x86/resctrl: Fix event counts regression in reused RMIDs
Subsystems (3)
PATCH v3 1/2 x86/resctrl: Fix event counts regression in reused RMIDs
When creating a new monitoring group, the RMID allocated for it may have been used by a group which was previously removed. In this case, the hardware counters will have non-zero values which should be deducted from what is reported in the new group's counts.
resctrl_arch_reset_rmid() initializes the prev_msr value for counters to 0, causing the initial count to be charged to the new group. Resurrect __rmid_read() and use it to initialize prev_msr correctly.
Unlike before, __rmid_read() checks for error bits in the MSR read so that callers don't need to.
---
v3:
- add changelog
- CC stable
v2:
- move error bit processing into __rmid_read()
v1: https://lore.kernel.org/lkml/[email protected]/
v2: https://lore.kernel.org/lkml/[email protected]/
--- arch/x86/kernel/cpu/resctrl/monitor.c | 49 ++++++++++++++++++--------- 1 file changed, 33 insertions(+), 16 deletions(-)
arch/x86/kernel/cpu/resctrl/monitor.c +33 -16
@@ -146,6 +146,30 @@ static inline struct rmid_entry *__rmid_entry(u32 rmid)return entry; } +static int __rmid_read (u32 rmid,enum resctrl_event_id eventid,u64 *val) +{ +u64 msr_val; + +/* +* As per the SDM, when IA32_QM_EVTSEL.EvtID (bits 7:0) is configured +* with a valid event code for supported resource type and the bits +* IA32_QM_EVTSEL.RMID (bits 41:32) are configured with valid RMID, +* IA32_QM_CTR.data (bits 61:0) reports the monitored data. +* IA32_QM_CTR.Error (bit 63) and IA32_QM_CTR.Unavailable (bit 62) +* are error bits. +*/ +wrmsr (MSR_IA32_QM_EVTSEL , eventid, rmid); +rdmsrl (MSR_IA32_QM_CTR , msr_val); + +if (msr_val &RMID_VAL_ERROR ) +return -EIO ; +if (msr_val &RMID_VAL_UNAVAIL ) +return -EINVAL ; + + *val = msr_val; +return 0 ; +} +static struct arch_mbm_state *get_arch_mbm_state (struct rdt_hw_domain *hw_dom,u32 rmid,enum resctrl_event_id eventid)
@@ -172,8 +196,12 @@ void resctrl_arch_reset_rmid(struct rdt_resource *r, struct rdt_domain *d,struct arch_mbm_state *am; am =get_arch_mbm_state (hw_dom, rmid, eventid); -if (am) +if (am) {memset (am,0 ,sizeof (*am)); + +/* Record any initial, non-zero count value. */ +__rmid_read (rmid, eventid, &am->prev_msr); + } }static u64 mbm_overflow_count (u64 prev_msr,u64 cur_msr,unsigned int width)
@@ -191,25 +219,14 @@ int resctrl_arch_rmid_read(struct rdt_resource *r, struct rdt_domain *d,struct rdt_hw_domain *hw_dom =resctrl_to_arch_dom (d);struct arch_mbm_state *am;u64 msr_val, chunks; +int ret;if (!cpumask_test_cpu (smp_processor_id (), &d->cpu_mask))return -EINVAL ; -/* -* As per the SDM, when IA32_QM_EVTSEL.EvtID (bits 7:0) is configured -* with a valid event code for supported resource type and the bits -* IA32_QM_EVTSEL.RMID (bits 41:32) are configured with valid RMID, -* IA32_QM_CTR.data (bits 61:0) reports the monitored data. -* IA32_QM_CTR.Error (bit 63) and IA32_QM_CTR.Unavailable (bit 62) -* are error bits. -*/ -wrmsr (MSR_IA32_QM_EVTSEL , eventid, rmid); -rdmsrl (MSR_IA32_QM_CTR , msr_val); - -if (msr_val &RMID_VAL_ERROR ) -return -EIO ; -if (msr_val &RMID_VAL_UNAVAIL ) -return -EINVAL ; + ret =__rmid_read (rmid, eventid, &msr_val); +if (ret) +return ret; am =get_arch_mbm_state (hw_dom, rmid, eventid);if (am) {
base-commit: 830b3c68c1fb1e9176028d02ef86f3cf76aa2476
Signature · 1 line
2.39.0.314.g84b9a713c41-goog
PATCH v3 2/2 x86/resctrl: Avoid redundant counter read in __mon_event_count()
__mon_event_count() does the per-RMID, per-domain work for user-initiated event count reads and the initialization of new monitor groups.
In the initialization case, after resctrl_arch_reset_rmid() calls __rmid_read() to record an initial count for a new monitor group, it immediately calls resctrl_arch_rmid_read(). This re-read of the hardware counter is unnecessary and the following computations are ignored by the caller during initialization.
Following return from resctrl_arch_reset_rmid(), just clear the mbm_state and return. This involves moving the mbm_state lookup into the rr->first case, as it's not needed for regular event count reads: the QOS_L3_OCCUP_EVENT_ID case was redundant with the accumulating logic at the end of the function.
---
v3:
- changelog clarifications suggested by Reinette
v2:
(patch introduced)
v2: https://lore.kernel.org/lkml/[email protected]/
--- arch/x86/kernel/cpu/resctrl/monitor.c | 43 ++++++++++++--------------- 1 file changed, 19 insertions(+), 24 deletions(-)
arch/x86/kernel/cpu/resctrl/monitor.c +19 -24
@@ -366,41 +366,36 @@ void free_rmid(u32 rmid)list_add_tail (&entry->list, &rmid_free_lru); } +static struct mbm_state *get_mbm_state (struct rdt_domain *d,u32 rmid, +enum resctrl_event_id evtid) +{ +switch (evtid) { +case QOS_L3_MBM_TOTAL_EVENT_ID : +return &d->mbm_total[rmid]; +case QOS_L3_MBM_LOCAL_EVENT_ID : +return &d->mbm_local[rmid]; +default : +return NULL ; + } +} +static int __mon_event_count (u32 rmid,struct rmid_read *rr) {struct mbm_state *m;u64 tval =0 ; -if (rr->first) +if (rr->first) {resctrl_arch_reset_rmid (rr->r, rr->d, rmid, rr->evtid); + m =get_mbm_state (rr->d, rmid, rr->evtid); +if (m) +memset (m,0 ,sizeof (struct mbm_state )); +return 0 ; + } rr->err =resctrl_arch_rmid_read (rr->r, rr->d, rmid, rr->evtid, &tval);if (rr->err)return rr->err; -switch (rr->evtid) { -case QOS_L3_OCCUP_EVENT_ID : - rr->val += tval; -return 0 ; -case QOS_L3_MBM_TOTAL_EVENT_ID : - m = &rr->d->mbm_total[rmid]; -break ; -case QOS_L3_MBM_LOCAL_EVENT_ID : - m = &rr->d->mbm_local[rmid]; -break ; -default : -/* -* Code would never reach here because an invalid -* event id would fail in resctrl_arch_rmid_read(). -*/ -return -EINVAL ; - } - -if (rr->first) { -memset (m,0 ,sizeof (struct mbm_state )); -return 0 ; - } - rr->val += tval;return 0 ;
Signature · 1 line
2.39.0.314.g84b9a713c41-goog
Re: [PATCH v3 2/2] x86/resctrl: Avoid redundant counter read in __mon_event_count()
Quoted 84 lines · Peter Newman
__mon_event_count()does the per-RMID, per-domain work for user-initiated event count reads and the initialization of new monitor groups.In the initialization case, after
resctrl_arch_reset_rmid()calls__rmid_read()to record an initial count for a new monitor group, it immediately callsresctrl_arch_rmid_read(). This re-read of the hardware counter is unnecessary and the following computations are ignored by the caller during initialization.Following return from
