cxl, acpi/hmat, node: Update CXL access coordinates to node directly
Subsystems (4)
Dave Jiang
maintainercover cxl, acpi/hmat, node: Update CXL access coordinates to node directly
I plan to take this series through the CXL tree when all the necessary tags are received.
Andrew or David, can you please ack patch 1/4 if you are okay with it.
Greg or Rafael, please ack patch 2/4 if you are satisfied with it. The helper function's usage is in patch 3/4.
Rafael, please ack patches 3/4 and 4/4 if you are happy with the changes.
Thank you all!
The series aim to clean up the current CXL memory region hotplug notifier by removing the update path through HMAT and updating the node access coordinates directly. With the existing implementation, the CXL memory hotplug notifier gets called first. It updates the HMAT target access coordinates. And then the HMAT notifier gets called and create the node sysfs attribs. The new implemenation flips the callback ordering and directly updates the sysfs attribs already created in the node and leaves HMAT data structures alone.
# Dave Jiang (4):
# mm/memory_hotplug: Update comment for hotplug memory callback priorities # drivers/base/node: Add a helper function node_update_perf_attrs() # cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT # acpi/hmat: Remove now unused hmat_update_target_coordinates() # # drivers/acpi/numa/hmat.c | 34 ---------------------------------- # drivers/base/node.c | 39 +++++++++++++++++++++++++++++++++++++++ # drivers/cxl/core/cdat.c | 11 ----------- # drivers/cxl/core/core.h | 3 --- # drivers/cxl/core/region.c | 10 ++-------- # include/linux/acpi.h | 12 ------------ # include/linux/memory.h | 4 ++-- # include/linux/node.h | 8 ++++++++ # 8 files changed, 51 insertions(+), 70 deletions(-)
# </REPLACE>
Dave Jiang (4):
mm/memory_hotplug: Update comment for hotplug memory callback
priorities
drivers/base/node: Add a helper function node_update_perf_attrs() cxl, acpi/hmat: Update CXL access coordinates directly instead of
through HMAT
acpi/hmat: Remove now unused hmat_update_target_coordinates()
drivers/acpi/numa/hmat.c | 34 ---------------------------------- drivers/base/node.c | 39 +++++++++++++++++++++++++++++++++++++++ drivers/cxl/core/cdat.c | 11 ----------- drivers/cxl/core/core.h | 3 --- drivers/cxl/core/region.c | 10 ++-------- include/linux/acpi.h | 12 ------------ include/linux/memory.h | 4 ++-- include/linux/node.h | 8 ++++++++ 8 files changed, 51 insertions(+), 70 deletions(-)
base-commit: 8f5ae30d69d7543eee0d70083daf4de8fe15d585
Signature · 1 line
2.50.1
Dave Jiang
maintainerPATCH 1/4 mm/memory_hotplug: Update comment for hotplug memory callback priorities
Add clarification to comment for memory hotplug callback ordering as the current comment does not provide clear language on which callback happens first.
--- include/linux/memory.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
include/linux/memory.h +1 -1
@@ -116,7 +116,7 @@ struct mem_section;/* * Priorities for the hotplug memory callback routines (stored in decreasing -* order in the callback chain) +* order in the callback chain). The callback ordering happens from high to low. */ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1
Signature · 1 line
2.50.1
Re: [PATCH 1/4] mm/memory_hotplug: Update comment for hotplug memory callback priorities
Quoted 14 lines · Dave Jiang
Add clarification to comment for memory hotplug callback ordering as the current comment does not provide clear language on which callback happens first.
Signed-off-by Dave Jiang <[email protected]>--- include/linux/memory.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/linux/memory.h +0
-0
include/linux/memory.h struct mem_section; · line 122
Quoted 8 lines
include/linux/memory.h +1
-1@@ -116,7 +116,7 @@ struct mem_section;/* * Priorities for the hotplug memory callback routines (stored in decreasing -* order in the callback chain) +* order in the callback chain). The callback ordering happens from high to low. */ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1
"stored in decreasing order in the callback chain"
is pretty clear? It's a chain after all that gets called.
Signature · 3 lines
Cheers David / dhildenb
Dave Jiang
maintainerRe: [PATCH 1/4] mm/memory_hotplug: Update comment for hotplug memory callback priorities
Quoted 25 lines · David Hildenbrand
Quoted 14 lines · Dave Jiang
Add clarification to comment for memory hotplug callback ordering as the current comment does not provide clear language on which callback happens first.
Signed-off-by Dave Jiang <[email protected]>---
include/linux/memory.h| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)include/linux/memory.h +0
-0include/linux/memory.h struct mem_section; · line 121
Quoted 7 lines
include/linux/memory.h +1
-1@@ -116,7 +116,7 @@ struct mem_section; /* * Priorities for the hotplug memory callback routines (stored in decreasing - * order in the callback chain) + * order in the callback chain). The callback ordering happens from high to low. */ #define DEFAULT_CALLBACK_PRI 0 #define SLAB_CALLBACK_PRI 1"stored in decreasing order in the callback chain"
Quoted 1 line
is pretty clear? It's a chain after all that gets called.
I can drop the patch. For some reason when I read it I'm thinking the opposite, and when Marc was also confused I started questioning things.
Quoted 1 line
Re: [PATCH 1/4] mm/memory_hotplug: Update comment for hotplug memory callback priorities
Quoted 28 lines · Dave Jiang
Quoted 25 lines · David Hildenbrand
Quoted 14 lines · Dave Jiang
Add clarification to comment for memory hotplug callback ordering as the current comment does not provide clear language on which callback happens first.
Signed-off-by Dave Jiang <[email protected]>--- include/linux/memory.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/linux/memory.h +0
-0include/linux/memory.h struct mem_section; · line 121
Quoted 7 lines
include/linux/memory.h +1
-1@@ -116,7 +116,7 @@ struct mem_section;/* * Priorities for the hotplug memory callback routines (stored in decreasing -* order in the callback chain) +* order in the callback chain). The callback ordering happens from high to low. */ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1"stored in decreasing order in the callback chain"
Quoted 4 lines
Quoted 1 line
is pretty clear? It's a chain after all that gets called.
I can drop the patch. For some reason when I read it I'm thinking the opposite, and when Marc was also confused I started questioning things.
I think we both found the current comment confusing (even together!) because:
- It very briefly alludes to an implementation detail (the chain) without really getting into detail. A "chain" could be bi-directional; why not? This one is... "most likely" not. Doubt.
- Higher priorities can have lower numbers, example: "P1 bugs". Not the case here, but this "double standards" situation makes _all_ priorities suspicious and confusing.
- Constants that come first in the file are called last.
I would go further than Dave and also drop the "chain" implementation detail because it makes the reader to think too much. Not needed and distracting at this particular point in the file.
/*
* Priorities for the hotplug memory callback routines.
