Add support for T-HEAD C9xx PMU extensions
v5 cover Add support for T-HEAD C9xx PMU extensions
The T-HEAD C9XX cores implement functionality very similar to SSCOFPMF. Instead of implementing a separate interface into SBI as done in v1, we can add the C9XX tidbits with quite minimal overhead and leaverage the existing SBI pmu interface including giving access to the firmware counters.
The current only hickup is the detection override in sbi_hart, where I still need to find out why the MHPMCOUNTERs are not writeable at _that_ point of the boot, but with this series on top of openSBI 1.1 and the matching kernel perf patch I get reasonable results for example for:
perf stat -e cycles -e instructions -e L1-icache-load-misses -e L1-icache-loads ls
Performance counter stats for 'ls':
17119496 cycles
2867765 instructions # 0.17 insn per cycle
55929 L1-icache-load-misses # 1.61% of all L1-icache accesses
3467346 L1-icache-loads 0.030156216 seconds time elapsed
0.000000000 seconds user
0.023496000 seconds syschanges in v5:
- use csr_set, csr_clear in allwinner pmu functions
- expose hart_features struct into header
- make hart_features param for extension_init callback to allow platforms to override specific settings
changes in v4:
- use the new sbi_pmu_device structure for holding the overrides and extend it where needed
changes in v3:
- follow Atish's advice and implement an abstraction to not pollute the core pmu with cpu-specific variants
changes in v2:
- don't implement a separate interface but instead modify the sbi-pmu to allow the c9xx to use standard pmu interface
Heiko Stuebner (5):
lib: sbi_pmu: move pmu irq information into pmu itself
lib: sbi_hart: move hart_features struct to a public location
lib: sbi_platform: expose hart_features to extension_init callback
platform: generic: add extensions_init handler and platform-override
platform: generic: allwinner: add support for c9xx pmu
include/sbi/sbi_hart.h | 11 ++ include/sbi/sbi_platform.h | 8 +- include/sbi/sbi_pmu.h | 8 ++ lib/sbi/sbi_hart.c | 43 +++---- lib/sbi/sbi_pmu.c | 12 ++ platform/generic/allwinner/sun20i-d1.c | 60 +++++++++ platform/generic/include/platform_override.h | 3 + platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++ platform/generic/platform.c | 10 ++ 9 files changed, 253 insertions(+), 29 deletions(-) create mode 100644 platform/generic/include/thead_c9xx.h
Signature · 1 line
2.35.1
PATCH v5 1/5 lib: sbi_pmu: move pmu irq information into pmu itself
Don't spread checking for pmu extensions through the code but instead introduce a sbi-pmu function that other code can call to get the correct information about the existence of the pmu interrupt.
Add a sbi_pmu_device override function to allow overridung this bit as well if needed.
--- include/sbi/sbi_pmu.h | 8 ++++++++ lib/sbi/sbi_hart.c | 4 ++-- lib/sbi/sbi_pmu.c | 12 ++++++++++++ 3 files changed, 22 insertions(+), 2 deletions(-)
include/sbi/sbi_pmu.h +8 -0
@@ -73,6 +73,11 @@ struct sbi_pmu_device {* Note: 0 <= counter_index < SBI_PMU_HW_CTR_MAX */ void (*hw_counter_disable_irq)(uint32_t counter_index); + +/** +* Custom function returning the machine-specific irq-bit. +*/ +int (*hw_counter_irq_bit)(void ); };/** Get the PMU platform device */
@@ -87,6 +92,9 @@ int sbi_pmu_init(struct sbi_scratch *scratch, bool cold_boot);/** Reset PMU during hart exit */ void sbi_pmu_exit (struct sbi_scratch *scratch); +/** Return the pmu irq bit depending on extension existence */ +int sbi_pmu_irq_bit (void ); +/** * Add the hardware event to counter mapping information. This should be called * from the platform code to update the mapping table.
lib/sbi/sbi_hart.c +2 -2
@@ -19,6 +19,7 @@#include <sbi/sbi_hart.h> #include <sbi/sbi_math.h> #include <sbi/sbi_platform.h> +#include <sbi/sbi_pmu.h> #include <sbi/sbi_string.h> #include <sbi/sbi_trap.h>
@@ -208,8 +209,7 @@ static int delegate_traps(struct sbi_scratch *scratch)/* Send M-mode interrupts and most exceptions to S-mode */ interrupts =MIP_SSIP |MIP_STIP |MIP_SEIP ; -if (sbi_hart_has_extension (scratch,SBI_HART_EXT_SSCOFPMF )) - interrupts |=MIP_LCOFIP ; + interrupts |=sbi_pmu_irq_bit (); exceptions = (1U <<CAUSE_MISALIGNED_FETCH ) | (1U <<CAUSE_BREAKPOINT ) | (1U <<CAUSE_USER_ECALL );
lib/sbi/sbi_pmu.c +12 -0
@@ -344,6 +344,18 @@ skip_inhibit_update:return 0 ; } +int sbi_pmu_irq_bit (void ) +{ +struct sbi_scratch *scratch =sbi_scratch_thishart_ptr (); + +if (sbi_hart_has_extension (scratch,SBI_HART_EXT_SSCOFPMF )) +return MIP_LCOFIP ; +if (pmu_dev && pmu_dev->hw_counter_irq_bit) +return pmu_dev->hw_counter_irq_bit (); + +return 0 ; +} +static int pmu_ctr_start_fw (uint32_t cidx,uint32_t event_code,uint64_t ival,bool ival_update) {
Signature · 1 line
2.35.1
PATCH v5 1/5 lib: sbi_pmu: move pmu irq information into pmu itself
Quoted 6 lines · Heiko Stuebner
Don't spread checking for pmu extensions through the code but instead introduce a sbi-pmu function that other code can call to get the correct information about the existence of the pmu interrupt.
Quoted 6 lines
Add a
sbi_pmu_deviceoverride function to allow overridung this bit as well if needed.Reviewed-by Atish Patra <[email protected]>Reviewed-by Guo Ren <[email protected]>Signed-off-by Heiko Stuebner <[email protected]>
Looks good to me.
