[PATCH 2/2] drm/xe/guc: Enable w/a 14022293748 and 22019794406
John Harrison
john.c.harrison at intel.com
Fri Aug 2 22:44:57 UTC 2024
On 8/1/2024 17:40, Julia Filipchuk wrote:
> Enable workarounds for HW bug where render engine reset fails. Check
> firmware version supports the workarounds.
>
> v2: Enable KLV correctly for either workaround (Lucas)
> v4: Add check for minimum supported GuC firmware version. Enable w/a for
> hw version 20.01 too. (Daniele)
>
> Signed-off-by: Julia Filipchuk <julia.filipchuk at intel.com>
> ---
> drivers/gpu/drm/xe/abi/guc_klvs_abi.h | 1 +
> drivers/gpu/drm/xe/xe_guc_ads.c | 12 ++++++++++++
> drivers/gpu/drm/xe/xe_wa_oob.rules | 4 ++++
> 3 files changed, 17 insertions(+)
>
> diff --git a/drivers/gpu/drm/xe/abi/guc_klvs_abi.h b/drivers/gpu/drm/xe/abi/guc_klvs_abi.h
> index 8f9f60b28306..6b30743a2f6c 100644
> --- a/drivers/gpu/drm/xe/abi/guc_klvs_abi.h
> +++ b/drivers/gpu/drm/xe/abi/guc_klvs_abi.h
> @@ -351,6 +351,7 @@ enum xe_guc_klv_ids {
> GUC_WORKAROUND_KLV_ID_GAM_PFQ_SHADOW_TAIL_POLLING = 0x9005,
> GUC_WORKAROUND_KLV_ID_DISABLE_MTP_DURING_ASYNC_COMPUTE = 0x9007,
> GUC_WA_KLV_NP_RD_WRITE_TO_CLEAR_RCSM_AT_CGP_LATE_RESTORE = 0x9008,
> + GUC_WORKAROUND_KLV_ID_BACK_TO_BACK_RCS_ENGINE_RESET = 0x9009,
> };
>
> #endif
> diff --git a/drivers/gpu/drm/xe/xe_guc_ads.c b/drivers/gpu/drm/xe/xe_guc_ads.c
> index 1c60b685dbc6..ba70e6a81d88 100644
> --- a/drivers/gpu/drm/xe/xe_guc_ads.c
> +++ b/drivers/gpu/drm/xe/xe_guc_ads.c
> @@ -24,6 +24,7 @@
> #include "xe_map.h"
> #include "xe_mmio.h"
> #include "xe_platform_types.h"
> +#include "xe_uc_fw.h"
> #include "xe_wa.h"
>
> /* Slack of a few additional entries per engine */
> @@ -337,9 +338,14 @@ static void guc_waklv_enable_simple(struct xe_guc_ads *ads,
> static void guc_waklv_init(struct xe_guc_ads *ads)
> {
> struct xe_gt *gt = ads_to_gt(ads);
> + struct xe_guc *guc = ads_to_guc(ads);
> u64 addr_ggtt;
> u32 offset, remain, size;
> + u64 fw_ver;
>
> + xe_gt_assert(gt, guc->fw.type == XE_UC_FW_TYPE_GUC);
Not sure if this is necessary? If the ads -> guc -> fw route is broken
to the extent of returning a HuC firmware or something then I think we
have bigger problems than version checking a w/a flag.
> +
> + fw_ver = GUC_FIRMWARE_VER(guc);
> offset = guc_ads_waklv_offset(ads);
> remain = guc_ads_waklv_size(ads);
>
> @@ -367,6 +373,12 @@ static void guc_waklv_init(struct xe_guc_ads *ads)
> 0xC40,
> &offset, &remain);
>
> + if (fw_ver >= MAKE_VER(70, 29, 0) &&
Given the other patch set to bump the minimum required version to
70.29.2, this version check is no longer required.
John.
> + (XE_WA(gt, 14022293748) || XE_WA(gt, 22019794406)))
> + guc_waklv_enable_simple(ads,
> + GUC_WORKAROUND_KLV_ID_BACK_TO_BACK_RCS_ENGINE_RESET,
> + &offset, &remain);
> +
> size = guc_ads_waklv_size(ads) - remain;
> if (!size)
> return;
> diff --git a/drivers/gpu/drm/xe/xe_wa_oob.rules b/drivers/gpu/drm/xe/xe_wa_oob.rules
> index 540d38603f32..5cf27ff27ce6 100644
> --- a/drivers/gpu/drm/xe/xe_wa_oob.rules
> +++ b/drivers/gpu/drm/xe/xe_wa_oob.rules
> @@ -27,6 +27,10 @@
> 16022287689 GRAPHICS_VERSION(2001)
> GRAPHICS_VERSION(2004)
> 13011645652 GRAPHICS_VERSION(2004)
> +14022293748 GRAPHICS_VERSION(2001)
> + GRAPHICS_VERSION(2004)
> +22019794406 GRAPHICS_VERSION(2001)
> + GRAPHICS_VERSION(2004)
> 22019338487 MEDIA_VERSION(2000)
> GRAPHICS_VERSION(2001)
> 22019338487_display PLATFORM(LUNARLAKE)
More information about the Intel-xe
mailing list