Am 21.12.18 um 19:19 schrieb Alex Deucher: > On Fri, Dec 21, 2018 at 1:15 PM Christian König > <ckoenig.leichtzumerken@xxxxxxxxx> wrote: >> Am 21.12.18 um 10:09 schrieb Deng, Emily: >>>> -----Original Message----- >>>> From: Michel Dänzer <michel@xxxxxxxxxxx> >>>> Sent: Friday, December 21, 2018 4:52 PM >>>> To: Deng, Emily <Emily.Deng@xxxxxxx> >>>> Cc: amd-gfx@xxxxxxxxxxxxxxxxxxxxx >>>> Subject: Re: [PATCH] drm/amdgpu/virtual_dce: Need to pin the fb's bo >>>> >>>> On 2018-12-21 9:45 a.m., Deng, Emily wrote: >>>>>> -----Original Message----- >>>>>> From: Michel Dänzer <michel@xxxxxxxxxxx> >>>>>> Sent: Friday, December 21, 2018 4:38 PM >>>>>> To: Deng, Emily <Emily.Deng@xxxxxxx> >>>>>> Cc: amd-gfx@xxxxxxxxxxxxxxxxxxxxx >>>>>> Subject: Re: [PATCH] drm/amdgpu/virtual_dce: Need to pin the fb's bo >>>>>> >>>>>> On 2018-12-21 8:26 a.m., Emily Deng wrote: >>>>>>> When the bo is used to set mode, the bo need to be pinned. >>>>>> On second thought, why does the BO need to be pinned? When using the >>>>>> display hardware, the BO needs to be pinned to prevent it from being >>>>>> moved while the hardware is scanning out from it, but that shouldn't be >>>> necessary here. >>>>> The pin here is used for scan out the buffer by remote display app. >>>> I still don't understand why pinning is needed. What mechanism does the remote >>>> display app use to access the BO contents? >>> Sorry, I am not familiar with the remote display app. Maybe it will use drm ioctl function to get the >>> current crtc's fb's information, and get the content in the fb's buffer object by mmap or translate the bo >>> to an OpenGL texture for next rendering. Maybe don't need to pin the bo here, as the use has no different with >>> other normal bos. So please ignore the patch, and will send another patch to remove the unpin the fb's bo code. >> Opening the BO and doing something with it in OpenGL should result in >> proper fencing of the BO, so this sounds like a workaround for a bug >> somewhere else. >> >> As long as this isn't explained further this patch is certainly a NAK. >> >> Pinning for physical displays is allowed because they are limited in >> number, but this is not necessary the case with a virtual output. > Well practically, it's the same as real displays. We limit the > virtual displays to 6 just like most hw. It really should overate the > same. We just don't need the pinning per se because there is no hw > actively scanning out of it. Yeah, but do we limit the number of pending flips like we do for real hardware as well? For real hard we have at maximum two BOs pinned, the current one and the next one. For the virtual scanout I'm not sure about that. Christian. > > Alex _______________________________________________ amd-gfx mailing list amd-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/amd-gfx
