GuC Scrub vol. 1
Subsystems (2)
v4 cover GuC Scrub vol. 1
General GuC/HuC cleanup simplifying logic, and moving chunks around, as the area got pretty rusty.
A lot of logic were extracted from intel_guc_load() to other functions - it not only did handle the actual loading but had WA implementations and had the code thatenabled submission baked it.
This is the first part of effort to clean it up.
v2: rebase after HuC merge + feedback
v3: even more renaming that aims to make things more semantic
v4: some naming improvements, some bikeshedding
Arkadiusz Hiler (8):
drm/i915/uc: Rename intel_?uc_{setup,load}() to _init_hw() drm/i915/uc: Drop superfluous externs in intel_uc.h drm/i915/huc: Add huc_to_i915
drm/i915/uc: Rename intel_?uc_init() to intel_?uc_fetch_fw() drm/i915/uc: Make intel_uc_fw_fetch() static drm/i915/guc: Extract param logic form guc_init drm/i915/guc: Simplify intel_guc_init_hw() drm/i915/uc: Simplify firmware path handling
drivers/gpu/drm/i915/i915_drv.c | 5 +- drivers/gpu/drm/i915/i915_drv.h | 5 + drivers/gpu/drm/i915/i915_gem.c | 2 +- drivers/gpu/drm/i915/intel_guc_loader.c | 354 ++++---------------------------- drivers/gpu/drm/i915/intel_huc.c | 57 +++-- drivers/gpu/drm/i915/intel_uc.c | 280 +++++++++++++++++++++++++ drivers/gpu/drm/i915/intel_uc.h | 27 ++- 7 files changed, 370 insertions(+), 360 deletions(-)
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 1/8 drm/i915/uc: Rename intel_?uc_{setup, load}() to _init_hw()
GuC historically has two "startup" functions called _init() and _setup()
Then HuC came with it's _init() and _load().
This commit renames intel_guc_setup() and intel_huc_load() to *uc_init_hw() as they called from the i915_gem_init_hw().
The aim is to be consistent in that entry points called during particular driver init phases (e.g. init_hw) are all suffixed by that phase. When reading the leaf functions, it should be clear at what stage during the driver load it is called and therefore what operations are legal at that point.
v2: commit message update (Chris Wilson)
--- drivers/gpu/drm/i915/i915_gem.c | 2 +- drivers/gpu/drm/i915/intel_guc_loader.c | 14 +++++++------- drivers/gpu/drm/i915/intel_huc.c | 6 +++--- drivers/gpu/drm/i915/intel_uc.h | 4 ++-- 4 files changed, 13 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +1 -1
@@ -4512,7 +4512,7 @@ int i915_gem_init_hw(struct drm_i915_private *dev_priv)intel_mocs_init_l3cc_table (dev_priv);/* We can't enable contexts until all firmware is loaded */ - ret =intel_guc_setup (dev_priv); + ret =intel_guc_init_hw (dev_priv);if (ret) goto out;
drivers/gpu/drm/i915/intel_guc_loader.c +7 -7
@@ -428,19 +428,19 @@ static int guc_hw_reset(struct drm_i915_private *dev_priv) }/** -* intel_guc_setup() - finish preparing the GuC for activity +* intel_guc_init_hw() - finish preparing the GuC for activity * @dev_priv: i915 device private * -* Called from gem_init_hw() during driver loading and also after a GPU reset. +* Called during driver loading and also after a GPU reset. * * The main action required here it to load the GuC uCode into the device. * The firmware image should have already been fetched into memory by the -* earlier call to intel_guc_init(), so here we need only check that worked, -* and then transfer the image to the h/w. +* earlier call to intel_guc_init(), so here we need only check that +* worked, and then transfer the image to the h/w. * * Return: non-zero code on error */ -int intel_guc_setup (struct drm_i915_private *dev_priv) +int intel_guc_init_hw (struct drm_i915_private *dev_priv) {struct intel_uc_fw *guc_fw = &dev_priv->guc.fw;const char *fw_path = guc_fw->path;
@@ -506,7 +506,7 @@ int intel_guc_setup(struct drm_i915_private *dev_priv)if (err) goto fail; -intel_huc_load (dev_priv); +intel_huc_init_hw (dev_priv); err =guc_ucode_xfer (dev_priv);if (!err)break ;
@@ -729,7 +729,7 @@ void intel_uc_fw_fetch(struct drm_i915_private *dev_priv,* Called early during driver load, but after GEM is initialised. * * The firmware will be transferred to the GuC's memory later, -* when intel_guc_setup() is called. +* when intel_guc_init_hw() is called. */ void intel_guc_init (struct drm_i915_private *dev_priv) {
drivers/gpu/drm/i915/intel_huc.c +3 -3
@@ -150,7 +150,7 @@ static int huc_ucode_xfer(struct drm_i915_private *dev_priv)* is not capable or driver yet support it. And there will be no error message * for INTEL_UC_FIRMWARE_NONE cases. * -* The DMA-copying to HW is done later when intel_huc_load() is called. +* The DMA-copying to HW is done later when intel_huc_init_hw() is called. */ void intel_huc_init (struct drm_i915_private *dev_priv) {
@@ -191,7 +191,7 @@ void intel_huc_init(struct drm_i915_private *dev_priv) }/** -* intel_huc_load() - load HuC uCode to device +* intel_huc_init_hw() - load HuC uCode to device * @dev_priv: the drm_i915_private device * * Called from guc_setup() during driver loading and also after a GPU reset.
@@ -203,7 +203,7 @@ void intel_huc_init(struct drm_i915_private *dev_priv)* * Return: non-zero code on error */ -int intel_huc_load (struct drm_i915_private *dev_priv) +int intel_huc_init_hw (struct drm_i915_private *dev_priv) {struct intel_uc_fw *huc_fw = &dev_priv->huc.fw;int err;
drivers/gpu/drm/i915/intel_uc.h +2 -2
@@ -190,7 +190,7 @@ int intel_guc_sample_forcewake(struct intel_guc *guc);/* intel_guc_loader.c */ extern void intel_guc_init (struct drm_i915_private *dev_priv); -extern int intel_guc_setup (struct drm_i915_private *dev_priv); +extern int intel_guc_init_hw (struct drm_i915_private *dev_priv);extern void intel_guc_fini (struct drm_i915_private *dev_priv);extern const char *intel_uc_fw_status_repr (enum intel_uc_fw_status status);extern int intel_guc_suspend (struct drm_i915_private *dev_priv);
@@ -225,7 +225,7 @@ static inline u32 guc_ggtt_offset(struct i915_vma *vma)/* intel_huc.c */ void intel_huc_init (struct drm_i915_private *dev_priv);void intel_huc_fini (struct drm_i915_private *dev_priv); -int intel_huc_load (struct drm_i915_private *dev_priv); +int intel_huc_init_hw (struct drm_i915_private *dev_priv);void intel_guc_auth_huc (struct drm_i915_private *dev_priv);#endif
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 2/8 drm/i915/uc: Drop superfluous externs in intel_uc.h
In reply to Arkadiusz HilerExterns are implicit and we generally try to avoid them.
--- drivers/gpu/drm/i915/intel_uc.h | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/gpu/drm/i915/intel_uc.h +6 -6
@@ -189,12 +189,12 @@ int intel_guc_send(struct intel_guc *guc, const u32 *action, u32 len);int intel_guc_sample_forcewake (struct intel_guc *guc);/* intel_guc_loader.c */ -extern void intel_guc_init (struct drm_i915_private *dev_priv); -extern int intel_guc_init_hw (struct drm_i915_private *dev_priv); -extern void intel_guc_fini (struct drm_i915_private *dev_priv); -extern const char *intel_uc_fw_status_repr (enum intel_uc_fw_status status); -extern int intel_guc_suspend (struct drm_i915_private *dev_priv); -extern int intel_guc_resume (struct drm_i915_private *dev_priv); +void intel_guc_init (struct drm_i915_private *dev_priv); +int intel_guc_init_hw (struct drm_i915_private *dev_priv); +void intel_guc_fini (struct drm_i915_private *dev_priv); +const char *intel_uc_fw_status_repr (enum intel_uc_fw_status status); +int intel_guc_suspend (struct drm_i915_private *dev_priv); +int intel_guc_resume (struct drm_i915_private *dev_priv);void intel_uc_fw_fetch (struct drm_i915_private *dev_priv,struct intel_uc_fw *uc_fw);u32 intel_guc_wopcm_size (struct drm_i915_private *dev_priv);
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 2/8] drm/i915/uc: Drop superfluous externs in intel_uc.h
Quoted 4 lines · Arkadiusz Hiler
Externs are implicit and we generally try to avoid them.
Cc Michal Wajdeczko <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>
Yeah, for actual "extern" there's EXPORT_SYMBOL (which is not needed here).
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 3/8 drm/i915/huc: Add huc_to_i915
In reply to Arkadiusz HilerUsed to obtain "dev_priv" from huc struct pointer. We already have similar thing for guc.
--- drivers/gpu/drm/i915/i915_drv.h | 5 +++++ 1 file changed, 5 insertions(+)
drivers/gpu/drm/i915/i915_drv.h +5 -0
@@ -2528,6 +2528,11 @@ static inline struct drm_i915_private *guc_to_i915(struct intel_guc *guc) return container_of(guc, struct drm_i915_private, guc); } +static inline struct drm_i915_private *huc_to_i915 (struct intel_huc *huc) +{ + return container_of(huc, struct drm_i915_private, huc); +} +/* Simple iterator over all initialised engines */ #define for_each_engine (engine__, dev_priv__, id__) \ for ((id__) = 0; \
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 4/8 drm/i915/uc: Rename intel_?uc_init() to intel_?uc_fetch_fw()
In reply to Arkadiusz HilerTrying to have subject_verb_object ordering and more descriptive names, the intel_huc_init() and intel_guc_init() functions are renamed:
* intel_guc is the subject, so those functions now take intel_guc structure, instead of the dev_priv
* fetch is the verb
* fw is the object which better describes the function's role
Same change is done for the huc counterpart.
