Summary: | ppc64el:Gallium / clover fails to compile | ||
---|---|---|---|
Product: | Mesa | Reporter: | Timothy Pearson <kb9vqf> |
Component: | Drivers/Gallium/r600 | Assignee: | Default DRI bug account <dri-devel> |
Status: | RESOLVED DUPLICATE | QA Contact: | Default DRI bug account <dri-devel> |
Severity: | major | ||
Priority: | medium | CC: | currojerez |
Version: | 11.0 | ||
Hardware: | PowerPC | ||
OS: | All | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: | clover_workaround_altivec_madness.patch |
Description
Timothy Pearson
2016-01-13 01:49:20 UTC
Created attachment 120997 [details] [review] clover_workaround_altivec_madness.patch The error is caused by the altivec.h system header (included on PPC systems by Khronos' CL headers we cannot change) which redefines standard C++ keywords. You may be able to use the attached hack to work around the issue, sigh... Since this is marked NOTOURBUG, is there somewhere I should be submitting a bug report to get this fixed? I'd really like to see OpenCL on the OpenPOWER hardware without having to manually patch / build, and without having to get a downstream patch into Debian / RHEL. Thanks! (In reply to Timothy Pearson from comment #3) > Since this is marked NOTOURBUG, is there somewhere I should be submitting a > bug report to get this fixed? I'd really like to see OpenCL on the > OpenPOWER hardware without having to manually patch / build, and without > having to get a downstream patch into Debian / RHEL. > > Thanks! Either GCC or Khronos -- I'd be inclined to blame GCC for this, since the altivec.h header is unusable from standard C++ code (assuming you don't hack around it by undef'ing "vector" and "bool" after inclusion or by preventing it from defining them in the first place somehow), however last time somebody reported this [1] the bug was closed as invalid, so apparently some people from the GCC community think that this is the (unfortunate) expected behaviour -- Assuming they're right, cl_platform.h should stop including altivec.h (or hack around it somehow) because it makes cl.h unusable from standard C++ code on PPC platforms. [1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58241 Bug reported here: https://www.khronos.org/bugzilla/show_bug.cgi?id=1449 Thanks for the hint, and link to the GCC bug report! |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.