* Invoked from high to low.
*/
=> Hopefully zero cognitive load.
Re: [PATCH 1/4] mm/memory_hotplug: Update comment for hotplug memory callback priorities
Quoted 36 lines · Marc Herbert
Quoted 28 lines · Dave Jiang
Quoted 25 lines · David Hildenbrand
Quoted 14 lines · Dave Jiang
Add clarification to comment for memory hotplug callback ordering as the current comment does not provide clear language on which callback happens first.
Signed-off-by Dave Jiang <[email protected]>--- include/linux/memory.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/linux/memory.h +0
-0include/linux/memory.h struct mem_section; · line 121
Quoted 7 lines
include/linux/memory.h +1
-1@@ -116,7 +116,7 @@ struct mem_section;/* * Priorities for the hotplug memory callback routines (stored in decreasing -* order in the callback chain) +* order in the callback chain). The callback ordering happens from high to low. */ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1"stored in decreasing order in the callback chain"
Quoted 4 lines
Quoted 1 line
is pretty clear? It's a chain after all that gets called.
I can drop the patch. For some reason when I read it I'm thinking the opposite, and when Marc was also confused I started questioning things.
Quoted 7 lines
I think we both found the current comment confusing (even together!) because:
- It very briefly alludes to an implementation detail (the chain) without really getting into detail. A "chain" could be bi-directional; why not? This one is... "most likely" not. Doubt.
Please note that the memory notifier is really just using the generic *notifier chain* mechanism as documented in include/linux/notifier.h.
Here is a good summary of that mechanism. I don't quite agree with the "implementation detail" part, but that information might indeed not be required here.
https://0xax.gitbooks.io/linux-insides/content/Concepts/linux-cpu-4.html
Quoted 4 lines
- Higher priorities can have lower numbers, example: "P1 bugs". Not the case here, but this "double standards" situation makes _all_ priorities suspicious and confusing.
Yes, "priorities" are handled differently in different context.
Quoted 1 line
- Constants that come first in the file are called last.
Yes, but that part makes perfect sense to me.
Quoted 10 lines
I would go further than Dave and also drop the "chain" implementation detail because it makes the reader to think too much. Not needed and distracting at this particular point in the file.
/*
* Priorities for the hotplug memory callback routines.
* Invoked from high to low.
*/=> Hopefully zero cognitive load.
Still confusion about how a high priority translates to a number, maybe?
/*
* Priorities for the hotplug memory callback routines. Invoked from
* high to low. Higher priorities corresponds to higher numbers. */
Signature · 3 lines
Cheers David / dhildenb
Dave Jiang
maintainerRe: [PATCH 1/4] mm/memory_hotplug: Update comment for hotplug memory callback priorities
Quoted 70 lines · David Hildenbrand
Quoted 36 lines · Marc Herbert
Quoted 28 lines · Dave Jiang
Quoted 25 lines · David Hildenbrand
Quoted 14 lines · Dave Jiang
Add clarification to comment for memory hotplug callback ordering as the current comment does not provide clear language on which callback happens first.
Signed-off-by Dave Jiang <[email protected]>---
include/linux/memory.h| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)include/linux/memory.h +0
-0include/linux/memory.h struct mem_section; · line 121
Quoted 7 lines
include/linux/memory.h +1
-1@@ -116,7 +116,7 @@ struct mem_section; /* * Priorities for the hotplug memory callback routines (stored in decreasing - * order in the callback chain) + * order in the callback chain). The callback ordering happens from high to low. */ #define DEFAULT_CALLBACK_PRI 0 #define SLAB_CALLBACK_PRI 1"stored in decreasing order in the callback chain"
Quoted 4 lines
Quoted 1 line
is pretty clear? It's a chain after all that gets called.
I can drop the patch. For some reason when I read it I'm thinking the opposite, and when Marc was also confused I started questioning things.
Quoted 7 lines
I think we both found the current comment confusing (even together!) because:
- It very briefly alludes to an implementation detail (the chain) without really getting into detail. A "chain" could be bi-directional; why not? This one is... "most likely" not. Doubt.
Please note that the memory notifier is really just using the generic *notifier chain* mechanism as documented in
include/linux/notifier.h.Here is a good summary of that mechanism. I don't quite agree with the "implementation detail" part, but that information might indeed not be required here.
https://0xax.gitbooks.io/linux-insides/content/Concepts/linux-cpu-4.html
Quoted 4 lines
- Higher priorities can have lower numbers, example: "P1 bugs". Not the case here, but this "double standards" situation makes _all_ priorities suspicious and confusing.
Yes, "priorities" are handled differently in different context.
Quoted 1 line
- Constants that come first in the file are called last.
Yes, but that part makes perfect sense to me.
> I would go further than Dave and also drop the "chain" implementationQuoted 2 lines
detail because it makes the reader to think too much. Not needed and distracting at this particular point in the file.
> /*
Quoted 4 lines
* Priorities for the hotplug memory callback routines.
* Invoked from high to low.
*/
Quoted 9 lines
Quoted 1 line
=> Hopefully zero cognitive load.
Still confusion about how a high priority translates to a number, maybe?
/*
* Priorities for the hotplug memory callback routines. Invoked from
* high to low. Higher priorities corresponds to higher numbers. */
This reads clear to me. I will adopt this comment if there are no additional objections.
David Hildenbrand
Acked-byRe: [PATCH 1/4] mm/memory_hotplug: Update comment for hotplug memory callback priorities
Quoted 77 lines · Dave Jiang
Quoted 74 lines · David Hildenbrand
Quoted 36 lines · Marc Herbert
Quoted 28 lines · Dave Jiang
Quoted 25 lines · David Hildenbrand
Quoted 14 lines · Dave Jiang
Add clarification to comment for memory hotplug callback ordering as the current comment does not provide clear language on which callback happens first.
Signed-off-by Dave Jiang <[email protected]>--- include/linux/memory.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/linux/memory.h +0
-0include/linux/memory.h struct mem_section; · line 121
Quoted 7 lines
include/linux/memory.h +1
-1@@ -116,7 +116,7 @@ struct mem_section; /* * Priorities for the hotplug memory callback routines (stored in decreasing - * order in the callback chain) + * order in the callback chain). The callback ordering happens from high to low. */ #define DEFAULT_CALLBACK_PRI 0 #define SLAB_CALLBACK_PRI 1"stored in decreasing order in the callback chain"
Quoted 4 lines
Quoted 1 line
is pretty clear? It's a chain after all that gets called.
I can drop the patch. For some reason when I read it I'm thinking the opposite, and when Marc was also confused I started questioning things.
Quoted 7 lines
I think we both found the current comment confusing (even together!) because:
- It very briefly alludes to an implementation detail (the chain) without really getting into detail. A "chain" could be bi-directional; why not? This one is... "most likely" not. Doubt.