Also we bulk call both functions from higher-level intel_uc_fetch_fw:
* intel being the subject (taking the dev_priv as param)
* fetch being the verb
* uc_fw being the subject
v2: settle on intel_uc_fetch_fw name (M. Wajdeczko)
--- drivers/gpu/drm/i915/i915_drv.c | 3 +-- drivers/gpu/drm/i915/intel_guc_loader.c | 30 +++++++++++++------------- drivers/gpu/drm/i915/intel_huc.c | 37 ++++++++++++++++----------------- drivers/gpu/drm/i915/intel_uc.c | 6 ++++++ drivers/gpu/drm/i915/intel_uc.h | 5 +++-- 5 files changed, 43 insertions(+), 38 deletions(-)
drivers/gpu/drm/i915/i915_drv.c +1 -2
@@ -609,8 +609,7 @@ static int i915_load_modeset_init(struct drm_device *dev)if (ret) goto cleanup_irq; -intel_huc_init (dev_priv); -intel_guc_init (dev_priv); +intel_uc_fetch_fw (dev_priv); ret =i915_gem_init (dev_priv);if (ret)
drivers/gpu/drm/i915/intel_guc_loader.c +15 -15
@@ -723,17 +723,17 @@ void intel_uc_fw_fetch(struct drm_i915_private *dev_priv, }/** -* intel_guc_init() - define parameters and fetch firmware -* @dev_priv: i915 device private +* intel_guc_fetch_fw() - define parameters and fetch firmware +* @guc: intel_guc struct * * Called early during driver load, but after GEM is initialised. * * The firmware will be transferred to the GuC's memory later, * when intel_guc_init_hw() is called. */ -void intel_guc_init (struct drm_i915_private *dev_priv) +void intel_guc_fetch_fw (struct intel_guc *guc) { -struct intel_uc_fw *guc_fw = &dev_priv->guc.fw; +struct drm_i915_private *dev_priv =guc_to_i915 (guc);const char *fw_path;if (!HAS_GUC (dev_priv)) {
@@ -751,23 +751,23 @@ void intel_guc_init(struct drm_i915_private *dev_priv) fw_path =NULL ; }else if (IS_SKYLAKE (dev_priv)) { fw_path =I915_SKL_GUC_UCODE ; - guc_fw->major_ver_wanted =SKL_FW_MAJOR ; - guc_fw->minor_ver_wanted =SKL_FW_MINOR ; + guc->fw.major_ver_wanted =SKL_FW_MAJOR ; + guc->fw.minor_ver_wanted =SKL_FW_MINOR ; }else if (IS_BROXTON (dev_priv)) { fw_path =I915_BXT_GUC_UCODE ; - guc_fw->major_ver_wanted =BXT_FW_MAJOR ; - guc_fw->minor_ver_wanted =BXT_FW_MINOR ; + guc->fw.major_ver_wanted =BXT_FW_MAJOR ; + guc->fw.minor_ver_wanted =BXT_FW_MINOR ; }else if (IS_KABYLAKE (dev_priv)) { fw_path =I915_KBL_GUC_UCODE ; - guc_fw->major_ver_wanted =KBL_FW_MAJOR ; - guc_fw->minor_ver_wanted =KBL_FW_MINOR ; + guc->fw.major_ver_wanted =KBL_FW_MAJOR ; + guc->fw.minor_ver_wanted =KBL_FW_MINOR ; }else { fw_path ="" ;/* unknown device */ } - guc_fw->path = fw_path; - guc_fw->fetch_status =INTEL_UC_FIRMWARE_NONE ; - guc_fw->load_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.path = fw_path; + guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ;/* Early (and silent) return if GuC loading is disabled */ if (!i915.enable_guc_loading)
@@ -777,9 +777,9 @@ void intel_guc_init(struct drm_i915_private *dev_priv)if (*fw_path =='\0' )return ; - guc_fw->fetch_status =INTEL_UC_FIRMWARE_PENDING ; + guc->fw.fetch_status =INTEL_UC_FIRMWARE_PENDING ;DRM_DEBUG_DRIVER ("GuC firmware pending, path %s\n" , fw_path); -intel_uc_fw_fetch (dev_priv, guc_fw); +intel_uc_fw_fetch (dev_priv, &guc->fw);/* status must now be FAIL or SUCCESS */ }
drivers/gpu/drm/i915/intel_huc.c +18 -19
@@ -141,8 +141,8 @@ static int huc_ucode_xfer(struct drm_i915_private *dev_priv) }/** -* intel_huc_init() - initiate HuC firmware loading request -* @dev_priv: the drm_i915_private device +* intel_huc_fetch_fw() - initiate HuC firmware loading request +* @huc: intel_huc struct * * Called early during driver load, but after GEM is initialised. The loading * will continue only when driver explicitly specify firmware name and version.
@@ -152,42 +152,41 @@ static int huc_ucode_xfer(struct drm_i915_private *dev_priv)* * The DMA-copying to HW is done later when intel_huc_init_hw() is called. */ -void intel_huc_init (struct drm_i915_private *dev_priv) +void intel_huc_fetch_fw (struct intel_huc *huc) { -struct intel_huc *huc = &dev_priv->huc; -struct intel_uc_fw *huc_fw = &huc->fw; +struct drm_i915_private *dev_priv =huc_to_i915 (huc);const char *fw_path =NULL ; - huc_fw->path =NULL ; - huc_fw->fetch_status =INTEL_UC_FIRMWARE_NONE ; - huc_fw->load_status =INTEL_UC_FIRMWARE_NONE ; - huc_fw->fw =INTEL_UC_FW_TYPE_HUC ; + huc->fw.path =NULL ; + huc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + huc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; + huc->fw.fw =INTEL_UC_FW_TYPE_HUC ;if (!HAS_HUC_UCODE (dev_priv))return ;if (IS_SKYLAKE (dev_priv)) { fw_path =I915_SKL_HUC_UCODE ; - huc_fw->major_ver_wanted =SKL_HUC_FW_MAJOR ; - huc_fw->minor_ver_wanted =SKL_HUC_FW_MINOR ; + huc->fw.major_ver_wanted =SKL_HUC_FW_MAJOR ; + huc->fw.minor_ver_wanted =SKL_HUC_FW_MINOR ; }else if (IS_BROXTON (dev_priv)) { fw_path =I915_BXT_HUC_UCODE ; - huc_fw->major_ver_wanted =BXT_HUC_FW_MAJOR ; - huc_fw->minor_ver_wanted =BXT_HUC_FW_MINOR ; + huc->fw.major_ver_wanted =BXT_HUC_FW_MAJOR ; + huc->fw.minor_ver_wanted =BXT_HUC_FW_MINOR ; }else if (IS_KABYLAKE (dev_priv)) { fw_path =I915_KBL_HUC_UCODE ; - huc_fw->major_ver_wanted =KBL_HUC_FW_MAJOR ; - huc_fw->minor_ver_wanted =KBL_HUC_FW_MINOR ; + huc->fw.major_ver_wanted =KBL_HUC_FW_MAJOR ; + huc->fw.minor_ver_wanted =KBL_HUC_FW_MINOR ; } - huc_fw->path = fw_path; - huc_fw->fetch_status =INTEL_UC_FIRMWARE_PENDING ; + huc->fw.path = fw_path; + huc->fw.fetch_status =INTEL_UC_FIRMWARE_PENDING ;DRM_DEBUG_DRIVER ("HuC firmware pending, path %s\n" , fw_path); -WARN (huc_fw->path ==NULL ,"HuC present but no fw path\n" ); +WARN (huc->fw.path ==NULL ,"HuC present but no fw path\n" ); -intel_uc_fw_fetch (dev_priv, huc_fw); +intel_uc_fw_fetch (dev_priv, &huc->fw); }/**
drivers/gpu/drm/i915/intel_uc.c +6 -0
@@ -30,6 +30,12 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv)mutex_init (&dev_priv->guc.send_mutex); } +void intel_uc_fetch_fw (struct drm_i915_private *dev_priv) +{ +intel_huc_fetch_fw (&dev_priv->huc); +intel_guc_fetch_fw (&dev_priv->guc); +} +/* * Read GuC command/status register (SOFT_SCRATCH_0) * Return true if it contains a response rather than a command
drivers/gpu/drm/i915/intel_uc.h +3 -2
@@ -185,11 +185,12 @@ struct intel_huc {/* intel_uc.c */ void intel_uc_init_early (struct drm_i915_private *dev_priv); +void intel_uc_fetch_fw (struct drm_i915_private *dev_priv);int intel_guc_send (struct intel_guc *guc,const u32 *action,u32 len);int intel_guc_sample_forcewake (struct intel_guc *guc);/* intel_guc_loader.c */ -void intel_guc_init (struct drm_i915_private *dev_priv); +void intel_guc_fetch_fw (struct intel_guc *guc);int intel_guc_init_hw (struct drm_i915_private *dev_priv);void intel_guc_fini (struct drm_i915_private *dev_priv);const char *intel_uc_fw_status_repr (enum intel_uc_fw_status status);
@@ -223,7 +224,7 @@ static inline u32 guc_ggtt_offset(struct i915_vma *vma) }/* intel_huc.c */ -void intel_huc_init (struct drm_i915_private *dev_priv); +void intel_huc_fetch_fw (struct intel_huc *huc);void intel_huc_fini (struct drm_i915_private *dev_priv);int intel_huc_init_hw (struct drm_i915_private *dev_priv);void intel_guc_auth_huc (struct drm_i915_private *dev_priv);
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Joonas Lahtinen
maintainerRe: [PATCH 4/8] drm/i915/uc: Rename intel_?uc_init() to intel_?uc_fetch_fw()
Quoted 22 lines · Arkadiusz Hiler
Trying to have
subject_verb_objectordering and more descriptive names, theintel_huc_init()andintel_guc_init()functions are renamed:*
intel_gucis the subject, so those functions now takeintel_gucstructure, instead of thedev_priv
* fetch is the verb
* fw is the object which better describes the function's roleSame change is done for the huc counterpart.
Also we bulk call both functions from higher-level
intel_uc_fetch_fw:
* intel being the subject (taking thedev_privas param)
* fetch being the verb
*uc_fwbeing the subjectv2: settle on
intel_uc_fetch_fwname (M. Wajdeczko)Cc Chris Wilson <[email protected]>Cc Joonas Lahtinen <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Michal Winiarski <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>
<SNIP>
drivers/gpu/drm/i915/i915_drv.c static int i915_load_modeset_init(struct drm_device *dev) · line 612
Quoted 7 lines
drivers/gpu/drm/i915/i915_drv.c +0
-0@@ -609,8 +609,7 @@ static int i915_load_modeset_init(struct drm_device *dev) if (ret) goto cleanup_irq;drivers/gpu/drm/i915/i915_drv.c +1
-2-intel_huc_init (dev_priv); -intel_guc_init (dev_priv); +intel_uc_fetch_fw (dev_priv);
intel_uc_init fits this context. (See below)
Quoted 11 lines
/**
- *intel_guc_init()- define parameters and fetch firmware
- * @dev_priv: i915 device private
+ *intel_guc_fetch_fw()- define parameters and fetch firmware
+ * @guc:intel_gucstruct
*
* Called early during driver load, but after GEM is initialised. *
* The firmware will be transferred to the GuC's memory later,
* whenintel_guc_init_hw()is called.
*/
"define parameters" is little vague, maybe "select and fetch firmware"?
Quoted 8 lines
drivers/gpu/drm/i915/intel_guc_loader.c +1
-1-void intel_guc_init (struct drm_i915_private *dev_priv) +void intel_guc_fetch_fw (struct intel_guc *guc){
drivers/gpu/drm/i915/intel_guc_loader.c +1
-1-struct intel_uc_fw *guc_fw = &dev_priv->guc.fw; +struct drm_i915_private *dev_priv =guc_to_i915 (guc);const char
*fw_path;
if (!HAS_GUC(dev_priv)) {
This parameter dance needs to be moved away from guc_fetch_fw function, into intel_sanitize_options (I'm pretty sure I've mentioned this earlier).
