OK, so with respect to that article, we've been working with our members to provide KMS drivers for their SoCs. The results are in varying states of completeness and availability, but I would expect that to improve in the coming cycles. You should be able to find DRM drivers for pandaboard (omapdrm: see http://www.phoronix.com/scan.php?page=news_item&px=MTA5MDg) and origen (exynos - this driver is already upstream in drivers/gpu/drm/exynos) with patches out for review on those all the time on the dri-devel list. Snowball support has not quite made it to igloocommunity, but that should be in progress.
It's a work in progress, but we are making that progress all the time. I hope this helps. cheers, Jesse On Tue, May 15, 2012 at 7:45 PM, Ilyes Gouta <ilyes.go...@gmail.com> wrote: > Hi Jesse, > > Here: http://www.phoronix.com/scan.php?page=news_item&px=OTI0MA and > http://lists.freedesktop.org/archives/wayland-devel/2011-March/000855.html > > It doesn't propose a unified interface for 2D accelerators on SoCs, though. > >> There's nothing I'm aware of that would define what you are asking >> (apart from the Xserver's EXA framework which certainly isn't new or > > Yes, but that's heavily geared towards X11. > >> in the kernel). Even the interfaces exported by DRM require user >> space code to orchestrate them (i.e., no kernel acceleration in the >> purest sense). > > It just that it's kind of common that SoCs have separate IPs for > handling 2D and 3D acceleration, not like PC world. And client code > could live on both kernel and user lands, hence a need for arbitration > and so on. > > I was just asking, out of curiosity. > > -Ilyes > > On Tue, May 15, 2012 at 6:11 PM, Jesse Barker <jesse.bar...@linaro.org> wrote: >> Can you point out the article you're referring to that mentioned the >> Linaro project? >> >> There's nothing I'm aware of that would define what you are asking >> (apart from the Xserver's EXA framework which certainly isn't new or >> in the kernel). Even the interfaces exported by DRM require user >> space code to orchestrate them (i.e., no kernel acceleration in the >> purest sense). >> >> cheers, >> Jesse >> >> On Tue, May 15, 2012 at 5:59 PM, Ilyes Gouta <ilyes.go...@gmail.com> wrote: >>> Hi Christian, >>> >>> Yes dma-buf is part of the picture, but rather if any work has been >>> done to define an interface for the device itself, not the buffers. >>> >>> I do know that these are mostly managed from user-space for >>> performance reasons, however I was curious to see if anything has been >>> in the works for kernel-space (kind of drm but for much more tailored >>> for 2D blitters). >>> >>> -Ilyes >>> >>> On Tue, May 15, 2012 at 3:14 PM, Christian Robottom Reis >>> <k...@linaro.org> wrote: >>>> On Mon, May 14, 2012 at 10:18:17PM +0100, Ilyes Gouta wrote: >>>>> I've previously read (probably on Phoronix) that Linaro is working out >>>>> a 'standard' kernel interface for 2D blitters IPs as commonly found on >>>>> SoCs. >>>>> >>>>> Has it ever been the case? If yes, are there any >>>>> documentation/references online? >>>> >>>> I wonder if you are talking about dma-buf and the other collection of >>>> work around Unified Memory Management: >>>> >>>> https://wiki.linaro.org/OfficeofCTO/MemoryManagement >>>> >>>> It's not really specific for 2D blitter IPs, but it does define how >>>> memory can be allocated and shared between different devices, >>>> particularly between the CPU, display controller and GPU IP. >>>> -- >>>> Christian Robottom Reis, Engineering VP >>>> Brazil (GMT-3) | [+55] 16 9112 6430 | [+1] 612 216 4935 >>>> Linaro.org: Open Source Software for ARM SoCs >>> >>> _______________________________________________ >>> linaro-dev mailing list >>> linaro-dev@lists.linaro.org >>> http://lists.linaro.org/mailman/listinfo/linaro-dev _______________________________________________ linaro-dev mailing list linaro-dev@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-dev
