Miscellaneous Intel uncore patches
Subsystems (2)
v6 cover Miscellaneous Intel uncore patches
This is mainly a RESEND of the patch set below, with patch v5 3/4 split into two patches.
I grouped these patches together for the convenience of the maintainers and bumped the whole series to V6.
Changes since v5:
- Split patch v5 3/4 into two patches for clearer scope.
- Patch 2/5: Add changelog to clarify that it doesn't support die ID gaps.
- Patch 3/5: Change title and changelog to reflect the reduced scope.
- Patch 4/5: New patch. Remove redundant snbep_pci2phy_map_init() calls.
Changes since v4:
- Remove unused die_id variable from patch 3.
V5:
https://lore.kernel.org/lkml/[email protected]/
v4:
https://lore.kernel.org/lkml/[email protected]/
Previous versions:
https://lore.kernel.org/lkml/[email protected]/T/
https://lore.kernel.org/lkml/[email protected]/
https://lore.kernel.org/lkml/[email protected]/
Zide Chen (5):
perf/x86/intel/uncore: Fix iounmap() leak on global_init failure perf/x86/intel/uncore: Skip discovery table for offline dies perf/x86/intel/uncore: Do not treat -1 die_id as error during UBOX
scan
perf/x86/intel/uncore: Fix PMON enumeration with NUMA disabled perf/x86/intel/uncore: Remove extra double quote mark
arch/x86/events/intel/uncore.c | 1 + arch/x86/events/intel/uncore_discovery.c | 17 +++-- arch/x86/events/intel/uncore_snbep.c | 83 +++++++++++------------- 3 files changed, 49 insertions(+), 52 deletions(-)
Signature · 1 line
2.53.0
PATCH v6 1/5 perf/x86/intel/uncore: Fix iounmap() leak on global_init failure
Kernel test robot reported:
Unverified Error/Warning (likely false positive, kindly check if interested):
arch/x86/events/intel/uncore_discovery.c:293:2-8:
ERROR: missing iounmap; ioremap on line 288 and execution via conditional on line 292
If domain->global_init() fails in __parse_discovery_table(), the ioremap'ed MMIO region is not released before returning, resulting in an MMIO mapping leak.
---
V2:
- As suggested by Markus, add an out label and use goto-based error handling to reduce duplicated iounmap() code.
- Add the original warning from the kernel test robot to the commit message.
- Trivial rewording of the commit message.
V3:
- Add Reviewed-by tag.
--- arch/x86/events/intel/uncore_discovery.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-)
arch/x86/events/intel/uncore_discovery.c +10 -5
@@ -264,6 +264,7 @@ static int __parse_discovery_table(struct uncore_discovery_domain *domain,struct uncore_unit_discovery unit;void __iomem *io_addr;unsigned long size; +int ret =0 ;int i; size =UNCORE_DISCOVERY_GLOBAL_MAP_SIZE ;
@@ -273,21 +274,23 @@ static int __parse_discovery_table(struct uncore_discovery_domain *domain,/* Read Global Discovery State */ memcpy_fromio (&global, io_addr,sizeof (struct uncore_global_discovery )); +iounmap (io_addr); +if (uncore_discovery_invalid_unit (global)) {pr_info ("Invalid Global Discovery State: 0x%llx 0x%llx 0x%llx\n" , global.table1, global.ctl, global.table3); -iounmap (io_addr);return -EINVAL ; } -iounmap (io_addr); size = (1 + global.max_units) * global.stride *8 ; io_addr =ioremap (addr, size);if (!io_addr)return -ENOMEM ; -if (domain->global_init && domain->global_init (global.ctl)) -return -ENODEV ; +if (domain->global_init && domain->global_init (global.ctl)) { + ret = -ENODEV ; + goto out; + }/* Parsing Unit Discovery State */ for (i =0 ; i < global.max_units; i++) {
@@ -307,8 +310,10 @@ static int __parse_discovery_table(struct uncore_discovery_domain *domain, } *parsed = true; + +out:iounmap (io_addr); -return 0 ; +return ret; } static int parse_discovery_table(struct uncore_discovery_domain *domain,
Signature · 1 line
2.53.0
PATCH v6 2/5 perf/x86/intel/uncore: Skip discovery table for offline dies
In reply to Zide ChenThis warning can be triggered if NUMA is disabled and the system boots with fewer CPUs than the number of CPUs in die 0.
WARNING: CPU: 9 PID: 7257 at uncore.c:1157 uncore_pci_pmu_register+0x136/0x160 [intel_uncore]
Currently, the discovery table continues to be parsed even if all CPUs in the associated die are offline. This can lead to an array overflow at "pmu->boxes[die] = box" in uncore_pci_pmu_register(), which may trigger the warning above or cause other issues.
---
V2:
- Add the Tested-by tag
- Rebase onto perf/core (base commit: a491c02c2770)
V3:
- Remove the overly cautious WARN_ON()
V4:
- Add Reviewed-by tag.
--- arch/x86/events/intel/uncore_discovery.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/events/intel/uncore_discovery.c +1 -1
@@ -371,7 +371,7 @@ static bool uncore_discovery_pci(struct uncore_discovery_domain *domain) (val &UNCORE_DISCOVERY_DVSEC2_BIR_MASK ) *UNCORE_DISCOVERY_BIR_STEP ; die =get_device_die_id (dev); -if (die <0 ) +if ((die <0 ) || (die >=uncore_max_dies ()))continue ;parse_discovery_table (domain, dev, die, bar_offset, &parsed);
Signature · 1 line
2.53.0
PATCH v6 3/5 perf/x86/intel/uncore: Do not treat -1 die_id as error during UBOX scan
In reply to Zide ChenIn snbep_pci2phy_map_init(), in the nr_node_ids > 8 path, uncore_device_to_die() may return -1 when all CPUs associated with the UBOX device are offline.
Remove the WARN_ON_ONCE(die_id == -1) check for two reasons:
- The current code breaks out of the loop. This is incorrect because pci_get_device() does not guarantee iteration in domain or bus order, so additional UBOX devices may be skipped during the scan.
