Re: [RFC] Host1x/TegraDRM UAPI

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

 



On 6/25/20 1:33 AM, Dmitry Osipenko wrote:
23.06.2020 15:09, Mikko Perttunen пишет:
struct drm_tegra_submit_relocation {
         /* [in] Index of GATHER or GATHER_UPTR command in commands. */
         __u32 gather_command_index;

         /*
          * [in] Mapping handle (obtained through CHANNEL_MAP) of the memory
          *   the address of which will be patched in.
          */
         __u32 mapping_id;

         /*
          * [in] Offset in the gather that will be patched.
          */
         __u64 gather_offset;

         /*
          * [in] Offset in target buffer whose paddr/IOVA will be written
          *   to the gather.
          */
         __u64 target_offset;

         /*
          * [in] Number of bits the resulting address will be logically
          *   shifted right before writing to gather.
          */
         __u32 shift;

         __u32 reserved[1];
};

We will also need read/write direction flag here for the
DMA-reservations set up, it will be used for the implicit BO fencing by
the job's scheduler.


Ideally on newer chips with context isolation, we no longer know what DMA-BUFs are being used by the job at submit time - they would just be pointers after being MAPped.

Do you know how other GPUs deal with DMA reservations - I expect separate MAP and SUBMIT steps would be pretty common? Do you have to pass the DMA-BUF to both steps (i.e. do IOMMU mapping during MAP, and manage reservations at SUBMIT)?

Mikko



[Index of Archives]     [ARM Kernel]     [Linux ARM]     [Linux ARM MSM]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux