drm/i915: Add fault injection support
Subsystems (2)
PATCH 1/1 drm/i915: Add fault injection support
Add support for forcing an error at selected places in the driver. As an example add 3 options to fail during driver loading.
Requested by Chris.
---
[This depends on
https://lists.freedesktop.org/archives/intel-gfx/2016-March/089597.html]
drivers/gpu/drm/i915/i915_dma.c | 9 +++++++++ drivers/gpu/drm/i915/i915_drv.h | 4 ++++ drivers/gpu/drm/i915/i915_params.c | 6 ++++++ drivers/gpu/drm/i915/i915_params.h | 1 + 4 files changed, 20 insertions(+)
drivers/gpu/drm/i915/i915_dma.c +9 -0
@@ -950,6 +950,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,struct intel_device_info *device_info;int ret =0 ; +if (i915.inject_fault &I915_FAULT_INIT_EARLY ) +return -ENODEV ; + dev_priv->dev = dev;/* Setup the write-once "constant" device info */
@@ -1064,6 +1067,9 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv)struct drm_device *dev = dev_priv->dev;int ret; +if (i915.inject_fault &I915_FAULT_INIT_MMIO ) +return -ENODEV ; +if (i915_get_bridge_dev (dev))return -EIO ;
@@ -1107,6 +1113,9 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)uint32_t aperture_size;int ret; +if (i915.inject_fault &I915_FAULT_INIT_HW ) +return -ENODEV ; +intel_device_info_runtime_init (dev); ret =i915_gem_gtt_init (dev);
drivers/gpu/drm/i915/i915_drv.h +4 -0
@@ -98,6 +98,10 @@#define I915_STATE_WARN_ON (x) \ I915_STATE_WARN((x), "%s", "WARN_ON(" __stringify(x) ")") +#define I915_FAULT_INIT_EARLY 0x0001 +#define I915_FAULT_INIT_MMIO 0x0002 +#define I915_FAULT_INIT_HW 0x0004 +static inline const char *yesno (bool v) {return v ?"yes" :"no" ;
drivers/gpu/drm/i915/i915_params.c +6 -0
@@ -56,6 +56,7 @@ struct i915_params i915 __read_mostly = { .edp_vswing = 0, .enable_guc_submission = false, .guc_log_level = -1, + .inject_fault = 0, };module_param_named (modeset, i915.modeset, int,0400 );
@@ -201,3 +202,8 @@ MODULE_PARM_DESC(enable_guc_submission, "Enable GuC submission (default:false)")module_param_named (guc_log_level, i915.guc_log_level, int,0400 );MODULE_PARM_DESC (guc_log_level,"GuC firmware logging level (-1:disabled (default), 0-3:enabled)" ); + +module_param_named (inject_fault, i915.inject_fault, uint,0600 ); +MODULE_PARM_DESC (inject_fault, +"Force an error at selected points " +"(0:disabled, 0x1:INIT_EARLY, 0x2:INIT_MMIO, 0x4:INIT_HW)" );
drivers/gpu/drm/i915/i915_params.h +1 -0
@@ -59,6 +59,7 @@ struct i915_params {bool enable_guc_submission;bool verbose_state_checks;bool nuclear_pageflip; +unsigned int inject_fault; };extern struct i915_params i915__read_mostly ;
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for drm/i915: Add fault injection support
== Series Details ==
== Series Details ==
Series: drm/i915: Add fault injection support
URL : https://patchwork.freedesktop.org/series/4385/
State : failure
== Summary ==
Series 4385v1 drm/i915: Add fault injection support
2016-03-11T22:53:14.152284 http://patchwork.freedesktop.org/api/1.0/series/4385/revisions/1/mbox/
Applying: drm/i915: Add fault injection support
Using index info to reconstruct a base tree...
M drivers/gpu/drm/i915/i915_dma.c
M drivers/gpu/drm/i915/i915_drv.h
Falling back to patching base and 3-way merge... Auto-merging drivers/gpu/drm/i915/i915_drv.h
Auto-merging drivers/gpu/drm/i915/i915_dma.c
CONFLICT (content): Merge conflict in drivers/gpu/drm/i915/i915_dma.c Patch failed at 0001 drm/i915: Add fault injection support
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 1/1 drm/i915: Add fault injection support
In reply to Imre DeakAdd support for forcing an error at selected places in the driver. As an example add 4 options to fail during driver loading.
Requested by Chris.
