[Buildroot] [PATCH 2/2] package/iozone: Fix parallel build problem

[Alexis Lothoré]

Hello Thomas,

On Fri Oct 31, 2025 at 4:00 PM CET, Thomas Perale wrote:
> Hi Alexis,
>
> On 10/31/25 10:50 AM, Alexis Lothoré wrote:
>> Hello,
>>
>> On Fri Sep 12, 2025 at 2:50 PM CEST, Baruch Siach via buildroot wrote:
>>> Hi Thomas,
>>>
>>> On Thu, Sep 11 2025, Thomas Perale via buildroot wrote:
>>>> In reply of:
>>>>> Apply a patch from meta-openembedded fixing a cross-build issue [1].
>>>>>
>>>>> [1]
>>>>> https://github.com/openembedded/meta-openembedded/blob/walnascar/meta-oe/recipes-benchmark/iozone3/iozone3/parallelism.patch
>>>>>
>>>>> Fixes:
>>>>> https://gitlab.com/buildroot.org/buildroot/-/jobs/11176774405 (TestIozone)
>>>>>
>>>>> Signed-off-by: Romain Naour <romain.naour at smile.fr>
>>>> Applied to 2025.02.x & 2025.05.x. Thanks
>>> Not in 2025.02.x or 2025.05.x as of a4ac42f4e7a74 ("Revert
>>> "package/firewalld: add missing nftables json runtime dependency"") and
>>> 02f4d62bbf95 ("package/python3: do build-time detection of non-working
>>> toolchain"), respectively.
>> I was investigating a build failure on the iozone package on buildroot
>> 2025.02.x, and eventually found this series which indeed fixes the issues
>> observed locally. As Baruch stated, it does not seem to be applied on
>> 2025.02.x, despite Thomas's notification.
>>
>> Any plan to eventually apply it on 2025.02.x ?
>>
>> Thanks,
>>
>> Alexis
>
> Are you talking about the parallelism issue ? We didn't ended up picking 
> it because the fix came after the version bump which could imply this 
> error only happens after the version bump.

My bad, I did make things confusing by replying to the thread involving the
parallel build fix, but I am indeed referring to the bump to get build
fixes related to GCC raising errors instead of warnings.

> For the bump and the gcc fixes, if i remember correctly I tried to build 
> it locally and it didn't fail. Since the commit message was not specific 
> enough about the error I didn't include it. But I may have done 
> something wrong.
>
> What's the error you are running into ? The bump looks to be small so we 
> might include it as well next week.

Here are the errors I am seeing:

libbif.c:205:1: error: return type defaults to 'int' [-Wimplicit-int]
  205 | create_xls(name)
      | ^~~~~~~~~~
make[2]: *** [<builtin>: libbif.o] Error 1
make[2]: *** Waiting for unfinished jobs....
libasync.c: In function 'end_async':
libasync.c:302:9: error: implicit declaration of function 'del_cache' [-Wimplicit-function-declaration]
  302 |         del_cache(mygc);
      |         ^~~~~~~~~
libasync.c: In function 'async_read':
libasync.c:432:17: error: implicit declaration of function 'takeoff_cache'; did you mean 'alloc_cache'? [-Wimplicit-function-declaration]
  432 |                 takeoff_cache(gc,ce);
      |                 ^~~~~~~~~~~~~
      |                 alloc_cache
libasync.c: At top level:
libasync.c:618:1: warning: conflicting types for 'takeoff_cache'; have 'void(struct cache *, struct cache_ent *)'
  618 | takeoff_cache(struct cache *gc, struct cache_ent *ce)
      | ^~~~~~~~~~~~~
libasync.c:432:17: note: previous implicit declaration of 'takeoff_cache' with type 'void(struct cache *, struct cache_ent *)'
  432 |                 takeoff_cache(gc,ce);
      |                 ^~~~~~~~~~~~~
libasync.c:692:1: warning: conflicting types for 'del_cache'; have 'void()'
  692 | del_cache(gc)
      | ^~~~~~~~~
libasync.c:302:9: note: previous implicit declaration of 'del_cache' with type 'void()'
  302 |         del_cache(mygc);
      |         ^~~~~~~~~
libasync.c: In function 'async_read_no_copy':
libasync.c:815:17: error: implicit declaration of function 'putoninuse' [-Wimplicit-function-declaration]
  815 |                 putoninuse(gc,ce);
      |                 ^~~~~~~~~~
libasync.c: In function 'async_release':
libasync.c:982:9: error: implicit declaration of function 'takeoffinuse' [-Wimplicit-function-declaration]
  982 |         takeoffinuse(gc);
      |         ^~~~~~~~~~~~
libasync.c: At top level:
libasync.c:996:1: warning: conflicting types for 'putoninuse'; have 'void()'
  996 | putoninuse(gc,entry)
      | ^~~~~~~~~~
libasync.c:815:17: note: previous implicit declaration of 'putoninuse' with type 'void()'
  815 |                 putoninuse(gc,ce);
      |                 ^~~~~~~~~~
libasync.c:1017:1: warning: conflicting types for 'takeoffinuse'; have 'void()'
 1017 | takeoffinuse(gc)
      | ^~~~~~~~~~~~
libasync.c:982:9: note: previous implicit declaration of 'takeoffinuse' with type 'void()'
  982 |         takeoffinuse(gc);
      |         ^~~~~~~~~~~~
make[2]: *** [<builtin>: libasync.o] Error 1
make[2]: Leaving directory '/workspace/build/out/output-diamond/build/iozone-3.506/src/current'
make[1]: *** [package/pkg-generic.mk:273: /workspace/build/out/output-diamond/build/iozone-3.506/.stamp_built] Error 2
make[1]: Leaving directory '/workspace/build/build/buildroot'
make: *** [/workspace/build/build/matchbox/buildroot.mk:48: iozone] Error 2

I managed to reproduce the build issue with the following minimal
defconfig on 2025.02.x

BR2_arm=y
BR2_cortex_a9=y
BR2_ARM_ENABLE_NEON=y
BR2_ARM_ENABLE_VFP=y
BR2_ARM_FPU_NEON=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN=y
BR2_PACKAGE_IOZONE=y

If it helps, would you like me to resend the original patch and enrich it
with those info ?

Alexis
>
> Thanks for the message,
> PERALE Thomas




-- 
Alexis Lothoré, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com