resctrl_arch_reset_rmid(), just clear thembm_stateand return. This involves moving thembm_statelookup into therr->firstcase, as it's not needed for regular event count reads: theQOS_L3_OCCUP_EVENT_IDcase was redundant with the accumulating logic at the end of the function.Signed-off-by Peter Newman <[email protected]>Reviewed-by Reinette Chatre <[email protected]>---
v3:
- changelog clarifications suggested by Reinette
v2:
(patch introduced)v2: https://lore.kernel.org/lkml/[email protected]/
--- arch/x86/kernel/cpu/resctrl/monitor.c | 43 ++++++++++++--------------- 1 file changed, 19 insertions(+), 24 deletions(-)arch/x86/kernel/cpu/resctrl/monitor.c +19
-17@@ -366,41 +366,36 @@ void free_rmid(u32 rmid)list_add_tail (&entry->list, &rmid_free_lru); } +static struct mbm_state *get_mbm_state (struct rdt_domain *d,u32 rmid, +enum resctrl_event_id evtid) +{ +switch (evtid) { +case QOS_L3_MBM_TOTAL_EVENT_ID : +return &d->mbm_total[rmid]; +case QOS_L3_MBM_LOCAL_EVENT_ID : +return &d->mbm_local[rmid]; +default : +return NULL ; + } +} +static int __mon_event_count (u32 rmid,struct rmid_read *rr) {struct mbm_state *m;u64 tval =0 ; -if (rr->first) +if (rr->first) {resctrl_arch_reset_rmid (rr->r, rr->d, rmid, rr->evtid); + m =get_mbm_state (rr->d, rmid, rr->evtid); +if (m) +memset (m,0 ,sizeof (struct mbm_state )); +return 0 ; + } rr->err =resctrl_arch_rmid_read (rr->r, rr->d, rmid, rr->evtid, &tval);if (rr->err)return rr->err; -switch (rr->evtid) { -case QOS_L3_OCCUP_EVENT_ID : - rr->val += tval; -return 0 ; -case QOS_L3_MBM_TOTAL_EVENT_ID : - m = &rr->d->mbm_total[rmid]; -break ; -case QOS_L3_MBM_LOCAL_EVENT_ID : - m = &rr->d->mbm_local[rmid]; -break ; -default : -/* -* Code would never reach here because an invalid -* event id would fail in resctrl_arch_rmid_read(). -*/ -return -EINVAL ;
arch/x86/kernel/cpu/resctrl/monitor.c void free_rmid(u32 rmid) · line 401
Quoted 10 lines
arch/x86/kernel/cpu/resctrl/monitor.c +0
-7- } - -if (rr->first) { -memset (m,0 ,sizeof (struct mbm_state )); -return 0 ; - } - rr->val += tval;return 0 ;
Tested the patches on AMD systems. Looks good.
Thanks
Babu Moger
Quoted 109 lines · Peter Newman
When creating a new monitoring group, the RMID allocated for it may have been used by a group which was previously removed. In this case, the hardware counters will have non-zero values which should be deducted from what is reported in the new group's counts.
