the steps from the offical docs unless I am missing a step i.e. does
> make install need to be done or not?

Done for what? It needs to be done to install the compiler you just
built, but not to test it.

> Don't know if this a reported issue or not.

You haven't said what the issue is, so we can't say.
Reply | Threaded
Open this post in threaded view
|

Re: Tests Failing on x86_64_gnu

Nicholas Krause


On 2018-10-29 4:44 p.m., Jonathan Wakely wrote:
> Please reply to the mailing list, not just to me.
>
>
Sorry thought it was sent to the list.

Nick

> On Mon, 29 Oct 2018 at 20:14, nick <[hidden email]> wrote:
>>
>>
>>
>> On 2018-10-29 3:31 p.m., Jonathan Wakely wrote:
>>> On Mon, 29 Oct 2018 at 18:05, nick <[hidden email]> wrote:
>>>>
>>>> Greetings,
>>>>
>>>> I just build,configuring and ran the tests from the upstream gcc git repo like this:
>>>> mdkir odbjir
>>>> cd obdjir
>>>> $PWD/../gcc/configure  --prefix=$HOME/GCC-deve
>>>> make -j8
>>>> make bootstrapq
>>>
>>> This is redundant, just saying "make" does the same thing.
>>>
>>>> make -k check
>>>>
>>>> Those are the steps from the offical docs unless I am missing a step i.e. does
>>>> make install need to be done or not?
>>>
>>> Done for what? It needs to be done to install the compiler you just
>>> built, but not to test it.
>>>
>>>> Don't know if this a reported issue or not.
>>>
>>> You haven't said what the issue is, so we can't say.
>>>
>> Jonathan,
>> The issues is after the above commands including make -k check it fails the tests on
>> on a Ubuntu 18.04, x86_64_gnu based system for the mainline git repo. I tried it
>> clean with the above steps and it fails including reporting a few unexpected failures.
>> Those steps above  are the steps I ran in order to build and test it on a clean gcc without
>> anything else and it fails. I will run the tests again just to see if something is using too much
>> memory and making it fail as that may be the case but I doubt it.
>>
>> Sorry for the confusion,
>> Nick
Reply | Threaded
Open this post in threaded view
|

Re: Tests Failing on x86_64_gnu

Jonathan Wakely-4
On Mon, 29 Oct 2018 at 21:40, nick <[hidden email]> wrote:

>
>
>
> On 2018-10-29 4:44 p.m., Jonathan Wakely wrote:
> > Please reply to the mailing list, not just to me.
> >
> >
> Sorry thought it was sent to the list.
>
> Nick
> > On Mon, 29 Oct 2018 at 20:14, nick <[hidden email]> wrote:
> >>
> >>
> >>
> >> On 2018-10-29 3:31 p.m., Jonathan Wakely wrote:
> >>> On Mon, 29 Oct 2018 at 18:05, nick <[hidden email]> wrote:
> >>>>
> >>>> Greetings,
> >>>>
> >>>> I just build,configuring and ran the tests from the upstream gcc git repo like this:
> >>>> mdkir odbjir
> >>>> cd obdjir
> >>>> $PWD/../gcc/configure  --prefix=$HOME/GCC-deve
> >>>> make -j8
> >>>> make bootstrapq
> >>>
> >>> This is redundant, just saying "make" does the same thing.
> >>>
> >>>> make -k check
> >>>>
> >>>> Those are the steps from the offical docs unless I am missing a step i.e. does
> >>>> make install need to be done or not?
> >>>
> >>> Done for what? It needs to be done to install the compiler you just
> >>> built, but not to test it.
> >>>
> >>>> Don't know if this a reported issue or not.
> >>>
> >>> You haven't said what the issue is, so we can't say.
> >>>
> >> Jonathan,
> >> The issues is after the above commands including make -k check it fails the tests on
> >> on a Ubuntu 18.04, x86_64_gnu based system for the mainline git repo.

Which tests? All of them? More than half of them? You're really not
giving us much to go on here.

You can compare your results with other people's results for recent
trunk builds at https://gcc.gnu.org/ml/gcc-testresults/2018-10/

For recent x86_64-pc-linux-gnu builds I see about 0.1% of tests
failing (a few for the compiler, and a few for libstdc++).