- Returning -EINVAL is incorrect, since marking offline buses with die_id == -1 is expected and should not be treated as an error.
---
V2:
- Fix the commit message to note that spr_update_device_location() is used by EMR, not GNR.
- Rewrite the commit message for clarity.
- Add a Tested-by tag.
V5:
- Removed unused die_id (Dapeng).
V6:
- Move the spr_update_device_location() change to a separate patch.
- Update the comit message and title to reflect the reduced scope of this patch.
- Remove Tested-by since the code has changed.
--- arch/x86/events/intel/uncore_snbep.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-)
arch/x86/events/intel/uncore_snbep.c +2 -7
@@ -1413,7 +1413,7 @@ static int topology_gidnid_map(int nodeid, u32 gidnid)static int snbep_pci2phy_map_init (int devid,int nodeid_loc,int idmap_loc,bool reverse) {struct pci_dev *ubox_dev =NULL ; -int i, bus, nodeid, segment, die_id; +int i, bus, nodeid, segment;struct pci2phy_map *map;int err =0 ;u32 config =0 ;
@@ -1458,14 +1458,9 @@ static int snbep_pci2phy_map_init(int devid, int nodeid_loc, int idmap_loc, boolbreak ; } - map->pbus_to_dieid[bus] = die_id =uncore_device_to_die (ubox_dev); + map->pbus_to_dieid[bus] =uncore_device_to_die (ubox_dev);raw_spin_unlock (&pci2phy_map_lock); - -if (WARN_ON_ONCE (die_id ==-1 )) { - err = -EINVAL ; -break ; - } } }
Signature · 1 line
2.53.0
Re: [PATCH V6 3/5] perf/x86/intel/uncore: Do not treat -1 die_id as error during UBOX scan
Quoted 53 lines · Zide Chen
In
snbep_pci2phy_map_init(), in thenr_node_ids> 8 path,uncore_device_to_die()may return -1 when all CPUs associated with the UBOX device are offline.Remove the
WARN_ON_ONCE(die_id == -1)check for two reasons:- The current code breaks out of the loop. This is incorrect because
pci_get_device()does not guarantee iteration in domain or bus order, so additional UBOX devices may be skipped during the scan.- Returning
-EINVALis incorrect, since marking offline buses withdie_id== -1 is expected and should not be treated as an error.Fixes 9a7832ce3d92 ("perf/x86/intel/uncore: With > 8 nodes, get pci bus die id from NUMA info")Signed-off-by Zide Chen <[email protected]>---
V2:
- Fix the commit message to note thatspr_update_device_location()is used by EMR, not GNR.
- Rewrite the commit message for clarity.
- Add a Tested-by tag.V5:
- Removed unuseddie_id(Dapeng).V6:
- Move thespr_update_device_location()change to a separate patch.
- Update the comit message and title to reflect the reduced scope of this patch.
- Remove Tested-by since the code has changed.--- arch/x86/events/intel/uncore_snbep.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-)arch/x86/events/intel/uncore_snbep.c +2
-2@@ -1413,7 +1413,7 @@ static int topology_gidnid_map(int nodeid, u32 gidnid)static int snbep_pci2phy_map_init (int devid,int nodeid_loc,int idmap_loc,bool reverse) {struct pci_dev *ubox_dev =NULL ; -int i, bus, nodeid, segment, die_id; +int i, bus, nodeid, segment;struct pci2phy_map *map;int err =0 ;u32 config =0 ;@@ -1458,14 +1458,9 @@ static int snbep_pci2phy_map_init(int devid, int nodeid_loc, int idmap_loc, boolbreak ; } - map->pbus_to_dieid[bus] = die_id =uncore_device_to_die (ubox_dev); + map->pbus_to_dieid[bus] =uncore_device_to_die (ubox_dev);
arch/x86/events/intel/uncore_snbep.c static int snbep_pci2phy_map_init(int devid, int nodeid_loc, int idmap_loc, bool · line 1466
Quoted 10 lines
arch/x86/events/intel/uncore_snbep.c +0
-5raw_spin_unlock (&pci2phy_map_lock); - -if (WARN_ON_ONCE (die_id ==-1 )) { - err = -EINVAL ; -break ; - } } }
PATCH v6 4/5 perf/x86/intel/uncore: Fix PMON enumeration with NUMA disabled
In reply to Zide ChenWhen NUMA is disabled on a NUMA-capable platform, UPI and M3UPI PMON units are not enumerated.
In this case, pcibus_to_node() always returns NUMA_NO_NODE, causing uncore_device_to_die() to return -1 for all PCI devices. As a result, the corresponding PMON units are not added to the RB tree.
These PMON units are per-die resources, and their utility when NUMA is disabled is limited. The driver does not prohibit their use, and the enumeration should still work correctly.
Fix this by using uncore_pcibus_to_dieid(), which works regardless of whether NUMA is enabled. This requires calling snbep_pci2phy_map_init() in spr_uncore_pci_init().
Since pci_init() is called before mmio_init(), remove the redundant snbep_pci2phy_map_init() call from spr_uncore_mmio_init(). If snbep_pci2phy_map_init() fails, uncore driver should be bailed out, so the fallback path in spr_uncore_mmio_init() can be removed.
---
V6:
- Split from patch v5 3/4.
- Remove the redundant call in spr_uncore_mmio_init().
- Update commit messages.
--- arch/x86/events/intel/uncore.c | 1 + arch/x86/events/intel/uncore_snbep.c | 26 +++++++++++--------------- 2 files changed, 12 insertions(+), 15 deletions(-)
arch/x86/events/intel/uncore.c +1 -0
@@ -67,6 +67,7 @@ int uncore_die_to_segment(int die)return bus ?pci_domain_nr (bus) : -EINVAL ; } +/* Note: This API can only be used when NUMA information is available. */ int uncore_device_to_die (struct pci_dev *dev) {int node =pcibus_to_node (dev->bus);
arch/x86/events/intel/uncore_snbep.c +11 -15
@@ -6415,7 +6415,7 @@ static void spr_update_device_location(int type_id)while ((dev =pci_get_device (PCI_VENDOR_ID_INTEL , device, dev)) !=NULL ) { - die =uncore_device_to_die (dev); + die =uncore_pcibus_to_dieid (dev->bus);if (die <0 )continue ;
@@ -6439,6 +6439,10 @@ static void spr_update_device_location(int type_id)int spr_uncore_pci_init (void ) { +int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); +if (ret) +return ret; +/* * The discovery table of UPI on some SPR variant is broken, * which impacts the detection of both UPI and M3UPI uncore PMON.
@@ -6460,21 +6464,13 @@ int spr_uncore_pci_init(void)void spr_uncore_mmio_init (void ) { -int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); + uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , +UNCORE_SPR_MMIO_EXTRA_UNCORES , + spr_mmio_uncores, +UNCORE_SPR_NUM_UNCORE_TYPES , + spr_uncores); -if (ret) { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO ,0 ,NULL , -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); - }else { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , -UNCORE_SPR_MMIO_EXTRA_UNCORES , - spr_mmio_uncores, -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); - - spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ; - } + spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ; }/* end of SPR uncore support */
Signature · 1 line
2.53.0
Re: [PATCH V6 4/5] perf/x86/intel/uncore: Fix PMON enumeration with NUMA disabled
Quoted 82 lines · Zide Chen
When NUMA is disabled on a NUMA-capable platform, UPI and M3UPI PMON units are not enumerated.
In this case,
pcibus_to_node()always returnsNUMA_NO_NODE, causinguncore_device_to_die()to return -1 for all PCI devices. As a result, the corresponding PMON units are not added to the RB tree.These PMON units are per-die resources, and their utility when NUMA is disabled is limited. The driver does not prohibit their use, and the enumeration should still work correctly.
Fix this by using
uncore_pcibus_to_dieid(), which works regardless of whether NUMA is enabled. This requires callingsnbep_pci2phy_map_init()inspr_uncore_pci_init().Since
pci_init()is called beforemmio_init(), remove the redundantsnbep_pci2phy_map_init()call fromspr_uncore_mmio_init(). Ifsnbep_pci2phy_map_init()fails, uncore driver should be bailed out, so the fallback path inspr_uncore_mmio_init()can be removed.Signed-off-by Zide Chen <[email protected]>---
V6:
- Split from patch v5 3/4.
- Remove the redundant call inspr_uncore_mmio_init().
- Update commit messages.--- arch/x86/events/intel/uncore.c | 1 + arch/x86/events/intel/uncore_snbep.c | 26 +++++++++++--------------- 2 files changed, 12 insertions(+), 15 deletions(-)arch/x86/events/intel/uncore.c +1
-0@@ -67,6 +67,7 @@ int uncore_die_to_segment(int die)return bus ?pci_domain_nr (bus) : -EINVAL ; } +/* Note: This API can only be used when NUMA information is available. */ int uncore_device_to_die (struct pci_dev *dev) {int node =pcibus_to_node (dev->bus);arch/x86/events/intel/uncore_snbep.c +10
-6@@ -6415,7 +6415,7 @@ static void spr_update_device_location(int type_id)while ((dev =pci_get_device (PCI_VENDOR_ID_INTEL , device, dev)) !=NULL ) { - die =uncore_device_to_die (dev); + die =uncore_pcibus_to_dieid (dev->bus);if (die <0 )continue ;@@ -6439,6 +6439,10 @@ static void spr_update_device_location(int type_id)int spr_uncore_pci_init (void ) { +int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); +if (ret) +return ret; +/* * The discovery table of UPI on some SPR variant is broken, * which impacts the detection of both UPI and M3UPI uncore PMON. @@ -6460,21 +6464,13 @@ int spr_uncore_pci_init(void)void spr_uncore_mmio_init (void ) { -int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); + uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , +UNCORE_SPR_MMIO_EXTRA_UNCORES , + spr_mmio_uncores, +UNCORE_SPR_NUM_UNCORE_TYPES , + spr_uncores); -if (ret) { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO ,0 ,NULL , -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores);
arch/x86/events/intel/uncore_snbep.c int spr_uncore_pci_init(void) · line 6473
Quoted 10 lines
arch/x86/events/intel/uncore_snbep.c +1
-9- }else { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , -UNCORE_SPR_MMIO_EXTRA_UNCORES , - spr_mmio_uncores, -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); - - spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ; - } + spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ;
I'm not sure if we can directly remove the snbep_pci2phy_map_init() call here. In theory, the snbep_pci2phy_map_init() call in spr_uncore_pci_init() could fail and then spr_uncore_mmio_init() doesn't know it and directly initializes MMIO PMU, then it could lead to the MMIO initialization fails.
Currently the PCI, CPU and MMIO initialization are totally independent, only when the 3 types initialization all fail, then uncore PMU can abort.
```
if (uncore_init->pci_init) {
pret = uncore_init->pci_init();
if (!pret)
pret = uncore_pci_init();
}
if (uncore_init->cpu_init) {
uncore_init->cpu_init();
cret = uncore_cpu_init();
}
if (uncore_init->mmio_init) {
uncore_init->mmio_init();
mret = uncore_mmio_init();
}
if (cret && pret && mret) {
ret = -ENODEV;
goto free_discovery;
}
```Quoted 3 lines
arch/x86/events/intel/uncore_snbep.c +0
-0}/* end of SPR uncore support */
Re: [PATCH V6 4/5] perf/x86/intel/uncore: Fix PMON enumeration with NUMA disabled
Quoted 91 lines · Dapeng
Quoted 89 lines · Zide Chen
When NUMA is disabled on a NUMA-capable platform, UPI and M3UPI PMON units are not enumerated.
