[Bug 2209759] New: Review Request: rocclr - ROCm Compute Language Runtime

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

 



https://bugzilla.redhat.com/show_bug.cgi?id=2209759

            Bug ID: 2209759
           Summary: Review Request: rocclr - ROCm Compute Language Runtime
           Product: Fedora
           Version: rawhide
          Hardware: All
                OS: Linux
            Status: NEW
         Component: Package Review
          Severity: medium
          Priority: medium
          Assignee: nobody@xxxxxxxxxxxxxxxxx
          Reporter: alexjnewt@xxxxxxxxxxxx
        QA Contact: extras-qa@xxxxxxxxxxxxxxxxx
                CC: package-review@xxxxxxxxxxxxxxxxxxxxxxx
  Target Milestone: ---
    Classification: Fedora



Spec URL: https://mystro256.fedorapeople.org/rocclr.spec
SRPM URL: https://mystro256.fedorapeople.org/rocclr-5.5.0-1.fc39.src.rpm
Description: ROCm Compute Language Runtime
Fedora Account System Username: mystro256
Copr Build:
https://copr.fedorainfracloud.org/coprs/mystro256/rocm-hip/build/5952345/

Background:
I maintain rocm-opencl in Fedora and wanted to introduce rocm-hip to Fedora
too. I noticed that rocm-hip shares a lot of code with rocm-opencl, so I was
working with upstream via email to help clarify the source. They have agreed
that for ROCm 5.7.0, they will merge all the source into one rocclr tree to
reduce confusion and allow for easier distribution of source code. As a result,
I think I makes more sense to introduce a ROCclr package with opencl/hip as
subpackages.

The "preview" of rocm 5.7's new source organisation is located here:
https://github.com/ROCm-Developer-Tools/clr

This spec file emulates the future source organisation, so it should be very
simple to transition to ROCm 5.7.0 when it release. Therefore when I update to
5.7.0 in the future, I should be able to drop Source 2 and 3, Patch 0 and 1, as
well as the extraction logic in %prep.

Notes:
- As per above, this will replace and supersede rocm-opencl, so a lot of it is
copied from the rocm-opencl spec file
- I will retire rocm-opencl after this is approved, so ignore any conflicts
with it
- HIP attempts to be a platform generic API, while hipamd is AMD's vendor
specific implementation, so I've wrote the spec file to allow easy separation
later if HIP becomes more separated/detached from ROCm (see hip-devel) or there
is value to Fedora keeping them separated

RPMlint output:
> rocm-hip.x86_64: E: shared-library-without-dependency-information /usr/lib64/libhiprtc-builtins.so.5.5.30201

GDB fails to extract debug data from this library. I believe it's because it's
in a format that's not supported. Let me know if this is a blocker for
inclusion.

> hip-devel.noarch: W: no-manual-page-for-binary hipcc
> hip-devel.noarch: W: no-manual-page-for-binary hipcc.pl
> hip-devel.noarch: W: no-manual-page-for-binary hipcc_cmake_linker_helper
> hip-devel.noarch: W: no-manual-page-for-binary hipconfig
> hip-devel.noarch: W: no-manual-page-for-binary hipconfig.pl
> hip-devel.noarch: W: no-manual-page-for-binary hipdemangleatp
> rocm-clinfo.x86_64: W: no-manual-page-for-binary rocm-clinfo
> rocm-hip-devel.x86_64: W: no-manual-page-for-binary roc-obj
> rocm-hip-devel.x86_64: W: no-manual-page-for-binary roc-obj-extract
> rocm-hip-devel.x86_64: W: no-manual-page-for-binary roc-obj-ls

I have expressed to upstream over email that I would like to contribute
manpages, but I'm a bit busy, so they might take some time.

> rocm-clinfo.x86_64: W: no-documentation
> rocm-hip-devel.x86_64: W: no-documentation
> rocm-opencl.x86_64: W: no-documentation
> rocm-opencl-devel.x86_64: W: no-documentation

Not applicable/no docs available.

> rocm-hip.x86_64: E: executable-stack /usr/lib64/libamdhip64.so.5.5.30201
> rocm-hip.x86_64: E: executable-stack /usr/lib64/libhiprtc-builtins.so.5.5.30201
> rocm-hip.x86_64: E: executable-stack /usr/lib64/libhiprtc.so.5.5.30201

I've looked through the source and don't see anything obviously wrong. I'll do
a deeper dive later, but I suspect it's just due to how hipamd is written, not
a mistake per say, so it would take some time to resolve. Let me know if it's a
blocking issue.


-- 
You are receiving this mail because:
You are always notified about changes to this product and component
You are on the CC list for the bug.
https://bugzilla.redhat.com/show_bug.cgi?id=2209759
_______________________________________________
package-review mailing list -- package-review@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to package-review-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/package-review@xxxxxxxxxxxxxxxxxxxxxxx
Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue




[Index of Archives]     [Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Yosemite Conditions]     [KDE Users]

  Powered by Linux