Please note that the memory notifier is really just using the generic *notifier chain* mechanism as documented in
include/linux/notifier.h.Here is a good summary of that mechanism. I don't quite agree with the "implementation detail" part, but that information might indeed not be required here.
https://0xax.gitbooks.io/linux-insides/content/Concepts/linux-cpu-4.html
Quoted 4 lines
- Higher priorities can have lower numbers, example: "P1 bugs". Not the case here, but this "double standards" situation makes _all_ priorities suspicious and confusing.
Yes, "priorities" are handled differently in different context.
Quoted 1 line
- Constants that come first in the file are called last.
Yes, but that part makes perfect sense to me.
> I would go further than Dave and also drop the "chain" implementationQuoted 2 lines
detail because it makes the reader to think too much. Not needed and distracting at this particular point in the file.
> /*
Quoted 5 lines
* Priorities for the hotplug memory callback routines.
* Invoked from high to low.
*/=> Hopefully zero cognitive load.
Still confusion about how a high priority translates to a number, maybe?
Quoted 7 lines
Quoted 5 lines
/*
* Priorities for the hotplug memory callback routines. Invoked from
* high to low. Higher priorities corresponds to higher numbers. */This reads clear to me. I will adopt this comment if there are no additional objections.
Feel free to add
Signature · 3 lines
Cheers David / dhildenb
Dave Jiang
maintainerPATCH 2/4 drivers/base/node: Add a helper function node_update_perf_attrs()
In reply to Dave JiangAdd helper function node_update_perf_attrs() to allow update of node access coordinates computed by an external agent such as CXL. The helper allows updating of coordinates after the attribute being created by HMAT.
--- drivers/base/node.c | 39 +++++++++++++++++++++++++++++++++++++++ include/linux/node.h | 8 ++++++++ 2 files changed, 47 insertions(+)
drivers/base/node.c +39 -0
@@ -248,6 +248,45 @@ void node_set_perf_attrs(unsigned int nid, struct access_coordinate *coord, }EXPORT_SYMBOL_GPL (node_set_perf_attrs); +/** +* node_update_perf_attrs - Update the performance values for given access class +* @nid: Node identifier to be updated +* @coord: Heterogeneous memory performance coordinates +* @access: The access class the for the given attributes +*/ +void node_update_perf_attrs (unsigned int nid,struct access_coordinate *coord, +enum access_coordinate_class access) +{ +struct node_access_nodes *access_node; +struct node *node; +int i; + +if (WARN_ON_ONCE (!node_online (nid))) +return ; + + node = node_devices[nid]; +list_for_each_entry (access_node, &node->access_list, list_node) { +if (access_node->access != access) +continue ; + + access_node->coord = *coord; +for (i =0 ; access_attrs[i]; i++) { +sysfs_notify (&access_node->dev.kobj, +NULL , access_attrs[i]->name); + } +break ; + } + +/* When setting CPU access coordinates, update mempolicy */ +if (access ==ACCESS_COORDINATE_CPU ) { +if (mempolicy_set_node_perf (nid, coord)) { +pr_info ("failed to set mempolicy attrs for node %d\n" , + nid); + } + } +} +EXPORT_SYMBOL_GPL (node_update_perf_attrs); +/** * struct node_cache_info - Internal tracking for memory node caches * @dev: Device represeting the cache level
include/linux/node.h +8 -0
@@ -85,6 +85,8 @@ struct node_cache_attrs {void node_add_cache (unsigned int nid,struct node_cache_attrs *cache_attrs);void node_set_perf_attrs (unsigned int nid,struct access_coordinate *coord,enum access_coordinate_class access); +void node_update_perf_attrs (unsigned int nid,struct access_coordinate *coord, +enum access_coordinate_class access);#else static inline void node_add_cache (unsigned int nid,struct node_cache_attrs *cache_attrs)
@@ -96,6 +98,12 @@ static inline void node_set_perf_attrs(unsigned int nid, enum access_coordinate_class access) { } + +static inline void node_update_perf_attrs (unsigned int nid, +struct access_coordinate *coord, +enum access_coordinate_class access) +{ +}#endif struct node {
Signature · 1 line
2.50.1
Re: [PATCH 2/4] drivers/base/node: Add a helper function node_update_perf_attrs()
Quoted 5 lines · On Thu, 14 Aug 2025 10:16:48 -0700 Dave Jiang <[email protected]> wrote:
Add helper function
node_update_perf_attrs()to allow update of node access coordinates computed by an external agent such as CXL. The helper allows updating of coordinates after the attribute being created by HMAT.Signed-off-by Dave Jiang <[email protected]>
Re: [PATCH 2/4] drivers/base/node: Add a helper function node_update_perf_attrs()
In reply to Dave JiangQuoted 15 lines · Dave Jiang
Add helper function
node_update_perf_attrs()to allow update of node access coordinates computed by an external agent such as CXL. The helper allows updating of coordinates after the attribute being created by HMAT.Signed-off-by Dave Jiang <[email protected]>--- drivers/base/node.c | 39 +++++++++++++++++++++++++++++++++++++++ include/linux/node.h | 8 ++++++++ 2 files changed, 47 insertions(+)drivers/base/node.c +0
-0
drivers/base/node.c void node_set_perf_attrs(unsigned int nid, struct access_coordinate *coord, · line 255
Quoted 8 lines
drivers/base/node.c +5
-0@@ -248,6 +248,45 @@ void node_set_perf_attrs(unsigned int nid, struct access_coordinate *coord, }EXPORT_SYMBOL_GPL (node_set_perf_attrs); +/** +* node_update_perf_attrs - Update the performance values for given access class +* @nid: Node identifier to be updated +* @coord: Heterogeneous memory performance coordinates +* @access: The access class the for the given attributes
"the for": there is probably something missing
Quoted 20 lines
drivers/base/node.c +20
-0+*/ +void node_update_perf_attrs (unsigned int nid,struct access_coordinate *coord, +enum access_coordinate_class access) +{ +struct node_access_nodes *access_node; +struct node *node; +int i; + +if (WARN_ON_ONCE (!node_online (nid))) +return ; + + node = node_devices[nid]; +list_for_each_entry (access_node, &node->access_list, list_node) { +if (access_node->access != access) +continue ; + + access_node->coord = *coord; +for (i =0 ; access_attrs[i]; i++) { +sysfs_notify (&access_node->dev.kobj, +NULL , access_attrs[i]->name);
drivers/base/node.c void node_set_perf_attrs(unsigned int nid, struct access_coordinate *coord, · line 285
Quoted 10 lines
drivers/base/node.c +10
-0+ } +break ; + } + +/* When setting CPU access coordinates, update mempolicy */ +if (access ==ACCESS_COORDINATE_CPU ) { +if (mempolicy_set_node_perf (nid, coord)) { +pr_info ("failed to set mempolicy attrs for node %d\n" , + nid); + }
if (access == ACCESS_COORDINATE_CPU &&
mempolicy_set_node_perf(nid, coord))
pr_info("failed to set mempolicy attrs for node %d\n", nid);or
if (access != ACCESS_COORDINATE_CPU)
return
if (mempolicy_set_node_perf(nid, coord))
pr_info("failed to set mempolicy attrs for node %d\n", nid);With both things sorted
Signature · 3 lines
Cheers David / dhildenb
Dave Jiang
maintainerRe: [PATCH 2/4] drivers/base/node: Add a helper function node_update_perf_attrs()
Quoted 11 lines · David Hildenbrand
Quoted 10 lines · Dave Jiang
Add helper function
node_update_perf_attrs()to allow update of node access coordinates computed by an external agent such as CXL. The helper allows updating of coordinates after the attribute being created by HMAT.Signed-off-by Dave Jiang <[email protected]>---
drivers/base/node.c | 39 +++++++++++++++++++++++++++++++++++++++include/linux/node.h| 8 ++++++++2 files changed, 47 insertions(+)
Quoted 14 lines
Quoted 7 lines
drivers/base/node.c +0
-0@@ -248,6 +248,45 @@ void node_set_perf_attrs(unsigned int nid, struct access_coordinate *coord, } EXPORT_SYMBOL_GPL(node_set_perf_attrs);Quoted 5 lines
+/**
+ *node_update_perf_attrs- Update the performance values for given access class
+ * @nid: Node identifier to be updated
+ * @coord: Heterogeneous memory performance coordinates
+ * @access: The access class the for the given attributes"the for": there is probably something missing
looks like extra 'the'
Quoted 35 lines
Quoted 21 lines
+ */
drivers/base/node.c +19
-0+void node_update_perf_attrs (unsigned int nid,struct access_coordinate *coord, + enum access_coordinate_class access) +{ + struct node_access_nodes *access_node; + struct node *node; + int i; + + if (WARN_ON_ONCE(!node_online(nid))) + return; + + node = node_devices[nid]; + list_for_each_entry(access_node, &node->access_list, list_node) { + if (access_node->access != access) + continue; + + access_node->coord = *coord; + for (i = 0; access_attrs[i]; i++) { + sysfs_notify(&access_node->dev.kobj, + NULL, access_attrs[i]->name);drivers/base/node.c void node_set_perf_attrs(unsigned int nid, struct access_coordinate *coord, · line 285
Quoted 10 lines
drivers/base/node.c +10
-0+ } + break; + } + + /* When setting CPU access coordinates, update mempolicy */ + if (access == ACCESS_COORDINATE_CPU) { + if (mempolicy_set_node_perf(nid, coord)) { + pr_info("failed to set mempolicy attrs for node %d\n", + nid); + }if (access ==
ACCESS_COORDINATE_CPU&&mempolicy_set_node_perf(nid, coord))pr_info("failed to set mempolicy attrs for node %d\n", nid);
Quoted 7 lines
or
if (access !=
ACCESS_COORDINATE_CPU)
return
if (mempolicy_set_node_perf(nid, coord))pr_info("failed to set mempolicy attrs for node %d\n", nid);
ok
Quoted 5 lines
With both things sorted
Thanks for the ack
Dave Jiang
maintainerPATCH 3/4 cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT
In reply to Dave JiangThe current implementation of CXL memory hotplug notifier gets called before the HMAT memory hotplug notifier. The CXL driver calculates the access coordinates (bandwidth and latency values) for the CXL end to end path (i.e. CPU to endpoint). When the CXL region is onlined, the CXL memory hotplug notifier writes the access coordinates to the HMAT target structs. Then the HMAT memory hotplug notifier is called and it creates the access coordinates for the node sysfs attributes.
The original intent of the 'ext_updated' flag in HMAT handling code was to stop HMAT memory hotplug callback from clobbering the access coordinates after CXL has injected its calculated coordinates and replaced the generic target access coordinates provided by the HMAT table in the HMAT target structs. However the flag is hacky at best and blocks the updates from other CXL regions that are onlined in the same node later on. Remove the 'ext_updated' flag usage and just update the access coordinates for the nodes directly without touching HMAT target data.
The hotplug memory callback ordering is changed. Instead of changing CXL, move HMAT back so there's room for the levels rather than have CXL share the same level as SLAB_CALLBACK_PRI. The change will resulting in the CXL callback to be executed after the HMAT callback.
With the change, the CXL hotplug memory notifier runs after the HMAT callback. The HMAT callback will create the node sysfs attributes for access coordinates. The CXL callback will write the access coordinates to the now created node sysfs attributes directly and will not pollute the HMAT target values.
--- drivers/acpi/numa/hmat.c | 6 ------ drivers/cxl/core/cdat.c | 5 ----- drivers/cxl/core/core.h | 1 - drivers/cxl/core/region.c | 10 ++-------- include/linux/memory.h | 2 +- 5 files changed, 3 insertions(+), 21 deletions(-)
drivers/acpi/numa/hmat.c +0 -6
@@ -74,7 +74,6 @@ struct memory_target {struct node_cache_attrs cache_attrs;u8 gen_port_device_handle[ACPI_SRAT_DEVICE_HANDLE_SIZE ];bool registered; -bool ext_updated;/* externally updated */ };struct memory_initiator {
@@ -391,7 +390,6 @@ int hmat_update_target_coordinates(int nid, struct access_coordinate *coord, coord->read_bandwidth, access);hmat_update_target_access (target,ACPI_HMAT_WRITE_BANDWIDTH , coord->write_bandwidth, access); - target->ext_updated = true;return 0 ; }
@@ -773,10 +771,6 @@ static void hmat_update_target_attrs(struct memory_target *target,u32 best =0 ;int i; -/* Don't update if an external agent has changed the data. */ -if (target->ext_updated) -return ; -/* Don't update for generic port if there's no device handle */ if ((access == NODE_ACCESS_CLASS_GENPORT_SINK_LOCAL || access == NODE_ACCESS_CLASS_GENPORT_SINK_CPU) &&
drivers/cxl/core/cdat.c +0 -5
@@ -1081,8 +1081,3 @@ int cxl_update_hmat_access_coordinates(int nid, struct cxl_region *cxlr, {return hmat_update_target_coordinates (nid, &cxlr->coord[access], access); } - -bool cxl_need_node_perf_attrs_update (int nid) -{ -return !acpi_node_backed_by_real_pxm (nid); -}
drivers/cxl/core/core.h +0 -1
@@ -139,7 +139,6 @@ long cxl_pci_get_latency(struct pci_dev *pdev);int cxl_pci_get_bandwidth (struct pci_dev *pdev,struct access_coordinate *c);int cxl_update_hmat_access_coordinates (int nid,struct cxl_region *cxlr,enum access_coordinate_class access); -bool cxl_need_node_perf_attrs_update (int nid);int cxl_port_get_switch_dport_bandwidth (struct cxl_port *port,struct access_coordinate *c);
drivers/cxl/core/region.c +2 -8
@@ -2442,14 +2442,8 @@ static bool cxl_region_update_coordinates(struct cxl_region *cxlr, int nid)for (int i =0 ; i <ACCESS_COORDINATE_MAX ; i++) {if (cxlr->coord[i].read_bandwidth) { - rc =0 ; -if (cxl_need_node_perf_attrs_update (nid)) -node_set_perf_attrs (nid, &cxlr->coord[i], i); -else - rc =cxl_update_hmat_access_coordinates (nid, cxlr, i); - -if (rc ==0 ) - cset++; +node_update_perf_attrs (nid, &cxlr->coord[i], i); + cset++; } }
include/linux/memory.h +1 -1
@@ -120,8 +120,8 @@ struct mem_section;*/ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1 -#define HMAT_CALLBACK_PRI 2#define CXL_CALLBACK_PRI 5 +#define HMAT_CALLBACK_PRI 6#define MM_COMPUTE_BATCH_PRI 10#define CPUSET_CALLBACK_PRI 10#define MEMTIER_HOTPLUG_PRI 100
Signature · 1 line
2.50.1
Re: [PATCH 3/4] cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT
Quoted 7 lines · Dave Jiang
The current implementation of CXL memory hotplug notifier gets called before the HMAT memory hotplug notifier. The CXL driver calculates the access coordinates (bandwidth and latency values) for the CXL end to end path (i.e. CPU to endpoint). When the CXL region is onlined, the CXL memory hotplug notifier writes the access coordinates to the HMAT target structs. Then the HMAT memory hotplug notifier is called and it creates the access coordinates for the node sysfs attributes.