In this case,
pcibus_to_node()always returnsNUMA_NO_NODE, causinguncore_device_to_die()to return -1 for all PCI devices. As a result, the corresponding PMON units are not added to the RB tree.These PMON units are per-die resources, and their utility when NUMA is disabled is limited. The driver does not prohibit their use, and the enumeration should still work correctly.
Fix this by using
uncore_pcibus_to_dieid(), which works regardless of whether NUMA is enabled. This requires callingsnbep_pci2phy_map_init()inspr_uncore_pci_init().Since
pci_init()is called beforemmio_init(), remove the redundantsnbep_pci2phy_map_init()call fromspr_uncore_mmio_init(). Ifsnbep_pci2phy_map_init()fails, uncore driver should be bailed out, so the fallback path inspr_uncore_mmio_init()can be removed.Signed-off-by Zide Chen <[email protected]>---
V6:
- Split from patch v5 3/4.
- Remove the redundant call inspr_uncore_mmio_init().
- Update commit messages.--- arch/x86/events/intel/uncore.c | 1 + arch/x86/events/intel/uncore_snbep.c | 26 +++++++++++--------------- 2 files changed, 12 insertions(+), 15 deletions(-)arch/x86/events/intel/uncore.c +1
-0@@ -67,6 +67,7 @@ int uncore_die_to_segment(int die)return bus ?pci_domain_nr (bus) : -EINVAL ; } +/* Note: This API can only be used when NUMA information is available. */ int uncore_device_to_die (struct pci_dev *dev) {int node =pcibus_to_node (dev->bus);arch/x86/events/intel/uncore_snbep.c +10
-13@@ -6415,7 +6415,7 @@ static void spr_update_device_location(int type_id)while ((dev =pci_get_device (PCI_VENDOR_ID_INTEL , device, dev)) !=NULL ) { - die =uncore_device_to_die (dev); + die =uncore_pcibus_to_dieid (dev->bus);if (die <0 )continue ;@@ -6439,6 +6439,10 @@ static void spr_update_device_location(int type_id)int spr_uncore_pci_init (void ) { +int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); +if (ret) +return ret; +/* * The discovery table of UPI on some SPR variant is broken, * which impacts the detection of both UPI and M3UPI uncore PMON. @@ -6460,21 +6464,13 @@ int spr_uncore_pci_init(void)void spr_uncore_mmio_init (void ) { -int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); + uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , +UNCORE_SPR_MMIO_EXTRA_UNCORES , + spr_mmio_uncores, +UNCORE_SPR_NUM_UNCORE_TYPES , + spr_uncores); -if (ret) { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO ,0 ,NULL , -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); - }else { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , -UNCORE_SPR_MMIO_EXTRA_UNCORES , - spr_mmio_uncores, -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); -
Quoted 8 lines
arch/x86/events/intel/uncore_snbep.c int spr_uncore_pci_init(void) · line 6473
Quoted 3 lines
arch/x86/events/intel/uncore_snbep.c +1
-2- spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ; - } + spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ;I'm not sure if we can directly remove the
snbep_pci2phy_map_init()call here. In theory, thesnbep_pci2phy_map_init()call inspr_uncore_pci_init()could fail and thenspr_uncore_mmio_init()doesn't know it and directly initializes MMIO PMU, then it could lead to the MMIO initialization fails.
Yes, this is true. But I would argue that the fix in this patch is correct, and the issue you pointed out is not new: the uncore driver registers a PMU device without guaranteeing it's functioning.
This is because the Intel uncore driver employs a lazy init approach. And when init_box() fails, it doesn't unregister the inaccessible PMU devices. For example, intel_generic_uncore_mmio_init_box() could fail for a number of reasons, making all associated PMU devices non-functional.
Originally the uncore driver tried to enumerate PCI/MSR/MMIO uncore independently, but evolving hardware complexity makes this more challenging. This patch is just one example, IMC Freerunning is MMIO-accessed but relies on PCI devices to read the die-specific MMIO base address. Explicitly gating sysfs node creation with PCI init code in mmio_init() is neither clean nor reliable.
To fix it, it seems reasonable to have init_box() return int and unregister the PMU device if deemed inaccessible — similar to what perf_event_ibs_init() does.
arch/x86/events/intel/uncore.h +2 -2
@@ -129,7 +129,7 @@ struct intel_uncore_type {#define events_group attr_groups[2] struct intel_uncore_ops { - void (*init_box)(struct intel_uncore_box *); + int (*init_box)(struct intel_uncore_box *); --- a/arch/x86/events/intel/uncore.c +++ b/arch/x86/events/intel/uncore.c
@@ -1155,7 +1155,8 @@ static int uncore_pci_pmu_register(struct pci_dev
*pdev,
box->dieid = die;
box->pci_dev = pdev;
box->pmu = pmu;- uncore_box_init(box);
+ ret = uncore_box_init(box);
+ if (ret)
+ return ret;
@@ -1598,8 +1599,10 @@ static int uncore_box_ref(struct intel_uncore_type **types,
pmu = type->pmus;
for (i = 0; i < type->num_boxes; i++, pmu++) {
box = pmu->boxes[id];- if (box && box->cpu >= 0 &&atomic_inc_return(&box->refcnt) == 1)
- uncore_box_init(box);
+ if (box && box->cpu >= 0 &&atomic_inc_return(&box->refcnt) == 1)
+ if (uncore_box_init(box))
+ uncore_pmu_unregister(pmu);
Quoted 31 lines
Currently the PCI, CPU and MMIO initialization are totally independent, only when the 3 types initialization all fail, then uncore PMU can abort.
``` if (uncore_init->pci_init) { pret = uncore_init->pci_init(); if (!pret) pret = uncore_pci_init(); } if (uncore_init->cpu_init) { uncore_init->cpu_init(); cret = uncore_cpu_init(); } if (uncore_init->mmio_init) { uncore_init->mmio_init(); mret = uncore_mmio_init(); } if (cret && pret && mret) { ret = -ENODEV; goto free_discovery; } ```Quoted 3 lines
arch/x86/events/intel/uncore_snbep.c +0
-0}/* end of SPR uncore support */
Re: [PATCH V6 4/5] perf/x86/intel/uncore: Fix PMON enumeration with NUMA disabled
Quoted 143 lines · Zide
Quoted 89 lines · Dapeng
Quoted 88 lines · Zide Chen
When NUMA is disabled on a NUMA-capable platform, UPI and M3UPI PMON units are not enumerated.