resctrl_arch_reset_rmid()initializes theprev_msrvalue for counters to 0, causing the initial count to be charged to the new group. Resurrect__rmid_read()and use it to initializeprev_msrcorrectly.Unlike before,
__rmid_read()checks for error bits in the MSR read so that callers don't need to.Fixes 1d81d15db39c ("x86/resctrl: Move mbm_overflow_count() into resctrl_arch_rmid_read()")Signed-off-by Peter Newman <[email protected]>Reviewed-by Reinette Chatre <[email protected]>---
v3:
- add changelog
- CC stable
v2:
- move error bit processing into__rmid_read()v1: https://lore.kernel.org/lkml/[email protected]/
v2: https://lore.kernel.org/lkml/[email protected]/--- arch/x86/kernel/cpu/resctrl/monitor.c | 49 ++++++++++++++++++--------- 1 file changed, 33 insertions(+), 16 deletions(-)arch/x86/kernel/cpu/resctrl/monitor.c +33
-16@@ -146,6 +146,30 @@ static inline struct rmid_entry *__rmid_entry(u32 rmid)return entry; } +static int __rmid_read (u32 rmid,enum resctrl_event_id eventid,u64 *val) +{ +u64 msr_val; + +/* +* As per the SDM, when IA32_QM_EVTSEL.EvtID (bits 7:0) is configured +* with a valid event code for supported resource type and the bits +* IA32_QM_EVTSEL.RMID (bits 41:32) are configured with valid RMID, +* IA32_QM_CTR.data (bits 61:0) reports the monitored data. +* IA32_QM_CTR.Error (bit 63) and IA32_QM_CTR.Unavailable (bit 62) +* are error bits. +*/ +wrmsr (MSR_IA32_QM_EVTSEL , eventid, rmid); +rdmsrl (MSR_IA32_QM_CTR , msr_val); + +if (msr_val &RMID_VAL_ERROR ) +return -EIO ; +if (msr_val &RMID_VAL_UNAVAIL ) +return -EINVAL ; + + *val = msr_val; +return 0 ; +} +static struct arch_mbm_state *get_arch_mbm_state (struct rdt_hw_domain *hw_dom,u32 rmid,enum resctrl_event_id eventid)@@ -172,8 +196,12 @@ void resctrl_arch_reset_rmid(struct rdt_resource *r, struct rdt_domain *d,struct arch_mbm_state *am; am =get_arch_mbm_state (hw_dom, rmid, eventid); -if (am) +if (am) {memset (am,0 ,sizeof (*am)); + +/* Record any initial, non-zero count value. */ +__rmid_read (rmid, eventid, &am->prev_msr); + } }static u64 mbm_overflow_count (u64 prev_msr,u64 cur_msr,unsigned int width)@@ -191,25 +219,14 @@ int resctrl_arch_rmid_read(struct rdt_resource *r, struct rdt_domain *d,struct rdt_hw_domain *hw_dom =resctrl_to_arch_dom (d);struct arch_mbm_state *am;u64 msr_val, chunks; +int ret;if (!cpumask_test_cpu (smp_processor_id (), &d->cpu_mask))return -EINVAL ; -/* -* As per the SDM, when IA32_QM_EVTSEL.EvtID (bits 7:0) is configured -* with a valid event code for supported resource type and the bits -* IA32_QM_EVTSEL.RMID (bits 41:32) are configured with valid RMID, -* IA32_QM_CTR.data (bits 61:0) reports the monitored data. -* IA32_QM_CTR.Error (bit 63) and IA32_QM_CTR.Unavailable (bit 62) -* are error bits. -*/ -wrmsr (MSR_IA32_QM_EVTSEL , eventid, rmid); -rdmsrl (MSR_IA32_QM_CTR , msr_val); - -if (msr_val &RMID_VAL_ERROR ) -return -EIO ; -if (msr_val &RMID_VAL_UNAVAIL ) -return -EINVAL ; + ret =__rmid_read (rmid, eventid, &msr_val); +if (ret) +return ret; am =get_arch_mbm_state (hw_dom, rmid, eventid);if (am) {
Quoted 1 line
base-commit:
830b3c68c1fb1e9176028d02ef86f3cf76aa2476
Tested the patches on AMD systems. Looks good.
Thanks
Babu Moger
The following commit has been merged into the x86/urgent branch of tip:
The following commit has been merged into the x86/urgent branch of tip:
Commit-ID: 2a81160d29d65b5876ab3f824fda99ae0219f05e Gitweb: https://git.kernel.org/tip/2a81160d29d65b5876ab3f824fda99ae0219f05e Author: Peter Newman <[email protected]> AuthorDate: Tue, 20 Dec 2022 17:41:31 +01:00 Committer: Borislav Petkov (AMD) <[email protected]> CommitterDate: Tue, 10 Jan 2023 19:51:59 +01:00
x86/resctrl: Fix event counts regression in reused RMIDs
When creating a new monitoring group, the RMID allocated for it may have been used by a group which was previously removed. In this case, the hardware counters will have non-zero values which should be deducted from what is reported in the new group's counts.
resctrl_arch_reset_rmid() initializes the prev_msr value for counters to 0, causing the initial count to be charged to the new group. Resurrect __rmid_read() and use it to initialize prev_msr correctly.
