[Mesa-dev] [PATCH] radv: Emit cache flushes before CP DMA.

Alex Smith asmith at feralinteractive.com
Wed Mar 15 07:46:35 UTC 2017


On 14 March 2017 at 20:56, Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>
wrote:

> The flushes could be due to TRANSFER barriers.
>
> Signed-off-by: Bas Nieuwenhuizen <basni at google.com>
> Cc: 17.0 <mesa-stable at lists.freedesktop.org>
>

Hi Bas,

Are these needed? si_cp_dma_prepare already does a flush. That's why I
didn't add a flush to this path on my last patch.

Thanks,
Alex


> ---
>  src/amd/vulkan/si_cmd_buffer.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/src/amd/vulkan/si_cmd_buffer.c b/src/amd/vulkan/si_cmd_
> buffer.c
> index 5d35287f8e3..b808052ddb2 100644
> --- a/src/amd/vulkan/si_cmd_buffer.c
> +++ b/src/amd/vulkan/si_cmd_buffer.c
> @@ -998,6 +998,7 @@ void si_cp_dma_buffer_copy(struct radv_cmd_buffer
> *cmd_buffer,
>         uint64_t main_src_va, main_dest_va;
>         uint64_t skipped_size = 0, realign_size = 0;
>
> +       si_emit_cache_flush(cmd_buffer);
>
>         if (cmd_buffer->device->physical_device->rad_info.family <=
> CHIP_CARRIZO ||
>             cmd_buffer->device->physical_device->rad_info.family ==
> CHIP_STONEY) {
> @@ -1061,6 +1062,8 @@ void si_cp_dma_clear_buffer(struct radv_cmd_buffer
> *cmd_buffer, uint64_t va,
>
>         assert(va % 4 == 0 && size % 4 == 0);
>
> +       si_emit_cache_flush(cmd_buffer);
> +
>         while (size) {
>                 unsigned byte_count = MIN2(size, CP_DMA_MAX_BYTE_COUNT);
>                 unsigned dma_flags = 0;
> --
> 2.12.0
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20170315/00979c4f/attachment.html>


More information about the mesa-dev mailing list