In this case,
pcibus_to_node()always returnsNUMA_NO_NODE, causinguncore_device_to_die()to return -1 for all PCI devices. As a result, the corresponding PMON units are not added to the RB tree.These PMON units are per-die resources, and their utility when NUMA is disabled is limited. The driver does not prohibit their use, and the enumeration should still work correctly.
Fix this by using
uncore_pcibus_to_dieid(), which works regardless of whether NUMA is enabled. This requires callingsnbep_pci2phy_map_init()inspr_uncore_pci_init().Since
pci_init()is called beforemmio_init(), remove the redundantsnbep_pci2phy_map_init()call fromspr_uncore_mmio_init(). Ifsnbep_pci2phy_map_init()fails, uncore driver should be bailed out, so the fallback path inspr_uncore_mmio_init()can be removed.Signed-off-by Zide Chen <[email protected]>---
V6:
- Split from patch v5 3/4.
- Remove the redundant call inspr_uncore_mmio_init().
- Update commit messages.--- arch/x86/events/intel/uncore.c | 1 + arch/x86/events/intel/uncore_snbep.c | 26 +++++++++++--------------- 2 files changed, 12 insertions(+), 15 deletions(-)arch/x86/events/intel/uncore.c +1
-0@@ -67,6 +67,7 @@ int uncore_die_to_segment(int die)return bus ?pci_domain_nr (bus) : -EINVAL ; } +/* Note: This API can only be used when NUMA information is available. */ int uncore_device_to_die (struct pci_dev *dev) {int node =pcibus_to_node (dev->bus);arch/x86/events/intel/uncore_snbep.c +10
-12@@ -6415,7 +6415,7 @@ static void spr_update_device_location(int type_id)while ((dev =pci_get_device (PCI_VENDOR_ID_INTEL , device, dev)) !=NULL ) { - die =uncore_device_to_die (dev); + die =uncore_pcibus_to_dieid (dev->bus);if (die <0 )continue ;@@ -6439,6 +6439,10 @@ static void spr_update_device_location(int type_id)int spr_uncore_pci_init (void ) { +int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); +if (ret) +return ret; +/* * The discovery table of UPI on some SPR variant is broken, * which impacts the detection of both UPI and M3UPI uncore PMON. @@ -6460,21 +6464,13 @@ int spr_uncore_pci_init(void)void spr_uncore_mmio_init (void ) { -int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); + uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , +UNCORE_SPR_MMIO_EXTRA_UNCORES , + spr_mmio_uncores, +UNCORE_SPR_NUM_UNCORE_TYPES , + spr_uncores); -if (ret) { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO ,0 ,NULL , -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); - }else { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , -UNCORE_SPR_MMIO_EXTRA_UNCORES , - spr_mmio_uncores, -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores);Quoted 8 lines
arch/x86/events/intel/uncore_snbep.c int spr_uncore_pci_init(void) · line 6473
Quoted 4 lines
arch/x86/events/intel/uncore_snbep.c +1
-3- - spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ; - } + spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ;I'm not sure if we can directly remove the
snbep_pci2phy_map_init()call here. In theory, thesnbep_pci2phy_map_init()call inspr_uncore_pci_init()could fail and thenspr_uncore_mmio_init()doesn't know it and directly initializes MMIO PMU, then it could lead to the MMIO initialization fails.Yes, this is true. But I would argue that the fix in this patch is correct, and the issue you pointed out is not new: the uncore driver registers a PMU device without guaranteeing it's functioning.
This is because the Intel uncore driver employs a lazy init approach. And when
init_box()fails, it doesn't unregister the inaccessible PMU devices. For example,intel_generic_uncore_mmio_init_box()could fail for a number of reasons, making all associated PMU devices non-functional.Originally the uncore driver tried to enumerate PCI/MSR/MMIO uncore independently, but evolving hardware complexity makes this more challenging. This patch is just one example, IMC Freerunning is MMIO-accessed but relies on PCI devices to read the die-specific MMIO base address. Explicitly gating sysfs node creation with PCI init code in
mmio_init()is neither clean nor reliable.To fix it, it seems reasonable to have
init_box()return int and unregister the PMU device if deemed inaccessible — similar to whatperf_event_ibs_init()does.arch/x86/events/intel/uncore.h +1
-1@@ -129,7 +129,7 @@ struct intel_uncore_type {#define events_group attr_groups[2]struct intel_uncore_ops { -void (*init_box)(struct intel_uncore_box *); +int (*init_box)(struct intel_uncore_box *);arch/x86/events/intel/uncore.c +3
-1@@ -1155,7 +1155,8 @@ static int uncore_pci_pmu_register(struct pci_dev *pdev, box->dieid = die; box->pci_dev = pdev; box->pmu = pmu; -uncore_box_init (box); + ret =uncore_box_init (box); +if (ret) +return ret;@@ -1598,8 +1599,10 @@ static int uncore_box_ref(struct intel_uncore_type **types,
arch/x86/events/intel/uncore.c static int uncore_box_ref(struct · line 1607
Quoted 10 lines
arch/x86/events/intel/uncore.c +3
-2pmu = type->pmus;for (i =0 ; i < type->num_boxes; i++, pmu++) { box = pmu->boxes[id]; -if (box && box->cpu >=0 && atomic_inc_return(&box->refcnt) == 1) -uncore_box_init (box); +if (box && box->cpu >=0 &&atomic_inc_return (&box->refcnt) ==1 ) +if (uncore_box_init (box)) +uncore_pmu_unregister (pmu);
Yes, I like this idea. The return value of init_box() should always be checked. I'm not quite sure if there are other resources need to be cleaned besides unregistering the corresponding uncore pmu, please double check. Thanks.
Quoted 33 lines
Quoted 33 lines
Currently the PCI, CPU and MMIO initialization are totally independent, only when the 3 types initialization all fail, then uncore PMU can abort.
``` if (uncore_init->pci_init) { pret = uncore_init->pci_init(); if (!pret) pret = uncore_pci_init(); } if (uncore_init->cpu_init) { uncore_init->cpu_init(); cret = uncore_cpu_init(); } if (uncore_init->mmio_init) { uncore_init->mmio_init(); mret = uncore_mmio_init(); } if (cret && pret && mret) { ret = -ENODEV; goto free_discovery; } ```Quoted 3 lines
arch/x86/events/intel/uncore_snbep.c +0
-0}/* end of SPR uncore support */
Re: [PATCH V6 4/5] perf/x86/intel/uncore: Fix PMON enumeration with NUMA disabled
Quoted 152 lines · Dapeng
Quoted 150 lines · Zide
Quoted 89 lines · Dapeng
Quoted 88 lines · Zide Chen
When NUMA is disabled on a NUMA-capable platform, UPI and M3UPI PMON units are not enumerated.
In this case,
pcibus_to_node()always returnsNUMA_NO_NODE, causinguncore_device_to_die()to return -1 for all PCI devices. As a result, the corresponding PMON units are not added to the RB tree.These PMON units are per-die resources, and their utility when NUMA is disabled is limited. The driver does not prohibit their use, and the enumeration should still work correctly.
Fix this by using
uncore_pcibus_to_dieid(), which works regardless of whether NUMA is enabled. This requires callingsnbep_pci2phy_map_init()inspr_uncore_pci_init().Since
pci_init()is called beforemmio_init(), remove the redundantsnbep_pci2phy_map_init()call fromspr_uncore_mmio_init(). Ifsnbep_pci2phy_map_init()fails, uncore driver should be bailed out, so the fallback path inspr_uncore_mmio_init()can be removed.Signed-off-by Zide Chen <[email protected]>---
V6:
- Split from patch v5 3/4.
- Remove the redundant call inspr_uncore_mmio_init().
- Update commit messages.--- arch/x86/events/intel/uncore.c | 1 + arch/x86/events/intel/uncore_snbep.c | 26 +++++++++++--------------- 2 files changed, 12 insertions(+), 15 deletions(-)arch/x86/events/intel/uncore.c +1
-0@@ -67,6 +67,7 @@ int uncore_die_to_segment(int die)return bus ?pci_domain_nr (bus) : -EINVAL ; } +/* Note: This API can only be used when NUMA information is available. */ int uncore_device_to_die (struct pci_dev *dev) {int node =pcibus_to_node (dev->bus);arch/x86/events/intel/uncore_snbep.c +10
-12@@ -6415,7 +6415,7 @@ static void spr_update_device_location(int type_id)while ((dev =pci_get_device (PCI_VENDOR_ID_INTEL , device, dev)) !=NULL ) { - die =uncore_device_to_die (dev); + die =uncore_pcibus_to_dieid (dev->bus);if (die <0 )continue ;@@ -6439,6 +6439,10 @@ static void spr_update_device_location(int type_id)int spr_uncore_pci_init (void ) { +int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); +if (ret) +return ret; +/* * The discovery table of UPI on some SPR variant is broken, * which impacts the detection of both UPI and M3UPI uncore PMON. @@ -6460,21 +6464,13 @@ int spr_uncore_pci_init(void)void spr_uncore_mmio_init (void ) { -int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); + uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , +UNCORE_SPR_MMIO_EXTRA_UNCORES , + spr_mmio_uncores, +UNCORE_SPR_NUM_UNCORE_TYPES , + spr_uncores); -if (ret) { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO ,0 ,NULL , -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); - }else { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , -UNCORE_SPR_MMIO_EXTRA_UNCORES , - spr_mmio_uncores, -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores);Quoted 8 lines
arch/x86/events/intel/uncore_snbep.c int spr_uncore_pci_init(void) · line 6473
Quoted 4 lines
arch/x86/events/intel/uncore_snbep.c +1
-3- - spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ; - } + spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ;I'm not sure if we can directly remove the
snbep_pci2phy_map_init()call here. In theory, thesnbep_pci2phy_map_init()call inspr_uncore_pci_init()could fail and thenspr_uncore_mmio_init()doesn't know it and directly initializes MMIO PMU, then it could lead to the MMIO initialization fails.Yes, this is true. But I would argue that the fix in this patch is correct, and the issue you pointed out is not new: the uncore driver registers a PMU device without guaranteeing it's functioning.
This is because the Intel uncore driver employs a lazy init approach. And when
init_box()fails, it doesn't unregister the inaccessible PMU devices. For example,intel_generic_uncore_mmio_init_box()could fail for a number of reasons, making all associated PMU devices non-functional.Originally the uncore driver tried to enumerate PCI/MSR/MMIO uncore independently, but evolving hardware complexity makes this more challenging. This patch is just one example, IMC Freerunning is MMIO-accessed but relies on PCI devices to read the die-specific MMIO base address. Explicitly gating sysfs node creation with PCI init code in
mmio_init()is neither clean nor reliable.To fix it, it seems reasonable to have
init_box()return int and unregister the PMU device if deemed inaccessible — similar to whatperf_event_ibs_init()does.arch/x86/events/intel/uncore.h +1
-1@@ -129,7 +129,7 @@ struct intel_uncore_type {#define events_group attr_groups[2]struct intel_uncore_ops { -void (*init_box)(struct intel_uncore_box *); +int (*init_box)(struct intel_uncore_box *);arch/x86/events/intel/uncore.c +4
-3@@ -1155,7 +1155,8 @@ static int uncore_pci_pmu_register(struct pci_dev *pdev, box->dieid = die; box->pci_dev = pdev; box->pmu = pmu; -uncore_box_init (box); + ret =uncore_box_init (box); +if (ret) +return ret;@@ -1598,8 +1599,10 @@ static int uncore_box_ref(structintel_uncore_type **types, pmu = type->pmus;for (i =0 ; i < type->num_boxes; i++, pmu++) { box = pmu->boxes[id]; -if (box && box->cpu >=0 && atomic_inc_return(&box->refcnt) == 1) -uncore_box_init (box); + if (box && box->cpu >= 0 &&
Quoted 8 lines
arch/x86/events/intel/uncore.c static int uncore_box_ref(struct · line 1607
Quoted 3 lines
arch/x86/events/intel/uncore.c +2
-0atomic_inc_return(&box->refcnt) == 1) +if (uncore_box_init (box)) +uncore_pmu_unregister (pmu);Yes, I like this idea. The return value of
init_box()should always be checked. I'm not quite sure if there are other resources need to be cleaned besides unregistering the corresponding uncore pmu, please double check. Thanks.
I'm thinking of removing this patch from this series and putting it together with the init_box() changes, where it will be a complete fix and I need more time to double check all the init_box() callbacks.
Quoted 33 lines
Quoted 33 lines
Quoted 33 lines
Currently the PCI, CPU and MMIO initialization are totally independent, only when the 3 types initialization all fail, then uncore PMU can abort.
``` if (uncore_init->pci_init) { pret = uncore_init->pci_init(); if (!pret) pret = uncore_pci_init(); } if (uncore_init->cpu_init) { uncore_init->cpu_init(); cret = uncore_cpu_init(); } if (uncore_init->mmio_init) { uncore_init->mmio_init(); mret = uncore_mmio_init(); } if (cret && pret && mret) { ret = -ENODEV; goto free_discovery; } ```Quoted 3 lines
arch/x86/events/intel/uncore_snbep.c +0
-0}/* end of SPR uncore support */
Re: [PATCH V6 4/5] perf/x86/intel/uncore: Fix PMON enumeration with NUMA disabled
Quoted 158 lines · Zide
Quoted 140 lines · Dapeng
Quoted 139 lines · Zide
Quoted 89 lines · Dapeng
Quoted 88 lines · Zide Chen
When NUMA is disabled on a NUMA-capable platform, UPI and M3UPI PMON units are not enumerated.
In this case,
pcibus_to_node()always returnsNUMA_NO_NODE, causinguncore_device_to_die()to return -1 for all PCI devices. As a result, the corresponding PMON units are not added to the RB tree.These PMON units are per-die resources, and their utility when NUMA is disabled is limited. The driver does not prohibit their use, and the enumeration should still work correctly.
Fix this by using
uncore_pcibus_to_dieid(), which works regardless of whether NUMA is enabled. This requires callingsnbep_pci2phy_map_init()inspr_uncore_pci_init().Since
pci_init()is called beforemmio_init(), remove the redundantsnbep_pci2phy_map_init()call fromspr_uncore_mmio_init(). Ifsnbep_pci2phy_map_init()fails, uncore driver should be bailed out, so the fallback path inspr_uncore_mmio_init()can be removed.Signed-off-by Zide Chen <[email protected]>---
V6:
- Split from patch v5 3/4.
- Remove the redundant call inspr_uncore_mmio_init().
- Update commit messages.--- arch/x86/events/intel/uncore.c | 1 + arch/x86/events/intel/uncore_snbep.c | 26 +++++++++++--------------- 2 files changed, 12 insertions(+), 15 deletions(-)arch/x86/events/intel/uncore.c +1
-0@@ -67,6 +67,7 @@ int uncore_die_to_segment(int die)return bus ?pci_domain_nr (bus) : -EINVAL ; } +/* Note: This API can only be used when NUMA information is available. */ int uncore_device_to_die (struct pci_dev *dev) {int node =pcibus_to_node (dev->bus);arch/x86/events/intel/uncore_snbep.c +10
-12@@ -6415,7 +6415,7 @@ static void spr_update_device_location(int type_id)while ((dev =pci_get_device (PCI_VENDOR_ID_INTEL , device, dev)) !=NULL ) { - die =uncore_device_to_die (dev); + die =uncore_pcibus_to_dieid (dev->bus);if (die <0 )continue ;@@ -6439,6 +6439,10 @@ static void spr_update_device_location(int type_id)int spr_uncore_pci_init (void ) { +int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); +if (ret) +return ret; +/* * The discovery table of UPI on some SPR variant is broken, * which impacts the detection of both UPI and M3UPI uncore PMON. @@ -6460,21 +6464,13 @@ int spr_uncore_pci_init(void)void spr_uncore_mmio_init (void ) { -int ret =snbep_pci2phy_map_init (0x3250 ,SKX_CPUNODEID ,SKX_GIDNIDMAP , true); + uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , +UNCORE_SPR_MMIO_EXTRA_UNCORES , + spr_mmio_uncores, +UNCORE_SPR_NUM_UNCORE_TYPES , + spr_uncores); -if (ret) { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO ,0 ,NULL , -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores); - }else { - uncore_mmio_uncores =uncore_get_uncores (UNCORE_ACCESS_MMIO , -UNCORE_SPR_MMIO_EXTRA_UNCORES , - spr_mmio_uncores, -UNCORE_SPR_NUM_UNCORE_TYPES , - spr_uncores);Quoted 8 lines
arch/x86/events/intel/uncore_snbep.c int spr_uncore_pci_init(void) · line 6473
Quoted 4 lines
arch/x86/events/intel/uncore_snbep.c +1
-3- - spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ; - } + spr_uncore_imc_free_running.num_boxes =uncore_type_max_boxes (uncore_mmio_uncores,UNCORE_SPR_IMC ) /2 ;I'm not sure if we can directly remove the
snbep_pci2phy_map_init()call here. In theory, thesnbep_pci2phy_map_init()call inspr_uncore_pci_init()could fail and thenspr_uncore_mmio_init()doesn't know it and directly initializes MMIO PMU, then it could lead to the MMIO initialization fails.Yes, this is true. But I would argue that the fix in this patch is correct, and the issue you pointed out is not new: the uncore driver registers a PMU device without guaranteeing it's functioning.
This is because the Intel uncore driver employs a lazy init approach. And when
init_box()fails, it doesn't unregister the inaccessible PMU devices. For example,intel_generic_uncore_mmio_init_box()could fail for a number of reasons, making all associated PMU devices non-functional.Originally the uncore driver tried to enumerate PCI/MSR/MMIO uncore independently, but evolving hardware complexity makes this more challenging. This patch is just one example, IMC Freerunning is MMIO-accessed but relies on PCI devices to read the die-specific MMIO base address. Explicitly gating sysfs node creation with PCI init code in
mmio_init()is neither clean nor reliable.To fix it, it seems reasonable to have
init_box()return int and unregister the PMU device if deemed inaccessible — similar to whatperf_event_ibs_init()does.arch/x86/events/intel/uncore.h +1
-1@@ -129,7 +129,7 @@ struct intel_uncore_type {#define events_group attr_groups[2]struct intel_uncore_ops { -void (*init_box)(struct intel_uncore_box *); +int (*init_box)(struct intel_uncore_box *);arch/x86/events/intel/uncore.c +3
-1@@ -1155,7 +1155,8 @@ static int uncore_pci_pmu_register(struct pci_dev *pdev, box->dieid = die; box->pci_dev = pdev; box->pmu = pmu; -uncore_box_init (box); + ret =uncore_box_init (box); +if (ret) +return ret;Quoted 16 lines
arch/x86/events/intel/uncore.c static int uncore_box_ref(struct · line 1607
Quoted 12 lines
arch/x86/events/intel/uncore.c +3
-2@@ -1598,8 +1599,10 @@ static int uncore_box_ref(structintel_uncore_type **types, pmu = type->pmus;for (i =0 ; i < type->num_boxes; i++, pmu++) { box = pmu->boxes[id]; -if (box && box->cpu >=0 && atomic_inc_return(&box->refcnt) == 1) -uncore_box_init (box); +if (box && box->cpu >=0 &&atomic_inc_return (&box->refcnt) ==1 ) +if (uncore_box_init (box)) +uncore_pmu_unregister (pmu);Yes, I like this idea. The return value of
init_box()should always be checked. I'm not quite sure if there are other resources need to be cleaned besides unregistering the corresponding uncore pmu, please double check. Thanks.
Quoted 3 lines
I'm thinking of removing this patch from this series and putting it together with the
init_box()changes, where it will be a complete fix and I need more time to double check all theinit_box()callbacks.
Yeah, It's a fundamental change and let's do more tests. Thanks.
Quoted 33 lines
Quoted 33 lines
Quoted 33 lines
Quoted 33 lines
Currently the PCI, CPU and MMIO initialization are totally independent, only when the 3 types initialization all fail, then uncore PMU can abort.
``` if (uncore_init->pci_init) { pret = uncore_init->pci_init(); if (!pret) pret = uncore_pci_init(); } if (uncore_init->cpu_init) { uncore_init->cpu_init(); cret = uncore_cpu_init(); } if (uncore_init->mmio_init) { uncore_init->mmio_init(); mret = uncore_mmio_init(); } if (cret && pret && mret) { ret = -ENODEV; goto free_discovery; } ```Quoted 3 lines
arch/x86/events/intel/uncore_snbep.c +0
-0}/* end of SPR uncore support */
PATCH v6 5/5 perf/x86/intel/uncore: Remove extra double quote mark
In reply to Zide ChenThe third argument in INTEL_UNCORE_FR_EVENT_DESC() is subject to __stringify(), and the extra double quote marks can result in the expansion "3.814697266e-6" in the sysfs knobs, instead of 3.814697266e-6.
This is incorrect, though it may still work for perf, e.g. perf stat -e uncore_iio_free_running_0/bw_in_port0/
---
V2: Added Reviewed-by tags.
--- arch/x86/events/intel/uncore_snbep.c | 48 ++++++++++++++-------------- 1 file changed, 24 insertions(+), 24 deletions(-)
arch/x86/events/intel/uncore_snbep.c +24 -24
@@ -6952,34 +6952,34 @@ static struct freerunning_counters dmr_iio_freerunning[] = {static struct uncore_event_desc dmr_uncore_iio_freerunning_events[] = {/* ITC Free Running Data BW counter for inbound traffic */ -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port0,0x10 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port1,0x11 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port2,0x12 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port3,0x13 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port4,0x14 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port5,0x15 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port6,0x16 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (inb_data_port7,0x17 ,"3.814697266e-6" ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port0,0x10 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port1,0x11 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port2,0x12 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port3,0x13 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port4,0x14 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port5,0x15 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port6,0x16 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (inb_data_port7,0x17 ,3.814697266e-6 ),/* ITC Free Running BW IN counters */ -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port0,0x20 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port1,0x21 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port2,0x22 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port3,0x23 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port4,0x24 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port5,0x25 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port6,0x26 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_in_port7,0x27 ,"3.814697266e-6" ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port0,0x20 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port1,0x21 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port2,0x22 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port3,0x23 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port4,0x24 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port5,0x25 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port6,0x26 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_in_port7,0x27 ,3.814697266e-6 ),/* ITC Free Running BW OUT counters */ -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port0,0x30 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port1,0x31 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port2,0x32 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port3,0x33 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port4,0x34 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port5,0x35 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port6,0x36 ,"3.814697266e-6" ), -INTEL_UNCORE_FR_EVENT_DESC (bw_out_port7,0x37 ,"3.814697266e-6" ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port0,0x30 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port1,0x31 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port2,0x32 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port3,0x33 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port4,0x34 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port5,0x35 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port6,0x36 ,3.814697266e-6 ), +INTEL_UNCORE_FR_EVENT_DESC (bw_out_port7,0x37 ,3.814697266e-6 ),/* Free Running Clock Counter */ INTEL_UNCORE_EVENT_DESC (clockticks,"event=0xff,umask=0x40" ),
Signature · 1 line
2.53.0