v2:
- Add fault point for modeset initialization
- Print debug message when injecting an error
--- drivers/gpu/drm/i915/i915_dma.c | 12 ++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 16 ++++++++++++++++ drivers/gpu/drm/i915/i915_params.c | 5 +++++ drivers/gpu/drm/i915/i915_params.h | 1 + 4 files changed, 34 insertions(+)
drivers/gpu/drm/i915/i915_dma.c +12 -0
@@ -370,6 +370,9 @@ static int i915_load_modeset_init(struct drm_device *dev)struct drm_i915_private *dev_priv = dev->dev_private;int ret; +if (i915_inject_fault (I915_FAULT_INIT_MODESET )) +return -ENODEV ; + ret =intel_bios_init (dev_priv);if (ret)DRM_INFO ("failed to find VBIOS tables\n" );
@@ -950,6 +953,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,struct intel_device_info *device_info;int ret =0 ; +if (i915_inject_fault (I915_FAULT_INIT_EARLY )) +return -ENODEV ; + dev_priv->dev = dev;/* Setup the write-once "constant" device info */
@@ -1064,6 +1070,9 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv)struct drm_device *dev = dev_priv->dev;int ret; +if (i915_inject_fault (I915_FAULT_INIT_MMIO )) +return -ENODEV ; +if (i915_get_bridge_dev (dev))return -EIO ;
@@ -1107,6 +1116,9 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)uint32_t aperture_size;int ret; +if (i915_inject_fault (I915_FAULT_INIT_HW )) +return -ENODEV ; +intel_device_info_runtime_init (dev); ret =i915_gem_gtt_init (dev);
drivers/gpu/drm/i915/i915_drv.h +16 -0
@@ -98,6 +98,22 @@#define I915_STATE_WARN_ON (x) \ I915_STATE_WARN((x), "%s", "WARN_ON(" __stringify(x) ")") +#define I915_FAULT_INIT_EARLY 1 +#define I915_FAULT_INIT_MMIO 2 +#define I915_FAULT_INIT_HW 4 +#define I915_FAULT_INIT_MODESET 8 + +static inline bool i915_inject_fault (unsigned int fault_mask) +{ +if (i915.inject_fault & fault_mask) { +DRM_DEBUG_DRIVER ("Injecting fault %08x\n" , fault_mask); + +return true; + } + +return false; +} +static inline const char *yesno (bool v) {return v ?"yes" :"no" ;
drivers/gpu/drm/i915/i915_params.c +5 -0
@@ -56,6 +56,7 @@ struct i915_params i915 __read_mostly = { .edp_vswing = 0, .enable_guc_submission = false, .guc_log_level = -1, + .inject_fault = 0, };module_param_named (modeset, i915.modeset, int,0400 );
@@ -201,3 +202,7 @@ MODULE_PARM_DESC(enable_guc_submission, "Enable GuC submission (default:false)")module_param_named (guc_log_level, i915.guc_log_level, int,0400 );MODULE_PARM_DESC (guc_log_level,"GuC firmware logging level (-1:disabled (default), 0-3:enabled)" ); + +module_param_named (inject_fault, i915.inject_fault, uint,0600 ); +MODULE_PARM_DESC (inject_fault, +"Force an error at selected points (0:disabled, 1:INIT_EARLY, 2:INIT_MMIO, 4:INIT_HW)" );
drivers/gpu/drm/i915/i915_params.h +1 -0
@@ -59,6 +59,7 @@ struct i915_params {bool enable_guc_submission;bool verbose_state_checks;bool nuclear_pageflip; +unsigned int inject_fault; };extern struct i915_params i915__read_mostly ;
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 1/1 drm/i915: Add fault injection support
Add support for forcing an error at selected places in the driver. As an example add 4 options to fail during driver loading.
Requested by Chris.
v2:
- Add fault point for modeset initialization
- Print debug message when injecting an error
v3:
- Rename inject_fault to inject_load_failure, rename the related macros and helper accordingly (Chris)
---
[This depends on
https://lists.freedesktop.org/archives/intel-gfx/2016-March/089597.html]
drivers/gpu/drm/i915/i915_dma.c | 12 ++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 16 ++++++++++++++++ drivers/gpu/drm/i915/i915_params.c | 5 +++++ drivers/gpu/drm/i915/i915_params.h | 1 + 4 files changed, 34 insertions(+)
drivers/gpu/drm/i915/i915_dma.c +12 -0
@@ -370,6 +370,9 @@ static int i915_load_modeset_init(struct drm_device *dev)struct drm_i915_private *dev_priv = dev->dev_private;int ret; +if (i915_inject_load_failure (I915_FAIL_INIT_MODESET )) +return -ENODEV ; + ret =intel_bios_init (dev_priv);if (ret)DRM_INFO ("failed to find VBIOS tables\n" );
@@ -950,6 +953,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,struct intel_device_info *device_info;int ret =0 ; +if (i915_inject_load_failure (I915_FAIL_INIT_EARLY )) +return -ENODEV ; + dev_priv->dev = dev;/* Setup the write-once "constant" device info */
@@ -1064,6 +1070,9 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv)struct drm_device *dev = dev_priv->dev;int ret; +if (i915_inject_load_failure (I915_FAIL_INIT_MMIO )) +return -ENODEV ; +if (i915_get_bridge_dev (dev))return -EIO ;
@@ -1107,6 +1116,9 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv)uint32_t aperture_size;int ret; +if (i915_inject_load_failure (I915_FAIL_INIT_HW )) +return -ENODEV ; +intel_device_info_runtime_init (dev); ret =i915_gem_gtt_init (dev);
drivers/gpu/drm/i915/i915_drv.h +16 -0