Regards,
Anup
Quoted 85 lines
--- include/sbi/sbi_pmu.h | 8 ++++++++ lib/sbi/sbi_hart.c | 4 ++-- lib/sbi/sbi_pmu.c | 12 ++++++++++++ 3 files changed, 22 insertions(+), 2 deletions(-)include/sbi/sbi_pmu.h +8
-0@@ -73,6 +73,11 @@ struct sbi_pmu_device {* Note: 0 <= counter_index < SBI_PMU_HW_CTR_MAX */ void (*hw_counter_disable_irq)(uint32_t counter_index); + +/** +* Custom function returning the machine-specific irq-bit. +*/ +int (*hw_counter_irq_bit)(void ); };/** Get the PMU platform device */ @@ -87,6 +92,9 @@ int sbi_pmu_init(struct sbi_scratch *scratch, bool cold_boot);/** Reset PMU during hart exit */ void sbi_pmu_exit (struct sbi_scratch *scratch); +/** Return the pmu irq bit depending on extension existence */ +int sbi_pmu_irq_bit (void ); +/** * Add the hardware event to counter mapping information. This should be called * from the platform code to update the mapping table. lib/sbi/sbi_hart.c +2
-2@@ -19,6 +19,7 @@#include <sbi/sbi_hart.h> #include <sbi/sbi_math.h> #include <sbi/sbi_platform.h> +#include <sbi/sbi_pmu.h> #include <sbi/sbi_string.h> #include <sbi/sbi_trap.h> @@ -208,8 +209,7 @@ static int delegate_traps(struct sbi_scratch *scratch)/* Send M-mode interrupts and most exceptions to S-mode */ interrupts =MIP_SSIP |MIP_STIP |MIP_SEIP ; -if (sbi_hart_has_extension (scratch,SBI_HART_EXT_SSCOFPMF )) - interrupts |=MIP_LCOFIP ; + interrupts |=sbi_pmu_irq_bit (); exceptions = (1U <<CAUSE_MISALIGNED_FETCH ) | (1U <<CAUSE_BREAKPOINT ) | (1U <<CAUSE_USER_ECALL );lib/sbi/sbi_pmu.c +12
-1@@ -344,6 +344,18 @@ skip_inhibit_update:return 0 ; } +int sbi_pmu_irq_bit (void ) +{ +struct sbi_scratch *scratch =sbi_scratch_thishart_ptr (); + +if (sbi_hart_has_extension (scratch,SBI_HART_EXT_SSCOFPMF )) +return MIP_LCOFIP ; +if (pmu_dev && pmu_dev->hw_counter_irq_bit) +return pmu_dev->hw_counter_irq_bit (); + +return 0 ; +} +static int pmu_ctr_start_fw (uint32_t cidx,uint32_t event_code,uint64_t ival,bool ival_update) { --2.35.1
--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
PATCH v5 2/5 lib: sbi_hart: move hart_features struct to a public location
In reply to Heiko StuebnerPlatforms may need to override auto-detected hart features in their override functions. So move the hart_features struct to the sbi_hart.h header allowing us to pass it over to platform-handlers.
--- include/sbi/sbi_hart.h | 11 +++++++++++ lib/sbi/sbi_hart.c | 36 +++++++++++++----------------------- 2 files changed, 24 insertions(+), 23 deletions(-)
include/sbi/sbi_hart.h +11 -0
@@ -41,6 +41,17 @@ enum sbi_hart_extensions { SBI_HART_EXT_MAX, }; +struct sbi_hart_features { +bool detected; +int priv_version; +unsigned long extensions; +unsigned int pmp_count; +unsigned int pmp_addr_bits; +unsigned long pmp_gran; +unsigned int mhpm_count; +unsigned int mhpm_bits; +}; +struct sbi_scratch ;int sbi_hart_reinit (struct sbi_scratch *scratch);
lib/sbi/sbi_hart.c +13 -23
@@ -28,16 +28,6 @@ extern void __sbi_expected_trap_hext(void);void (*sbi_hart_expected_trap)(void ) = &__sbi_expected_trap; -struct hart_features { -bool detected; -int priv_version; -unsigned long extensions; -unsigned int pmp_count; -unsigned int pmp_addr_bits; -unsigned long pmp_gran; -unsigned int mhpm_count; -unsigned int mhpm_bits; -};static unsigned long hart_features_offset;static void mstatus_init (struct sbi_scratch *scratch)
@@ -254,7 +244,7 @@ void sbi_hart_delegation_dump(struct sbi_scratch *scratch,unsigned int sbi_hart_mhpm_count (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->mhpm_count;
@@ -262,7 +252,7 @@ unsigned int sbi_hart_mhpm_count(struct sbi_scratch *scratch)unsigned int sbi_hart_pmp_count (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->pmp_count;
@@ -270,7 +260,7 @@ unsigned int sbi_hart_pmp_count(struct sbi_scratch *scratch)unsigned long sbi_hart_pmp_granularity (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->pmp_gran;
@@ -278,7 +268,7 @@ unsigned long sbi_hart_pmp_granularity(struct sbi_scratch *scratch)unsigned int sbi_hart_pmp_addrbits (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->pmp_addr_bits;
@@ -286,7 +276,7 @@ unsigned int sbi_hart_pmp_addrbits(struct sbi_scratch *scratch)unsigned int sbi_hart_mhpm_bits (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->mhpm_bits;
@@ -336,7 +326,7 @@ int sbi_hart_pmp_configure(struct sbi_scratch *scratch)int sbi_hart_priv_version (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->priv_version;
@@ -346,7 +336,7 @@ void sbi_hart_get_priv_version_str(struct sbi_scratch *scratch, char *version_str, int nvstr) { char *temp; -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);switch (hfeatures->priv_version) {
@@ -368,7 +358,7 @@ void sbi_hart_get_priv_version_str(struct sbi_scratch *scratch, }static inline void __sbi_hart_update_extension ( -struct hart_features *hfeatures, +struct sbi_hart_features *hfeatures,enum sbi_hart_extensions ext,bool enable) {
@@ -389,7 +379,7 @@ void sbi_hart_update_extension(struct sbi_scratch *scratch, enum sbi_hart_extensions ext, bool enable) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);__sbi_hart_update_extension (hfeatures, ext, enable);
@@ -405,7 +395,7 @@ void sbi_hart_update_extension(struct sbi_scratch *scratch,bool sbi_hart_has_extension (struct sbi_scratch *scratch,enum sbi_hart_extensions ext) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);if (hfeatures->extensions &BIT (ext))
@@ -453,7 +443,7 @@ static inline char *sbi_hart_extension_id2string(int ext)void sbi_hart_get_extensions_str (struct sbi_scratch *scratch,char *extensions_str,int nestr) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);int offset =0 , ext =0 ;char *temp;
@@ -539,7 +529,7 @@ static int hart_pmu_get_allowed_bits(void)static int hart_detect_features (struct sbi_scratch *scratch) {struct sbi_trap_info trap = {0 }; -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);unsigned long val, oldval;int rc;
@@ -718,7 +708,7 @@ int sbi_hart_init(struct sbi_scratch *scratch, bool cold_boot) sbi_hart_expected_trap = &__sbi_expected_trap_hext; hart_features_offset =sbi_scratch_alloc_offset ( -sizeof (struct hart_features )); +sizeof (struct sbi_hart_features ));if (!hart_features_offset)return SBI_ENOMEM ; }
Signature · 1 line
2.35.1
PATCH v5 2/5 lib: sbi_hart: move hart_features struct to a public location
Quoted 8 lines · Heiko Stuebner
Platforms may need to override auto-detected hart features in their override functions. So move the
hart_featuresstruct to thesbi_hart.hheader allowing us to pass it over to platform-handlers.Suggested-by Atish Patra <[email protected]>Signed-off-by Heiko Stuebner <[email protected]>
Looks good to me.
Regards,
Anup
Quoted 173 lines
--- include/sbi/sbi_hart.h | 11 +++++++++++ lib/sbi/sbi_hart.c | 36 +++++++++++++----------------------- 2 files changed, 24 insertions(+), 23 deletions(-)include/sbi/sbi_hart.h +11
-0@@ -41,6 +41,17 @@ enum sbi_hart_extensions { SBI_HART_EXT_MAX, }; +struct sbi_hart_features { +bool detected; +int priv_version; +unsigned long extensions; +unsigned int pmp_count; +unsigned int pmp_addr_bits; +unsigned long pmp_gran; +unsigned int mhpm_count; +unsigned int mhpm_bits; +}; +struct sbi_scratch ;int sbi_hart_reinit (struct sbi_scratch *scratch);lib/sbi/sbi_hart.c +13
-24@@ -28,16 +28,6 @@ extern void __sbi_expected_trap_hext(void);void (*sbi_hart_expected_trap)(void ) = &__sbi_expected_trap; -struct hart_features { -bool detected; -int priv_version; -unsigned long extensions; -unsigned int pmp_count; -unsigned int pmp_addr_bits; -unsigned long pmp_gran; -unsigned int mhpm_count; -unsigned int mhpm_bits; -};static unsigned long hart_features_offset;static void mstatus_init (struct sbi_scratch *scratch)@@ -254,7 +244,7 @@ void sbi_hart_delegation_dump(struct sbi_scratch *scratch,unsigned int sbi_hart_mhpm_count (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->mhpm_count;@@ -262,7 +252,7 @@ unsigned int sbi_hart_mhpm_count(struct sbi_scratch *scratch)unsigned int sbi_hart_pmp_count (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->pmp_count;@@ -270,7 +260,7 @@ unsigned int sbi_hart_pmp_count(struct sbi_scratch *scratch)unsigned long sbi_hart_pmp_granularity (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->pmp_gran;@@ -278,7 +268,7 @@ unsigned long sbi_hart_pmp_granularity(struct sbi_scratch *scratch)unsigned int sbi_hart_pmp_addrbits (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->pmp_addr_bits;@@ -286,7 +276,7 @@ unsigned int sbi_hart_pmp_addrbits(struct sbi_scratch *scratch)unsigned int sbi_hart_mhpm_bits (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->mhpm_bits;@@ -336,7 +326,7 @@ int sbi_hart_pmp_configure(struct sbi_scratch *scratch)int sbi_hart_priv_version (struct sbi_scratch *scratch) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);return hfeatures->priv_version;@@ -346,7 +336,7 @@ void sbi_hart_get_priv_version_str(struct sbi_scratch *scratch, char *version_str, int nvstr) { char *temp; -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);switch (hfeatures->priv_version) {@@ -368,7 +358,7 @@ void sbi_hart_get_priv_version_str(struct sbi_scratch *scratch, }static inline void __sbi_hart_update_extension ( -struct hart_features *hfeatures, +struct sbi_hart_features *hfeatures,enum sbi_hart_extensions ext,bool enable) {@@ -389,7 +379,7 @@ void sbi_hart_update_extension(struct sbi_scratch *scratch, enum sbi_hart_extensions ext, bool enable) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);__sbi_hart_update_extension (hfeatures, ext, enable);@@ -405,7 +395,7 @@ void sbi_hart_update_extension(struct sbi_scratch *scratch,bool sbi_hart_has_extension (struct sbi_scratch *scratch,enum sbi_hart_extensions ext) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);if (hfeatures->extensions &BIT (ext))@@ -453,7 +443,7 @@ static inline char *sbi_hart_extension_id2string(int ext)void sbi_hart_get_extensions_str (struct sbi_scratch *scratch,char *extensions_str,int nestr) { -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);int offset =0 , ext =0 ;char *temp;@@ -539,7 +529,7 @@ static int hart_pmu_get_allowed_bits(void)static int hart_detect_features (struct sbi_scratch *scratch) {struct sbi_trap_info trap = {0 }; -struct hart_features *hfeatures = +struct sbi_hart_features *hfeatures =sbi_scratch_offset_ptr (scratch, hart_features_offset);unsigned long val, oldval;int rc;@@ -718,7 +708,7 @@ int sbi_hart_init(struct sbi_scratch *scratch, bool cold_boot) sbi_hart_expected_trap = &__sbi_expected_trap_hext; hart_features_offset =sbi_scratch_alloc_offset ( -sizeof (struct hart_features )); +sizeof (struct sbi_hart_features ));if (!hart_features_offset)return SBI_ENOMEM ; } --2.35.1
--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
PATCH v5 3/5 lib: sbi_platform: expose hart_features to extension_init callback
In reply to Heiko StuebnerThe platform-specific extension_init callback is supposed to set specific things for the platform opensbi is running on.
So it's also the right place to override specific hart_features if needed - when it's know that autodetection has provided wrong results for example.
--- include/sbi/sbi_platform.h | 8 +++++--- lib/sbi/sbi_hart.c | 3 ++- 2 files changed, 7 insertions(+), 4 deletions(-)
include/sbi/sbi_platform.h +5 -3
@@ -48,6 +48,7 @@struct sbi_domain_memregion ;struct sbi_trap_info ;struct sbi_trap_regs ; +struct sbi_hart_features ;/** Possible feature flags of a platform */ enum sbi_platform_features {
@@ -90,7 +91,7 @@ struct sbi_platform_operations {int (*misa_get_xlen)(void );/** Initialize (or populate) HART extensions for the platform */ -int (*extensions_init)(void ); +int (*extensions_init)(struct sbi_hart_features *hfeatures);/** Initialize (or populate) domains for the platform */ int (*domains_init)(void );
@@ -464,10 +465,11 @@ static inline int sbi_platform_misa_xlen(const struct sbi_platform *plat)* @return 0 on success and negative error code on failure */ static inline int sbi_platform_extensions_init ( -const struct sbi_platform *plat) +const struct sbi_platform *plat, +struct sbi_hart_features *hfeatures) {if (plat &&sbi_platform_ops (plat)->extensions_init) -return sbi_platform_ops (plat)->extensions_init (); +return sbi_platform_ops (plat)->extensions_init (hfeatures);return 0 ; }
lib/sbi/sbi_hart.c +2 -1
@@ -672,7 +672,8 @@ __mhpm_skip: }/* Let platform populate extensions */ - rc =sbi_platform_extensions_init (sbi_platform_thishart_ptr ()); + rc =sbi_platform_extensions_init (sbi_platform_thishart_ptr (), + hfeatures);if (rc)return rc;
Signature · 1 line
2.35.1
PATCH v5 3/5 lib: sbi_platform: expose hart_features to extension_init callback
Quoted 10 lines · Heiko Stuebner
The platform-specific
extension_initcallback is supposed to set specific things for the platform opensbi is running on.So it's also the right place to override specific
hart_featuresif needed - when it's know that autodetection has provided wrong results for example.Suggested-by Atish Patra <[email protected]>Signed-off-by Heiko Stuebner <[email protected]>
Looks good to me.
Regards,
Anup
Quoted 62 lines
--- include/sbi/sbi_platform.h | 8 +++++--- lib/sbi/sbi_hart.c | 3 ++- 2 files changed, 7 insertions(+), 4 deletions(-)include/sbi/sbi_platform.h +5
-3@@ -48,6 +48,7 @@struct sbi_domain_memregion ;struct sbi_trap_info ;struct sbi_trap_regs ; +struct sbi_hart_features ;/** Possible feature flags of a platform */ enum sbi_platform_features {@@ -90,7 +91,7 @@ struct sbi_platform_operations {int (*misa_get_xlen)(void );/** Initialize (or populate) HART extensions for the platform */ -int (*extensions_init)(void ); +int (*extensions_init)(struct sbi_hart_features *hfeatures);/** Initialize (or populate) domains for the platform */ int (*domains_init)(void );@@ -464,10 +465,11 @@ static inline int sbi_platform_misa_xlen(const struct sbi_platform *plat)* @return 0 on success and negative error code on failure */ static inline int sbi_platform_extensions_init ( -const struct sbi_platform *plat) +const struct sbi_platform *plat, +struct sbi_hart_features *hfeatures) {if (plat &&sbi_platform_ops (plat)->extensions_init) -return sbi_platform_ops (plat)->extensions_init (); +return sbi_platform_ops (plat)->extensions_init (hfeatures);return 0 ; }lib/sbi/sbi_hart.c +2
-2@@ -672,7 +672,8 @@ __mhpm_skip: }/* Let platform populate extensions */ - rc =sbi_platform_extensions_init (sbi_platform_thishart_ptr ()); + rc =sbi_platform_extensions_init (sbi_platform_thishart_ptr (), + hfeatures);if (rc)return rc; --2.35.1
--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
PATCH v5 4/5 platform: generic: add extensions_init handler and platform-override
In reply to Heiko StuebnerInit of non-standard extensions is a platform-specific thing, so allow generic platforms to do this via a platform-override.
--- platform/generic/include/platform_override.h | 3 +++ platform/generic/platform.c | 10 ++++++++++ 2 files changed, 13 insertions(+)
platform/generic/include/platform_override.h +3 -0
@@ -10,6 +10,7 @@#ifndef __PLATFORM_OVERRIDE_H__#define __PLATFORM_OVERRIDE_H__ +#include <sbi/sbi_hart.h> #include <sbi/sbi_types.h> #include <sbi/sbi_trap.h>
@@ -22,6 +23,8 @@ struct platform_override {void (*early_exit)(const struct fdt_match *match);void (*final_exit)(const struct fdt_match *match);int (*fdt_fixup)(void *fdt,const struct fdt_match *match); +int (*extensions_init)(const struct fdt_match *match, +struct sbi_hart_features *hfeatures);int (*vendor_ext_check)(long extid,const struct fdt_match *match); int (*vendor_ext_provider)(long extid, long funcid, const struct sbi_trap_regs *regs,
platform/generic/platform.c +10 -0
@@ -204,6 +204,15 @@ static void generic_final_exit(void) generic_plat->final_exit (generic_plat_match); } +static int generic_extensions_init (struct sbi_hart_features *hfeatures) +{ +if (generic_plat && generic_plat->extensions_init) +return generic_plat->extensions_init (generic_plat_match, + hfeatures); + +return 0 ; +} +static int generic_domains_init (void ) {return fdt_domains_populate (fdt_get_address ());
@@ -257,6 +266,7 @@ const struct sbi_platform_operations platform_ops = { .final_init = generic_final_init, .early_exit = generic_early_exit, .final_exit = generic_final_exit, + .extensions_init = generic_extensions_init, .domains_init = generic_domains_init, .console_init = generic_console_init, .irqchip_init = fdt_irqchip_init,
Signature · 1 line
2.35.1
PATCH v5 4/5 platform: generic: add extensions_init handler and platform-override
Quoted 5 lines · Heiko Stuebner
Init of non-standard extensions is a platform-specific thing, so allow generic platforms to do this via a platform-override.
Signed-off-by Heiko Stuebner <[email protected]>
Looks good to me.
Regards,
Anup
Quoted 62 lines
--- platform/generic/include/platform_override.h | 3 +++ platform/generic/platform.c | 10 ++++++++++ 2 files changed, 13 insertions(+)platform/generic/include/platform_override.h +3
-0@@ -10,6 +10,7 @@#ifndef __PLATFORM_OVERRIDE_H__#define __PLATFORM_OVERRIDE_H__ +#include <sbi/sbi_hart.h> #include <sbi/sbi_types.h> #include <sbi/sbi_trap.h> @@ -22,6 +23,8 @@ struct platform_override {void (*early_exit)(const struct fdt_match *match);void (*final_exit)(const struct fdt_match *match);int (*fdt_fixup)(void *fdt,const struct fdt_match *match); +int (*extensions_init)(const struct fdt_match *match, +struct sbi_hart_features *hfeatures);int (*vendor_ext_check)(long extid,const struct fdt_match *match); int (*vendor_ext_provider)(long extid, long funcid, const struct sbi_trap_regs *regs,platform/generic/platform.c +10
-1@@ -204,6 +204,15 @@ static void generic_final_exit(void) generic_plat->final_exit (generic_plat_match); } +static int generic_extensions_init (struct sbi_hart_features *hfeatures) +{ +if (generic_plat && generic_plat->extensions_init) +return generic_plat->extensions_init (generic_plat_match, + hfeatures); + +return 0 ; +} +static int generic_domains_init (void ) {return fdt_domains_populate (fdt_get_address ());@@ -257,6 +266,7 @@ const struct sbi_platform_operations platform_ops = { .final_init = generic_final_init, .early_exit = generic_early_exit, .final_exit = generic_final_exit, + .extensions_init = generic_extensions_init, .domains_init = generic_domains_init, .console_init = generic_console_init, .irqchip_init = fdt_irqchip_init, --2.35.1
--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
PATCH v5 5/5 platform: generic: allwinner: add support for c9xx pmu
In reply to Heiko StuebnerWith the T-HEAD C9XX cores being designed before or during ratification of the SSCOFPMF extension, they implement a PMU extension that behaves very similar but not equal to it by providing overflow interrupts though in a slightly different registers format.
The sun20i-d1 is using this core. So implement the necessary overrides to allow its pmu to be used via the standard sbi-pmu extension.
For now it's also the only soc using this core, so keep the additional code in the d1-space for now.
--- platform/generic/allwinner/sun20i-d1.c | 60 ++++++++++++ platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++++++++ 2 files changed, 187 insertions(+) create mode 100644 platform/generic/include/thead_c9xx.h
platform/generic/allwinner/sun20i-d1.c +60 -0
@@ -5,11 +5,13 @@*/ #include <platform_override.h> +#include <thead_c9xx.h> #include <sbi/riscv_io.h> #include <sbi/sbi_bitops.h> #include <sbi/sbi_ecall_interface.h> #include <sbi/sbi_error.h> #include <sbi/sbi_hsm.h> +#include <sbi/sbi_pmu.h> #include <sbi_utils/fdt/fdt_helper.h> #include <sbi_utils/irqchip/fdt_irqchip_plic.h>
@@ -199,6 +201,63 @@ static int sun20i_d1_final_init(bool cold_boot, const struct fdt_match *match)return 0 ; } +#include <sbi/sbi_console.h> + +static void thead_c9xx_pmu_ctr_enable_irq (uint32_t ctr_idx) +{ +unsigned long mip_val; + +if (ctr_idx >=SBI_PMU_HW_CTR_MAX ) +return ; + + mip_val =csr_read (CSR_MIP ); +/** +* Clear out the OF bit so that next interrupt can be enabled. +* This should be done only when the corresponding overflow interrupt +* bit is cleared. That indicates that software has already handled the +* previous interrupts or the hardware yet to set an overflow interrupt. +* Otherwise, there will be race conditions where we may clear the bit +* the software is yet to handle the interrupt. +*/ +if (!(mip_val &THEAD_C9XX_MIP_MOIP )) +csr_clear (THEAD_C9XX_CSR_MCOUNTEROF ,BIT (ctr_idx)); + +/** +* SSCOFPMF uses the OF bit for enabling/disabling the interrupt, +* while the C9XX has designated enable bits. +* So enable per-counter interrupt on C9xx here. +*/ +csr_set (THEAD_C9XX_CSR_MCOUNTERINTEN ,BIT (ctr_idx)); +} + +static void thead_c9xx_pmu_ctr_disable_irq (uint32_t ctr_idx) +{ +csr_clear (THEAD_C9XX_CSR_MCOUNTERINTEN ,BIT (ctr_idx)); +} + +static int thead_c9xx_pmu_irq_bit (void ) +{ +return THEAD_C9XX_MIP_MOIP ; +} + +const struct sbi_pmu_device thead_c9xx_pmu_device = { + .hw_counter_enable_irq = thead_c9xx_pmu_ctr_enable_irq, + .hw_counter_disable_irq = thead_c9xx_pmu_ctr_disable_irq, + .hw_counter_irq_bit = thead_c9xx_pmu_irq_bit, +}; + +static int sun20i_d1_extensions_init (const struct fdt_match *match, +struct sbi_hart_features *hfeatures) +{ +sbi_pmu_set_device (&thead_c9xx_pmu_device); + +/* auto-detection doesn't work on t-head c9xx cores */ + hfeatures->mhpm_count =29 ; + hfeatures->mhpm_bits =64 ; + +return 0 ; +} +static const struct fdt_match sun20i_d1_match[] = { { .compatible ="allwinner,sun20i-d1" }, { },
@@ -207,4 +266,5 @@ static const struct fdt_match sun20i_d1_match[] = {const struct platform_override sun20i_d1 = { .match_table = sun20i_d1_match, .final_init = sun20i_d1_final_init, + .extensions_init = sun20i_d1_extensions_init, };
platform/generic/include/thead_c9xx.h new file +127 -0
127 lines not on this page · raw message
Signature · 1 line
2.35.1
PATCH v5 5/5 platform: generic: allwinner: add support for c9xx pmu
Quoted 6 lines · Heiko Stuebner
With the T-HEAD C9XX cores being designed before or during ratification of the SSCOFPMF extension, they implement a PMU extension that behaves very similar but not equal to it by providing overflow interrupts though in a slightly different registers format.
Quoted 8 lines
The sun20i-d1 is using this core. So implement the necessary overrides to allow its pmu to be used via the standard sbi-pmu extension.
For now it's also the only soc using this core, so keep the additional code in the d1-space for now.
Reviewed-by Guo Ren <[email protected]>Signed-off-by Heiko Stuebner <[email protected]>
Looks good to me.
Regards,
Anup
Quoted 235 lines
--- platform/generic/allwinner/sun20i-d1.c | 60 ++++++++++++ platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++++++++ 2 files changed, 187 insertions(+) create mode 100644 platform/generic/include/thead_c9xx.hplatform/generic/allwinner/sun20i-d1.c +60
-0@@ -5,11 +5,13 @@*/ #include <platform_override.h> +#include <thead_c9xx.h> #include <sbi/riscv_io.h> #include <sbi/sbi_bitops.h> #include <sbi/sbi_ecall_interface.h> #include <sbi/sbi_error.h> #include <sbi/sbi_hsm.h> +#include <sbi/sbi_pmu.h> #include <sbi_utils/fdt/fdt_helper.h> #include <sbi_utils/irqchip/fdt_irqchip_plic.h> @@ -199,6 +201,63 @@ static int sun20i_d1_final_init(bool cold_boot, const struct fdt_match *match)return 0 ; } +#include <sbi/sbi_console.h> + +static void thead_c9xx_pmu_ctr_enable_irq (uint32_t ctr_idx) +{ +unsigned long mip_val; + +if (ctr_idx >=SBI_PMU_HW_CTR_MAX ) +return ; + + mip_val =csr_read (CSR_MIP ); +/** +* Clear out the OF bit so that next interrupt can be enabled. +* This should be done only when the corresponding overflow interrupt +* bit is cleared. That indicates that software has already handled the +* previous interrupts or the hardware yet to set an overflow interrupt. +* Otherwise, there will be race conditions where we may clear the bit +* the software is yet to handle the interrupt. +*/ +if (!(mip_val &THEAD_C9XX_MIP_MOIP )) +csr_clear (THEAD_C9XX_CSR_MCOUNTEROF ,BIT (ctr_idx)); + +/** +* SSCOFPMF uses the OF bit for enabling/disabling the interrupt, +* while the C9XX has designated enable bits. +* So enable per-counter interrupt on C9xx here. +*/ +csr_set (THEAD_C9XX_CSR_MCOUNTERINTEN ,BIT (ctr_idx)); +} + +static void thead_c9xx_pmu_ctr_disable_irq (uint32_t ctr_idx) +{ +csr_clear (THEAD_C9XX_CSR_MCOUNTERINTEN ,BIT (ctr_idx)); +} + +static int thead_c9xx_pmu_irq_bit (void ) +{ +return THEAD_C9XX_MIP_MOIP ; +} + +const struct sbi_pmu_device thead_c9xx_pmu_device = { + .hw_counter_enable_irq = thead_c9xx_pmu_ctr_enable_irq, + .hw_counter_disable_irq = thead_c9xx_pmu_ctr_disable_irq, + .hw_counter_irq_bit = thead_c9xx_pmu_irq_bit, +}; + +static int sun20i_d1_extensions_init (const struct fdt_match *match, +struct sbi_hart_features *hfeatures) +{ +sbi_pmu_set_device (&thead_c9xx_pmu_device); + +/* auto-detection doesn't work on t-head c9xx cores */ + hfeatures->mhpm_count =29 ; + hfeatures->mhpm_bits =64 ; + +return 0 ; +} +static const struct fdt_match sun20i_d1_match[] = { { .compatible ="allwinner,sun20i-d1" }, { },@@ -207,4 +266,5 @@ static const struct fdt_match sun20i_d1_match[] = {const struct platform_override sun20i_d1 = { .match_table = sun20i_d1_match, .final_init = sun20i_d1_final_init, + .extensions_init = sun20i_d1_extensions_init, };platform/generic/include/thead_c9xx.h new file +127
-1@@ -0,0 +1,127 @@ +#ifndef __RISCV_THEAD_C9XX_H____ +#define __RISCV_THEAD_C9XX_H____ + +/* T-HEAD C9xx M mode CSR. */ +#define THEAD_C9XX_CSR_MXSTATUS 0x7c0 +#define THEAD_C9XX_CSR_MHCR 0x7c1 +#define THEAD_C9XX_CSR_MCOR 0x7c2 +#define THEAD_C9XX_CSR_MCCR2 0x7c3 +#define THEAD_C9XX_CSR_MCER2 0x7c4 +#define THEAD_C9XX_CSR_MHINT 0x7c5 +#define THEAD_C9XX_CSR_MRMR 0x7c6 +#define THEAD_C9XX_CSR_MRVBR 0x7c7 +#define THEAD_C9XX_CSR_MCER 0x7c8 +#define THEAD_C9XX_CSR_MCOUNTERWEN 0x7c9 +#define THEAD_C9XX_CSR_MCOUNTERINTEN 0x7ca +#define THEAD_C9XX_CSR_MCOUNTEROF 0x7cb +#define THEAD_C9XX_CSR_MHINT2 0x7cc +#define THEAD_C9XX_CSR_MHINT3 0x7cd +#define THEAD_C9XX_CSR_MRADDR 0x7e0 +#define THEAD_C9XX_CSR_MEXSTATUS 0x7e1 +#define THEAD_C9XX_CSR_MNMICAUSE 0x7e2 +#define THEAD_C9XX_CSR_MNMIPC 0x7e3 +#define THEAD_C9XX_CSR_MHPMCR 0x7f0 +#define THEAD_C9XX_CSR_MHPMSR 0x7f1 +#define THEAD_C9XX_CSR_MHPMER 0x7f2 +#define THEAD_C9XX_CSR_MSMPR 0x7f3 +#define THEAD_C9XX_CSR_MTEECFG 0x7f4 +#define THEAD_C9XX_CSR_MZONEID 0x7f5 +#define THEAD_C9XX_CSR_ML2CPID 0x7f6 +#define THEAD_C9XX_CSR_ML2WP 0x7f7 +#define THEAD_C9XX_CSR_MDTCMCR 0x7f8 +#define THEAD_C9XX_CSR_USP 0x7d1 +#define THEAD_C9XX_CSR_MCINS 0x7d2 +#define THEAD_C9XX_CSR_MCINDEX 0x7d3 +#define THEAD_C9XX_CSR_MCDATA0 0x7d4 +#define THEAD_C9XX_CSR_MCDATA1 0x7d5 +#define THEAD_C9XX_CSR_MEICR 0x7d6 +#define THEAD_C9XX_CSR_MEICR2 0x7d7 +#define THEAD_C9XX_CSR_MBEADDR 0x7d8 +#define THEAD_C9XX_CSR_MCPUID 0xfc0 +#define THEAD_C9XX_CSR_MAPBADDR 0xfc1 +#define THEAD_C9XX_CSR_MWMSR 0xfc2 +#define THEAD_C9XX_CSR_MHALTCAUSE 0xfe0 +#define THEAD_C9XX_CSR_MDBGINFO 0xfe1 +#define THEAD_C9XX_CSR_MPCFIFO 0xfe2 + +/* T-HEAD C9xx S mode CSR. */ +#define THEAD_C9XX_CSR_SXSTATUS 0x5c0 +#define THEAD_C9XX_CSR_SHCR 0x5c1 +#define THEAD_C9XX_CSR_SCER2 0x5c2 +#define THEAD_C9XX_CSR_SCER 0x5c3 +#define THEAD_C9XX_CSR_SCOUNTERINTEN 0x5c4 +#define THEAD_C9XX_CSR_SCOUNTEROF 0x5c5 +#define THEAD_C9XX_CSR_SHINT 0x5c6 +#define THEAD_C9XX_CSR_SHINT2 0x5c7 +#define THEAD_C9XX_CSR_SHPMINHIBIT 0x5c8 +#define THEAD_C9XX_CSR_SHPMCR 0x5c9 +#define THEAD_C9XX_CSR_SHPMSR 0x5ca +#define THEAD_C9XX_CSR_SHPMER 0x5cb +#define THEAD_C9XX_CSR_SL2CPID 0x5cc +#define THEAD_C9XX_CSR_SL2WP 0x5cd +#define THEAD_C9XX_CSR_SBEADDR 0x5d0 +#define THEAD_C9XX_CSR_SCYCLE 0x5e0 +#define THEAD_C9XX_CSR_SHPMCOUNTER1 0x5e1 +#define THEAD_C9XX_CSR_SHPMCOUNTER2 0x5e2 +#define THEAD_C9XX_CSR_SHPMCOUNTER3 0x5e3 +#define THEAD_C9XX_CSR_SHPMCOUNTER4 0x5e4 +#define THEAD_C9XX_CSR_SHPMCOUNTER5 0x5e5 +#define THEAD_C9XX_CSR_SHPMCOUNTER6 0x5e6 +#define THEAD_C9XX_CSR_SHPMCOUNTER7 0x5e7 +#define THEAD_C9XX_CSR_SHPMCOUNTER8 0x5e8 +#define THEAD_C9XX_CSR_SHPMCOUNTER9 0x5e9 +#define THEAD_C9XX_CSR_SHPMCOUNTER10 0x5ea +#define THEAD_C9XX_CSR_SHPMCOUNTER11 0x5eb +#define THEAD_C9XX_CSR_SHPMCOUNTER12 0x5ec +#define THEAD_C9XX_CSR_SHPMCOUNTER13 0x5ed +#define THEAD_C9XX_CSR_SHPMCOUNTER14 0x5ee +#define THEAD_C9XX_CSR_SHPMCOUNTER15 0x5ef +#define THEAD_C9XX_CSR_SHPMCOUNTER16 0x5f0 +#define THEAD_C9XX_CSR_SHPMCOUNTER17 0x5f1 +#define THEAD_C9XX_CSR_SHPMCOUNTER18 0x5f2 +#define THEAD_C9XX_CSR_SHPMCOUNTER19 0x5f3 +#define THEAD_C9XX_CSR_SHPMCOUNTER20 0x5f4 +#define THEAD_C9XX_CSR_SHPMCOUNTER21 0x5f5 +#define THEAD_C9XX_CSR_SHPMCOUNTER22 0x5f6 +#define THEAD_C9XX_CSR_SHPMCOUNTER23 0x5f7 +#define THEAD_C9XX_CSR_SHPMCOUNTER24 0x5f8 +#define THEAD_C9XX_CSR_SHPMCOUNTER25 0x5f9 +#define THEAD_C9XX_CSR_SHPMCOUNTER26 0x5fa +#define THEAD_C9XX_CSR_SHPMCOUNTER27 0x5fb +#define THEAD_C9XX_CSR_SHPMCOUNTER28 0x5fc +#define THEAD_C9XX_CSR_SHPMCOUNTER29 0x5fd +#define THEAD_C9XX_CSR_SHPMCOUNTER30 0x5fe +#define THEAD_C9XX_CSR_SHPMCOUNTER31 0x5ff + +/* T-HEAD C9xx U mode CSR. */ +#define THEAD_C9XX_CSR_FXCR 0x800 + +/* T-HEAD C9xx MMU extentions. */ +#define THEAD_C9XX_CSR_SMIR 0x9c0 +#define THEAD_C9XX_CSR_SMEL 0x9c1 +#define THEAD_C9XX_CSR_SMEH 0x9c2 +#define THEAD_C9XX_CSR_SMCIR 0x9c3 + +/* T-HEAD C9xx Security CSR(May be droped). */ +#define THEAD_C9XX_CSR_MEBR 0xbe0 +#define THEAD_C9XX_CSR_NT_MSTATUS 0xbe1 +#define THEAD_C9XX_CSR_NT_MIE 0xbe2 +#define THEAD_C9XX_CSR_NT_MTVEC 0xbe3 +#define THEAD_C9XX_CSR_NT_MTVT 0xbe4 +#define THEAD_C9XX_CSR_NT_MEPC 0xbe5 +#define THEAD_C9XX_CSR_NT_MCAUSE 0xbe6 +#define THEAD_C9XX_CSR_NT_MIP 0xbe7 +#define THEAD_C9XX_CSR_NT_MINTSTATE 0xbe8 +#define THEAD_C9XX_CSR_NT_MXSTATUS 0xbe9 +#define THEAD_C9XX_CSR_NT_MEBR 0xbea +#define THEAD_C9XX_CSR_NT_MSP 0xbeb +#define THEAD_C9XX_CSR_T_USP 0xbec +#define THEAD_C9XX_CSR_T_MDCR 0xbed +#define THEAD_C9XX_CSR_T_MPCR 0xbee +#define THEAD_C9XX_CSR_PMPTEECFG 0xbef + +/* T-HEAD C9xx MIP CSR extension */ +#define THEAD_C9XX_IRQ_PMU_OVF 17 +#define THEAD_C9XX_MIP_MOIP (_UL(1) << THEAD_C9XX_IRQ_PMU_OVF) + +#endif --2.35.1
--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
v5 cover Add support for T-HEAD C9xx PMU extensions
In reply to Heiko StuebnerQuoted 52 lines · Heiko Stuebner
The T-HEAD C9XX cores implement functionality very similar to SSCOFPMF. Instead of implementing a separate interface into SBI as done in v1, we can add the C9XX tidbits with quite minimal overhead and leaverage the existing SBI pmu interface including giving access to the firmware counters.
The current only hickup is the detection override in
sbi_hart, where I still need to find out why the MHPMCOUNTERs are not writeable at _that_ point of the boot, but with this series on top of openSBI 1.1 and the matching kernel perf patch I get reasonable results for example for:perf stat -e cycles -e instructions -e L1-icache-load-misses -e L1-icache-loads ls
Performance counter stats for 'ls':
17119496 cycles 2867765 instructions # 0.17 insn per cycle 55929 L1-icache-load-misses # 1.61% of all L1-icache accesses 3467346 L1-icache-loads0.030156216 seconds time elapsed 0.000000000 seconds user 0.023496000 seconds syschanges in v5:
- usecsr_set,csr_clearin allwinner pmu functions
- exposehart_featuresstruct into header
- makehart_featuresparam forextension_initcallback to allow platforms to override specific settingschanges in v4:
- use the newsbi_pmu_devicestructure for holding the overrides and extend it where neededchanges in v3:
- follow Atish's advice and implement an abstraction to not pollute the core pmu with cpu-specific variantschanges in v2:
- don't implement a separate interface but instead modify the sbi-pmu to allow the c9xx to use standard pmu interfaceHeiko Stuebner (5):
lib:sbi_pmu: move pmu irq information into pmu itself
lib:sbi_hart: movehart_featuresstruct to a public location
lib:sbi_platform: exposehart_featurestoextension_initcallback
platform: generic: addextensions_inithandler and platform-override
platform: generic: allwinner: add support for c9xx pmu
Quoted 20 lines
include/sbi/sbi_hart.h | 11 ++ include/sbi/sbi_platform.h | 8 +- include/sbi/sbi_pmu.h | 8 ++ lib/sbi/sbi_hart.c | 43 +++---- lib/sbi/sbi_pmu.c | 12 ++ platform/generic/allwinner/sun20i-d1.c | 60 +++++++++ platform/generic/include/platform_override.h | 3 + platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++ platform/generic/platform.c | 10 ++ 9 files changed, 253 insertions(+), 29 deletions(-) create mode 100644 platform/generic/include/thead_c9xx.hSignature · 7 lines
2.35.1 -- opensbi mailing list opensbi at lists.infradead.org http://lists.infradead.org/mailman/listinfo/opensbi
For the series
v5 cover Add support for T-HEAD C9xx PMU extensions
For the series
Quoted 78 lines · Andrew Jones
Quoted 64 lines · Heiko Stuebner
The T-HEAD C9XX cores implement functionality very similar to SSCOFPMF. Instead of implementing a separate interface into SBI as done in v1, we can add the C9XX tidbits with quite minimal overhead and leaverage the existing SBI pmu interface including giving access to the firmware counters.
The current only hickup is the detection override in
sbi_hart, where I still need to find out why the MHPMCOUNTERs are not writeable at _that_ point of the boot, but with this series on top of openSBI 1.1 and the matching kernel perf patch I get reasonable results for example for:perf stat -e cycles -e instructions -e L1-icache-load-misses -e L1-icache-loads ls
Performance counter stats for 'ls':
17119496 cycles 2867765 instructions # 0.17 insn per cycle 55929 L1-icache-load-misses # 1.61% of all L1-icache accesses 3467346 L1-icache-loads0.030156216 seconds time elapsed 0.000000000 seconds user 0.023496000 seconds syschanges in v5:
- usecsr_set,csr_clearin allwinner pmu functions
- exposehart_featuresstruct into header
- makehart_featuresparam forextension_initcallback to allow platforms to override specific settingschanges in v4:
- use the newsbi_pmu_devicestructure for holding the overrides and extend it where neededchanges in v3:
- follow Atish's advice and implement an abstraction to not pollute the core pmu with cpu-specific variantschanges in v2:
- don't implement a separate interface but instead modify the sbi-pmu to allow the c9xx to use standard pmu interfaceHeiko Stuebner (5):
lib:sbi_pmu: move pmu irq information into pmu itself
lib:sbi_hart: movehart_featuresstruct to a public location
lib:sbi_platform: exposehart_featurestoextension_initcallback
platform: generic: addextensions_inithandler and platform-override
platform: generic: allwinner: add support for c9xx pmuinclude/sbi/sbi_hart.h | 11 ++ include/sbi/sbi_platform.h | 8 +- include/sbi/sbi_pmu.h | 8 ++ lib/sbi/sbi_hart.c | 43 +++---- lib/sbi/sbi_pmu.c | 12 ++ platform/generic/allwinner/sun20i-d1.c | 60 +++++++++ platform/generic/include/platform_override.h | 3 + platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++ platform/generic/platform.c | 10 ++ 9 files changed, 253 insertions(+), 29 deletions(-) create mode 100644 platform/generic/include/thead_c9xx.hQuoted 8 lines
--
2.35.1--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbiFor the series
Reviewed-by Andrew Jones <[email protected]>
Signature · 2 lines
Best Regards Guo Ren
Anup Patel
Appliedv5 cover Add support for T-HEAD C9xx PMU extensions
In reply to Heiko StuebnerQuoted 46 lines · Heiko Stuebner
The T-HEAD C9XX cores implement functionality very similar to SSCOFPMF. Instead of implementing a separate interface into SBI as done in v1, we can add the C9XX tidbits with quite minimal overhead and leaverage the existing SBI pmu interface including giving access to the firmware counters.
The current only hickup is the detection override in
sbi_hart, where I still need to find out why the MHPMCOUNTERs are not writeable at _that_ point of the boot, but with this series on top of openSBI 1.1 and the matching kernel perf patch I get reasonable results for example for:perf stat -e cycles -e instructions -e L1-icache-load-misses -e L1-icache-loads ls
Performance counter stats for 'ls':
17119496 cycles 2867765 instructions # 0.17 insn per cycle 55929 L1-icache-load-misses # 1.61% of all L1-icache accesses 3467346 L1-icache-loads0.030156216 seconds time elapsed 0.000000000 seconds user 0.023496000 seconds syschanges in v5:
- usecsr_set,csr_clearin allwinner pmu functions
- exposehart_featuresstruct into header
- makehart_featuresparam forextension_initcallback to allow platforms to override specific settingschanges in v4:
- use the newsbi_pmu_devicestructure for holding the overrides and extend it where neededchanges in v3:
- follow Atish's advice and implement an abstraction to not pollute the core pmu with cpu-specific variantschanges in v2:
- don't implement a separate interface but instead modify the sbi-pmu to allow the c9xx to use standard pmu interface
Quoted 6 lines
Heiko Stuebner (5):
lib:sbi_pmu: move pmu irq information into pmu itself
lib:sbi_hart: movehart_featuresstruct to a public location
lib:sbi_platform: exposehart_featurestoextension_initcallback
platform: generic: addextensions_inithandler and platform-override
platform: generic: allwinner: add support for c9xx pmu
Applied this series to the riscv/opensbi repo.
Thanks,
Anup
Quoted 21 lines
include/sbi/sbi_hart.h | 11 ++ include/sbi/sbi_platform.h | 8 +- include/sbi/sbi_pmu.h | 8 ++ lib/sbi/sbi_hart.c | 43 +++---- lib/sbi/sbi_pmu.c | 12 ++ platform/generic/allwinner/sun20i-d1.c | 60 +++++++++ platform/generic/include/platform_override.h | 3 + platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++ platform/generic/platform.c | 10 ++ 9 files changed, 253 insertions(+), 29 deletions(-) create mode 100644 platform/generic/include/thead_c9xx.h--
2.35.1--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
v5 cover Add support for T-HEAD C9xx PMU extensions
Am Donnerstag, 13. Oktober 2022, 06:26:43 CEST schrieb Anup Patel:
Quoted 47 lines · Heiko Stuebner
Quoted 46 lines
The T-HEAD C9XX cores implement functionality very similar to SSCOFPMF. Instead of implementing a separate interface into SBI as done in v1, we can add the C9XX tidbits with quite minimal overhead and leaverage the existing SBI pmu interface including giving access to the firmware counters.
The current only hickup is the detection override in
sbi_hart, where I still need to find out why the MHPMCOUNTERs are not writeable at _that_ point of the boot, but with this series on top of openSBI 1.1 and the matching kernel perf patch I get reasonable results for example for:perf stat -e cycles -e instructions -e L1-icache-load-misses -e L1-icache-loads ls
Performance counter stats for 'ls':
17119496 cycles 2867765 instructions # 0.17 insn per cycle 55929 L1-icache-load-misses # 1.61% of all L1-icache accesses 3467346 L1-icache-loads0.030156216 seconds time elapsed 0.000000000 seconds user 0.023496000 seconds syschanges in v5:
- usecsr_set,csr_clearin allwinner pmu functions
- exposehart_featuresstruct into header
- makehart_featuresparam forextension_initcallback to allow platforms to override specific settingschanges in v4:
- use the newsbi_pmu_devicestructure for holding the overrides and extend it where neededchanges in v3:
- follow Atish's advice and implement an abstraction to not pollute the core pmu with cpu-specific variantschanges in v2:
- don't implement a separate interface but instead modify the sbi-pmu to allow the c9xx to use standard pmu interface
Quoted 8 lines
Quoted 6 lines
Heiko Stuebner (5):
lib:sbi_pmu: move pmu irq information into pmu itself
lib:sbi_hart: movehart_featuresstruct to a public location
lib:sbi_platform: exposehart_featurestoextension_initcallback
platform: generic: addextensions_inithandler and platform-override
platform: generic: allwinner: add support for c9xx pmuApplied this series to the riscv/opensbi repo.
Very cool, thanks a lot for picking this up :-)
Thanks
Heiko
Quoted 21 lines
Quoted 20 lines
include/sbi/sbi_hart.h | 11 ++ include/sbi/sbi_platform.h | 8 +- include/sbi/sbi_pmu.h | 8 ++ lib/sbi/sbi_hart.c | 43 +++---- lib/sbi/sbi_pmu.c | 12 ++ platform/generic/allwinner/sun20i-d1.c | 60 +++++++++ platform/generic/include/platform_override.h | 3 + platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++ platform/generic/platform.c | 10 ++ 9 files changed, 253 insertions(+), 29 deletions(-) create mode 100644 platform/generic/include/thead_c9xx.h--
2.35.1--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
v5 cover Add support for T-HEAD C9xx PMU extensions
Great! Thx Anup, and Heiko.
Quoted 89 lines · Heiko Stuebner
Am Donnerstag, 13. Oktober 2022, 06:26:43 CEST schrieb Anup Patel:
Quoted 47 lines · Heiko Stuebner
Quoted 46 lines
The T-HEAD C9XX cores implement functionality very similar to SSCOFPMF. Instead of implementing a separate interface into SBI as done in v1, we can add the C9XX tidbits with quite minimal overhead and leaverage the existing SBI pmu interface including giving access to the firmware counters.
The current only hickup is the detection override in
sbi_hart, where I still need to find out why the MHPMCOUNTERs are not writeable at _that_ point of the boot, but with this series on top of openSBI 1.1 and the matching kernel perf patch I get reasonable results for example for:perf stat -e cycles -e instructions -e L1-icache-load-misses -e L1-icache-loads ls
Performance counter stats for 'ls':
17119496 cycles 2867765 instructions # 0.17 insn per cycle 55929 L1-icache-load-misses # 1.61% of all L1-icache accesses 3467346 L1-icache-loads0.030156216 seconds time elapsed 0.000000000 seconds user 0.023496000 seconds syschanges in v5:
- usecsr_set,csr_clearin allwinner pmu functions
- exposehart_featuresstruct into header
- makehart_featuresparam forextension_initcallback to allow platforms to override specific settingschanges in v4:
- use the newsbi_pmu_devicestructure for holding the overrides and extend it where neededchanges in v3:
- follow Atish's advice and implement an abstraction to not pollute the core pmu with cpu-specific variantschanges in v2:
- don't implement a separate interface but instead modify the sbi-pmu to allow the c9xx to use standard pmu interfaceQuoted 8 lines
Quoted 6 lines
Heiko Stuebner (5):
lib:sbi_pmu: move pmu irq information into pmu itself
lib:sbi_hart: movehart_featuresstruct to a public location
lib:sbi_platform: exposehart_featurestoextension_initcallback
platform: generic: addextensions_inithandler and platform-override
platform: generic: allwinner: add support for c9xx pmuApplied this series to the riscv/opensbi repo.
Very cool, thanks a lot for picking this up :-)
Thanks
HeikoQuoted 21 lines
Quoted 20 lines
include/sbi/sbi_hart.h | 11 ++ include/sbi/sbi_platform.h | 8 +- include/sbi/sbi_pmu.h | 8 ++ lib/sbi/sbi_hart.c | 43 +++---- lib/sbi/sbi_pmu.c | 12 ++ platform/generic/allwinner/sun20i-d1.c | 60 +++++++++ platform/generic/include/platform_override.h | 3 + platform/generic/include/thead_c9xx.h | 127 +++++++++++++++++++ platform/generic/platform.c | 10 ++ 9 files changed, 253 insertions(+), 29 deletions(-) create mode 100644 platform/generic/include/thead_c9xx.h--
2.35.1--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
Signature · 2 lines
Best Regards Guo Ren