Message ID | 20220719172825.3155696-2-alan.previn.teres.alexis@intel.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | drm/i915/guc: Remove more GuC error capture noise | expand |
On 7/19/2022 10:28, Alan Previn wrote: > Remove the CONFIG_DRM_I915_DEBUG_GUC version of the > __out macro. The original thought was we have additional > dmesg entries in the event that the last gpu_coredump > error capture state was never retrieved, we don't > lose the new capture. These additional messages only > when CONFIG_DRM_I915_DEBUG_GUC is on. However it should > have been a drm_dbg instead of drm_warn. Additionally, > upon further inspection, it became clear we don't really > need this additional messages to align with execlist > as well as remove some more unncessary noise. > > Signed-off-by: Alan Previn <alan.previn.teres.alexis@intel.com> > --- > drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c | 8 -------- > 1 file changed, 8 deletions(-) > > diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c > index 75257bd20ff0..a9910962d2dc 100644 > --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c > +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c > @@ -1365,16 +1365,8 @@ guc_capture_reg_to_str(const struct intel_guc *guc, u32 owner, u32 type, > return NULL; > } > > -#ifdef CONFIG_DRM_I915_DEBUG_GUC > -#define __out(a, ...) \ > - do { \ > - drm_warn((&(a)->i915->drm), __VA_ARGS__); \ > - i915_error_printf((a), __VA_ARGS__); \ > - } while (0) > -#else > #define __out(a, ...) \ > i915_error_printf(a, __VA_ARGS__) > -#endif Is there any point in keeping the _out wrapper? Why not just call i915_error_printf directly? Seems like an unnecessary level of obfuscation now. John. > > #define GCAP_PRINT_INTEL_ENG_INFO(ebuf, eng) \ > do { \
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c index 75257bd20ff0..a9910962d2dc 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c @@ -1365,16 +1365,8 @@ guc_capture_reg_to_str(const struct intel_guc *guc, u32 owner, u32 type, return NULL; } -#ifdef CONFIG_DRM_I915_DEBUG_GUC -#define __out(a, ...) \ - do { \ - drm_warn((&(a)->i915->drm), __VA_ARGS__); \ - i915_error_printf((a), __VA_ARGS__); \ - } while (0) -#else #define __out(a, ...) \ i915_error_printf(a, __VA_ARGS__) -#endif #define GCAP_PRINT_INTEL_ENG_INFO(ebuf, eng) \ do { \
Remove the CONFIG_DRM_I915_DEBUG_GUC version of the __out macro. The original thought was we have additional dmesg entries in the event that the last gpu_coredump error capture state was never retrieved, we don't lose the new capture. These additional messages only when CONFIG_DRM_I915_DEBUG_GUC is on. However it should have been a drm_dbg instead of drm_warn. Additionally, upon further inspection, it became clear we don't really need this additional messages to align with execlist as well as remove some more unncessary noise. Signed-off-by: Alan Previn <alan.previn.teres.alexis@intel.com> --- drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c | 8 -------- 1 file changed, 8 deletions(-)