@@ -98,6 +98,22 @@#define I915_STATE_WARN_ON (x) \ I915_STATE_WARN((x), "%s", "WARN_ON(" __stringify(x) ")") +#define I915_FAIL_INIT_EARLY BIT(0) +#define I915_FAIL_INIT_MMIO BIT(1) +#define I915_FAIL_INIT_HW BIT(2) +#define I915_FAIL_INIT_MODESET BIT(3) + +static inline bool i915_inject_load_failure (unsigned int fail_mask) +{ +if (i915.inject_load_failure & fail_mask) { +DRM_DEBUG_DRIVER ("Injecting failure %08x\n" , fail_mask); + +return true; + } + +return false; +} +static inline const char *yesno (bool v) {return v ?"yes" :"no" ;
drivers/gpu/drm/i915/i915_params.c +5 -0
@@ -56,6 +56,7 @@ struct i915_params i915 __read_mostly = { .edp_vswing = 0, .enable_guc_submission = false, .guc_log_level = -1, + .inject_load_failure = 0, };module_param_named (modeset, i915.modeset, int,0400 );
@@ -201,3 +202,7 @@ MODULE_PARM_DESC(enable_guc_submission, "Enable GuC submission (default:false)")module_param_named (guc_log_level, i915.guc_log_level, int,0400 );MODULE_PARM_DESC (guc_log_level,"GuC firmware logging level (-1:disabled (default), 0-3:enabled)" ); + +module_param_named (inject_load_failure, i915.inject_load_failure, uint,0600 ); +MODULE_PARM_DESC (inject_load_failure, +"Force an error at selected points (0:disabled, 0x1:INIT_EARLY, 0x2:INIT_MMIO, 0x4:INIT_HW, 0x8:INIT_MODESET)" );
drivers/gpu/drm/i915/i915_params.h +1 -0
@@ -59,6 +59,7 @@ struct i915_params {bool enable_guc_submission;bool verbose_state_checks;bool nuclear_pageflip; +unsigned int inject_load_failure; };extern struct i915_params i915__read_mostly ;
Signature · 6 lines
2.5.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Joonas Lahtinen
maintainerQuoted 128 lines · Imre Deak
Add support for forcing an error at selected places in the driver. As an
example add 4 options to fail during driver loading.Requested by Chris.
v2:
- Add fault point for modeset initialization
- Print debug message when injecting an error
v3:
- Renameinject_faulttoinject_load_failure, rename the related macros
and helper accordingly (Chris)CC Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>---
[This depends on
https://lists.freedesktop.org/archives/intel-gfx/2016-March/089597.h
tml]drivers/gpu/drm/i915/i915_dma.c | 12 ++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 16 ++++++++++++++++
drivers/gpu/drm/i915/i915_params.c| 5 +++++drivers/gpu/drm/i915/i915_params.h| 1 +4 files changed, 34 insertions(+)drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -370,6 +370,9 @@ static int i915_load_modeset_init(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; int ret;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_MODESET )) +return -ENODEV ; +ret =
intel_bios_init(dev_priv);
if (ret)
DRM_INFO("failed to find VBIOS tables\n");drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -950,6 +953,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv, struct intel_device_info *device_info; int ret = 0;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_EARLY )) +return -ENODEV ; +
dev_priv->dev= dev;
/* Setup the write-once "constant" device info */drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -1064,6 +1070,9 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv) struct drm_device *dev = dev_priv->dev; int ret;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_MMIO )) +return -ENODEV ; +if (
i915_get_bridge_dev(dev))
return-EIO;drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -1107,6 +1116,9 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv) uint32_t aperture_size; int ret;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_HW )) +return -ENODEV ; +
intel_device_info_runtime_init(dev);
ret =i915_gem_gtt_init(dev);drivers/gpu/drm/i915/i915_drv.h +0
-0@@ -98,6 +98,22 @@ #define I915_STATE_WARN_ON(x) \ I915_STATE_WARN((x), "%s", "WARN_ON(" __stringify(x) ")")drivers/gpu/drm/i915/i915_drv.h +9
-0+#define I915_FAIL_INIT_EARLY BIT(0) +#define I915_FAIL_INIT_MMIO BIT(1) +#define I915_FAIL_INIT_HW BIT(2) +#define I915_FAIL_INIT_MODESET BIT(3) + +static inline bool i915_inject_load_failure (unsigned int fail_mask) +{ +if (i915.inject_load_failure & fail_mask) { + DRM_DEBUG_DRIVER("Injecting failure %08x\n",
fail_mask);drivers/gpu/drm/i915/i915_drv.h +7
-0+ +return true; + } + +return false; +} +static inline const char *yesno(bool v)
{
return v ? "yes" : "no";drivers/gpu/drm/i915/i915_params.c +1
-0@@ -56,6 +56,7 @@ struct i915_params i915 __read_mostly = { .edp_vswing = 0, .enable_guc_submission = false, .guc_log_level = -1, + .inject_load_failure = 0, };
module_param_named(modeset, i915.modeset, int, 0400);drivers/gpu/drm/i915/i915_params.c +0
-0@@ -201,3 +202,7 @@ MODULE_PARM_DESC(enable_guc_submission, "Enable GuC submission (default:false)") module_param_named(guc_log_level, i915.guc_log_level, int, 0400); MODULE_PARM_DESC(guc_log_level,"GuC firmware logging level (-1:disabled (default), 0- 3:enabled)");
drivers/gpu/drm/i915/i915_params.c +2
-0+ +module_param_named(inject_load_failure, i915.inject_load_failure,
Quoted 1 line
uint, 0600);
This most definitely should be module_param_named_unsafe.
I think I'd also hope to see some kind of compile time option to enable this. I don't think average user wants this by default.
drivers/gpu/drm/i915/i915_params.h struct i915_params { · line 62
Quoted 13 lines
drivers/gpu/drm/i915/i915_params.c +2
-0+MODULE_PARM_DESC(inject_load_failure, + "Force an error at selected points (0:disabled,0x1:INIT_EARLY, 0x2:INIT_MMIO, 0x4:INIT_HW, 0x8:INIT_MODESET)");
drivers/gpu/drm/i915/i915_params.h +1
-0@@ -59,6 +59,7 @@ struct i915_params { bool enable_guc_submission; bool verbose_state_checks; bool nuclear_pageflip; +unsigned int inject_load_failure;
Duh, add it above the bools, this struct was cleaned once already :P
Regards, Joonas
Quoted 3 lines
drivers/gpu/drm/i915/i915_params.h +0
-0};
extern structi915_paramsi915__read_mostly;
Signature · 8 lines
Joonas Lahtinen Open Source Technology Center Intel Corporation _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 8 lines · Joonas Lahtinen
Quoted 2 lines · Imre Deak
+module_param_named(inject_load_failure, i915.inject_load_failure, uint, 0600);
This most definitely should be module_param_named_unsafe.
I think I'd also hope to see some kind of compile time option to enable this. I don't think average user wants this by default.
That's not a bad idea. We can get rid of most of the dangerous params this way. On the other hand, we need to keep the debug/quirk params around so that we can test options in the wild (bug reporters). -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
Quoted 5 lines · Chris Wilson
Quoted 4 lines · Joonas Lahtinen
Quoted 2 lines · Imre Deak
+module_param_named(inject_load_failure, i915.inject_load_failure, uint, 0600);
Quoted 8 lines
Quoted 4 lines
This most definitely should be module_param_named_unsafe.
I think I'd also hope to see some kind of compile time option to enable this. I don't think average user wants this by default.
That's not a bad idea. We can get rid of most of the dangerous params this way. On the other hand, we need to keep the debug/quirk params around so that we can test options in the wild (bug reporters).
Yeah, most of the _unsafe stuff is _really_ useful for quick bug triage. Definitely don't want to disable those by default. Maybe an i915.fry_my_kernel option? But even that is too confusing I think for testers.
-Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 127 lines · Joonas Lahtinen
Quoted 126 lines · Imre Deak
Add support for forcing an error at selected places in the driver. As
an
example add 4 options to fail during driver loading.Requested by Chris.
v2:
- Add fault point for modeset initialization
- Print debug message when injecting an error
v3:
- Renameinject_faulttoinject_load_failure, rename the related macros
and helper accordingly (Chris)CC Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>---
[This depends on
https://lists.freedesktop.org/archives/intel-gfx/2016-March/089597
.h
tml]drivers/gpu/drm/i915/i915_dma.c | 12 ++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 16 ++++++++++++++++
drivers/gpu/drm/i915/i915_params.c| 5 +++++drivers/gpu/drm/i915/i915_params.h| 1 +4 files changed, 34 insertions(+)drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -370,6 +370,9 @@ static int i915_load_modeset_init(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; int ret;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_MODESET )) +return -ENODEV ; +ret =
intel_bios_init(dev_priv);
if (ret)
DRM_INFO("failed to find VBIOS tables\n");drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -950,6 +953,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv, struct intel_device_info *device_info; int ret = 0;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_EARLY )) +return -ENODEV ; +
dev_priv->dev= dev;
/* Setup the write-once "constant" device info */drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -1064,6 +1070,9 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv) struct drm_device *dev = dev_priv->dev; int ret;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_MMIO )) +return -ENODEV ; +if (
i915_get_bridge_dev(dev))
return-EIO;drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -1107,6 +1116,9 @@ static int i915_driver_init_hw(struct drm_i915_private *dev_priv) uint32_t aperture_size; int ret;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_HW )) +return -ENODEV ; +
intel_device_info_runtime_init(dev);
ret =i915_gem_gtt_init(dev);drivers/gpu/drm/i915/i915_drv.h +0
-0@@ -98,6 +98,22 @@ #define I915_STATE_WARN_ON(x) \ I915_STATE_WARN((x), "%s", "WARN_ON(" __stringify(x) ")")drivers/gpu/drm/i915/i915_drv.h +6
-0+#define I915_FAIL_INIT_EARLY BIT(0) +#define I915_FAIL_INIT_MMIO BIT(1) +#define I915_FAIL_INIT_HW BIT(2) +#define I915_FAIL_INIT_MODESET BIT(3) + +static inline bool i915_inject_load_failure(unsigned int
fail_mask)drivers/gpu/drm/i915/i915_drv.h +3
-0+{ +if (i915.inject_load_failure & fail_mask) { + DRM_DEBUG_DRIVER("Injecting failure %08x\n",
fail_mask);drivers/gpu/drm/i915/i915_drv.h +7
-0+ +return true; + } + +return false; +} +static inline const char *yesno(bool v)
{
return v ? "yes" : "no";drivers/gpu/drm/i915/i915_params.c +1
-0@@ -56,6 +56,7 @@ struct i915_params i915 __read_mostly = { .edp_vswing = 0, .enable_guc_submission = false, .guc_log_level = -1, + .inject_load_failure = 0, };
module_param_named(modeset, i915.modeset, int, 0400);drivers/gpu/drm/i915/i915_params.c +0
-0@@ -201,3 +202,7 @@ MODULE_PARM_DESC(enable_guc_submission, "Enable GuC submission (default:false)") module_param_named(guc_log_level, i915.guc_log_level, int, 0400);
Quoted 8 lines
Quoted 6 lines
drivers/gpu/drm/i915/i915_params.c +0
-0MODULE_PARM_DESC(guc_log_level,"GuC firmware logging level (-1:disabled (default), 0- 3:enabled)");
drivers/gpu/drm/i915/i915_params.c +2
-0+ +module_param_named(inject_load_failure, i915.inject_load_failure,uint, 0600);
This most definitely should be module_param_named_unsafe.
