Re: [PATCH v3 0/5] Guestperf: miscellaneous refinement and enrichment

[Fabiano Rosas <farosas@suse.de>]

Daniel P. Berrangé <berrange@redhat.com> writes:

> On Tue, Dec 03, 2024 at 09:20:44AM -0300, Fabiano Rosas wrote:
>> Daniel P. Berrangé <berrange@redhat.com> writes:
>> 
>> > On Mon, Dec 02, 2024 at 02:49:33PM -0500, Peter Xu wrote:
>> >> On Sat, Nov 30, 2024 at 10:01:43PM +0800, Yong Huang wrote:
>> >> > On Fri, Nov 29, 2024 at 9:44 PM Fabiano Rosas <farosas@suse.de> wrote:
>> >> For the stress.c warnings/error, I saw another one warning with gcc when
>> >> built it locally, then I found Fabiano's CI run also has it:
>> >> 
>> >> https://gitlab.com/farosas/qemu/-/jobs/8504697347
>> >> 
>> >> Which has:
>> >> 
>> >> [3381/3611] Linking target tests/migration/stress
>> >> /usr/bin/ld: /usr/lib64/libglib-2.0.a(gutils.c.o): in function `g_get_user_database_entry':
>> >> (.text+0xeb): warning: Using 'getpwnam_r' in statically linked applications requires at runtime the shared libraries from the glibc version used for linkign
>> >> /usr/bin/ld: (.text+0x2be): warning: Using 'getpwuid' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
>> >> /usr/bin/ld: (.text+0x134): warning: Using 'getpwuid_r' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
>> >> 
>> >> Feel free to look at it too if you like. Nothing should be relevant to your
>> >> changes, so I think it could be there for a while when compilers upgrade.
>> >
>> > Those messages are mostly harmless, especially in the stress scenario, and
>> > can't practically be eliminated:
>> >
>> >   https://gitlab.gnome.org/GNOME/glib/-/issues/2019
>> 
>> Although harmless, we can't have them showing up during the
>> build. Should we go back to not building stress.c by default? Another
>> option (which I prefer) would be to stop using glib in stress.c, it
>> doesn't look like it would be much work to do that. 
>
> FYI they already show up in QEMU builds if you are building qemu-user as
> static binaries, which is what you want for binfmt usage.  The main
> diff is that in this case they'll show up by default for anyone who has
> glib-static available.  Is that a big problem ?

We shouldn't be adding warnings to the build like that. When building
static binaries, I'd assume the person at least knows there's a -static
in there somewhere. If you're just building the system binaries and
warnings start to show up, that's not good. Since this is just a side
script that's very infrequently used, I don't think it justifies the
extra warning.

Maybe let's see what the other build issues were, I'm guessing it will
be more work to fix them than to disable stress.c build again.

>
>
> With regards,
> Daniel