Tomasz, > -----Original Message----- > From: mesa-dev [mailto:[email protected]] On Behalf > Of Tomasz Figa > Sent: Friday, September 1, 2017 9:53 AM > To: Marathe, Yogesh <[email protected]> > Cc: Gao, Shuo <[email protected]>; Liu, Zhiquan <[email protected]>; > [email protected]; [email protected]; Antognolli, Rafael > <[email protected]>; [email protected]; Emil Velikov > <[email protected]>; Wu, Zhongmin <[email protected]>; > [email protected]; [email protected]; > [email protected]; Kondapally, Kalyan <[email protected]>; > [email protected]; [email protected] > Subject: Re: [Mesa-dev] [PATCH v6.2] egl: Allow creation of per surface out > fence > > On Thu, Aug 31, 2017 at 2:18 AM, Marathe, Yogesh > <[email protected]> wrote: > >> -----Original Message----- > >> From: mesa-dev [mailto:[email protected]] On > >> Behalf Of Emil Velikov > >> Sent: Wednesday, August 30, 2017 9:44 PM > >> To: Marathe, Yogesh <[email protected]> > >> Cc: Gao, Shuo <[email protected]>; Liu, Zhiquan > >> <[email protected]>; [email protected]; > >> [email protected]; Antognolli, Rafael > >> <[email protected]>; [email protected]; Wu, Zhongmin > >> <[email protected]>; [email protected]; [email protected]; > >> Kondapally, Kalyan <[email protected]>; > >> [email protected]; [email protected]; > >> [email protected]; [email protected] > >> Subject: Re: [Mesa-dev] [PATCH v6.2] egl: Allow creation of per > >> surface out fence > >> > >> On 30 August 2017 at 15:39, Marathe, Yogesh > >> <[email protected]> > >> wrote: > >> > >> > > >> > Thank you, Tomasz and all involved for the help and guidance. > >> > > >> Our excitement was short lived - see commit > >> 8c9df0daf20206fafb7df77b1edcbc41b8e91372. > >> > >> Seems the patch was not run through the Intel CI, though I'm should > >> not have assumed that you're aware of if. > >> Please get in touch with Mark Janes (Cc'd here, janesma on IRC). He > >> can set you up and/or run a branch for you. > >> > > > > No problem. I will contact Mark. > > > > Primarily looks like platform / kernel version issue. > > intel_get_boolean() for I915_PARAM_HAS_EXEC_FENCE is false, but I see > > following in i915_drv.c:915_getparam in kernel, no clue why that would > > come false in UMD. > > > > ... > > case I915_PARAM_HAS_EXEC_FENCE: > > /* For the time being all of these are always true; > > * if some supported hardware does not have one of these > > * features this value needs to be provided from > > * INTEL_INFO(), a feature macro, or similar. > > */ > > value = 1; > > break; > > ... > > Which kernel are you looking at? Remember that not everyone uses current > upstream master. There is a number of upstream stable releases and > downstream forks. Grepping for I915_PARAM_HAS_EXEC_FENCE on > http://elixir.free-electrons.com, shows that it was only added in Linux 4.12. >
I'm on 4.9.x but I see my kernel tree has following patch, so this looks like it is done for android (cherry picked / backport). That’s why it worked for me always! commit f0683754f03fa308a2657cb1dadbf235c9607188 Author: Chris Wilson <[email protected]> Date: Fri Jan 27 09:40:08 2017 +0000 drm/i915: Support explicit fencing for execbuf Nonetheless, as you mentioned, I've synced up with Mark and we've created a separate branch where CTS / intel mesa CI can run. Let me try to fix this. Caveat: To have flatland running on android there was another issue in kernel which needed a fix. Details - https://bugs.freedesktop.org/show_bug.cgi?id=101656 > Best regards, > Tomasz > _______________________________________________ > mesa-dev mailing list > [email protected] > https://lists.freedesktop.org/mailman/listinfo/mesa-dev _______________________________________________ mesa-dev mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/mesa-dev
