lib: pmu: allow to use the highest available counter
PATCH v3 1/1 lib: pmu: allow to use the highest available counter
OpenSBI explicitly assumes that there is no pmu hardware counter with index 1: hardware uses that bit for TM control. So OpenSBI filters out that index in sanity checks. However OpenSBI also excludes that counter when reports total amount of hardware counters to Linux. As a result, Linux uses incomplete counters mask excluding the highest available counter.
Return accurate number of counters, update the firmware counter starting index, fix range checks that include num_hw_ctrs.
The simple test is to make sure that there is no counter multiplexing in the following command:
$ perf stat -e \ r8000000000000000,r8000000000000001,r8000000000000002,r8000000000000003, \ r8000000000000004,r8000000000000005,r8000000000000006,r8000000000000007, \ r8000000000000008,r8000000000000009,r800000000000000a,r800000000000000b, \ r800000000000000c,r800000000000000d,r800000000000000e,r800000000000000f \ ls
Note that 16 firmware events with 16 counters won't require multiplexing.
---
Hi all,
Tested on QEMU with and without OpenSBI firmware counters enabled.
v2 -> v3:
- Fix range checks that include num_hw_ctrs
- Update commit message
v1 -> v2:
- Rework: report accurate number of hardware counters to OS instead of handling timer counter in a special way. Special handling should be done in sbi_pmu_ctr_cfg_match
Regards,
Sergey
lib/sbi/sbi_pmu.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)
lib/sbi/sbi_pmu.c +7 -7
@@ -297,7 +297,7 @@ static int pmu_ctr_start_hw(uint32_t cidx, uint64_t ival, bool ival_update)unsigned long mctr_inhbt;/* Make sure the counter index lies within the range and is not TM bit */ -if (cidx > num_hw_ctrs || cidx ==1 ) +if (cidx >= num_hw_ctrs || cidx ==1 )return SBI_EINVAL ;if (sbi_hart_priv_version (scratch) <SBI_HART_PRIV_VER_1_11 )
@@ -378,7 +378,7 @@ static int pmu_ctr_stop_hw(uint32_t cidx) mctr_inhbt =csr_read (CSR_MCOUNTINHIBIT );/* Make sure the counter index lies within the range and is not TM bit */ -if (cidx > num_hw_ctrs || cidx ==1 ) +if (cidx >= num_hw_ctrs || cidx ==1 )return SBI_EINVAL ;if (!__test_bit (cidx, &mctr_inhbt)) {
@@ -516,7 +516,7 @@ static int pmu_ctr_find_hw(unsigned long cbase, unsigned long cmask, unsigned lou32 hartid =current_hartid ();struct sbi_scratch *scratch =sbi_scratch_thishart_ptr (); -if (cbase > num_hw_ctrs) +if (cbase >= num_hw_ctrs)return SBI_EINVAL ;/**
@@ -593,8 +593,8 @@ static int pmu_ctr_find_fw(unsigned long cbase, unsigned long cmask, u32 hartid)int fw_base;unsigned long ctr_mask = cmask << cbase; -if (cbase <= num_hw_ctrs) - fw_base = num_hw_ctrs +1 ; +if (cbase < num_hw_ctrs) + fw_base = num_hw_ctrs;else fw_base = cbase;
@@ -694,7 +694,7 @@ int sbi_pmu_ctr_get_info(uint32_t cidx, unsigned long *ctr_info)return SBI_EINVAL ;/* We have 31 HW counters with 31 being the last index(MHPMCOUNTER31) */ -if (cidx <= num_hw_ctrs) { +if (cidx < num_hw_ctrs) { cinfo.type =SBI_PMU_CTR_TYPE_HW ; cinfo.csr =CSR_CYCLE + cidx;/* mcycle & minstret are always 64 bit */
@@ -749,7 +749,7 @@ int sbi_pmu_init(struct sbi_scratch *scratch, bool cold_boot)sbi_platform_pmu_init (plat);/* mcycle & minstret is available always */ - num_hw_ctrs =sbi_hart_mhpm_count (scratch) +2 ; + num_hw_ctrs =sbi_hart_mhpm_count (scratch) +3 ; total_ctrs = num_hw_ctrs +SBI_PMU_FW_CTR_MAX ; }
Signature · 1 line
2.36.1
PATCH v3 1/1 lib: pmu: allow to use the highest available counter
Quoted 110 lines · Sergey Matyukevich
OpenSBI explicitly assumes that there is no pmu hardware counter with index 1: hardware uses that bit for TM control. So OpenSBI filters out that index in sanity checks. However OpenSBI also excludes that counter when reports total amount of hardware counters to Linux. As a result, Linux uses incomplete counters mask excluding the highest available counter.
Return accurate number of counters, update the firmware counter starting index, fix range checks that include
num_hw_ctrs.The simple test is to make sure that there is no counter multiplexing in the following command:
$ perf stat -e \ r8000000000000000,r8000000000000001,r8000000000000002,r8000000000000003, \ r8000000000000004,r8000000000000005,r8000000000000006,r8000000000000007, \ r8000000000000008,r8000000000000009,r800000000000000a,r800000000000000b, \ r800000000000000c,r800000000000000d,r800000000000000e,r800000000000000f \ lsNote that 16 firmware events with 16 counters won't require multiplexing.
Signed-off-by Sergey Matyukevich <[email protected]>Signed-off-by Atish Patra <[email protected]>---
Hi all,
Tested on QEMU with and without OpenSBI firmware counters enabled.
v2 -> v3:
- Fix range checks that includenum_hw_ctrs
- Update commit messagev1 -> v2:
- Rework: report accurate number of hardware counters to OS instead of handling timer counter in a special way. Special handling should be done insbi_pmu_ctr_cfg_matchRegards,
Sergeylib/sbi/sbi_pmu.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-)lib/sbi/sbi_pmu.c +7
-8@@ -297,7 +297,7 @@ static int pmu_ctr_start_hw(uint32_t cidx, uint64_t ival, bool ival_update)unsigned long mctr_inhbt;/* Make sure the counter index lies within the range and is not TM bit */ -if (cidx > num_hw_ctrs || cidx ==1 ) +if (cidx >= num_hw_ctrs || cidx ==1 )return SBI_EINVAL ;if (sbi_hart_priv_version (scratch) <SBI_HART_PRIV_VER_1_11 )@@ -378,7 +378,7 @@ static int pmu_ctr_stop_hw(uint32_t cidx) mctr_inhbt =csr_read (CSR_MCOUNTINHIBIT );/* Make sure the counter index lies within the range and is not TM bit */ -if (cidx > num_hw_ctrs || cidx ==1 ) +if (cidx >= num_hw_ctrs || cidx ==1 )return SBI_EINVAL ;if (!__test_bit (cidx, &mctr_inhbt)) {@@ -516,7 +516,7 @@ static int pmu_ctr_find_hw(unsigned long cbase, unsigned long cmask, unsigned lou32 hartid =current_hartid ();struct sbi_scratch *scratch =sbi_scratch_thishart_ptr (); -if (cbase > num_hw_ctrs) +if (cbase >= num_hw_ctrs)return SBI_EINVAL ;/** @@ -593,8 +593,8 @@ static int pmu_ctr_find_fw(unsigned long cbase, unsigned long cmask, u32 hartid)int fw_base;unsigned long ctr_mask = cmask << cbase; -if (cbase <= num_hw_ctrs) - fw_base = num_hw_ctrs +1 ; +if (cbase < num_hw_ctrs) + fw_base = num_hw_ctrs;else fw_base = cbase;@@ -694,7 +694,7 @@ int sbi_pmu_ctr_get_info(uint32_t cidx, unsigned long *ctr_info)return SBI_EINVAL ;/* We have 31 HW counters with 31 being the last index(MHPMCOUNTER31) */ -if (cidx <= num_hw_ctrs) { +if (cidx < num_hw_ctrs) { cinfo.type =SBI_PMU_CTR_TYPE_HW ; cinfo.csr =CSR_CYCLE + cidx;/* mcycle & minstret are always 64 bit */ @@ -749,7 +749,7 @@ int sbi_pmu_init(struct sbi_scratch *scratch, bool cold_boot)sbi_platform_pmu_init (plat);/* mcycle & minstret is available always */ - num_hw_ctrs =sbi_hart_mhpm_count (scratch) +2 ; + num_hw_ctrs =sbi_hart_mhpm_count (scratch) +3 ; total_ctrs = num_hw_ctrs +SBI_PMU_FW_CTR_MAX ; } --2.36.1
Quoted 4 lines
--
opensbi mailing list
opensbi at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
Thanks for fixing the issue. LGTM.
Signature · 2 lines
Regards, Atish