[PATCH umr] Fix VMID of chained IBs

Christian König ckoenig.leichtzumerken at gmail.com
Tue Apr 10 15:29:24 UTC 2018


Am 10.04.2018 um 17:23 schrieb Tom St Denis:
> We were using the VMID field literally when inside an IB it's inherited instead
>
> Signed-off-by: Tom St Denis <tom.stdenis at amd.com>

Acked-by: Christian König <christian.koenig at amd.com>

> ---
>   src/lib/dump_ib.c     | 8 ++++----
>   src/lib/ring_decode.c | 2 +-
>   2 files changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/src/lib/dump_ib.c b/src/lib/dump_ib.c
> index 74a3241c309e..cbf859e3633b 100644
> --- a/src/lib/dump_ib.c
> +++ b/src/lib/dump_ib.c
> @@ -39,10 +39,10 @@ void umr_dump_ib(struct umr_asic *asic, struct umr_ring_decoder *decoder)
>   	if (decoder->src.ib_addr == 0)
>   		printf("ring[%s%u%s]", BLUE, (unsigned)decoder->src.addr, RST);
>   	else
> -		printf("IB[%s%u%s] at %s%d%s@%s0x%llx%s",
> -			BLUE, (unsigned)decoder->src.addr, RST,
> -			YELLOW, (int)decoder->src.vmid, RST,
> -			YELLOW, (unsigned long long)decoder->src.ib_addr, RST);
> +		printf("IB[%s%u%s@%s0x%llx%s + %s0x%x%s]",
> +			BLUE, (int)decoder->src.vmid, RST,
> +			YELLOW, (unsigned long long)decoder->src.ib_addr, RST,
> +			YELLOW, (unsigned)decoder->src.addr * 4, RST);
>   
>   	printf("\n");
>   
> diff --git a/src/lib/ring_decode.c b/src/lib/ring_decode.c
> index c1d6bcb98bae..42265e0a74c9 100644
> --- a/src/lib/ring_decode.c
> +++ b/src/lib/ring_decode.c
> @@ -540,7 +540,7 @@ static void print_decode_pm4_pkt3(struct umr_asic *asic, struct umr_ring_decoder
>   					break;
>   				case 2: printf("IB_SIZE:%s%lu%s, VMID: %s%lu%s", BLUE, BITS(ib, 0, 20), RST, BLUE, BITS(ib, 24, 32), RST);
>   					decoder->pm4.next_ib_state.ib_size = BITS(ib, 0, 20) * 4;
> -					decoder->pm4.next_ib_state.ib_vmid = BITS(ib, 24, 32);
> +					decoder->pm4.next_ib_state.ib_vmid = decoder->next_ib_info.vmid ? decoder->next_ib_info.vmid : BITS(ib, 24, 32);
>   					add_ib_pm4(decoder);
>   					break;
>   				default: printf("Invalid word for opcode 0x%02lx", (unsigned long)decoder->pm4.cur_opcode);



More information about the amd-gfx mailing list