[PATCH 4/5] drm/amdgpu: Query boot status if discovery failed

[Zhang, Hawking]

[AMD Official Use Only - General]

RE - I'm not sure about hard-coding 4 instances here. The code you dropped in patch 1 was using adev->aid_mask. But I guess that's not even initialized correctly if IP discovery failed. Will this work correctly on the APU version?

Yes aid_mask is not initialized. IP_DISCOVERY_VERSION is the only available fuse setting that can be used to identify or equivalent to 4 instances of aid in such case. We switched to a common mailbox reg that works for both APU and dGPU. The expectation is for APU, driver still reports fw boot status, while it gives next level information on the failures if boot fails on dGPU.

Regards,
Hawking

-----Original Message-----
From: Kuehling, Felix <Felix.Kuehling at amd.com>
Sent: Wednesday, January 3, 2024 01:49
To: Zhang, Hawking <Hawking.Zhang at amd.com>; amd-gfx at lists.freedesktop.org; Zhou1, Tao <Tao.Zhou1 at amd.com>; Yang, Stanley <Stanley.Yang at amd.com>; Wang, Yang(Kevin) <KevinYang.Wang at amd.com>; Chai, Thomas <YiPeng.Chai at amd.com>; Li, Candice <Candice.Li at amd.com>
Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Ma, Le <Le.Ma at amd.com>; Lazar, Lijo <Lijo.Lazar at amd.com>
Subject: Re: [PATCH 4/5] drm/amdgpu: Query boot status if discovery failed


On 2024-01-02 09:07, Hawking Zhang wrote:
> Check and report boot status if discovery failed.
>
> Signed-off-by: Hawking Zhang <Hawking.Zhang at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 6 +++++-
>   1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> index b8fde08aec8e..302b71e9f1e2 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> @@ -27,6 +27,7 @@
>   #include "amdgpu_discovery.h"
>   #include "soc15_hw_ip.h"
>   #include "discovery.h"
> +#include "amdgpu_ras.h"
>
>   #include "soc15.h"
>   #include "gfx_v9_0.h"
> @@ -98,6 +99,7 @@
>   #define FIRMWARE_IP_DISCOVERY "amdgpu/ip_discovery.bin"
>   MODULE_FIRMWARE(FIRMWARE_IP_DISCOVERY);
>
> +#define mmIP_DISCOVERY_VERSION  0x16A00
>   #define mmRCC_CONFIG_MEMSIZE        0xde3
>   #define mmMP0_SMN_C2PMSG_33 0x16061
>   #define mmMM_INDEX          0x0
> @@ -518,7 +520,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev)
>   out:
>       kfree(adev->mman.discovery_bin);
>       adev->mman.discovery_bin = NULL;
> -
> +     if ((amdgpu_discovery != 2) &&
> +         (RREG32(mmIP_DISCOVERY_VERSION) == 4))
> +             amdgpu_ras_query_boot_status(adev, 4);
I'm not sure about hard-coding 4 instances here. The code you dropped in patch 1 was using adev->aid_mask. But I guess that's not even initialized correctly if IP discovery failed. Will this work correctly on the APU version?

Regards,
   Felix


>       return r;
>   }
>