Ok, will change that.
Quoted 4 lines
I think I'd also hope to see some kind of compile time option to enable
this. I don't think average user wants this by default.
I agree with Chris and Daniel, so I'll leave this out for now. It can be added later in any case.
Quoted 16 lines
Quoted 10 lines
drivers/gpu/drm/i915/i915_params.c +2
-0+MODULE_PARM_DESC(inject_load_failure, + "Force an error at selected points (0:disabled,0x1:INIT_EARLY, 0x2:INIT_MMIO, 0x4:INIT_HW, 0x8:INIT_MODESET)");
drivers/gpu/drm/i915/i915_params.h +0
-0drivers/gpu/drm/i915/i915_params.h struct i915_params { · line 62
Quoted 4 lines
drivers/gpu/drm/i915/i915_params.h +1
-0@@ -59,6 +59,7 @@ struct i915_params { bool enable_guc_submission; bool verbose_state_checks; bool nuclear_pageflip; +unsigned int inject_load_failure;Duh, add it above the bools, this struct was cleaned once already :P
Ah, missed the code comment about this. Will move it thanks.
--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 50 lines · Imre Deak
Add support for forcing an error at selected places in the driver. As an example add 4 options to fail during driver loading.
Requested by Chris.
v2:
- Add fault point for modeset initialization
- Print debug message when injecting an error
v3:
- Renameinject_faulttoinject_load_failure, rename the related macros and helper accordingly (Chris)CC Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>---
[This depends on
https://lists.freedesktop.org/archives/intel-gfx/2016-March/089597.html]drivers/gpu/drm/i915/i915_dma.c | 12 ++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 16 ++++++++++++++++ drivers/gpu/drm/i915/i915_params.c | 5 +++++ drivers/gpu/drm/i915/i915_params.h | 1 + 4 files changed, 34 insertions(+)drivers/gpu/drm/i915/i915_dma.c +6
-0@@ -370,6 +370,9 @@ static int i915_load_modeset_init(struct drm_device *dev)struct drm_i915_private *dev_priv = dev->dev_private;int ret; +if (i915_inject_load_failure (I915_FAIL_INIT_MODESET )) +return -ENODEV ; + ret =intel_bios_init (dev_priv);if (ret)DRM_INFO ("failed to find VBIOS tables\n" );@@ -950,6 +953,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv,struct intel_device_info *device_info;int ret =0 ; +if (i915_inject_load_failure (I915_FAIL_INIT_EARLY )) +return -ENODEV ; + dev_priv->dev = dev;/* Setup the write-once "constant" device info */
drivers/gpu/drm/i915/i915_dma.c static int i915_driver_init_mmio(struct drm_i915_private *dev_priv) · line 1077
Quoted 8 lines
drivers/gpu/drm/i915/i915_dma.c +3
-0@@ -1064,6 +1070,9 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv)struct drm_device *dev = dev_priv->dev;int ret; +if (i915_inject_load_failure (I915_FAIL_INIT_MMIO )) +return -ENODEV ; +if (i915_get_bridge_dev (dev))return -EIO ;
Ok, thought a bit more about this, and if we really want to check all the load failure paths then this will become extremely verbose. Since we'd need to have a bitflag for every return -EFAIL. So maybe another look at lib/fault-inject.c is warranted, plus then some macro magic that we could to wrap _all_ the checks in our load code like this:
if (i915_load_failure(i915_get_bridge_dev(dev)))
return -EIO;
i915_load_failure ofc needs to fail if it's argument is false, but it could also increment a counter every time it's called and then use that counter to inquire the fault injection framework with should_fail(i915_load_failures, counter). Abusing a counter for the size would allow us to easily restrict fault injection to a certain range of faults. We could also expose the final value of that counter in debugfs, so that the igt can tune itself.
Anyway, this is kinda a big plan, but the part I think we should ponder is how to make the fault injection less noise and intrusive. A macro to wrap existing checks seems like a good idea.
-Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 62 lines · Daniel Vetter
Quoted 61 lines · Imre Deak
Add support for forcing an error at selected places in the driver. As an
example add 4 options to fail during driver loading.Requested by Chris.
v2:
- Add fault point for modeset initialization
- Print debug message when injecting an error
v3:
- Renameinject_faulttoinject_load_failure, rename the related macros
and helper accordingly (Chris)CC Chris Wilson <[email protected]>Signed-off-by Imre Deak <[email protected]>---
[This depends on
https://lists.freedesktop.org/archives/intel-gfx/2016-March/089597
.html]drivers/gpu/drm/i915/i915_dma.c | 12 ++++++++++++ drivers/gpu/drm/i915/i915_drv.h | 16 ++++++++++++++++
drivers/gpu/drm/i915/i915_params.c| 5 +++++drivers/gpu/drm/i915/i915_params.h| 1 +4 files changed, 34 insertions(+)drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -370,6 +370,9 @@ static int i915_load_modeset_init(struct drm_device *dev) struct drm_i915_private *dev_priv = dev->dev_private; int ret;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_MODESET )) +return -ENODEV ; +ret =
intel_bios_init(dev_priv);
if (ret)
DRM_INFO("failed to find VBIOS tables\n");drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -950,6 +953,9 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv, struct intel_device_info *device_info; int ret = 0;drivers/gpu/drm/i915/i915_dma.c +3
-0+if (i915_inject_load_failure (I915_FAIL_INIT_EARLY )) +return -ENODEV ; +
dev_priv->dev= dev;
/* Setup the write-once "constant" device info */drivers/gpu/drm/i915/i915_dma.c +0
-0@@ -1064,6 +1070,9 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv) struct drm_device *dev = dev_priv->dev; int ret;drivers/gpu/drm/i915/i915_dma.c +1
-0+if (i915_inject_load_failure (I915_FAIL_INIT_MMIO ))
Quoted 8 lines
Quoted 4 lines
drivers/gpu/drm/i915/i915_dma.c +2
-0+return -ENODEV ; +if (
i915_get_bridge_dev(dev))
return-EIO;Ok, thought a bit more about this, and if we really want to check all the load failure paths then this will become extremely verbose. Since we'd need to have a bitflag for every return
-EFAIL.
I'm not sure if you want to check all failure paths, I think for that the existing failslab etc. mechanisms are better suited. This new option would be used at relatively few well defined places. The option is a mask since Chris wanted the possibility to mix failures (which makes sense when injecting a non-fatal failure somewhere). If he doesn't insist on that possibility I can convert the mask option to a counter based one identifying a single failure spot instead as you suggest. Chris?
Quoted 6 lines
So maybe another look at
lib/fault-inject.cis warranted, plus then some macro magic that we could to wrap _all_ the checks in our load code like this:if (i915_load_failure(i915_get_bridge_dev(dev))) return
-EIO;
The above will not work when we want to propagate the error and I haven't found a nice way to do that with such a helper taking the function call as argument. I could move the check to the call site instead of having it inside the called function, if that's what you'd prefer:
if ((ret = i915_load_failure(...)) < 0 ||
(ret = i915_driver_init_mmio(...)) < 0) return ret;
Although I'm not sure if this makes things clearer. Other suggestions are welcome.
Quoted 7 lines
i915_load_failureofc needs to fail if it's argument is false, but it could also increment a counter every time it's called and then use that counter to inquire the fault injection framework withshould_fail(i915_load_failures, counter). Abusing a counter for the size would allow us to easily restrict fault injection to a certain range of faults. We could also expose the final value of that counter in debugfs, so that the igt can tune itself.
Yes, I could do this provided Chris doesn't oppose the idea of the counter based approach.
Quoted 3 lines
Anyway, this is kinda a big plan, but the part I think we should ponder is how to make the fault injection less noise and intrusive. A macro to wrap existing checks seems like a good idea.
I can switch to using a counter instead of the mask and if there is a good suggestion about the macro I can use that instead.
--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 8 lines · Imre Deak
I'm not sure if you want to check all failure paths, I think for that the existing failslab etc. mechanisms are better suited. This new option would be used at relatively few well defined places. The option is a mask since Chris wanted the possibility to mix failures (which makes sense when injecting a non-fatal failure somewhere). If he doesn't insist on that possibility I can convert the mask option to a counter based one identifying a single failure spot instead as you suggest. Chris?
We can extend the counter mechanism by having multiple counters behind i915.inject_load_failure (i.e. =gem:4,driver:10,modeset:1) so extensibility for more testing seems fine.
-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
Joonas Lahtinen
maintainerQuoted 10 lines · Chris Wilson
Quoted 9 lines · Imre Deak
I'm not sure if you want to check all failure paths, I think for that the existing failslab etc. mechanisms are better suited. This new option would be used at relatively few well defined places. The option is a mask since Chris wanted the possibility to mix failures (which makes sense when injecting a non-fatal failure somewhere). If he doesn't insist on that possibility I can convert the mask option to a counter based one identifying a single failure spot instead as you suggest. Chris?
Quoted 2 lines
We can extend the counter mechanism by having multiple counters behind i915.inject_load_failure (i.e. =gem:4,driver:10,modeset:1)
Now that there's a series to split down the init functions nicely, one could use the function names directly. By stripping parts of it if needed to shorten them.
Regards, Joonas
Quoted 3 lines
so extensibility for more testing seems fine.