Quoted 24 lines
drivers/gpu/drm/i915/intel_guc_loader.c +6
-6@@ -751,23 +751,23 @@ void intel_guc_init(struct drm_i915_private *dev_priv) fw_path = NULL; } else if (IS_SKYLAKE(dev_priv)) { fw_path = I915_SKL_GUC_UCODE; - guc_fw->major_ver_wanted =SKL_FW_MAJOR ; - guc_fw->minor_ver_wanted =SKL_FW_MINOR ; + guc->fw.major_ver_wanted =SKL_FW_MAJOR ; + guc->fw.minor_ver_wanted =SKL_FW_MINOR ; } else if (IS_BROXTON(dev_priv)) { fw_path = I915_BXT_GUC_UCODE; - guc_fw->major_ver_wanted =BXT_FW_MAJOR ; - guc_fw->minor_ver_wanted =BXT_FW_MINOR ; + guc->fw.major_ver_wanted =BXT_FW_MAJOR ; + guc->fw.minor_ver_wanted =BXT_FW_MINOR ; } else if (IS_KABYLAKE(dev_priv)) { fw_path = I915_KBL_GUC_UCODE; - guc_fw->major_ver_wanted =KBL_FW_MAJOR ; - guc_fw->minor_ver_wanted =KBL_FW_MINOR ; + guc->fw.major_ver_wanted =KBL_FW_MAJOR ; + guc->fw.minor_ver_wanted =KBL_FW_MINOR ; } else {
fw_path= ""; /* unknown device */
}
drivers/gpu/drm/i915/intel_guc_loader.c void intel_guc_init(struct drm_i915_private *dev_priv) · line 768
Quoted 4 lines
drivers/gpu/drm/i915/intel_guc_loader.c +1
-3- guc_fw->path = fw_path; - guc_fw->fetch_status =INTEL_UC_FIRMWARE_NONE ; - guc_fw->load_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.path = fw_path;
Just get rid of fw_path variable and assign directly, also hoist the warning to the else branch, which can then do "return;"
drivers/gpu/drm/i915/intel_guc_loader.c void intel_guc_init(struct drm_i915_private *dev_priv) · line 770
Quoted 2 lines
drivers/gpu/drm/i915/intel_guc_loader.c +2
-0+ guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ;
Hoist this assignment before the if block, so no need to special for the early return from else branch.
<SNIP>
Quoted 4 lines
/**
- *intel_huc_init()- initiate HuC firmware loading request
- * @dev_priv: thedrm_i915_privatedevice
+ *intel_huc_fetch_fw()- initiate HuC firmware loading request
Correct this commit too to be more descriptive.
Quoted 5 lines
+ * @huc:
intel_hucstruct
*
* Called early during driver load, but after GEM is initialised. The loading
* will continue only when driver explicitly specify firmware name and version.
drivers/gpu/drm/i915/intel_huc.c static int huc_ucode_xfer(struct drm_i915_private *dev_priv) · line 158
Quoted 10 lines
drivers/gpu/drm/i915/intel_huc.c +2
-3@@ -152,42 +152,41 @@ static int huc_ucode_xfer(struct drm_i915_private *dev_priv)* * The DMA-copying to HW is done later when intel_huc_init_hw() is called. */ -void intel_huc_init (struct drm_i915_private *dev_priv) +void intel_huc_fetch_fw (struct intel_huc *huc) { -struct intel_huc *huc = &dev_priv->huc; -struct intel_uc_fw *huc_fw = &huc->fw; +struct drm_i915_private *dev_priv =huc_to_i915 (huc); const char *fw_path = NULL;
Similarly arrange to get rid of fw_path here.
<SNIP>
Quoted 5 lines
drivers/gpu/drm/i915/intel_uc.c +0
-0@@ -30,6 +30,12 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv) mutex_init(&dev_priv->guc.send_mutex); }
+voidintel_uc_fetch_fw(struct drm_i915_private *dev_priv)
This function might be worth calling intel_uc_init (See above), if the need comes to add other stuff. But either way.
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/8] drm/i915/uc: Rename intel_?uc_init() to intel_?uc_fetch_fw()
Quoted 26 lines · Joonas Lahtinen
Quoted 22 lines · Arkadiusz Hiler
Trying to have
subject_verb_objectordering and more descriptive names, theintel_huc_init()andintel_guc_init()functions are renamed:*
intel_gucis the subject, so those functions now takeintel_gucstructure, instead of thedev_priv
* fetch is the verb
* fw is the object which better describes the function's roleSame change is done for the huc counterpart.
Also we bulk call both functions from higher-level
intel_uc_fetch_fw:
* intel being the subject (taking thedev_privas param)
* fetch being the verb
*uc_fwbeing the subjectv2: settle on
intel_uc_fetch_fwname (M. Wajdeczko)Cc Chris Wilson <[email protected]>Cc Joonas Lahtinen <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Michal Winiarski <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]><SNIP>
Quoted 9 lines
drivers/gpu/drm/i915/i915_drv.c static int i915_load_modeset_init(struct drm_device *dev) · line 612
Quoted 7 lines
drivers/gpu/drm/i915/i915_drv.c +0
-0@@ -609,8 +609,7 @@ static int i915_load_modeset_init(struct drm_device *dev) if (ret) goto cleanup_irq;drivers/gpu/drm/i915/i915_drv.c +1
-2-intel_huc_init (dev_priv); -intel_guc_init (dev_priv); +intel_uc_fetch_fw (dev_priv);
intel_uc_initfits this context. (See below)
Answer below.
Quoted 13 lines
Quoted 12 lines
/**
- *intel_guc_init()- define parameters and fetch firmware
- * @dev_priv: i915 device private
+ *intel_guc_fetch_fw()- define parameters and fetch firmware
+ * @guc:intel_gucstruct
*
* Called early during driver load, but after GEM is initialised. *
* The firmware will be transferred to the GuC's memory later,
* whenintel_guc_init_hw()is called.
*/
Quoted 1 line
"define parameters" is little vague, maybe "select and fetch firmware"?
I like those verbs. Let start using it through the whole thing!
Quoted 13 lines
Quoted 9 lines
drivers/gpu/drm/i915/intel_guc_loader.c +1
-1-void intel_guc_init (struct drm_i915_private *dev_priv) +void intel_guc_fetch_fw (struct intel_guc *guc){
drivers/gpu/drm/i915/intel_guc_loader.c +1
-1-struct intel_uc_fw *guc_fw = &dev_priv->guc.fw; +struct drm_i915_private *dev_priv =guc_to_i915 (guc);const char
*fw_path;
if (!HAS_GUC(dev_priv)) {This parameter dance needs to be moved away from
guc_fetch_fwfunction, into intel_sanitize_options (I'm pretty sure I've mentioned this earlier).
This is removed in patch 8, as the fetch_fw is called conditionally.
Quoted 23 lines
Quoted 23 lines
drivers/gpu/drm/i915/intel_guc_loader.c +6
-6@@ -751,23 +751,23 @@ void intel_guc_init(struct drm_i915_private *dev_priv) fw_path = NULL; } else if (IS_SKYLAKE(dev_priv)) { fw_path = I915_SKL_GUC_UCODE; - guc_fw->major_ver_wanted =SKL_FW_MAJOR ; - guc_fw->minor_ver_wanted =SKL_FW_MINOR ; + guc->fw.major_ver_wanted =SKL_FW_MAJOR ; + guc->fw.minor_ver_wanted =SKL_FW_MINOR ; } else if (IS_BROXTON(dev_priv)) { fw_path = I915_BXT_GUC_UCODE; - guc_fw->major_ver_wanted =BXT_FW_MAJOR ; - guc_fw->minor_ver_wanted =BXT_FW_MINOR ; + guc->fw.major_ver_wanted =BXT_FW_MAJOR ; + guc->fw.minor_ver_wanted =BXT_FW_MINOR ; } else if (IS_KABYLAKE(dev_priv)) { fw_path = I915_KBL_GUC_UCODE; - guc_fw->major_ver_wanted =KBL_FW_MAJOR ; - guc_fw->minor_ver_wanted =KBL_FW_MINOR ; + guc->fw.major_ver_wanted =KBL_FW_MAJOR ; + guc->fw.minor_ver_wanted =KBL_FW_MINOR ; } else {
fw_path= ""; /* unknown device */
}
Quoted 8 lines
drivers/gpu/drm/i915/intel_guc_loader.c void intel_guc_init(struct drm_i915_private *dev_priv) · line 768
Quoted 5 lines
drivers/gpu/drm/i915/intel_guc_loader.c +1
-3- guc_fw->path = fw_path; - guc_fw->fetch_status =INTEL_UC_FIRMWARE_NONE ; - guc_fw->load_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.path = fw_path;Just get rid of
fw_pathvariable and assign directly, also hoist the warning to the else branch, which can then do "return;"
This is done done in patch 8.
Quoted 5 lines
drivers/gpu/drm/i915/intel_guc_loader.c void intel_guc_init(struct drm_i915_private *dev_priv) · line 770
Quoted 2 lines
drivers/gpu/drm/i915/intel_guc_loader.c +2
-0+ guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ;Hoist this assignment before the if block, so no need to special for the early return from else branch.
This is done done in patch 8.
Quoted 8 lines
<SNIP>
Quoted 4 lines
/**
- *intel_huc_init()- initiate HuC firmware loading request
- * @dev_priv: thedrm_i915_privatedevice
+ *intel_huc_fetch_fw()- initiate HuC firmware loading requestCorrect this commit too to be more descriptive.
Okay.
Quoted 9 lines
Quoted 9 lines
+ * @huc:
intel_hucstruct
*
* Called early during driver load, but after GEM is initialised. The loading
* will continue only when driver explicitly specify firmware name and version.drivers/gpu/drm/i915/intel_huc.c +0
-1@@ -152,42 +152,41 @@ static int huc_ucode_xfer(struct drm_i915_private *dev_priv)* * The DMA-copying to HW is done later when intel_huc_init_hw() is called. */ -void intel_huc_init (struct drm_i915_private *dev_priv)
Quoted 8 lines
drivers/gpu/drm/i915/intel_huc.c static int huc_ucode_xfer(struct drm_i915_private *dev_priv) · line 158
Quoted 6 lines
drivers/gpu/drm/i915/intel_huc.c +2
-2+void intel_huc_fetch_fw (struct intel_huc *huc) { -struct intel_huc *huc = &dev_priv->huc; -struct intel_uc_fw *huc_fw = &huc->fw; +struct drm_i915_private *dev_priv =huc_to_i915 (huc); const char *fw_path = NULL;Similarly arrange to get rid of
fw_pathhere.
Patch 8 in the series addresses that issue as well. Maybe I should move them around?
Quoted 10 lines
<SNIP>
Quoted 5 lines
drivers/gpu/drm/i915/intel_uc.c +0
-0@@ -30,6 +30,12 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv) mutex_init(&dev_priv->guc.send_mutex); }
+voidintel_uc_fetch_fw(struct drm_i915_private *dev_priv)This function might be worth calling
intel_uc_init(See above), if the need comes to add other stuff. But either way.
This is quite confusing now. I was fine it being named init, someone suggested to be more descriptive with the name, as it is not general enough to be "init". Seemed reasonable enough for me, so I incorporated that in the respin.
This is turning into some heavy bikeshedding now...
I agree that it's more than fetch, it actually selects + fetches + populates the structures.
I'll gladly ignore previous feedback on being to vague with name and just go with init, but let give the _fw postfix one last chance:
intel_guc_init_fw {
intel_guc_select_fw
if (NULL != guc.fw.path)
intel_uc_prepare_fw
}Where select does what the guc's fetch fw does sans the uc_fetch call.
Also intel_{g,h}uc_select_fw can be made part of the sanitize options, but I think it better belongs here.
That's is basing on your suggestions for the other patch.
