GPU: Provide general GPUChkErr functionality also externally and several unrelated changes

[davidrohr]

@mconcas : Sorry this took so long :(.
This adds a GPUCommonChkErr.h header, providing my GPUChkErr macros.
Currently, they must be used upon a GPUReconstruction object.
The reason is that I need to dispatch to the backend to get the errorcodes, so I could not easily make it static.

Would this work for you?

I could probably create a static version, which would automatically detect which GPUReconstruction class is created first, and attach to that, and then dispatch to that.
That should in principle cover all use cases, except for if someone would use multiple backends in the same application, which is in principle supported.

[github-actions]

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2023-pbpb-apass4
async-2023-pp-apass4
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0
async-2024-PbPb-apass1
async-2024-ppRef-apass1
async-2024-PbPb-apass2
async-2023-PbPb-apass5

[alibuild]

Error while checking build/O2/fullCI_slc9 for 061b880 at 2025-03-14 09:52:

## sw/BUILD/O2-latest/log
ld.lld: error: undefined symbol: o2::gpu::GPUReconstructionHIPBackend::GPUChkErrStatic(long, char const*, int)
clang++: error: linker command failed with exit code 1 (use -v to see invocation)
ninja: build stopped: subcommand failed.

Full log here.

[davidrohr]

@mconcas : You should be able to use GPUChkErrS and GPUChkErrSI statically, if you link against the external provider.

[mconcas]

Thx, I'll try it!

[davidrohr]

@ktf @singiamtel : Could you check the MacOS builder that ran for this PR, this error seems pretty weird:

Call Stack (most recent call first):
  dependencies/CMakeLists.txt:13 (include)
  CMakeLists.txt:65 (include)
This warning is for project developers.  Use -Wno-dev to suppress it.

-- Input BUILD_SIMULATION=
CMake Error at dependencies/O2Dependencies.cmake:161 (include):
  include could not find requested file:

[ktf]

Yes, it looks indeed quite strange. I assume the issue is:

[0/1] Re-running CMake...
CMake Error: The source directory "/Volumes/build/alice-ci-workdir/o2/sw/SOURCES/O2/14062/0" does not exist.
Specify --help for usage, or press the help button on the CMake GUI.

I am looking into it.

[ktf]

I restarted the CI script on the builder and it seems to work. I suspect a recent change in upstream python might have left the SOURCES area in a bad state and while @singiamtel fixed the issue in alisw/alibuild#913 not all the builders where restarted to a clean slate.

[ktf]

@singiamtel I have the impression there is still something wrong with the mac builders.

[ktf]

Ok, after another cleanup of the build machine the error seems to be a genuine complaint about a missing:

dependencies/O2SimulationDependencies.cmake

any idea of how that might happen?

[davidrohr]

Now one of the Mac CIs passed O2, and then just stopped.
The other failed with CMake Error: The source directory "/Volumes/build/alice-ci-workdir/o2/sw/SOURCES/O2/14062/0" does not exist.

Do you think these issues are related to my PR? Or is this a general problem of the MacCI?