-Chris
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
Quoted 11 lines · Joonas Lahtinen
Quoted 10 lines · Chris Wilson
Quoted 9 lines · Imre Deak
I'm not sure if you want to check all failure paths, I think for that the existing failslab etc. mechanisms are better suited. This new option would be used at relatively few well defined places. The option is a mask since Chris wanted the possibility to mix failures (which makes sense when injecting a non-fatal failure somewhere). If he doesn't insist on that possibility I can convert the mask option to a counter based one identifying a single failure spot instead as you suggest. Chris?
Quoted 6 lines
Quoted 2 lines
We can extend the counter mechanism by having multiple counters behind i915.inject_load_failure (i.e. =gem:4,driver:10,modeset:1)
Now that there's a series to split down the init functions nicely, one could use the function names directly. By stripping parts of it if needed to shorten them.
That's nice. Advantage for using counters is that we can write a test to iterate over the first thousand and have it run against future faultpointers. Advantage for using names is that it is readable and easily extensible.
What we could do for testing is record the available fault injection points for debugfs and so have the tests automatically adjust (given a working start point). Overkill?
-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
Quoted 34 lines · Chris Wilson
Quoted 18 lines · Joonas Lahtinen
Quoted 17 lines · Chris Wilson
Quoted 16 lines · Imre Deak
I'm not sure if you want to check all failure paths, I think for that
the existing failslab etc. mechanisms are better suited. This new
option would be used at relatively few well defined places. The option
is a mask since Chris wanted the possibility to mix failures (which
makes sense when injecting a non-fatal failure somewhere). If he
doesn't insist on that possibility I can convert the mask option to a
counter based one identifying a single failure spot instead as you
suggest. Chris?Quoted 8 lines
Quoted 3 lines
We can extend the counter mechanism by having multiple counters behind
i915.inject_load_failure (i.e. =gem:4,driver:10,modeset:1)Now that there's a series to split down the init functions nicely, one
could use the function names directly. By stripping parts of it if needed to shorten them.That's nice. Advantage for using counters is that we can write a test to
iterate over the first thousand and have it run against future faultpointers. Advantage for using names is that it is readable and easily extensible.
Quoted 4 lines
What we could do for testing is record the available fault injection points for debugfs and so have the tests automatically adjust (given a
working start point). Overkill?
I'd prefer if we could fine grain things once we have more failure points. Atm the ones we defined cause driver init to fail, so separate counters wouldn't bring us much. I suggest that we keep the module option as a simple integer for now and extend it later to support the above tagged counter mechanism and debugfs interface. Here is what I planned to submit to the list:
https://github.com/ideak/linux/commits/driver_init_refactor
Comments?
--Imre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 40 lines · Imre Deak
Quoted 34 lines · Chris Wilson
Quoted 18 lines · Joonas Lahtinen
Quoted 17 lines · Chris Wilson
Quoted 16 lines · Imre Deak
I'm not sure if you want to check all failure paths, I think for that
the existing failslab etc. mechanisms are better suited. This new
option would be used at relatively few well defined places. The option
is a mask since Chris wanted the possibility to mix failures (which
makes sense when injecting a non-fatal failure somewhere). If he
doesn't insist on that possibility I can convert the mask option to a
counter based one identifying a single failure spot instead as you
suggest. Chris?Quoted 8 lines
Quoted 3 lines
We can extend the counter mechanism by having multiple counters behind
i915.inject_load_failure (i.e. =gem:4,driver:10,modeset:1)Now that there's a series to split down the init functions nicely, one
could use the function names directly. By stripping parts of it if needed to shorten them.That's nice. Advantage for using counters is that we can write a test to
iterate over the first thousand and have it run against future faultpointers. Advantage for using names is that it is readable and easily extensible.Quoted 4 lines
What we could do for testing is record the available fault injection points for debugfs and so have the tests automatically adjust (given a
working start point). Overkill?
Quoted 8 lines
I'd prefer if we could fine grain things once we have more failure points. Atm the ones we defined cause driver init to fail, so separate counters wouldn't bring us much. I suggest that we keep the module option as a simple integer for now and extend it later to support the above tagged counter mechanism and debugfs interface. Here is what I planned to submit to the list:
Move the inline into __i915_inject_load_failure, and yes I am happy with the simple first step. Please use __func__ or __FUNCTION__ though as that is more descriptive than file, especially when they move. -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
Quoted 64 lines · Chris Wilson
Quoted 63 lines · Imre Deak
Quoted 45 lines · Chris Wilson
Quoted 29 lines · Joonas Lahtinen
Quoted 23 lines · Chris Wilson
Quoted 14 lines · Imre Deak
I'm not sure if you want to check all failure paths, I think
for that
the existing failslab etc. mechanisms are better suited. This
new
option would be used at relatively few well defined places. The
option
is a mask since Chris wanted the possibility to mix failures
(which
makes sense when injecting a non-fatal failure somewhere).Quoted 8 lines
If
he
doesn't insist on that possibility I can convert the mask option to a
counter based one identifying a single failure spot instead as
you
suggest. Chris?Quoted 4 lines
We can extend the counter mechanism by having multiple counters
behind
i915.inject_load_failure (i.e. =gem:4,driver:10,modeset:1)Quoted 6 lines
Now that there's a series to split down the init functions nicely,
one
could use the function names directly. By stripping parts of it if
needed to shorten them.That's nice. Advantage for using counters is that we can write a test
to
iterate over the first thousand and have it run against future faultpointers. Advantage for using names is that it is readable and
easily extensible.Quoted 6 lines
What we could do for testing is record the available fault injection
points for debugfs and so have the tests automatically adjust (given
a
working start point). Overkill?I'd prefer if we could fine grain things once we have more failure points. Atm the ones we defined cause driver init to fail, so separate
counters wouldn't bring us much. I suggest that we keep the module option as a simple integer for now and extend it later to support the
above tagged counter mechanism and debugfs interface. Here is what I
planned to submit to the list:
Quoted 3 lines
Move the inline into
__i915_inject_load_failure
hm, I have the macro in the header file to optimize for the failure_count=0 case. Or did you mean to
bring __i915_inject_load_failure to the header file too/merge it into the macro?