Signature · 6 lines
Cheers, Arek _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Joonas Lahtinen
maintainerRe: [PATCH 4/8] drm/i915/uc: Rename intel_?uc_init() to intel_?uc_fetch_fw()
Quoted 20 lines · Arkadiusz Hiler
Quoted 12 lines · Joonas Lahtinen
Quoted 12 lines
+ * @huc:
intel_hucstruct
*
* Called early during driver load, but after GEM is initialised. The loading
* will continue only when driver explicitly specify firmware name and version.drivers/gpu/drm/i915/intel_huc.c +1
-1@@ -152,42 +152,41 @@ static int huc_ucode_xfer(struct drm_i915_private *dev_priv)* * The DMA-copying to HW is done later when intel_huc_init_hw() is called. */ -void intel_huc_init (struct drm_i915_private *dev_priv) +void intel_huc_fetch_fw (struct intel_huc *huc) {Quoted 6 lines
Review static int huc_ucode_xfer(struct drm_i915_private *dev_priv) · line 160
Quoted 4 lines
(quoted diff) +0
-0> > > - struct intel_huc *huc = &dev_priv->huc; > > > - struct intel_uc_fw *huc_fw = &huc->fw; > > > + struct drm_i915_private *dev_priv = huc_to_i915(huc); > > > const char *fw_path = NULL;Similarly arrange to get rid of
fw_pathhere.
Quoted 2 lines
Patch 8 in the series addresses that issue as well. Maybe I should move them around?
Nah, it's fine, the intermediary steps need to be working (for bisecting), but not necessarily 100% pretty. If it's addressed later, it's good.
Quoted 9 lines
Quoted 8 lines
Quoted 5 lines
drivers/gpu/drm/i915/intel_uc.c +0
-0@@ -30,6 +30,12 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv) mutex_init(&dev_priv->guc.send_mutex); }
+voidintel_uc_fetch_fw(struct drm_i915_private *dev_priv)This function might be worth calling
intel_uc_init(See above), if the need comes to add other stuff. But either way.
Quoted 6 lines
This is quite confusing now. I was fine it being named init, someone suggested to be more descriptive with the name, as it is not general enough to be "init". Seemed reasonable enough for me, so I incorporated that in the respin.
This is turning into some heavy bikeshedding now...
That's why actual code in the mailing list is the only right way, discussion in IRC can be misleading :)
Quoted 11 lines
I agree that it's more than fetch, it actually selects + fetches + populates the structures.
I'll gladly ignore previous feedback on being to vague with name and just go with init, but let give the
_fwpostfix one last chance:
intel_guc_init_fw{
intel_guc_select_fw
if (NULL != guc.fw.path)
if (guc.fw.patch) to stick to coding style.
Quoted 4 lines
intel_uc_prepare_fw
}Where select does what the guc's fetch fw does sans the
uc_fetchcall.
Sounds good to me.
Quoted 4 lines
Also intel_{g,h}uc_select_fw can be made part of the sanitize options, but I think it better belongs here.
That's is basing on your suggestions for the other patch.
Thats, correct, select_fw should be here.
if (!HAS_GUC(dev_priv)) {
i915.enable_guc_loading = 0;
i915.enable_guc_submission = 0;
} else {
/* A negative value means "use platform default" */
if (i915.enable_guc_loading < 0)
i915.enable_guc_loading = HAS_GUC_UCODE(dev_priv);
if (i915.enable_guc_submission < 0)
i915.enable_guc_submission = HAS_GUC_SCHED(dev_priv);
}This part is a perfect fit to the sanitize_options function, because that's what it does, makes sure we don't try to enable something we don't have.
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 5/8 drm/i915/uc: Make intel_uc_fw_fetch() static
In reply to Arkadiusz Hilerintel_uc_fw_fetch() is confusingly named in the light of recent changes.
It's also in the wrong place - 'guc_loader.h' - it's used for both guc and huc, which was reflected in name, but not it's location, so let's move it to 'intel_uc.c'.
We can make a intel_uc_fw callback out of it, to avoid leaking it outside intel_uc.c
--- drivers/gpu/drm/i915/intel_guc_loader.c | 137 +------------------------------ drivers/gpu/drm/i915/intel_huc.c | 2 +- drivers/gpu/drm/i915/intel_uc.c | 141 ++++++++++++++++++++++++++++++++ drivers/gpu/drm/i915/intel_uc.h | 4 +- 4 files changed, 145 insertions(+), 139 deletions(-)
drivers/gpu/drm/i915/intel_guc_loader.c +1 -136
@@ -26,7 +26,6 @@* Dave Gordon <[email protected]> * Alex Dai <[email protected]> */ -#include <linux/firmware.h> #include "i915_drv.h" #include "intel_uc.h"
@@ -587,140 +586,6 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)return ret; } -void intel_uc_fw_fetch (struct drm_i915_private *dev_priv, -struct intel_uc_fw *uc_fw) -{ -struct pci_dev *pdev = dev_priv->drm.pdev; -struct drm_i915_gem_object *obj; -const struct firmware *fw =NULL ; -struct uc_css_header *css; -size_t size; -int err; - -DRM_DEBUG_DRIVER ("before requesting firmware: uC fw fetch status %s\n" , -intel_uc_fw_status_repr (uc_fw->fetch_status)); - - err =request_firmware (&fw, uc_fw->path, &pdev->dev); -if (err) - goto fail; -if (!fw) - goto fail; - -DRM_DEBUG_DRIVER ("fetch uC fw from %s succeeded, fw %p\n" , - uc_fw->path, fw); - -/* Check the size of the blob before examining buffer contents */ -if (fw->size <sizeof (struct uc_css_header )) { -DRM_NOTE ("Firmware header is missing\n" ); - goto fail; - } - - css = (struct uc_css_header *)fw->data; - -/* Firmware bits always start from header */ - uc_fw->header_offset =0 ; - uc_fw->header_size = (css->header_size_dw - css->modulus_size_dw - - css->key_size_dw - css->exponent_size_dw) *sizeof (u32); - -if (uc_fw->header_size !=sizeof (struct uc_css_header )) { -DRM_NOTE ("CSS header definition mismatch\n" ); - goto fail; - } - -/* then, uCode */ - uc_fw->ucode_offset = uc_fw->header_offset + uc_fw->header_size; - uc_fw->ucode_size = (css->size_dw - css->header_size_dw) *sizeof (u32); - -/* now RSA */ -if (css->key_size_dw !=UOS_RSA_SCRATCH_MAX_COUNT ) { -DRM_NOTE ("RSA key size is bad\n" ); - goto fail; - } - uc_fw->rsa_offset = uc_fw->ucode_offset + uc_fw->ucode_size; - uc_fw->rsa_size = css->key_size_dw *sizeof (u32); - -/* At least, it should have header, uCode and RSA. Size of all three. */ - size = uc_fw->header_size + uc_fw->ucode_size + uc_fw->rsa_size; -if (fw->size < size) { -DRM_NOTE ("Missing firmware components\n" ); - goto fail; - } - -/* -* The GuC firmware image has the version number embedded at a well-known -* offset within the firmware blob; note that major / minor version are -* TWO bytes each (i.e. u16), although all pointers and offsets are defined -* in terms of bytes (u8). -*/ -switch (uc_fw->fw) { -case INTEL_UC_FW_TYPE_GUC : -/* Header and uCode will be loaded to WOPCM. Size of the two. */ - size = uc_fw->header_size + uc_fw->ucode_size; - -/* Top 32k of WOPCM is reserved (8K stack + 24k RC6 context). */ -if (size >intel_guc_wopcm_size (dev_priv)) { -DRM_ERROR ("Firmware is too large to fit in WOPCM\n" ); - goto fail; - } - uc_fw->major_ver_found = css->guc.sw_version >>16 ; - uc_fw->minor_ver_found = css->guc.sw_version &0xFFFF ; -break ; - -case INTEL_UC_FW_TYPE_HUC : - uc_fw->major_ver_found = css->huc.sw_version >>16 ; - uc_fw->minor_ver_found = css->huc.sw_version &0xFFFF ; -break ; - -default : -DRM_ERROR ("Unknown firmware type %d\n" , uc_fw->fw); - err = -ENOEXEC ; - goto fail; - } - -if (uc_fw->major_ver_found != uc_fw->major_ver_wanted || - uc_fw->minor_ver_found < uc_fw->minor_ver_wanted) { -DRM_NOTE ("uC firmware version %d.%d, required %d.%d\n" , - uc_fw->major_ver_found, uc_fw->minor_ver_found, - uc_fw->major_ver_wanted, uc_fw->minor_ver_wanted); - err = -ENOEXEC ; - goto fail; - } - -DRM_DEBUG_DRIVER ("firmware version %d.%d OK (minimum %d.%d)\n" , - uc_fw->major_ver_found, uc_fw->minor_ver_found, - uc_fw->major_ver_wanted, uc_fw->minor_ver_wanted); - -mutex_lock (&dev_priv->drm.struct_mutex); - obj =i915_gem_object_create_from_data (dev_priv, fw->data, fw->size); -mutex_unlock (&dev_priv->drm.struct_mutex); -if (IS_ERR_OR_NULL (obj)) { - err = obj ?PTR_ERR (obj) : -ENOMEM ; - goto fail; - } - - uc_fw->obj = obj; - uc_fw->size = fw->size; - -DRM_DEBUG_DRIVER ("uC fw fetch status SUCCESS, obj %p\n" , - uc_fw->obj); - -release_firmware (fw); - uc_fw->fetch_status =INTEL_UC_FIRMWARE_SUCCESS ; -return ; - -fail: -DRM_WARN ("Failed to fetch valid uC firmware from %s (error %d)\n" , - uc_fw->path, err); -DRM_DEBUG_DRIVER ("uC fw fetch status FAIL; err %d, fw %p, obj %p\n" , - err, fw, uc_fw->obj); - - obj =fetch_and_zero (&uc_fw->obj); -if (obj) -i915_gem_object_put (obj); - -release_firmware (fw);/* OK even if fw is NULL */ - uc_fw->fetch_status =INTEL_UC_FIRMWARE_FAIL ; -}/** * intel_guc_fetch_fw() - define parameters and fetch firmware
@@ -779,7 +644,7 @@ void intel_guc_fetch_fw(struct intel_guc *guc) guc->fw.fetch_status =INTEL_UC_FIRMWARE_PENDING ;DRM_DEBUG_DRIVER ("GuC firmware pending, path %s\n" , fw_path); -intel_uc_fw_fetch (dev_priv, &guc->fw); + guc->fw.fetch (dev_priv, &guc->fw);/* status must now be FAIL or SUCCESS */ }
drivers/gpu/drm/i915/intel_huc.c +1 -1
@@ -186,7 +186,7 @@ void intel_huc_fetch_fw(struct intel_huc *huc)WARN (huc->fw.path ==NULL ,"HuC present but no fw path\n" ); -intel_uc_fw_fetch (dev_priv, &huc->fw); + huc->fw.fetch (dev_priv, &huc->fw); }/**
drivers/gpu/drm/i915/intel_uc.c +141 -0
157 lines not on this page · raw message
drivers/gpu/drm/i915/intel_uc.h +2 -2
@@ -110,6 +110,8 @@ enum intel_uc_fw_type {* of fetching, caching, and loading the firmware image into the GuC. */ struct intel_uc_fw { +void (*fetch)(struct drm_i915_private *dev_priv, +struct intel_uc_fw *uc_fw);const char *path;size_t size;struct drm_i915_gem_object *obj;
@@ -196,8 +198,6 @@ void intel_guc_fini(struct drm_i915_private *dev_priv);const char *intel_uc_fw_status_repr (enum intel_uc_fw_status status);int intel_guc_suspend (struct drm_i915_private *dev_priv);int intel_guc_resume (struct drm_i915_private *dev_priv); -void intel_uc_fw_fetch (struct drm_i915_private *dev_priv, -struct intel_uc_fw *uc_fw);u32 intel_guc_wopcm_size (struct drm_i915_private *dev_priv);/* i915_guc_submission.c */
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Joonas Lahtinen
maintainerRe: [PATCH 5/8] drm/i915/uc: Make intel_uc_fw_fetch() static
Quoted 12 lines · Arkadiusz Hiler
intel_uc_fw_fetch()is confusingly named in the light of recent changes.It's also in the wrong place - '
guc_loader.h' - it's used for both guc and huc, which was reflected in name, but not it's location, so let's move it to 'intel_uc.c'.We can make a
intel_uc_fwcallback out of it, to avoid leaking it outsideintel_uc.cCc Joonas Lahtinen <[email protected]>Cc Michal Winiarski <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>
<SNIP>
drivers/gpu/drm/i915/intel_uc.c void intel_uc_init_early(struct drm_i915_private *dev_priv) · line 42
Quoted 8 lines
(quoted diff) +0
-0@@ -32,7 +36,10 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv)
voidintel_uc_fetch_fw(struct drm_i915_private *dev_priv){drivers/gpu/drm/i915/intel_uc.c +1
-0+ dev_priv->huc.fw.fetch = uc_fetch_fw;
intel_huc_fetch_fw(&dev_priv->huc);drivers/gpu/drm/i915/intel_uc.c +2
-0+ + dev_priv->guc.fw.fetch = uc_fetch_fw;
I'm bit confused, I was under impression these functions were going to be different for each uC? If they're not, then it most definitely should not be a function pointer.