Perhaps summarize quickly here the before and after of sysfs, so people know if they are impacted by this bug, and backporters can verify they fixed it?
Quoted 14 lines
The original intent of the '
ext_updated' flag in HMAT handling code was to stop HMAT memory hotplug callback from clobbering the access coordinates after CXL has injected its calculated coordinates and replaced the generic target access coordinates provided by the HMAT table in the HMAT target structs. However the flag is hacky at best and blocks the updates from other CXL regions that are onlined in the same node later on. Remove the 'ext_updated' flag usage and just update the access coordinates for the nodes directly without touching HMAT target data.The hotplug memory callback ordering is changed. Instead of changing CXL, move HMAT back so there's room for the levels rather than have CXL share the same level as
SLAB_CALLBACK_PRI. The change will resulting in the CXL callback to be executed after the HMAT callback.
Quoted 7 lines
With the change, the CXL hotplug memory notifier runs after the HMAT callback. The HMAT callback will create the node sysfs attributes for access coordinates. The CXL callback will write the access coordinates to the now created node sysfs attributes directly and will not pollute the HMAT target values.
Fixes debdce20c4f2 ("cxl/region: Deal with numa nodes not enumerated by SRAT")
Why that one and not?
067353a46d8c cxl/region: Add memory hotplug notifier for cxl region
It is the ext_updated machinery that is the main problem that messes up sysfs, right?
...and per the backport concern this should be cc: stable as well.
Other than that you can add:
Dave Jiang
maintainerRe: [PATCH 3/4] cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT
Quoted 9 lines · On 8/14/25 3:33 PM, [email protected] wrote:
Quoted 7 lines · Dave Jiang
The current implementation of CXL memory hotplug notifier gets called before the HMAT memory hotplug notifier. The CXL driver calculates the access coordinates (bandwidth and latency values) for the CXL end to end path (i.e. CPU to endpoint). When the CXL region is onlined, the CXL memory hotplug notifier writes the access coordinates to the HMAT target structs. Then the HMAT memory hotplug notifier is called and it creates the access coordinates for the node sysfs attributes.
Quoted 3 lines
Perhaps summarize quickly here the before and after of sysfs, so people know if they are impacted by this bug, and backporters can verify they fixed it?
ok
Quoted 21 lines
Quoted 21 lines
The original intent of the '
ext_updated' flag in HMAT handling code was to stop HMAT memory hotplug callback from clobbering the access coordinates after CXL has injected its calculated coordinates and replaced the generic target access coordinates provided by the HMAT table in the HMAT target structs. However the flag is hacky at best and blocks the updates from other CXL regions that are onlined in the same node later on. Remove the 'ext_updated' flag usage and just update the access coordinates for the nodes directly without touching HMAT target data.The hotplug memory callback ordering is changed. Instead of changing CXL, move HMAT back so there's room for the levels rather than have CXL share the same level as
SLAB_CALLBACK_PRI. The change will resulting in the CXL callback to be executed after the HMAT callback.With the change, the CXL hotplug memory notifier runs after the HMAT callback. The HMAT callback will create the node sysfs attributes for access coordinates. The CXL callback will write the access coordinates to the now created node sysfs attributes directly and will not pollute the HMAT target values.
Quoted 5 lines
Quoted 1 line
Fixes debdce20c4f2 ("cxl/region: Deal with numa nodes not enumerated by SRAT")Why that one and not?
067353a46d8ccxl/region: Add memory hotplug notifier for cxl region
I think I grabbed the wrong line for 'git blame'.
Quoted 9 lines
It is the
ext_updatedmachinery that is the main problem that messes up sysfs, right?...and per the backport concern this should be cc: stable as well.
Other than that you can add:
Reviewed-by Dan Williams <[email protected]>
Marc Herbert
Tested-byRe: [PATCH 3/4] cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT
In reply to [email protected]Quoted 9 lines · On 2025-08-14 15:33, [email protected] wrote:
Quoted 1 line
Fixes debdce20c4f2 ("cxl/region: Deal with numa nodes not enumerated by SRAT")Why that one and not?
067353a46d8ccxl/region: Add memory hotplug notifier for cxl regionIt is the
ext_updatedmachinery that is the main problem that messes up sysfs, right?
For sure 067353a46d8c is where my git bisect unambiguously landed.
(the series as a whole)
Re: [PATCH 3/4] cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT
In reply to Dave JiangQuoted 8 lines · On Thu, 14 Aug 2025 10:16:49 -0700 Dave Jiang <[email protected]> wrote:
The current implementation of CXL memory hotplug notifier gets called before the HMAT memory hotplug notifier. The CXL driver calculates the access coordinates (bandwidth and latency values) for the CXL end to end path (i.e. CPU to endpoint). When the CXL region is onlined, the CXL memory hotplug notifier writes the access coordinates to the HMAT target structs. Then the HMAT memory hotplug notifier is called and it creates the access coordinates for the node sysfs attributes.
Quoted 6 lines
The original intent of the '
ext_updated' flag in HMAT handling code was to stop HMAT memory hotplug callback from clobbering the access coordinates after CXL has injected its calculated coordinates and replaced the generic target access coordinates provided by the HMAT table in the HMAT target structs. However the flag is hacky at best and blocks the updates from other CXL regions that are onlined in the same node later on.
After all removed, or when a second region onlined in that node whilst the first is still online? In that second case I think we should not update the access properties as that would surprise anything already using the earlier one.
Jonathan
Quoted 117 lines
Remove the
'ext_updated' flag usage and just update the access coordinates for the nodes directly without touching HMAT target data.The hotplug memory callback ordering is changed. Instead of changing CXL, move HMAT back so there's room for the levels rather than have CXL share the same level as
SLAB_CALLBACK_PRI. The change will resulting in the CXL callback to be executed after the HMAT callback.With the change, the CXL hotplug memory notifier runs after the HMAT callback. The HMAT callback will create the node sysfs attributes for access coordinates. The CXL callback will write the access coordinates to the now created node sysfs attributes directly and will not pollute the HMAT target values.
Fixes debdce20c4f2 ("cxl/region: Deal with numa nodes not enumerated by SRAT")Tested-by Marc Herbert <[email protected]>Signed-off-by Dave Jiang <[email protected]>--- drivers/acpi/numa/hmat.c | 6 ------ drivers/cxl/core/cdat.c | 5 ----- drivers/cxl/core/core.h | 1 - drivers/cxl/core/region.c | 10 ++-------- include/linux/memory.h | 2 +- 5 files changed, 3 insertions(+), 21 deletions(-)drivers/acpi/numa/hmat.c +0
-6@@ -74,7 +74,6 @@ struct memory_target {struct node_cache_attrs cache_attrs;u8 gen_port_device_handle[ACPI_SRAT_DEVICE_HANDLE_SIZE ];bool registered; -bool ext_updated;/* externally updated */ };struct memory_initiator {@@ -391,7 +390,6 @@ int hmat_update_target_coordinates(int nid, struct access_coordinate *coord, coord->read_bandwidth, access);hmat_update_target_access (target,ACPI_HMAT_WRITE_BANDWIDTH , coord->write_bandwidth, access); - target->ext_updated = true;return 0 ; }@@ -773,10 +771,6 @@ static void hmat_update_target_attrs(struct memory_target *target,u32 best =0 ;int i; -/* Don't update if an external agent has changed the data. */ -if (target->ext_updated) -return ; -/* Don't update for generic port if there's no device handle */ if ((access == NODE_ACCESS_CLASS_GENPORT_SINK_LOCAL || access == NODE_ACCESS_CLASS_GENPORT_SINK_CPU) &&drivers/cxl/core/cdat.c +0
-5@@ -1081,8 +1081,3 @@ int cxl_update_hmat_access_coordinates(int nid, struct cxl_region *cxlr, {return hmat_update_target_coordinates (nid, &cxlr->coord[access], access); } - -bool cxl_need_node_perf_attrs_update (int nid) -{ -return !acpi_node_backed_by_real_pxm (nid); -}drivers/cxl/core/core.h +0
-1@@ -139,7 +139,6 @@ long cxl_pci_get_latency(struct pci_dev *pdev);int cxl_pci_get_bandwidth (struct pci_dev *pdev,struct access_coordinate *c);int cxl_update_hmat_access_coordinates (int nid,struct cxl_region *cxlr,enum access_coordinate_class access); -bool cxl_need_node_perf_attrs_update (int nid);int cxl_port_get_switch_dport_bandwidth (struct cxl_port *port,struct access_coordinate *c);drivers/cxl/core/region.c +2
-8@@ -2442,14 +2442,8 @@ static bool cxl_region_update_coordinates(struct cxl_region *cxlr, int nid)for (int i =0 ; i <ACCESS_COORDINATE_MAX ; i++) {if (cxlr->coord[i].read_bandwidth) { - rc =0 ; -if (cxl_need_node_perf_attrs_update (nid)) -node_set_perf_attrs (nid, &cxlr->coord[i], i); -else - rc =cxl_update_hmat_access_coordinates (nid, cxlr, i); - -if (rc ==0 ) - cset++; +node_update_perf_attrs (nid, &cxlr->coord[i], i); + cset++; } }include/linux/memory.h +1
-1@@ -120,8 +120,8 @@ struct mem_section;*/ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1 -#define HMAT_CALLBACK_PRI 2#define CXL_CALLBACK_PRI 5 +#define HMAT_CALLBACK_PRI 6#define MM_COMPUTE_BATCH_PRI 10#define CPUSET_CALLBACK_PRI 10#define MEMTIER_HOTPLUG_PRI 100
Dave Jiang
maintainerRe: [PATCH 3/4] cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT
Quoted 18 lines · Jonathan Cameron
Quoted 8 lines · On Thu, 14 Aug 2025 10:16:49 -0700 Dave Jiang <[email protected]> wrote:
The current implementation of CXL memory hotplug notifier gets called before the HMAT memory hotplug notifier. The CXL driver calculates the access coordinates (bandwidth and latency values) for the CXL end to end path (i.e. CPU to endpoint). When the CXL region is onlined, the CXL memory hotplug notifier writes the access coordinates to the HMAT target structs. Then the HMAT memory hotplug notifier is called and it creates the access coordinates for the node sysfs attributes.
Quoted 6 lines
The original intent of the '
ext_updated' flag in HMAT handling code was to stop HMAT memory hotplug callback from clobbering the access coordinates after CXL has injected its calculated coordinates and replaced the generic target access coordinates provided by the HMAT table in the HMAT target structs. However the flag is hacky at best and blocks the updates from other CXL regions that are onlined in the same node later on.
Quoted 4 lines
After all removed, or when a second region onlined in that node whilst the first is still online? In that second case I think we should not update the access properties as that would surprise anything already using the earlier one.
Are you thinking we'll need some sort of ref counting on the node?
DJ
Quoted 121 lines
Jonathan
Quoted 117 lines
Remove the
'ext_updated' flag usage and just update the access coordinates for the nodes directly without touching HMAT target data.The hotplug memory callback ordering is changed. Instead of changing CXL, move HMAT back so there's room for the levels rather than have CXL share the same level as
SLAB_CALLBACK_PRI. The change will resulting in the CXL callback to be executed after the HMAT callback.With the change, the CXL hotplug memory notifier runs after the HMAT callback. The HMAT callback will create the node sysfs attributes for access coordinates. The CXL callback will write the access coordinates to the now created node sysfs attributes directly and will not pollute the HMAT target values.
Fixes debdce20c4f2 ("cxl/region: Deal with numa nodes not enumerated by SRAT")Tested-by Marc Herbert <[email protected]>Signed-off-by Dave Jiang <[email protected]>--- drivers/acpi/numa/hmat.c | 6 ------ drivers/cxl/core/cdat.c | 5 ----- drivers/cxl/core/core.h | 1 - drivers/cxl/core/region.c | 10 ++-------- include/linux/memory.h | 2 +- 5 files changed, 3 insertions(+), 21 deletions(-)drivers/acpi/numa/hmat.c +0
-6@@ -74,7 +74,6 @@ struct memory_target {struct node_cache_attrs cache_attrs;u8 gen_port_device_handle[ACPI_SRAT_DEVICE_HANDLE_SIZE ];bool registered; -bool ext_updated;/* externally updated */ };struct memory_initiator {@@ -391,7 +390,6 @@ int hmat_update_target_coordinates(int nid, struct access_coordinate *coord, coord->read_bandwidth, access);hmat_update_target_access (target,ACPI_HMAT_WRITE_BANDWIDTH , coord->write_bandwidth, access); - target->ext_updated = true;return 0 ; }@@ -773,10 +771,6 @@ static void hmat_update_target_attrs(struct memory_target *target,u32 best =0 ;int i; -/* Don't update if an external agent has changed the data. */ -if (target->ext_updated) -return ; -/* Don't update for generic port if there's no device handle */ if ((access == NODE_ACCESS_CLASS_GENPORT_SINK_LOCAL || access == NODE_ACCESS_CLASS_GENPORT_SINK_CPU) &&drivers/cxl/core/cdat.c +0
-5@@ -1081,8 +1081,3 @@ int cxl_update_hmat_access_coordinates(int nid, struct cxl_region *cxlr, {return hmat_update_target_coordinates (nid, &cxlr->coord[access], access); } - -bool cxl_need_node_perf_attrs_update (int nid) -{ -return !acpi_node_backed_by_real_pxm (nid); -}drivers/cxl/core/core.h +0
-1@@ -139,7 +139,6 @@ long cxl_pci_get_latency(struct pci_dev *pdev);int cxl_pci_get_bandwidth (struct pci_dev *pdev,struct access_coordinate *c);int cxl_update_hmat_access_coordinates (int nid,struct cxl_region *cxlr,enum access_coordinate_class access); -bool cxl_need_node_perf_attrs_update (int nid);int cxl_port_get_switch_dport_bandwidth (struct cxl_port *port,struct access_coordinate *c);drivers/cxl/core/region.c +2
-8@@ -2442,14 +2442,8 @@ static bool cxl_region_update_coordinates(struct cxl_region *cxlr, int nid)for (int i =0 ; i <ACCESS_COORDINATE_MAX ; i++) {if (cxlr->coord[i].read_bandwidth) { - rc =0 ; -if (cxl_need_node_perf_attrs_update (nid)) -node_set_perf_attrs (nid, &cxlr->coord[i], i); -else - rc =cxl_update_hmat_access_coordinates (nid, cxlr, i); - -if (rc ==0 ) - cset++; +node_update_perf_attrs (nid, &cxlr->coord[i], i); + cset++; } }include/linux/memory.h +1
-1@@ -120,8 +120,8 @@ struct mem_section;*/ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1 -#define HMAT_CALLBACK_PRI 2#define CXL_CALLBACK_PRI 5 +#define HMAT_CALLBACK_PRI 6#define MM_COMPUTE_BATCH_PRI 10#define CPUSET_CALLBACK_PRI 10#define MEMTIER_HOTPLUG_PRI 100
Re: [PATCH 3/4] cxl, acpi/hmat: Update CXL access coordinates directly instead of through HMAT
Quoted 19 lines · On Fri, 15 Aug 2025 08:35:31 -0700 Dave Jiang <[email protected]> wrote:
Quoted 18 lines · Jonathan Cameron
Quoted 14 lines · On Thu, 14 Aug 2025 10:16:49 -0700 Dave Jiang <[email protected]> wrote:
The current implementation of CXL memory hotplug notifier gets called before the HMAT memory hotplug notifier. The CXL driver calculates the access coordinates (bandwidth and latency values) for the CXL end to end path (i.e. CPU to endpoint). When the CXL region is onlined, the CXL memory hotplug notifier writes the access coordinates to the HMAT target structs. Then the HMAT memory hotplug notifier is called and it creates the access coordinates for the node sysfs attributes.
The original intent of the '
ext_updated' flag in HMAT handling code was to stop HMAT memory hotplug callback from clobbering the access coordinates after CXL has injected its calculated coordinates and replaced the generic target access coordinates provided by the HMAT table in the HMAT target structs. However the flag is hacky at best and blocks the updates from other CXL regions that are onlined in the same node later on.
Quoted 6 lines
Quoted 4 lines
After all removed, or when a second region onlined in that node whilst the first is still online? In that second case I think we should not update the access properties as that would surprise anything already using the earlier one.
Are you thinking we'll need some sort of ref counting on the node?
Was more a question on the intent, but indeed some sort of refcount would be needed to tear down cleanly. Might get messy though as what do we do if two regions are added with different characteristics then the first one is removed. Do we update?
Meh. This is a corner case and if we get different properties in one CFMWS as a common thing then we need to figure out how to spin more NUMA nodes. So I'm not sure this isn't a 'won't fix until the case turns out to matter' thing.
Jonathan
Quoted 125 lines
DJ
Quoted 121 lines
Jonathan
Quoted 117 lines
Remove the
'ext_updated' flag usage and just update the access coordinates for the nodes directly without touching HMAT target data.The hotplug memory callback ordering is changed. Instead of changing CXL, move HMAT back so there's room for the levels rather than have CXL share the same level as
SLAB_CALLBACK_PRI. The change will resulting in the CXL callback to be executed after the HMAT callback.With the change, the CXL hotplug memory notifier runs after the HMAT callback. The HMAT callback will create the node sysfs attributes for access coordinates. The CXL callback will write the access coordinates to the now created node sysfs attributes directly and will not pollute the HMAT target values.
Fixes debdce20c4f2 ("cxl/region: Deal with numa nodes not enumerated by SRAT")Tested-by Marc Herbert <[email protected]>Signed-off-by Dave Jiang <[email protected]>--- drivers/acpi/numa/hmat.c | 6 ------ drivers/cxl/core/cdat.c | 5 ----- drivers/cxl/core/core.h | 1 - drivers/cxl/core/region.c | 10 ++-------- include/linux/memory.h | 2 +- 5 files changed, 3 insertions(+), 21 deletions(-)drivers/acpi/numa/hmat.c +0
-6@@ -74,7 +74,6 @@ struct memory_target {struct node_cache_attrs cache_attrs;u8 gen_port_device_handle[ACPI_SRAT_DEVICE_HANDLE_SIZE ];bool registered; -bool ext_updated;/* externally updated */ };struct memory_initiator {@@ -391,7 +390,6 @@ int hmat_update_target_coordinates(int nid, struct access_coordinate *coord, coord->read_bandwidth, access);hmat_update_target_access (target,ACPI_HMAT_WRITE_BANDWIDTH , coord->write_bandwidth, access); - target->ext_updated = true;return 0 ; }@@ -773,10 +771,6 @@ static void hmat_update_target_attrs(struct memory_target *target,u32 best =0 ;int i; -/* Don't update if an external agent has changed the data. */ -if (target->ext_updated) -return ; -/* Don't update for generic port if there's no device handle */ if ((access == NODE_ACCESS_CLASS_GENPORT_SINK_LOCAL || access == NODE_ACCESS_CLASS_GENPORT_SINK_CPU) &&drivers/cxl/core/cdat.c +0
-5@@ -1081,8 +1081,3 @@ int cxl_update_hmat_access_coordinates(int nid, struct cxl_region *cxlr, {return hmat_update_target_coordinates (nid, &cxlr->coord[access], access); } - -bool cxl_need_node_perf_attrs_update (int nid) -{ -return !acpi_node_backed_by_real_pxm (nid); -}drivers/cxl/core/core.h +0
-1@@ -139,7 +139,6 @@ long cxl_pci_get_latency(struct pci_dev *pdev);int cxl_pci_get_bandwidth (struct pci_dev *pdev,struct access_coordinate *c);int cxl_update_hmat_access_coordinates (int nid,struct cxl_region *cxlr,enum access_coordinate_class access); -bool cxl_need_node_perf_attrs_update (int nid);int cxl_port_get_switch_dport_bandwidth (struct cxl_port *port,struct access_coordinate *c);drivers/cxl/core/region.c +2
-8@@ -2442,14 +2442,8 @@ static bool cxl_region_update_coordinates(struct cxl_region *cxlr, int nid)for (int i =0 ; i <ACCESS_COORDINATE_MAX ; i++) {if (cxlr->coord[i].read_bandwidth) { - rc =0 ; -if (cxl_need_node_perf_attrs_update (nid)) -node_set_perf_attrs (nid, &cxlr->coord[i], i); -else - rc =cxl_update_hmat_access_coordinates (nid, cxlr, i); - -if (rc ==0 ) - cset++; +node_update_perf_attrs (nid, &cxlr->coord[i], i); + cset++; } }include/linux/memory.h +1
-1@@ -120,8 +120,8 @@ struct mem_section;*/ #define DEFAULT_CALLBACK_PRI 0#define SLAB_CALLBACK_PRI 1 -#define HMAT_CALLBACK_PRI 2#define CXL_CALLBACK_PRI 5 +#define HMAT_CALLBACK_PRI 6#define MM_COMPUTE_BATCH_PRI 10#define CPUSET_CALLBACK_PRI 10#define MEMTIER_HOTPLUG_PRI 100
Dave Jiang
maintainerPATCH 4/4 acpi/hmat: Remove now unused hmat_update_target_coordinates()
In reply to Dave JiangRemove deadcode since CXL no longer calls hmat_update_target_coordinates().
--- drivers/acpi/numa/hmat.c | 28 ---------------------------- drivers/cxl/core/cdat.c | 6 ------ drivers/cxl/core/core.h | 2 -- include/linux/acpi.h | 12 ------------ 4 files changed, 48 deletions(-)
drivers/acpi/numa/hmat.c +0 -28
@@ -367,34 +367,6 @@ static void hmat_update_target_access(struct memory_target *target, } } -int hmat_update_target_coordinates (int nid,struct access_coordinate *coord, -enum access_coordinate_class access) -{ -struct memory_target *target; -int pxm; - -if (nid ==NUMA_NO_NODE ) -return -EINVAL ; - - pxm =node_to_pxm (nid); -guard (mutex)(&target_lock); - target =find_mem_target (pxm); -if (!target) -return -ENODEV ; - -hmat_update_target_access (target,ACPI_HMAT_READ_LATENCY , - coord->read_latency, access); -hmat_update_target_access (target,ACPI_HMAT_WRITE_LATENCY , - coord->write_latency, access); -hmat_update_target_access (target,ACPI_HMAT_READ_BANDWIDTH , - coord->read_bandwidth, access); -hmat_update_target_access (target,ACPI_HMAT_WRITE_BANDWIDTH , - coord->write_bandwidth, access); - -return 0 ; -} -EXPORT_SYMBOL_GPL (hmat_update_target_coordinates); -static __init void hmat_add_locality (struct acpi_hmat_locality *hmat_loc) {struct memory_locality *loc;
drivers/cxl/core/cdat.c +0 -6
@@ -1075,9 +1075,3 @@ void cxl_region_perf_data_calculate(struct cxl_region *cxlr, cxlr->coord[i].write_bandwidth += perf->coord[i].write_bandwidth; } } - -int cxl_update_hmat_access_coordinates (int nid,struct cxl_region *cxlr, -enum access_coordinate_class access) -{ -return hmat_update_target_coordinates (nid, &cxlr->coord[access], access); -}
drivers/cxl/core/core.h +0 -2
@@ -137,8 +137,6 @@ enum cxl_poison_trace_type {long cxl_pci_get_latency (struct pci_dev *pdev);int cxl_pci_get_bandwidth (struct pci_dev *pdev,struct access_coordinate *c); -int cxl_update_hmat_access_coordinates (int nid,struct cxl_region *cxlr, -enum access_coordinate_class access);int cxl_port_get_switch_dport_bandwidth (struct cxl_port *port,struct access_coordinate *c);
include/linux/acpi.h +0 -12
@@ -1595,18 +1595,6 @@ static inline void acpi_use_parent_companion(struct device *dev)ACPI_COMPANION_SET (dev,ACPI_COMPANION (dev->parent)); } -#ifdef CONFIG_ACPI_HMAT -int hmat_update_target_coordinates (int nid,struct access_coordinate *coord, -enum access_coordinate_class access); -#else -static inline int hmat_update_target_coordinates (int nid, -struct access_coordinate *coord, -enum access_coordinate_class access) -{ -return -EOPNOTSUPP ; -} -#endif -#ifdef CONFIG_ACPI_NUMA bool acpi_node_backed_by_real_pxm (int nid);#else
Signature · 1 line
2.50.1
Re: [PATCH 4/4] acpi/hmat: Remove now unused hmat_update_target_coordinates()
Quoted 3 lines · On Thu, 14 Aug 2025 10:16:50 -0700 Dave Jiang <[email protected]> wrote:
Remove deadcode since CXL no longer calls
hmat_update_target_coordinates().Signed-off-by Dave Jiang <[email protected]>
Hard to argue with this one ;)