Quoted 3 lines
, and yes I am happy with
the simple first step. Please use__func__or__FUNCTION__though as that is more descriptive than file, especially when they move.
Ok.
--Imre
Quoted 2 lines
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 65 lines · Imre Deak
Quoted 64 lines · Chris Wilson
Quoted 63 lines · Imre Deak
Quoted 45 lines · Chris Wilson
Quoted 29 lines · Joonas Lahtinen
Quoted 23 lines · Chris Wilson
Quoted 14 lines · Imre Deak
I'm not sure if you want to check all failure paths, I think
for that
the existing failslab etc. mechanisms are better suited. This
new
option would be used at relatively few well defined places. The
option
is a mask since Chris wanted the possibility to mix failures
(which
makes sense when injecting a non-fatal failure somewhere).Quoted 8 lines
If
he
doesn't insist on that possibility I can convert the mask option to a
counter based one identifying a single failure spot instead as
you
suggest. Chris?Quoted 4 lines
We can extend the counter mechanism by having multiple counters
behind
i915.inject_load_failure (i.e. =gem:4,driver:10,modeset:1)Quoted 6 lines
Now that there's a series to split down the init functions nicely,
one
could use the function names directly. By stripping parts of it if
needed to shorten them.That's nice. Advantage for using counters is that we can write a test
to
iterate over the first thousand and have it run against future faultpointers. Advantage for using names is that it is readable and
easily extensible.Quoted 6 lines
What we could do for testing is record the available fault injection
points for debugfs and so have the tests automatically adjust (given
a
working start point). Overkill?I'd prefer if we could fine grain things once we have more failure points. Atm the ones we defined cause driver init to fail, so separate
counters wouldn't bring us much. I suggest that we keep the module option as a simple integer for now and extend it later to support the
above tagged counter mechanism and debugfs interface. Here is what I
planned to submit to the list:
Quoted 8 lines
Quoted 3 lines
Move the inline into
__i915_inject_load_failurehm, I have the macro in the header file to optimize for the failure_count=0 case. Or did you mean to
bring __i915_inject_load_failure to the header file too/merge it into the macro?
Just the extra inlined conditional need not be in the header. -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
Patchwork
CI✗ Fi.CI.BAT: failure for drm/i915: Add fault injection support (rev2)
In reply to Imre Deak== Series Details ==
== Series Details ==
Series: drm/i915: Add fault injection support (rev2)
URL : https://patchwork.freedesktop.org/series/4385/
State : failure
== Summary ==
Series 4385v2 drm/i915: Add fault injection support
2016-03-14T08:58:17.234391 http://patchwork.freedesktop.org/api/1.0/series/4385/revisions/2/mbox/
Applying: drm/i915: Add fault injection support
Using index info to reconstruct a base tree...
M drivers/gpu/drm/i915/i915_dma.c
M drivers/gpu/drm/i915/i915_drv.h
Falling back to patching base and 3-way merge... Auto-merging drivers/gpu/drm/i915/i915_drv.h
Auto-merging drivers/gpu/drm/i915/i915_dma.c
CONFLICT (content): Merge conflict in drivers/gpu/drm/i915/i915_dma.c Patch failed at 0001 drm/i915: Add fault injection support
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for drm/i915: Add fault injection support (rev3)
In reply to Imre Deak== Series Details ==
== Series Details ==
Series: drm/i915: Add fault injection support (rev3)
URL : https://patchwork.freedesktop.org/series/4385/
State : failure
== Summary ==
Series 4385v3 drm/i915: Add fault injection support
2016-03-14T14:37:24.542491 http://patchwork.freedesktop.org/api/1.0/series/4385/revisions/3/mbox/
Applying: drm/i915: Add fault injection support
Using index info to reconstruct a base tree...
M drivers/gpu/drm/i915/i915_dma.c
M drivers/gpu/drm/i915/i915_drv.h
Falling back to patching base and 3-way merge... Auto-merging drivers/gpu/drm/i915/i915_drv.h
Auto-merging drivers/gpu/drm/i915/i915_dma.c
CONFLICT (content): Merge conflict in drivers/gpu/drm/i915/i915_dma.c Patch failed at 0001 drm/i915: Add fault injection support
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx