[PATCH 08/14] drm/amdgpu: add gfx v9.4.3 ACA support

[Wang, Yang(Kevin)]

[AMD Official Use Only - General]

Thanks, will update it in next version.

Best Regards,
Kevin

-----Original Message-----
From: Zhang, Hawking <Hawking.Zhang at amd.com>
Sent: Wednesday, January 3, 2024 7:45 PM
To: Wang, Yang(Kevin) <KevinYang.Wang at amd.com>; amd-gfx at lists.freedesktop.org
Cc: Zhou1, Tao <Tao.Zhou1 at amd.com>; Chai, Thomas <YiPeng.Chai at amd.com>
Subject: RE: [PATCH 08/14] drm/amdgpu: add gfx v9.4.3 ACA support

[AMD Official Use Only - General]

Simiar as patch #9 and #10, let's use macro to define the magic numbers

+       case 0x36430400: /* SMNAID XCD 0 */
+       case 0x38430400: /* SMNAID XCD 1 */
+       case 0x40430400: /* SMNXCD XCD 0, NOTE: FIXME: fix this error
+ later */

Regards,
Hawking

-----Original Message-----
From: Wang, Yang(Kevin) <KevinYang.Wang at amd.com>
Sent: Wednesday, January 3, 2024 16:02
To: amd-gfx at lists.freedesktop.org
Cc: Zhang, Hawking <Hawking.Zhang at amd.com>; Zhou1, Tao <Tao.Zhou1 at amd.com>; Chai, Thomas <YiPeng.Chai at amd.com>; Wang, Yang(Kevin) <KevinYang.Wang at amd.com>
Subject: [PATCH 08/14] drm/amdgpu: add gfx v9.4.3 ACA support

add gfx v9.4.3 ACA driver support

Signed-off-by: Yang Wang <kevinyang.wang at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c | 85 +++++++++++++++++++++++++
 1 file changed, 85 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
index 131cddbdda0d..18b8e4dbe9a7 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@@ -38,6 +38,7 @@

 #include "gfx_v9_4_3.h"
 #include "amdgpu_xcp.h"
+#include "amdgpu_aca.h"

 MODULE_FIRMWARE("amdgpu/gc_9_4_3_mec.bin");
 MODULE_FIRMWARE("amdgpu/gc_9_4_3_rlc.bin");
@@ -675,6 +676,67 @@ static const struct amdgpu_gfx_funcs gfx_v9_4_3_gfx_funcs = {
        .ih_node_to_logical_xcc = &gfx_v9_4_3_ih_to_xcc_inst,  };

+static int gfx_v9_4_3_aca_bank_generate_report(struct aca_handle *handle,
+                                              struct aca_bank *bank, enum aca_error_type type,
+                                              struct aca_bank_report *report, void *data) {
+       u64 status, misc0;
+       u32 instlo;
+       int ret;
+
+       status = bank->regs[ACA_REG_IDX_STATUS];
+       if ((type == ACA_ERROR_TYPE_UE &&
+            ACA_REG__STATUS__ERRORCODEEXT(status) == ACA_EXTERROR_CODE_FAULT) ||
+           (type == ACA_ERROR_TYPE_CE &&
+            ACA_REG__STATUS__ERRORCODEEXT(status) ==
+ ACA_EXTERROR_CODE_CE)) {
+
+               ret = aca_bank_info_decode(bank, &report->info);
+               if (ret)
+                       return ret;
+
+               /* NOTE: overwrite info.die_id with xcd id for gfx */
+               instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+               instlo &= GENMASK(31, 1);
+               report->info.die_id = instlo == 0x36430400 ? 0 : 1;
+
+               misc0 = bank->regs[ACA_REG_IDX_MISC0];
+               report->count = ACA_REG__MISC0__ERRCNT(misc0);
+               report->type = type;
+       }
+
+       return 0;
+}
+
+static bool gfx_v9_4_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank,
+                                        enum aca_error_type type, void *data) {
+       u32 instlo;
+
+       instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]);
+       instlo &= GENMASK(31, 1);
+       switch (instlo) {
+       case 0x36430400: /* SMNAID XCD 0 */
+       case 0x38430400: /* SMNAID XCD 1 */
+       case 0x40430400: /* SMNXCD XCD 0, NOTE: FIXME: fix this error later */
+               return true;
+       default:
+               return false;
+       }
+
+       return false;
+}
+
+static const struct aca_bank_ops gfx_v9_4_3_aca_bank_ops = {
+       .aca_bank_generate_report = gfx_v9_4_3_aca_bank_generate_report,
+       .aca_bank_is_valid = gfx_v9_4_3_aca_bank_is_valid, };
+
+static const struct aca_info gfx_v9_4_3_aca_info = {
+       .hwip = ACA_HWIP_TYPE_SMU,
+       .mask = ACA_ERROR_UE_MASK | ACA_ERROR_CE_MASK,
+       .bank_ops = &gfx_v9_4_3_aca_bank_ops, };
+
 static int gfx_v9_4_3_gpu_early_init(struct amdgpu_device *adev)  {
        u32 gb_addr_config;
@@ -4242,9 +4304,32 @@ struct amdgpu_ras_block_hw_ops  gfx_v9_4_3_ras_ops = {
        .reset_ras_error_count = &gfx_v9_4_3_reset_ras_error_count,
 };

+static int gfx_v9_4_3_ras_late_init(struct amdgpu_device *adev, struct
+ras_common_if *ras_block) {
+       int r;
+
+       r = amdgpu_ras_block_late_init(adev, ras_block);
+       if (r)
+               return r;
+
+       r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__GFX,
+                               &gfx_v9_4_3_aca_info,
+                               NULL);
+       if (r)
+               goto late_fini;
+
+       return 0;
+
+late_fini:
+       amdgpu_ras_block_late_fini(adev, ras_block);
+
+       return r;
+}
+
 struct amdgpu_gfx_ras gfx_v9_4_3_ras = {
        .ras_block = {
                .hw_ops = &gfx_v9_4_3_ras_ops,
+               .ras_late_init = &gfx_v9_4_3_ras_late_init,
        },
        .enable_watchdog_timer = &gfx_v9_4_3_enable_watchdog_timer,
 };
--
2.34.1