Unlike before, __rmid_read() checks for error bits in the MSR read so that callers don't need to.
--- arch/x86/kernel/cpu/resctrl/monitor.c | 49 +++++++++++++++++--------- 1 file changed, 33 insertions(+), 16 deletions(-)
arch/x86/kernel/cpu/resctrl/monitor.c +33 -16
@@ -146,6 +146,30 @@ static inline struct rmid_entry *__rmid_entry(u32 rmid)return entry; } +static int __rmid_read (u32 rmid,enum resctrl_event_id eventid,u64 *val) +{ +u64 msr_val; + +/* +* As per the SDM, when IA32_QM_EVTSEL.EvtID (bits 7:0) is configured +* with a valid event code for supported resource type and the bits +* IA32_QM_EVTSEL.RMID (bits 41:32) are configured with valid RMID, +* IA32_QM_CTR.data (bits 61:0) reports the monitored data. +* IA32_QM_CTR.Error (bit 63) and IA32_QM_CTR.Unavailable (bit 62) +* are error bits. +*/ +wrmsr (MSR_IA32_QM_EVTSEL , eventid, rmid); +rdmsrl (MSR_IA32_QM_CTR , msr_val); + +if (msr_val &RMID_VAL_ERROR ) +return -EIO ; +if (msr_val &RMID_VAL_UNAVAIL ) +return -EINVAL ; + + *val = msr_val; +return 0 ; +} +static struct arch_mbm_state *get_arch_mbm_state (struct rdt_hw_domain *hw_dom,u32 rmid,enum resctrl_event_id eventid)
@@ -172,8 +196,12 @@ void resctrl_arch_reset_rmid(struct rdt_resource *r, struct rdt_domain *d,struct arch_mbm_state *am; am =get_arch_mbm_state (hw_dom, rmid, eventid); -if (am) +if (am) {memset (am,0 ,sizeof (*am)); + +/* Record any initial, non-zero count value. */ +__rmid_read (rmid, eventid, &am->prev_msr); + } }static u64 mbm_overflow_count (u64 prev_msr,u64 cur_msr,unsigned int width)
@@ -191,25 +219,14 @@ int resctrl_arch_rmid_read(struct rdt_resource *r, struct rdt_domain *d,struct rdt_hw_domain *hw_dom =resctrl_to_arch_dom (d);struct arch_mbm_state *am;u64 msr_val, chunks; +int ret;if (!cpumask_test_cpu (smp_processor_id (), &d->cpu_mask))return -EINVAL ; -/* -* As per the SDM, when IA32_QM_EVTSEL.EvtID (bits 7:0) is configured -* with a valid event code for supported resource type and the bits -* IA32_QM_EVTSEL.RMID (bits 41:32) are configured with valid RMID, -* IA32_QM_CTR.data (bits 61:0) reports the monitored data. -* IA32_QM_CTR.Error (bit 63) and IA32_QM_CTR.Unavailable (bit 62) -* are error bits. -*/ -wrmsr (MSR_IA32_QM_EVTSEL , eventid, rmid); -rdmsrl (MSR_IA32_QM_CTR , msr_val); - -if (msr_val &RMID_VAL_ERROR ) -return -EIO ; -if (msr_val &RMID_VAL_UNAVAIL ) -return -EINVAL ; + ret =__rmid_read (rmid, eventid, &msr_val); +if (ret) +return ret; am =get_arch_mbm_state (hw_dom, rmid, eventid);if (am) {