Re: Android i.mx6 drm_hwcomposer

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

 



Hey Martin,

On Fri, 2017-11-17 at 15:27 +0100, Martin Fuzzey wrote:
> Hi,
> 
> I'm trying to get Android graphics working on i.MX6  using upstream 
> versions:
> 
> Kernel 4.14
> 
> Mesa: mesa-17.3.0-rc2
> 
> Libdrm: libdrm-2.4.88
> 
> drm_hwcomposer (git head)
> 
> gbm_gralloc (git head from git://github.com/robherring)
> 
> 
> I want to run on Android 5, which only has HWC1, the code for that
> is 
> still in drm_hwcomposer but is not built anymore.
> So I added that back and fixed up a few minor build errors.
> 
> The first problem I ran into was
>      E/GRALLOC-GBM(  437): failed to create BO, size=1024x600,
> fmt=2, 
> usage=5
>      W/GraphicBufferAllocator(  437): alloc(1024, 600, 2, 00001a00,
> ...) 
> failed -13 (Permission denied)
> 
> This was due to an GBM attempting to issue DRM_IOCTL_MODE_CREATE_DUMB
> on 
> a render node.
> 
> Setting gralloc.gbm.device /dev/dri/card1  (the imx-drm node) instead
> of 
> the default /dev/dri/renderD128 fixed that but then HWC couldn't
> become 
> master on the DRI node to issue KMS commands since gralloc had
> already 
> opened the device.
> 
> Adding a patch similar to:
> 
> commit 4f4cb902517f5caacea075fb9724e4ce0c435e3d
> Author: Robert Foss <robert.foss@xxxxxxxxxxxxx>
> Date:   Wed May 17 18:08:14 2017 -0400
> 
>      drm_hwcomposer: Get KMS FD from gbm_gralloc
> 
> To make drm_hwc and gbm_gralloc use the same FD fixed that.
> 
> 
> But the problem I now have that when HWC tries to import the first 
> client buffers (from bootanimation) there are errors like:
> 
> 01-01 17:27:02.282   443   443 E hwc-platform-drm-generic: failed to 
> import prime fd 25 ret=-1
> 
> This is because the imx-drm kernel driver wants physically
> contiguous 
> buffers for import (it uses drm_gem_cma_prime_import_sg_table()) but
> the 
> buffer allocated by drm etnaviv isn't contiguous...
> 
> Not sure what to do about that one.
> 
> Any help appreciated.
> 
> Martin
> 

Remarkably this is an issue have not encountered. I would suggest
poking the fine folks on the etnaviv ML, like Lucas/Wladimir/Christian.


Rob.

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel

[Index of Archives]     [Linux DRI Users]     [Linux Intel Graphics]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]
  Powered by Linux