It's not unusual for there to be a few failures at this point in the
development cycle. If you want a stable compiler with no issues, don't
use the development trunk.

> >> I tried it
> >> clean with the above steps and it fails including reporting a few unexpected failures.
> >> Those steps above  are the steps I ran in order to build and test it on a clean gcc without
> >> anything else and it fails. I will run the tests again just to see if something is using too much
> >> memory and making it fail as that may be the case but I doubt it.
> >>
> >> Sorry for the confusion,
> >> Nick
Reply | Threaded
Open this post in threaded view
|

Re: Tests Failing on x86_64_gnu

Jonathan Wakely-4
CCing the list again.
On Mon, 29 Oct 2018 at 22:13, nick <[hidden email]> wrote:

>
>
>
> On 2018-10-29 5:45 p.m., Jonathan Wakely wrote:
> > On Mon, 29 Oct 2018 at 21:40, nick <[hidden email]> wrote:
> >>
> >>
> >>
> >> On 2018-10-29 4:44 p.m., Jonathan Wakely wrote:
> >>> Please reply to the mailing list, not just to me.
> >>>
> >>>
> >> Sorry thought it was sent to the list.
> >>
> >> Nick
> >>> On Mon, 29 Oct 2018 at 20:14, nick <[hidden email]> wrote:
> >>>>
> >>>>
> >>>>
> >>>> On 2018-10-29 3:31 p.m., Jonathan Wakely wrote:
> >>>>> On Mon, 29 Oct 2018 at 18:05, nick <[hidden email]> wrote:
> >>>>>>
> >>>>>> Greetings,
> >>>>>>
> >>>>>> I just build,configuring and ran the tests from the upstream gcc git repo like this:
> >>>>>> mdkir odbjir
> >>>>>> cd obdjir
> >>>>>> $PWD/../gcc/configure  --prefix=$HOME/GCC-deve
> >>>>>> make -j8
> >>>>>> make bootstrapq
> >>>>>
> >>>>> This is redundant, just saying "make" does the same thing.
> >>>>>
> >>>>>> make -k check
> >>>>>>
> >>>>>> Those are the steps from the offical docs unless I am missing a step i.e. does
> >>>>>> make install need to be done or not?
> >>>>>
> >>>>> Done for what? It needs to be done to install the compiler you just
> >>>>> built, but not to test it.
> >>>>>
> >>>>>> Don't know if this a reported issue or not.
> >>>>>
> >>>>> You haven't said what the issue is, so we can't say.
> >>>>>
> >>>> Jonathan,
> >>>> The issues is after the above commands including make -k check it fails the tests on
> >>>> on a Ubuntu 18.04, x86_64_gnu based system for the mainline git repo.
> >
> > Which tests? All of them? More than half of them? You're really not
> > giving us much to go on here.
> >
> > You can compare your results with other people's results for recent
> > trunk builds at https://gcc.gnu.org/ml/gcc-testresults/2018-10/
> >
> > For recent x86_64-pc-linux-gnu builds I see about 0.1% of tests
> > failing (a few for the compiler, and a few for libstdc++).
> >
> > It's not unusual for there to be a few failures at this point in the
> > development cycle. If you want a stable compiler with no issues, don't
> > use the development trunk.
> >
>
> Sure I was aware of that and seems so after checking the build logs, thanks for those.
> Anyhow my last question is I fixed a bug and was wondering if even with failing tests
> before the patch was applied i.e. a clean gcc tree will it be able to be merged or
> not as that's the standard I was aware of for patches.

If tests are failing without your patch, you don't need to worry about
them. You're looking for changes caused by your patch, not
pre-existing failures. This is documented at
https://gcc.gnu.org/contribute.html#testing

"You must also perform regression tests to ensure that your patch does
not break anything else. Typically, this means comparing post-patch
test results to pre-patch results by testing twice or comparing with
recent posts to the gcc-testresults list."

If there had to be zero test failures before any patch is committed
then there wouldn't be dozens of commits per day:
https://gcc.gnu.org/ml/gcc-cvs/2018-10/