[igt-dev] ✗ Fi.CI.BAT: failure for lib/igt_dummyload: Use mapping selection to allow run batch from lmem

[Zbigniew Kempczyński]

On Fri, Dec 13, 2019 at 10:45:54AM +0000, Chris Wilson wrote:
> Quoting Patchwork (2019-12-13 10:35:28)
> > == Series Details ==
> > 
> > Series: lib/igt_dummyload: Use mapping selection to allow run batch from lmem
> > URL   : https://patchwork.freedesktop.org/series/70872/
> > State : failure
> > 
> > == Summary ==
> > 
> > CI Bug Log - changes from IGT_5349 -> IGTPW_3860
> > ====================================================
> > 
> > Summary
> > -------
> > 
> >   **FAILURE**
> > 
> >   Serious unknown changes coming with IGTPW_3860 absolutely need to be
> >   verified manually.
> >   
> >   If you think the reported changes have nothing to do with the changes
> >   introduced in IGTPW_3860, please notify your bug team to allow them
> >   to document this new failure mode, which will reduce false positives in CI.
> > 
> >   External URL: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_3860/index.html
> > 
> > Possible new issues
> > -------------------
> > 
> >   Here are the unknown changes that may have been introduced in IGTPW_3860:
> > 
> > ### IGT changes ###
> > 
> > #### Possible regressions ####
> > 
> >   * igt at gem_busy@busy-all:
> >     - fi-glk-dsi:         [PASS][1] -> [FAIL][2] +1 similar issue
> >    [1]: https://intel-gfx-ci.01.org/tree/drm-tip/IGT_5349/fi-glk-dsi/igt@gem_busy@busy-all.html
> >    [2]: https://intel-gfx-ci.01.org/tree/drm-tip/IGTPW_3860/fi-glk-dsi/igt@gem_busy@busy-all.html
> 
> Well that's conclusive, our device-coherent mapping is not.
> -Chris

For GEM_MMAP and GEM_MMAP_OFFSET we're assigning pgprot_writecombine to vma.
I see no difference between these two WC mappings apart of ioctl path.
Am I missed something?

--
Zbigniew