int ret; ret = intel_huc_select_fw(dev_priv->huc); if (ret) goto err; ret = intel_uc_fw_prepare(&dev_priv->huc->fw); if (ret) goto err; ret = intel_guc_select_fw(dev_priv->guc); if (ret) goto err_huc; ret = intel_uc_fw_prepare(&dev_priv->guc->fw); if (ret)
goto err_huc;
return 0;
err_huc:
intel_uc_fw_unprepare(&dev_priv->huc->fw); err: return ret;
By always involving proper onion teardown, no dangling objects are left around.
Quoted 7 lines
intel_guc_fetch_fw(&dev_priv->guc);
}(quoted diff) +0
-0@@ -120,3 +127,137 @@ int intel_guc_sample_forcewake(struct intel_guc *guc) > return intel_guc_send(guc, action, ARRAY_SIZE(action)); }
drivers/gpu/drm/i915/intel_uc.c int intel_guc_sample_forcewake(struct intel_guc *guc) · line 138
Quoted 9 lines
drivers/gpu/drm/i915/intel_uc.c +9
-0+static void uc_fetch_fw (struct drm_i915_private *dev_priv, + struct intel_uc_fw *uc_fw) +{ +struct pci_dev *pdev = dev_priv->drm.pdev; +struct drm_i915_gem_object *obj; +const struct firmware *fw =NULL ; +struct uc_css_header *css; +size_t size; +int err;
This function should be named differently, because it doesn't simply fetch it, but also validates and makes an object of it (which may be left dangling).
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 5/8] drm/i915/uc: Make intel_uc_fw_fetch() static
Quoted 16 lines · Joonas Lahtinen
Quoted 12 lines · Arkadiusz Hiler
intel_uc_fw_fetch()is confusingly named in the light of recent changes.It's also in the wrong place - '
guc_loader.h' - it's used for both guc and huc, which was reflected in name, but not it's location, so let's move it to 'intel_uc.c'.We can make a
intel_uc_fwcallback out of it, to avoid leaking it outsideintel_uc.cCc Joonas Lahtinen <[email protected]>Cc Michal Winiarski <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]><SNIP>
Quoted 12 lines
drivers/gpu/drm/i915/intel_uc.c void intel_uc_init_early(struct drm_i915_private *dev_priv) · line 42
Quoted 8 lines
(quoted diff) +0
-0@@ -32,7 +36,10 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv)
voidintel_uc_fetch_fw(struct drm_i915_private *dev_priv){drivers/gpu/drm/i915/intel_uc.c +1
-0+ dev_priv->huc.fw.fetch = uc_fetch_fw;
intel_huc_fetch_fw(&dev_priv->huc);drivers/gpu/drm/i915/intel_uc.c +2
-0+ + dev_priv->guc.fw.fetch = uc_fetch_fw;I'm bit confused, I was under impression these functions were going to be different for each uC? If they're not, then it most definitely should not be a function pointer.
The issue I presented on the IRC (maybe not clearly enough) was just placement and naming of that function. The proposition with callback seemd odd, but since it was backed by couple of people I commited to it.
Glad you spoted that here.
Quoted 19 lines
int ret; ret = intel_huc_select_fw(dev_priv->huc); if (ret) goto err; ret = intel_uc_fw_prepare(&dev_priv->huc->fw); if (ret) goto err; ret = intel_guc_select_fw(dev_priv->guc); if (ret) goto err_huc; ret = intel_uc_fw_prepare(&dev_priv->guc->fw); if (ret)goto
err_huc;return 0;
Quoted 7 lines
err_huc:intel_uc_fw_unprepare(&dev_priv->huc->fw); err: return ret;By always involving proper onion teardown, no dangling objects are left around.
That's exactly what I was missing in my approach. Thanks!
Quoted 5 lines
Quoted 5 lines
intel_guc_fetch_fw(&dev_priv->guc);
}(quoted diff) +0
-0@@ -120,3 +127,137 @@ int intel_guc_sample_forcewake(struct intel_guc *guc) > return intel_guc_send(guc, action, ARRAY_SIZE(action));
Quoted 15 lines
drivers/gpu/drm/i915/intel_uc.c int intel_guc_sample_forcewake(struct intel_guc *guc) · line 138
Quoted 11 lines
(quoted diff) +0
-0}drivers/gpu/drm/i915/intel_uc.c +9
-0+static void uc_fetch_fw (struct drm_i915_private *dev_priv, + struct intel_uc_fw *uc_fw) +{ +struct pci_dev *pdev = dev_priv->drm.pdev; +struct drm_i915_gem_object *obj; +const struct firmware *fw =NULL ; +struct uc_css_header *css; +size_t size; +int err;This function should be named differently, because it doesn't simply fetch it, but also validates and makes an object of it (which may be left dangling).
intel_uc_fw_prepare you've used in the example above seems like a good name.
Onto fixing this mess. Thanks for the input :-)
Signature · 6 lines
Cheers, Arek _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 6/8 drm/i915/guc: Extract param logic form guc_init
In reply to Arkadiusz HilerLet intel_guc_fetch_fw() focus on determining and fetching the correct firmware.
This patch introduces intel_uc_sanitize_options() that is called from intel_sanitize_options().
Then, if we have GuC, we can call intel_guc_fetch_fw() conditionally and we do not have to do the internal checks.
v2: fix comment, notify when nuking GuC explicitly enabled (M. Wajdeczko)
v3: fix comment again, change the nuke message (M. Wajdeczko)
--- drivers/gpu/drm/i915/i915_drv.c | 2 ++ drivers/gpu/drm/i915/intel_guc_loader.c | 16 +--------------- drivers/gpu/drm/i915/intel_uc.c | 27 ++++++++++++++++++++++++++- drivers/gpu/drm/i915/intel_uc.h | 1 + 4 files changed, 30 insertions(+), 16 deletions(-)
drivers/gpu/drm/i915/i915_drv.c +2 -0
@@ -993,6 +993,8 @@ static void intel_sanitize_options(struct drm_i915_private *dev_priv) i915.semaphores =intel_sanitize_semaphores (dev_priv, i915.semaphores);DRM_DEBUG_DRIVER ("use GPU semaphores? %s\n" ,yesno (i915.semaphores)); + +intel_uc_sanitize_options (dev_priv); }/**
drivers/gpu/drm/i915/intel_guc_loader.c +1 -15
@@ -588,7 +588,7 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)/** -* intel_guc_fetch_fw() - define parameters and fetch firmware +* intel_guc_fetch_fw() - determine and fetch firmware * @guc: intel_guc struct * * Called early during driver load, but after GEM is initialised.
@@ -601,17 +601,6 @@ void intel_guc_fetch_fw(struct intel_guc *guc)struct drm_i915_private *dev_priv =guc_to_i915 (guc);const char *fw_path; -if (!HAS_GUC (dev_priv)) { - i915.enable_guc_loading =0 ; - i915.enable_guc_submission =0 ; - }else { -/* A negative value means "use platform default" */ -if (i915.enable_guc_loading <0 ) - i915.enable_guc_loading =HAS_GUC_UCODE (dev_priv); -if (i915.enable_guc_submission <0 ) - i915.enable_guc_submission =HAS_GUC_SCHED (dev_priv); - } -if (!HAS_GUC_UCODE (dev_priv)) { fw_path =NULL ; }else if (IS_SKYLAKE (dev_priv)) {
@@ -634,9 +623,6 @@ void intel_guc_fetch_fw(struct intel_guc *guc) guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; -/* Early (and silent) return if GuC loading is disabled */ -if (!i915.enable_guc_loading) -return ;if (fw_path ==NULL )return ;if (*fw_path =='\0' )
drivers/gpu/drm/i915/intel_uc.c +26 -1
@@ -29,6 +29,27 @@static void uc_fetch_fw (struct drm_i915_private *dev_priv,struct intel_uc_fw *uc_fw); +void intel_uc_sanitize_options (struct drm_i915_private *dev_priv) +{ +if (!HAS_GUC (dev_priv)) { +if (i915.enable_guc_loading >0 ) +DRM_INFO ("Ignoring GuC options, no hardware" ); + + i915.enable_guc_loading =0 ; + i915.enable_guc_submission =0 ; + }else { +/* A negative value means "use platform default" */ +if (i915.enable_guc_loading <0 ) + i915.enable_guc_loading =HAS_GUC_UCODE (dev_priv); +if (i915.enable_guc_submission <0 ) + i915.enable_guc_submission =HAS_GUC_SCHED (dev_priv); + +/* can't enable guc submission without guc loaded */ +if (!i915.enable_guc_loading) + i915.enable_guc_submission =0 ; + } +} +void intel_uc_init_early (struct drm_i915_private *dev_priv) {mutex_init (&dev_priv->guc.send_mutex);
@@ -36,8 +57,12 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv)void intel_uc_fetch_fw (struct drm_i915_private *dev_priv) { +if (!i915.enable_guc_loading) +return ; + dev_priv->huc.fw.fetch = uc_fetch_fw; -intel_huc_fetch_fw (&dev_priv->huc); +if (HAS_HUC_UCODE (dev_priv)) +intel_huc_fetch_fw (&dev_priv->huc); dev_priv->guc.fw.fetch = uc_fetch_fw;intel_guc_fetch_fw (&dev_priv->guc);
drivers/gpu/drm/i915/intel_uc.h +1 -0
@@ -186,6 +186,7 @@ struct intel_huc { };/* intel_uc.c */ +void intel_uc_sanitize_options (struct drm_i915_private *dev_priv);void intel_uc_init_early (struct drm_i915_private *dev_priv);void intel_uc_fetch_fw (struct drm_i915_private *dev_priv);int intel_guc_send (struct intel_guc *guc,const u32 *action,u32 len);
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 6/8] drm/i915/guc: Extract param logic form guc_init
Quoted 9 lines · Arkadiusz Hiler
Let
intel_guc_fetch_fw()focus on determining and fetching the correct firmware.This patch introduces
intel_uc_sanitize_options()that is called fromintel_sanitize_options().Then, if we have GuC, we can call
intel_guc_fetch_fw()conditionally and we do not have to do the internal checks.
Quoted 7 lines
v2: fix comment, notify when nuking GuC explicitly enabled (M. Wajdeczko)
v3: fix comment again, change the nuke message (M. Wajdeczko)Cc Michal Winiarski <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>
Ah, you already had this patch :)
One note below.
<SNIP>
drivers/gpu/drm/i915/intel_uc.c void intel_uc_init_early(struct drm_i915_private *dev_priv) · line 65
Quoted 11 lines
(quoted diff) +0
-0@@ -36,8 +57,12 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv)
voidintel_uc_fetch_fw(struct drm_i915_private *dev_priv){drivers/gpu/drm/i915/intel_uc.c +3
-0+if (!i915.enable_guc_loading) +return ; +
dev_priv->huc.fw.fetch=uc_fetch_fw;drivers/gpu/drm/i915/intel_uc.c +2
-1-intel_huc_fetch_fw (&dev_priv->huc); +if (HAS_HUC_UCODE (dev_priv)) +intel_huc_fetch_fw (&dev_priv->huc);
With the addition of select_fw, the check would become
if (dev_priv->huc->fw.path)
to allow specifying the firmware from command line.
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 7/8 drm/i915/guc: Simplify intel_guc_init_hw()
In reply to Arkadiusz HilerCurrent version of intel_guc_init_hw() does a lot:
- cares about submission
- loads huc
- implement WA
This change offloads some of the logic to intel_uc_load(), which now cares about the above.
v2: rename guc_hw_reset and fix typo in define name (M. Wajdeczko)
v3: rename once again
--- drivers/gpu/drm/i915/i915_gem.c | 2 +- drivers/gpu/drm/i915/intel_guc_loader.c | 136 +++----------------------------- drivers/gpu/drm/i915/intel_uc.c | 107 +++++++++++++++++++++++++ drivers/gpu/drm/i915/intel_uc.h | 5 ++ 4 files changed, 125 insertions(+), 125 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +1 -1
@@ -4512,7 +4512,7 @@ int i915_gem_init_hw(struct drm_i915_private *dev_priv)intel_mocs_init_l3cc_table (dev_priv);/* We can't enable contexts until all firmware is loaded */ - ret =intel_guc_init_hw (dev_priv); + ret =intel_uc_init_hw (dev_priv);if (ret) goto out;
drivers/gpu/drm/i915/intel_guc_loader.c +12 -124
204 lines not on this page · raw message
drivers/gpu/drm/i915/intel_uc.c +107 -0
119 lines not on this page · raw message
drivers/gpu/drm/i915/intel_uc.h +5 -0
@@ -29,6 +29,8 @@#include "intel_ringbuffer.h" #include "i915_vma.h" +/* WaEnableGuCBootHashCheckNotSet:skl,bxt */ +#define GUC_WA_HASH_CHECK_NOT_SET_ATTEMPTS 3struct drm_i915_gem_request ;
@@ -189,6 +191,7 @@ struct intel_huc {void intel_uc_sanitize_options (struct drm_i915_private *dev_priv);void intel_uc_init_early (struct drm_i915_private *dev_priv);void intel_uc_fetch_fw (struct drm_i915_private *dev_priv); +int intel_uc_init_hw (struct drm_i915_private *dev_priv);int intel_guc_send (struct intel_guc *guc,const u32 *action,u32 len);int intel_guc_sample_forcewake (struct intel_guc *guc);
@@ -199,6 +202,8 @@ void intel_guc_fini(struct drm_i915_private *dev_priv);const char *intel_uc_fw_status_repr (enum intel_uc_fw_status status);int intel_guc_suspend (struct drm_i915_private *dev_priv);int intel_guc_resume (struct drm_i915_private *dev_priv); +void intel_guc_interrupts_release (struct drm_i915_private *dev_priv); +void intel_guc_interrupts_capture (struct drm_i915_private *dev_priv);u32 intel_guc_wopcm_size (struct drm_i915_private *dev_priv);/* i915_guc_submission.c */
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 7/8] drm/i915/guc: Simplify intel_guc_init_hw()
Quoted 16 lines · Arkadiusz Hiler
Current version of
intel_guc_init_hw()does a lot:
- cares about submission
- loads huc
- implement WAThis change offloads some of the logic to
intel_uc_load(), which now cares about the above.v2: rename
guc_hw_resetand fix typo in define name (M. Wajdeczko)
v3: rename once againCc Anusha Srivatsa <[email protected]>Cc Michal Winiarski <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>
<SNIP>
Quoted 12 lines
drivers/gpu/drm/i915/intel_guc_loader.c +1
-1@@ -443,147 +425,53 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv) { struct intel_uc_fw *guc_fw = &dev_priv->guc.fw; const char *fw_path = guc_fw->path; -int retries, ret, err; +int ret;
DRM_DEBUG_DRIVER("GuC fw status: path %s, fetch %s, load %s\n",fw_path,intel_uc_fw_status_repr(guc_fw->fetch_status),intel_uc_fw_status_repr(guc_fw->load_status));
drivers/gpu/drm/i915/intel_guc_loader.c int intel_guc_init_hw(struct drm_i915_private *dev_priv) · line 435
Quoted 6 lines
drivers/gpu/drm/i915/intel_guc_loader.c +1
-5-/* Loading forbidden, or no firmware to load? */ -if (!i915.enable_guc_loading) { - err =0 ; - goto fail; - }else if (fw_path ==NULL ) { +if (fw_path ==NULL ) {
if (!fw_path)
Quoted 1 line
/* Device is known to have no uCode (e.g. no GuC) */
Spurious comment.
Quoted 5 lines
drivers/gpu/drm/i915/intel_guc_loader.c +1
-2- err = -ENXIO ; - goto fail; +return -ENXIO ;} else if (
*fw_path== '\0') {
/* Device has a GuC but we don't know what f/w to load? */
Even more spurious comment, as there's WARN :)
Quoted 7 lines
WARN(1, "No GuC firmware known for this platform!\n");drivers/gpu/drm/i915/intel_guc_loader.c +1
-2- err = -ENODEV ; - goto fail; +return -ENODEV ;}
/* Fetch failed, or already fetched but failed to load? */
Code is again self-documenting here.
Quoted 9 lines
if (
guc_fw->fetch_status!=INTEL_UC_FIRMWARE_SUCCESS) {drivers/gpu/drm/i915/intel_guc_loader.c +1
-2- err = -EIO ; - goto fail; +return -EIO ;} else if (
guc_fw->load_status==INTEL_UC_FIRMWARE_FAIL) {drivers/gpu/drm/i915/intel_guc_loader.c +1
-2- err = -ENOEXEC ; - goto fail; +return -ENOEXEC ;}
Single line braces can be dropped.
drivers/gpu/drm/i915/intel_uc.c void intel_uc_fetch_fw(struct drm_i915_private *dev_priv) · line 93
Quoted 2 lines
drivers/gpu/drm/i915/intel_uc.c +2
-0+int intel_uc_init_hw (struct drm_i915_private *dev_priv) +{
This function is still lacking onion tear-down (because it's so convoluted), but it's in the better direction (and it was previously so).
+++ b/drivers/gpu/drm/i915/intel_uc.h
drivers/gpu/drm/i915/intel_uc.h line 33
Quoted 6 lines
drivers/gpu/drm/i915/intel_uc.h +0
-0@@ -29,6 +29,8 @@ #include "intel_ringbuffer.h"
#include "i915_vma.h"drivers/gpu/drm/i915/intel_uc.h +2
-0+/* WaEnableGuCBootHashCheckNotSet:skl,bxt */ +#define GUC_WA_HASH_CHECK_NOT_SET_ATTEMPTS 3
As this is a W/A only, I'd keep the code local to single spot.
With above stylistic nitpicks;
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 8/8 drm/i915/uc: Simplify firmware path handling
In reply to Arkadiusz HilerCurrently fw->path values can represent one of three possible states:
1) NULL - device without the uC
2) '\0' - device with the uC but have no firmware
3) else - device with the uC and we have firmware
Second case is used only to WARN at a later stage.
We can WARN right away and merge cases 1 and 2.
Code can be even further simplified and common (HuC/GuC logic) happening right before the fetch can be offloaded to the common function.
v2: fewer temporary variables, more straightforward flow (M. Wajdeczko)
v3: DRM_ERROR instead of WARN (M. Wajdeczko)
--- drivers/gpu/drm/i915/intel_guc_loader.c | 39 +++++++++++---------------------- drivers/gpu/drm/i915/intel_huc.c | 20 +++++------------ drivers/gpu/drm/i915/intel_uc.c | 5 +++-- 3 files changed, 22 insertions(+), 42 deletions(-)
drivers/gpu/drm/i915/intel_guc_loader.c +13 -26
@@ -433,12 +433,8 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv) intel_uc_fw_status_repr(guc_fw->load_status));if (fw_path ==NULL ) { -/* Device is known to have no uCode (e.g. no GuC) */ +/* We do not have uCode for the device */ return -ENXIO ; - }else if (*fw_path =='\0' ) { -/* Device has a GuC but we don't know what f/w to load? */ -WARN (1 ,"No GuC firmware known for this platform!\n" ); -return -ENODEV ; }/* Fetch failed, or already fetched but failed to load? */
@@ -474,7 +470,6 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)return 0 ; } -/** * intel_guc_fetch_fw() - determine and fetch firmware * @guc: intel_guc struct
@@ -487,39 +482,31 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)void intel_guc_fetch_fw (struct intel_guc *guc) {struct drm_i915_private *dev_priv =guc_to_i915 (guc); -const char *fw_path; -if (!HAS_GUC_UCODE (dev_priv)) { - fw_path =NULL ; - }else if (IS_SKYLAKE (dev_priv)) { - fw_path =I915_SKL_GUC_UCODE ; + guc->fw.path =NULL ; + guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.fw =INTEL_UC_FW_TYPE_GUC ; + +if (IS_SKYLAKE (dev_priv)) { + guc->fw.path =I915_SKL_GUC_UCODE ; guc->fw.major_ver_wanted =SKL_FW_MAJOR ; guc->fw.minor_ver_wanted =SKL_FW_MINOR ; }else if (IS_BROXTON (dev_priv)) { - fw_path =I915_BXT_GUC_UCODE ; + guc->fw.path =I915_BXT_GUC_UCODE ; guc->fw.major_ver_wanted =BXT_FW_MAJOR ; guc->fw.minor_ver_wanted =BXT_FW_MINOR ; }else if (IS_KABYLAKE (dev_priv)) { - fw_path =I915_KBL_GUC_UCODE ; + guc->fw.path =I915_KBL_GUC_UCODE ; guc->fw.major_ver_wanted =KBL_FW_MAJOR ; guc->fw.minor_ver_wanted =KBL_FW_MINOR ; }else { - fw_path ="" ;/* unknown device */ +DRM_ERROR ("No GuC firmware known for platform with GuC!\n" ); + i915.enable_guc_loading =0 ; +return ; } - guc->fw.path = fw_path; - guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; - guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; - -if (fw_path ==NULL ) -return ; -if (*fw_path =='\0' ) -return ; - - guc->fw.fetch_status =INTEL_UC_FIRMWARE_PENDING ; -DRM_DEBUG_DRIVER ("GuC firmware pending, path %s\n" , fw_path); guc->fw.fetch (dev_priv, &guc->fw); -/* status must now be FAIL or SUCCESS */ }/**
drivers/gpu/drm/i915/intel_huc.c +6 -14
@@ -155,37 +155,29 @@ static int huc_ucode_xfer(struct drm_i915_private *dev_priv)void intel_huc_fetch_fw (struct intel_huc *huc) {struct drm_i915_private *dev_priv =huc_to_i915 (huc); -const char *fw_path =NULL ; huc->fw.path =NULL ; huc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; huc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; huc->fw.fw =INTEL_UC_FW_TYPE_HUC ; -if (!HAS_HUC_UCODE (dev_priv)) -return ; -if (IS_SKYLAKE (dev_priv)) { - fw_path =I915_SKL_HUC_UCODE ; + huc->fw.path =I915_SKL_HUC_UCODE ; huc->fw.major_ver_wanted =SKL_HUC_FW_MAJOR ; huc->fw.minor_ver_wanted =SKL_HUC_FW_MINOR ; }else if (IS_BROXTON (dev_priv)) { - fw_path =I915_BXT_HUC_UCODE ; + huc->fw.path =I915_BXT_HUC_UCODE ; huc->fw.major_ver_wanted =BXT_HUC_FW_MAJOR ; huc->fw.minor_ver_wanted =BXT_HUC_FW_MINOR ; }else if (IS_KABYLAKE (dev_priv)) { - fw_path =I915_KBL_HUC_UCODE ; + huc->fw.path =I915_KBL_HUC_UCODE ; huc->fw.major_ver_wanted =KBL_HUC_FW_MAJOR ; huc->fw.minor_ver_wanted =KBL_HUC_FW_MINOR ; + }else { +DRM_ERROR ("No HuC firmware known for platform with HuC!\n" ); +return ; } - huc->fw.path = fw_path; - huc->fw.fetch_status =INTEL_UC_FIRMWARE_PENDING ; - -DRM_DEBUG_DRIVER ("HuC firmware pending, path %s\n" , fw_path); - -WARN (huc->fw.path ==NULL ,"HuC present but no fw path\n" ); - huc->fw.fetch (dev_priv, &huc->fw); }
drivers/gpu/drm/i915/intel_uc.c +3 -2
@@ -269,8 +269,9 @@ static void uc_fetch_fw(struct drm_i915_private *dev_priv,size_t size;int err; -DRM_DEBUG_DRIVER ("before requesting firmware: uC fw fetch status %s\n" , -intel_uc_fw_status_repr (uc_fw->fetch_status)); + uc_fw->fetch_status =INTEL_UC_FIRMWARE_PENDING ; + +DRM_DEBUG_DRIVER ("uC firmware pending, path %s\n" , uc_fw->path); err =request_firmware (&fw, uc_fw->path, &pdev->dev);if (err)
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 8/8] drm/i915/uc: Simplify firmware path handling
Quoted 77 lines · Arkadiusz Hiler
Currently
fw->pathvalues can represent one of three possible states:1) NULL - device without the uC
2) '\0' - device with the uC but have no firmware
3) else - device with the uC and we have firmwareSecond case is used only to WARN at a later stage.
We can WARN right away and merge cases 1 and 2.
Code can be even further simplified and common (HuC/GuC logic) happening right before the fetch can be offloaded to the common function.
v2: fewer temporary variables, more straightforward flow (M. Wajdeczko)
v3:DRM_ERRORinstead of WARN (M. Wajdeczko)Cc Anusha Srivatsa <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Michal Winiarski <[email protected]>Cc Michal Wajdeczko <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>--- drivers/gpu/drm/i915/intel_guc_loader.c | 39 +++++++++++---------------------- drivers/gpu/drm/i915/intel_huc.c | 20 +++++------------ drivers/gpu/drm/i915/intel_uc.c | 5 +++-- 3 files changed, 22 insertions(+), 42 deletions(-)drivers/gpu/drm/i915/intel_guc_loader.c +9
-12@@ -433,12 +433,8 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv) intel_uc_fw_status_repr(guc_fw->load_status));if (fw_path ==NULL ) { -/* Device is known to have no uCode (e.g. no GuC) */ +/* We do not have uCode for the device */ return -ENXIO ; - }else if (*fw_path =='\0' ) { -/* Device has a GuC but we don't know what f/w to load? */ -WARN (1 ,"No GuC firmware known for this platform!\n" ); -return -ENODEV ; }/* Fetch failed, or already fetched but failed to load? */ @@ -474,7 +470,6 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)return 0 ; } -/** * intel_guc_fetch_fw() - determine and fetch firmware * @guc: intel_guc struct @@ -487,39 +482,31 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)void intel_guc_fetch_fw (struct intel_guc *guc) {struct drm_i915_private *dev_priv =guc_to_i915 (guc); -const char *fw_path; -if (!HAS_GUC_UCODE (dev_priv)) { - fw_path =NULL ; - }else if (IS_SKYLAKE (dev_priv)) { - fw_path =I915_SKL_GUC_UCODE ; + guc->fw.path =NULL ; + guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.fw =INTEL_UC_FW_TYPE_GUC ; + +if (IS_SKYLAKE (dev_priv)) { + guc->fw.path =I915_SKL_GUC_UCODE ; guc->fw.major_ver_wanted =SKL_FW_MAJOR ; guc->fw.minor_ver_wanted =SKL_FW_MINOR ; }else if (IS_BROXTON (dev_priv)) { - fw_path =I915_BXT_GUC_UCODE ; + guc->fw.path =I915_BXT_GUC_UCODE ; guc->fw.major_ver_wanted =BXT_FW_MAJOR ; guc->fw.minor_ver_wanted =BXT_FW_MINOR ;
drivers/gpu/drm/i915/intel_guc_loader.c int intel_guc_init_hw(struct drm_i915_private *dev_priv) · line 506
Quoted 10 lines
drivers/gpu/drm/i915/intel_guc_loader.c +4
-2}else if (IS_KABYLAKE (dev_priv)) { - fw_path =I915_KBL_GUC_UCODE ; + guc->fw.path =I915_KBL_GUC_UCODE ; guc->fw.major_ver_wanted =KBL_FW_MAJOR ; guc->fw.minor_ver_wanted =KBL_FW_MINOR ; }else { - fw_path ="" ;/* unknown device */ +DRM_ERROR ("No GuC firmware known for platform with GuC!\n" ); + i915.enable_guc_loading =0 ; +return ;
Now plan for having fw_path overriden by a i915_param.guc_firmware.
Perhaps something like
if (i915_param.guc_firmware) {
guc->fw.path = i915_param.guc_firmware; /* needs 0400! */
guc->fw.major_ver_wanted = -1;
guc->fw.minor_ver_wanted = -1;
} else if (IS_SKYLAKE....works?
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 8/8] drm/i915/uc: Simplify firmware path handling
Quoted 91 lines · Chris Wilson
Quoted 77 lines · Arkadiusz Hiler
Currently
fw->pathvalues can represent one of three possible states:1) NULL - device without the uC
2) '\0' - device with the uC but have no firmware
3) else - device with the uC and we have firmwareSecond case is used only to WARN at a later stage.
We can WARN right away and merge cases 1 and 2.
Code can be even further simplified and common (HuC/GuC logic) happening right before the fetch can be offloaded to the common function.
v2: fewer temporary variables, more straightforward flow (M. Wajdeczko)
v3:DRM_ERRORinstead of WARN (M. Wajdeczko)Cc Anusha Srivatsa <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Michal Winiarski <[email protected]>Cc Michal Wajdeczko <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>--- drivers/gpu/drm/i915/intel_guc_loader.c | 39 +++++++++++---------------------- drivers/gpu/drm/i915/intel_huc.c | 20 +++++------------ drivers/gpu/drm/i915/intel_uc.c | 5 +++-- 3 files changed, 22 insertions(+), 42 deletions(-)drivers/gpu/drm/i915/intel_guc_loader.c +9
-12@@ -433,12 +433,8 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv) intel_uc_fw_status_repr(guc_fw->load_status));if (fw_path ==NULL ) { -/* Device is known to have no uCode (e.g. no GuC) */ +/* We do not have uCode for the device */ return -ENXIO ; - }else if (*fw_path =='\0' ) { -/* Device has a GuC but we don't know what f/w to load? */ -WARN (1 ,"No GuC firmware known for this platform!\n" ); -return -ENODEV ; }/* Fetch failed, or already fetched but failed to load? */ @@ -474,7 +470,6 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)return 0 ; } -/** * intel_guc_fetch_fw() - determine and fetch firmware * @guc: intel_guc struct @@ -487,39 +482,31 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)void intel_guc_fetch_fw (struct intel_guc *guc) {struct drm_i915_private *dev_priv =guc_to_i915 (guc); -const char *fw_path; -if (!HAS_GUC_UCODE (dev_priv)) { - fw_path =NULL ; - }else if (IS_SKYLAKE (dev_priv)) { - fw_path =I915_SKL_GUC_UCODE ; + guc->fw.path =NULL ; + guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.fw =INTEL_UC_FW_TYPE_GUC ; + +if (IS_SKYLAKE (dev_priv)) { + guc->fw.path =I915_SKL_GUC_UCODE ; guc->fw.major_ver_wanted =SKL_FW_MAJOR ; guc->fw.minor_ver_wanted =SKL_FW_MINOR ; }else if (IS_BROXTON (dev_priv)) { - fw_path =I915_BXT_GUC_UCODE ; + guc->fw.path =I915_BXT_GUC_UCODE ; guc->fw.major_ver_wanted =BXT_FW_MAJOR ; guc->fw.minor_ver_wanted =BXT_FW_MINOR ;drivers/gpu/drm/i915/intel_guc_loader.c int intel_guc_init_hw(struct drm_i915_private *dev_priv) · line 506
Quoted 10 lines
drivers/gpu/drm/i915/intel_guc_loader.c +4
-2}else if (IS_KABYLAKE (dev_priv)) { - fw_path =I915_KBL_GUC_UCODE ; + guc->fw.path =I915_KBL_GUC_UCODE ; guc->fw.major_ver_wanted =KBL_FW_MAJOR ; guc->fw.minor_ver_wanted =KBL_FW_MINOR ; }else { - fw_path ="" ;/* unknown device */ +DRM_ERROR ("No GuC firmware known for platform with GuC!\n" ); + i915.enable_guc_loading =0 ; +return ;Now plan for having
fw_pathoverriden by a i915_param.guc_firmware.
Quoted 7 lines
Perhaps something like
if (i915_param.guc_firmware) { guc->fw.path = i915_param.guc_firmware; /* needs 0400! */ guc->fw.major_ver_wanted = -1; guc->fw.minor_ver_wanted = -1; } else if (IS_SKYLAKE....works?
Sorry, I do not quite understand the comment. Can you elaborate?
Signature · 6 lines
Cheers, Arek _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 8/8] drm/i915/uc: Simplify firmware path handling
Quoted 100 lines · Arkadiusz Hiler
Quoted 91 lines · Chris Wilson
Quoted 77 lines · Arkadiusz Hiler
Currently
fw->pathvalues can represent one of three possible states:1) NULL - device without the uC
2) '\0' - device with the uC but have no firmware
3) else - device with the uC and we have firmwareSecond case is used only to WARN at a later stage.
We can WARN right away and merge cases 1 and 2.
Code can be even further simplified and common (HuC/GuC logic) happening right before the fetch can be offloaded to the common function.
v2: fewer temporary variables, more straightforward flow (M. Wajdeczko)
v3:DRM_ERRORinstead of WARN (M. Wajdeczko)Cc Anusha Srivatsa <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Michal Winiarski <[email protected]>Cc Michal Wajdeczko <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>--- drivers/gpu/drm/i915/intel_guc_loader.c | 39 +++++++++++---------------------- drivers/gpu/drm/i915/intel_huc.c | 20 +++++------------ drivers/gpu/drm/i915/intel_uc.c | 5 +++-- 3 files changed, 22 insertions(+), 42 deletions(-)drivers/gpu/drm/i915/intel_guc_loader.c +9
-12@@ -433,12 +433,8 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv) intel_uc_fw_status_repr(guc_fw->load_status));if (fw_path ==NULL ) { -/* Device is known to have no uCode (e.g. no GuC) */ +/* We do not have uCode for the device */ return -ENXIO ; - }else if (*fw_path =='\0' ) { -/* Device has a GuC but we don't know what f/w to load? */ -WARN (1 ,"No GuC firmware known for this platform!\n" ); -return -ENODEV ; }/* Fetch failed, or already fetched but failed to load? */ @@ -474,7 +470,6 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)return 0 ; } -/** * intel_guc_fetch_fw() - determine and fetch firmware * @guc: intel_guc struct @@ -487,39 +482,31 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv)void intel_guc_fetch_fw (struct intel_guc *guc) {struct drm_i915_private *dev_priv =guc_to_i915 (guc); -const char *fw_path; -if (!HAS_GUC_UCODE (dev_priv)) { - fw_path =NULL ; - }else if (IS_SKYLAKE (dev_priv)) { - fw_path =I915_SKL_GUC_UCODE ; + guc->fw.path =NULL ; + guc->fw.fetch_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.load_status =INTEL_UC_FIRMWARE_NONE ; + guc->fw.fw =INTEL_UC_FW_TYPE_GUC ; + +if (IS_SKYLAKE (dev_priv)) { + guc->fw.path =I915_SKL_GUC_UCODE ; guc->fw.major_ver_wanted =SKL_FW_MAJOR ; guc->fw.minor_ver_wanted =SKL_FW_MINOR ; }else if (IS_BROXTON (dev_priv)) { - fw_path =I915_BXT_GUC_UCODE ; + guc->fw.path =I915_BXT_GUC_UCODE ; guc->fw.major_ver_wanted =BXT_FW_MAJOR ; guc->fw.minor_ver_wanted =BXT_FW_MINOR ;drivers/gpu/drm/i915/intel_guc_loader.c int intel_guc_init_hw(struct drm_i915_private *dev_priv) · line 506
Quoted 10 lines
drivers/gpu/drm/i915/intel_guc_loader.c +4
-2}else if (IS_KABYLAKE (dev_priv)) { - fw_path =I915_KBL_GUC_UCODE ; + guc->fw.path =I915_KBL_GUC_UCODE ; guc->fw.major_ver_wanted =KBL_FW_MAJOR ; guc->fw.minor_ver_wanted =KBL_FW_MINOR ; }else { - fw_path ="" ;/* unknown device */ +DRM_ERROR ("No GuC firmware known for platform with GuC!\n" ); + i915.enable_guc_loading =0 ; +return ;Now plan for having
fw_pathoverriden by a i915_param.guc_firmware.Quoted 7 lines
Perhaps something like
if (i915_param.guc_firmware) { guc->fw.path = i915_param.guc_firmware; /* needs 0400! */ guc->fw.major_ver_wanted = -1; guc->fw.minor_ver_wanted = -1; } else if (IS_SKYLAKE....works?
Quoted 1 line
Sorry, I do not quite understand the comment. Can you elaborate?
Nevermind, got it.
LGTM, but for it to fully work we need to make uc_fetch_fw (or however it will end up being named) aware that -1 have special meaning.
Now the version cross-check looks like that:
if (uc_fw->major_ver_found != uc_fw->major_ver_wanted ||
uc_fw->minor_ver_found < uc_fw->minor_ver_wanted) {Signature · 6 lines
Cheers, Arek _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Joonas Lahtinen
maintainerRe: [PATCH 8/8] drm/i915/uc: Simplify firmware path handling
Quoted 18 lines · Arkadiusz Hiler
Quoted 13 lines · Arkadiusz Hiler
Quoted 10 lines · Chris Wilson
Now plan for having
fw_pathoverriden by a i915_param.guc_firmware.Perhaps something like
if (i915_param.guc_firmware) { guc->fw.path = i915_param.guc_firmware; /* needs 0400! */ guc->fw.major_ver_wanted = -1; guc->fw.minor_ver_wanted = -1; } else if (IS_SKYLAKE....works?
Quoted 1 line
Sorry, I do not quite understand the comment. Can you elaborate?
Nevermind, got it.
Quoted 7 lines
LGTM, but for it to fully work we need to make
uc_fetch_fw(or however it will end up being named) aware that -1 have special meaning.Now the version cross-check looks like that:
if (
uc_fw->major_ver_found!=uc_fw->major_ver_wanted||uc_fw->minor_ver_found<uc_fw->minor_ver_wanted) {
What Chris proposed, together with tweaking the check sounds good.
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 8/8] drm/i915/uc: Simplify firmware path handling
In reply to Arkadiusz HilerQuoted 21 lines · Arkadiusz Hiler
Currently
fw->pathvalues can represent one of three possible states:1) NULL - device without the uC
2) '\0' - device with the uC but have no firmware
3) else - device with the uC and we have firmwareSecond case is used only to WARN at a later stage.
We can WARN right away and merge cases 1 and 2.
Code can be even further simplified and common (HuC/GuC logic) happening right before the fetch can be offloaded to the common function.
v2: fewer temporary variables, more straightforward flow (M. Wajdeczko)
v3:DRM_ERRORinstead of WARN (M. Wajdeczko)Cc Anusha Srivatsa <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Michal Winiarski <[email protected]>Cc Michal Wajdeczko <[email protected]>Signed-off-by Arkadiusz Hiler <[email protected]>
<SNIP>
Quoted 3 lines
drivers/gpu/drm/i915/intel_guc_loader.c +0
-0@@ -433,12 +433,8 @@ int intel_guc_init_hw(struct drm_i915_private *dev_priv) intel_uc_fw_status_repr(guc_fw->load_status));
if (!guc_fw->path)
return -ENXIO;
Quoted 9 lines
if (
fw_path== NULL) {drivers/gpu/drm/i915/intel_guc_loader.c +1
-1-/* Device is known to have no uCode (e.g. no GuC) */ +/* We do not have uCode for the device */ return
-ENXIO;drivers/gpu/drm/i915/intel_guc_loader.c +0
-4- }else if (*fw_path =='\0' ) { -/* Device has a GuC but we don't know what f/w to load? */ -WARN (1 ,"No GuC firmware known for this platform!\n" ); -return -ENODEV ;}
<SNIP>
Quoted 4 lines
drivers/gpu/drm/i915/intel_guc_loader.c +0
-3- - guc->fw.fetch_status =INTEL_UC_FIRMWARE_PENDING ; -DRM_DEBUG_DRIVER ("GuC firmware pending, path %s\n" , fw_path);
guc->fw.fetch(dev_priv, &guc->fw);
This was function pointer was discussed in other thread.
With above addressed;
Regards, Joonas
Signature · 7 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✓ Fi.CI.BAT: success for GuC Scrub vol. 1 (rev4)
In reply to Arkadiusz Hiler== Series Details ==
== Series Details ==
Series: GuC Scrub vol. 1 (rev4)
URL : https://patchwork.freedesktop.org/series/16856/
State : success
== Summary ==
Series 16856v4 GuC Scrub vol. 1
https://patchwork.freedesktop.org/api/1.0/series/16856/revisions/4/mbox/
fi-bdw-5557u total:253 pass:242 dwarn:0 dfail:0 fail:0 skip:11 fi-bsw-n3050 total:253 pass:214 dwarn:0 dfail:0 fail:0 skip:39 fi-bxt-j4205 total:253 pass:234 dwarn:0 dfail:0 fail:0 skip:19 fi-bxt-t5700 total:83 pass:70 dwarn:0 dfail:0 fail:0 skip:12 fi-byt-j1900 total:253 pass:226 dwarn:0 dfail:0 fail:0 skip:27 fi-byt-n2820 total:253 pass:222 dwarn:0 dfail:0 fail:0 skip:31 fi-hsw-4770 total:253 pass:237 dwarn:0 dfail:0 fail:0 skip:16 fi-hsw-4770r total:253 pass:237 dwarn:0 dfail:0 fail:0 skip:16 fi-ilk-650 total:253 pass:203 dwarn:0 dfail:0 fail:0 skip:50 fi-ivb-3520m total:253 pass:235 dwarn:0 dfail:0 fail:0 skip:18 fi-ivb-3770 total:253 pass:235 dwarn:0 dfail:0 fail:0 skip:18 fi-kbl-7500u total:253 pass:235 dwarn:0 dfail:0 fail:0 skip:18 fi-skl-6260u total:253 pass:243 dwarn:0 dfail:0 fail:0 skip:10 fi-skl-6700hq total:253 pass:236 dwarn:0 dfail:0 fail:0 skip:17 fi-skl-6700k total:253 pass:231 dwarn:4 dfail:0 fail:0 skip:18 fi-skl-6770hq total:253 pass:243 dwarn:0 dfail:0 fail:0 skip:10 fi-snb-2520m total:253 pass:225 dwarn:0 dfail:0 fail:0 skip:28 fi-snb-2600 total:253 pass:224 dwarn:0 dfail:0 fail:0 skip:29
bf89ec45d0822835b03910371ac0baf46c4efa2d drm-tip: 2017y-02m-22d-14h-30m-04s UTC integration manifest de86964 drm/i915/uc: Simplify firmware path handling
3bd35f1 drm/i915/guc: Simplify intel_guc_init_hw()
9d55149 drm/i915/guc: Extract param logic form guc_init
e5b5068 drm/i915/uc: Make intel_uc_fw_fetch() static
f32b9e4 drm/i915/uc: Rename intel_?uc_init() to intel_?uc_fetch_fw()
d2ca92a drm/i915/huc: Add huc_to_i915
20c4531 drm/i915/uc: Drop superfluous externs in intel_uc.h
437508e drm/i915/uc: Rename intel_?uc_{setup, load}() to _init_hw()
== Logs ==
For more details see: https://intel-gfx-ci.01.org/CI/Patchwork_3929/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx