* [PATCH 0/5] cacheinfo, powerpc: Move powerpc to generic cacheinfo
@ 2026-11-18 16:36 Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 1/5] cacheinfo: Allow architectures to provide custom cache IDs Shrikanth Hegde
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: Shrikanth Hegde @ 2026-11-18 16:36 UTC (permalink / raw)
To: linuxppc-dev, linux-kernel, driver-core
Cc: sshegde, christophe.leroy, maddy, gregkh, rafael, dakr, johan,
sudeep.holla, ritesh.list, tim.c.chen, yu.c.chen
This is another attempt to move powerpc towards generic cacheinfo
framework. This is spiritual successor to earlier attempts by Sudeep
Holla [1]
The context around this time was, when trying to enable Cache Aware
scheduling on powerpc using a llc_override kernel parameter[2], it was
pointed out that the llc_bytes which is retreived using
get_effective_llc_bytes fails and hence the cache aware scheduling is
effectively a NOP, moment the process has a visible footprint.
Instead of adding more hacks around with custom arch_llc_bytes, this is
an attempt to move powerpc to generic cacheinfo instead. This removes
the powerpc maintaining a separate sysfs entries for all caches.
A few modifications to the generic code is necessary due to some of the
conventions followed in the device tree specifications. The details are
in the respective patches.
If this approach does not work, then I am afraid we might have to fallback to
arch_llc_bytes.
Note that initial draft of the work[code,changelog] was carried out by LLM.
Provided the device tree blob and both powerpc,generic cacheinfo code.
With few iterations, LLM did generate the right code with few tweaks.
Only the cover-letter and Patch 5 are completely from me. Thanks to
LLM, this series is in shape in few days otherwise would have taken
more time.
Tested the values compatibilty on Power11 Dedicated and Shared LPAR's.
All the values are same, except id and uevent which was not there
before. That should help rather than hurt any tool.
This may need more broader checks on different boards/configuration
to ensure it is same. Also DLPAR CPU add/remove is not tested yet.
Please test and report if your board/platform shows up any different
values in those cache entries.
[1]: https://lore.kernel.org/all/1424715500-21725-1-git-send-email-sudeep.holla@arm.com/#t
[2]: https://lore.kernel.org/all/20261023131646.2907938-1-sshegde@linux.ibm.com/
This series is needed llc_override patch[2] series to work. A subsequent
version of that will be sent out soon.
PS: I am working with 90% coffee and 10% sleep.
Shrikanth Hegde (5):
cacheinfo: Allow architectures to provide custom cache IDs
cacheinfo: Support unified caches described by d-cache properties
cacheinfo: Add teardown and rebuild helpers
powerpc/cacheinfo: Move to generic cacheinfo
sched/topology: Remove stale comment around llc_bytes on powerpc
arch/powerpc/kernel/cacheinfo.c | 994 +++-------------------
arch/powerpc/kernel/cacheinfo.h | 13 -
arch/powerpc/kernel/sysfs.c | 3 -
arch/powerpc/platforms/pseries/mobility.c | 2 +-
drivers/base/cacheinfo.c | 111 ++-
include/linux/cacheinfo.h | 3 +
kernel/sched/topology.c | 7 -
7 files changed, 194 insertions(+), 939 deletions(-)
delete mode 100644 arch/powerpc/kernel/cacheinfo.h
--
2.54.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 1/5] cacheinfo: Allow architectures to provide custom cache IDs
2026-11-18 16:36 [PATCH 0/5] cacheinfo, powerpc: Move powerpc to generic cacheinfo Shrikanth Hegde
@ 2026-11-18 16:36 ` Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 2/5] cacheinfo: Support unified caches described by d-cache properties Shrikanth Hegde
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Shrikanth Hegde @ 2026-11-18 16:36 UTC (permalink / raw)
To: linuxppc-dev, linux-kernel, driver-core
Cc: sshegde, christophe.leroy, maddy, gregkh, rafael, dakr, johan,
sudeep.holla, ritesh.list, tim.c.chen, yu.c.chen
On PowerPC, a firmware cache node does not always correspond to a single
hardware cache instance.
For example, a Power11 CPU device-tree node can describe eight threads
and refer to one L2 and L3 cache-node chain:
CPU node (0-7)
l2-cache = <L2 node>
L2 node
l2-cache = <L3 node>
At the same time, the ibm,thread-groups property divides those threads
into two groups, such as:
0,2,4,6
1,3,5,7
Each group represents a separate hardware cache instance, even though
both groups refer to the same firmware L2 and L3 cache nodes.
Generic cacheinfo currently derives the cache ID from the firmware cache
node. It would therefore assign the same ID to both groups and incorrectly
represent all eight threads as sharing one cache.
Allow an architecture to provide the cache ID for a CPU and cache level.
When the architecture supplies an ID, use it instead of deriving one from
the firmware cache node. Otherwise, retain the existing generic behavior.
[Both draft changelog and code are written by LLM]
Assisted-by: LLM
Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
---
drivers/base/cacheinfo.c | 43 +++++++++++++++++++++++++++++++--------
include/linux/cacheinfo.h | 1 +
2 files changed, 35 insertions(+), 9 deletions(-)
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
index eb8a907f6d50..10a9bbce899e 100644
--- a/drivers/base/cacheinfo.c
+++ b/drivers/base/cacheinfo.c
@@ -243,22 +243,46 @@ static bool match_cache_node(struct device_node *cpu,
#define arch_compact_of_hwid(_x) (_x)
#endif
+/*
+ * Return an architecture-specific cache ID for @cpu and @level.
+ *
+ * Return 0 and store the ID in @id when the architecture provides the
+ * cache identity. Return an error to use the generic firmware-derived
+ * cache ID instead.
+ *
+ * The ID must uniquely identify a cache among caches of the same level
+ * and type.
+ */
+int __weak cacheinfo_get_arch_id(unsigned int cpu, unsigned int level,
+ unsigned int *id)
+{
+ return -ENOENT;
+}
+
static void cache_of_set_id(struct cacheinfo *this_leaf,
- struct device_node *cache_node)
+ struct device_node *cache_node,
+ unsigned int cpu)
{
- struct device_node *cpu;
+ struct device_node *cpu_node;
u32 min_id = ~0;
+ unsigned int arch_id;
+
+ if (!cacheinfo_get_arch_id(cpu, this_leaf->level, &arch_id)) {
+ this_leaf->id = arch_id;
+ this_leaf->attributes |= CACHE_ID;
+ return;
+ }
- for_each_of_cpu_node(cpu) {
- u64 id = of_get_cpu_hwid(cpu, 0);
+ for_each_of_cpu_node(cpu_node) {
+ u64 id = of_get_cpu_hwid(cpu_node, 0);
id = arch_compact_of_hwid(id);
if (FIELD_GET(GENMASK_ULL(63, 32), id)) {
- of_node_put(cpu);
+ of_node_put(cpu_node);
return;
}
- if (match_cache_node(cpu, cache_node))
+ if (match_cache_node(cpu_node, cache_node))
min_id = min(min_id, id);
}
@@ -269,7 +293,8 @@ static void cache_of_set_id(struct cacheinfo *this_leaf,
}
static void cache_of_set_props(struct cacheinfo *this_leaf,
- struct device_node *np)
+ struct device_node *np,
+ unsigned int cpu)
{
/*
* init_cache_level must setup the cache level correctly
@@ -283,7 +308,7 @@ static void cache_of_set_props(struct cacheinfo *this_leaf,
cache_get_line_size(this_leaf, np);
cache_nr_sets(this_leaf, np);
cache_associativity(this_leaf);
- cache_of_set_id(this_leaf, np);
+ cache_of_set_id(this_leaf, np, cpu);
}
static int cache_setup_of_node(unsigned int cpu)
@@ -309,7 +334,7 @@ static int cache_setup_of_node(unsigned int cpu)
if (!np)
break;
}
- cache_of_set_props(this_leaf, np);
+ cache_of_set_props(this_leaf, np, cpu);
this_leaf->fw_token = np;
index++;
}
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
index 56fa646df0d1..d3557a1b464b 100644
--- a/include/linux/cacheinfo.h
+++ b/include/linux/cacheinfo.h
@@ -93,6 +93,7 @@ bool last_level_cache_is_shared(unsigned int cpu_x, unsigned int cpu_y);
struct cacheinfo *get_cpu_cacheinfo_llc(unsigned int cpu);
int fetch_cache_info(unsigned int cpu);
int detect_cache_attributes(unsigned int cpu);
+int cacheinfo_get_arch_id(unsigned int cpu, unsigned int level, unsigned int *id);
#ifndef CONFIG_ACPI_PPTT
/*
* acpi_get_cache_info() is only called on ACPI enabled
--
2.54.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 2/5] cacheinfo: Support unified caches described by d-cache properties
2026-11-18 16:36 [PATCH 0/5] cacheinfo, powerpc: Move powerpc to generic cacheinfo Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 1/5] cacheinfo: Allow architectures to provide custom cache IDs Shrikanth Hegde
@ 2026-11-18 16:36 ` Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 3/5] cacheinfo: Add teardown and rebuild helpers Shrikanth Hegde
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Shrikanth Hegde @ 2026-11-18 16:36 UTC (permalink / raw)
To: linuxppc-dev, linux-kernel, driver-core
Cc: sshegde, christophe.leroy, maddy, gregkh, rafael, dakr, johan,
sudeep.holla, ritesh.list, tim.c.chen, yu.c.chen
PowerPC device trees commonly describe a unified cache using d-cache-*
properties.
For example, Power11 L2 and L3 cache nodes have cache-unified together
with properties such as:
d-cache-size
d-cache-line-size
d-cache-sets
They do not provide the corresponding cache-size, cache-line-size and
cache-sets properties expected by generic cacheinfo for a unified cache.
When a unified cache node does not contain cache-size but contains
d-cache-size, select the data-cache property set. This only changes which
device-tree properties are read. The cache leaf remains
CACHE_TYPE_UNIFIED.
[Both draft changelog and code are written by LLM]
Assisted-by: LLM
Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
---
drivers/base/cacheinfo.c | 24 +++++++++++++++++-------
1 file changed, 17 insertions(+), 7 deletions(-)
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
index 10a9bbce899e..4a23377e9312 100644
--- a/drivers/base/cacheinfo.c
+++ b/drivers/base/cacheinfo.c
@@ -148,11 +148,21 @@ static const struct cache_type_info cache_type_info[] = {
},
};
-static inline int get_cacheinfo_idx(enum cache_type type)
+static inline int get_cacheinfo_idx(enum cache_type type, struct device_node *np)
{
- if (type == CACHE_TYPE_UNIFIED)
- return 0;
- return type;
+ if (type != CACHE_TYPE_UNIFIED)
+ return type;
+
+ /*
+ * Some device-tree bindings describe unified caches using
+ * d-cache-* properties. Select the data-cache property set
+ * without changing the cache leaf type.
+ */
+ if (!of_property_present(np, "cache-size") &&
+ of_property_present(np, "d-cache-size"))
+ return CACHE_TYPE_DATA;
+
+ return 0;
}
static void cache_size(struct cacheinfo *this_leaf, struct device_node *np)
@@ -160,7 +170,7 @@ static void cache_size(struct cacheinfo *this_leaf, struct device_node *np)
const char *propname;
int ct_idx;
- ct_idx = get_cacheinfo_idx(this_leaf->type);
+ ct_idx = get_cacheinfo_idx(this_leaf->type, np);
propname = cache_type_info[ct_idx].size_prop;
of_property_read_u32(np, propname, &this_leaf->size);
@@ -172,7 +182,7 @@ static void cache_get_line_size(struct cacheinfo *this_leaf,
{
int i, lim, ct_idx;
- ct_idx = get_cacheinfo_idx(this_leaf->type);
+ ct_idx = get_cacheinfo_idx(this_leaf->type, np);
lim = ARRAY_SIZE(cache_type_info[ct_idx].line_size_props);
for (i = 0; i < lim; i++) {
@@ -194,7 +204,7 @@ static void cache_nr_sets(struct cacheinfo *this_leaf, struct device_node *np)
const char *propname;
int ct_idx;
- ct_idx = get_cacheinfo_idx(this_leaf->type);
+ ct_idx = get_cacheinfo_idx(this_leaf->type, np);
propname = cache_type_info[ct_idx].nr_sets_prop;
of_property_read_u32(np, propname, &this_leaf->number_of_sets);
--
2.54.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 3/5] cacheinfo: Add teardown and rebuild helpers
2026-11-18 16:36 [PATCH 0/5] cacheinfo, powerpc: Move powerpc to generic cacheinfo Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 1/5] cacheinfo: Allow architectures to provide custom cache IDs Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 2/5] cacheinfo: Support unified caches described by d-cache properties Shrikanth Hegde
@ 2026-11-18 16:36 ` Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 4/5] powerpc/cacheinfo: Move to generic cacheinfo Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 5/5] sched/topology: Remove stale comment around llc_bytes on powerpc Shrikanth Hegde
4 siblings, 0 replies; 6+ messages in thread
From: Shrikanth Hegde @ 2026-11-18 16:36 UTC (permalink / raw)
To: linuxppc-dev, linux-kernel, driver-core
Cc: sshegde, christophe.leroy, maddy, gregkh, rafael, dakr, johan,
sudeep.holla, ritesh.list, tim.c.chen, yu.c.chen
The PowerPC cacheinfo implementation provides teardown and rebuild
operations for pSeries mobility.
During partition migration, firmware cache nodes may be replaced as part
of the device-tree update. Cacheinfo must first remove its sysfs objects
and release references to the old firmware nodes. After the device tree
has been updated, it must discover the new cache hierarchy and recreate
the cache information for each online CPU.
Provide equivalent helpers in generic cacheinfo so architectures moving
from private cacheinfo implementations can retain this behavior.
Both helpers require the CPU hotplug lock to be held. The teardown helper
removes cache sysfs devices and sharing-map membership before freeing the
per-CPU cache leaves. The rebuild helper rediscovers and recreates
cacheinfo for every online CPU.
[Both draft changelog and code are written by LLM]
Assisted-by: LLM
Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
---
drivers/base/cacheinfo.c | 44 +++++++++++++++++++++++++++++++++++++++
include/linux/cacheinfo.h | 2 ++
2 files changed, 46 insertions(+)
diff --git a/drivers/base/cacheinfo.c b/drivers/base/cacheinfo.c
index 4a23377e9312..b13046863ea3 100644
--- a/drivers/base/cacheinfo.c
+++ b/drivers/base/cacheinfo.c
@@ -1120,6 +1120,50 @@ static int cacheinfo_cpu_pre_down(unsigned int cpu)
return 0;
}
+void cacheinfo_teardown(void)
+{
+ unsigned int cpu;
+
+ lockdep_assert_cpus_held();
+
+ /*
+ * Remove sysfs entries and detach online CPUs from the
+ * cache-sharing maps before releasing cache leaves that may
+ * reference firmware nodes.
+ */
+ for_each_online_cpu(cpu)
+ cacheinfo_cpu_pre_down(cpu);
+
+ for_each_possible_cpu(cpu) {
+ struct cpu_cacheinfo *ci = ci_cacheinfo(cpu);
+
+ kfree(ci->info_list);
+ ci->info_list = NULL;
+ ci->num_levels = 0;
+ ci->num_leaves = 0;
+ ci->per_cpu_data_slice_size = 0;
+ ci->cpu_map_populated = false;
+ ci->early_ci_levels = false;
+ }
+}
+
+int cacheinfo_rebuild(void)
+{
+ unsigned int cpu;
+ int first_err = 0;
+ int ret;
+
+ lockdep_assert_cpus_held();
+
+ for_each_online_cpu(cpu) {
+ ret = cacheinfo_cpu_online(cpu);
+ if (ret && !first_err)
+ first_err = ret;
+ }
+
+ return first_err;
+}
+
static int __init cacheinfo_sysfs_init(void)
{
return cpuhp_setup_state(CPUHP_AP_BASE_CACHEINFO_ONLINE,
diff --git a/include/linux/cacheinfo.h b/include/linux/cacheinfo.h
index d3557a1b464b..89f2f9fcc88d 100644
--- a/include/linux/cacheinfo.h
+++ b/include/linux/cacheinfo.h
@@ -94,6 +94,8 @@ struct cacheinfo *get_cpu_cacheinfo_llc(unsigned int cpu);
int fetch_cache_info(unsigned int cpu);
int detect_cache_attributes(unsigned int cpu);
int cacheinfo_get_arch_id(unsigned int cpu, unsigned int level, unsigned int *id);
+void cacheinfo_teardown(void);
+int cacheinfo_rebuild(void);
#ifndef CONFIG_ACPI_PPTT
/*
* acpi_get_cache_info() is only called on ACPI enabled
--
2.54.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 4/5] powerpc/cacheinfo: Move to generic cacheinfo
2026-11-18 16:36 [PATCH 0/5] cacheinfo, powerpc: Move powerpc to generic cacheinfo Shrikanth Hegde
` (2 preceding siblings ...)
2026-11-18 16:36 ` [PATCH 3/5] cacheinfo: Add teardown and rebuild helpers Shrikanth Hegde
@ 2026-11-18 16:36 ` Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 5/5] sched/topology: Remove stale comment around llc_bytes on powerpc Shrikanth Hegde
4 siblings, 0 replies; 6+ messages in thread
From: Shrikanth Hegde @ 2026-11-18 16:36 UTC (permalink / raw)
To: linuxppc-dev, linux-kernel, driver-core
Cc: sshegde, christophe.leroy, maddy, gregkh, rafael, dakr, johan,
sudeep.holla, ritesh.list, tim.c.chen, yu.c.chen
PowerPC maintains an architecture-specific implementation of the CPU
cache sysfs interface. Replace it with the generic cacheinfo
infrastructure.
PowerPC requires some architecture-specific handling when identifying
hardware cache instances. On Power11, one firmware CPU node can represent
eight threads and refer to one L2 and L3 cache-node chain. However, the
ibm,thread-groups property divides those threads into two cache-sharing
groups, for example:
0,2,4,6
1,3,5,7
The firmware node alone therefore cannot distinguish the two hardware
cache instances. Implement cacheinfo_get_arch_id() using the PowerPC
thread-group masks and use the first CPU in each mask as the cache ID.
This gives the two groups distinct IDs while assigning the same ID to all
CPUs that share a cache.
Initialize the generic cache leaves by walking the firmware cache chain.
A node with cache-unified produces one unified leaf. Other nodes produce
separate data and instruction leaves. Generic cacheinfo then populates
the size, line size, number of sets, associativity, firmware token and
shared CPU map.
The existing PowerPC cache sysfs attributes retain their values. Generic
cacheinfo additionally exposes the standard cache id attribute. The IDs
can be sparse because they are derived from the first CPU in each
thread-group mask, but CPUs sharing a cache have the same ID and distinct
cache instances have different IDs.
Remove the PowerPC-specific sysfs and CPU-hotplug handling. Update
pSeries mobility to use the generic cacheinfo teardown and rebuild
helpers.
On a dedicated Power11 LPAR, the generic implementation preserves the
existing L1 data, L1 instruction, L2 and L3 sizes, associativity, set
counts and shared CPU maps. The two SMT4 groups continue to appear as
separate cache instances.
[Both draft changelog and code are written by LLM]
Assisted-by: LLM
Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
---
arch/powerpc/kernel/cacheinfo.c | 994 +++-------------------
arch/powerpc/kernel/cacheinfo.h | 13 -
arch/powerpc/kernel/sysfs.c | 3 -
arch/powerpc/platforms/pseries/mobility.c | 2 +-
4 files changed, 96 insertions(+), 916 deletions(-)
delete mode 100644 arch/powerpc/kernel/cacheinfo.h
diff --git a/arch/powerpc/kernel/cacheinfo.c b/arch/powerpc/kernel/cacheinfo.c
index cb0933061006..f3046c7c56d3 100644
--- a/arch/powerpc/kernel/cacheinfo.c
+++ b/arch/powerpc/kernel/cacheinfo.c
@@ -1,955 +1,151 @@
// SPDX-License-Identifier: GPL-2.0-only
/*
- * Processor cache information made available to userspace via sysfs;
- * intended to be compatible with x86 intel_cacheinfo implementation.
+ * PowerPC cacheinfo support using generic cacheinfo
+ *
+ * Copyright 2008,2026 IBM Corporation
*
- * Copyright 2008 IBM Corporation
* Author: Nathan Lynch
+ * Author: Shrikanth Hegde
*/
-#define pr_fmt(fmt) "cacheinfo: " fmt
-
-#include <linux/cpu.h>
+#include <linux/cacheinfo.h>
#include <linux/cpumask.h>
-#include <linux/kernel.h>
-#include <linux/kobject.h>
-#include <linux/list.h>
-#include <linux/notifier.h>
+#include <linux/errno.h>
#include <linux/of.h>
#include <linux/percpu.h>
-#include <linux/slab.h>
-#include <linux/sysfs.h>
+
#include <asm/cputhreads.h>
#include <asm/smp.h>
-#include "cacheinfo.h"
-
-/* per-cpu object for tracking:
- * - a "cache" kobject for the top-level directory
- * - a list of "index" objects representing the cpu's local cache hierarchy
- */
-struct cache_dir {
- struct kobject *kobj; /* bare (not embedded) kobject for cache
- * directory */
- struct cache_index_dir *index; /* list of index objects */
-};
-
-/* "index" object: each cpu's cache directory has an index
- * subdirectory corresponding to a cache object associated with the
- * cpu. This object's lifetime is managed via the embedded kobject.
- */
-struct cache_index_dir {
- struct kobject kobj;
- struct cache_index_dir *next; /* next index in parent directory */
- struct cache *cache;
-};
-
-/* Template for determining which OF properties to query for a given
- * cache type */
-struct cache_type_info {
- const char *name;
- const char *size_prop;
-
- /* Allow for both [di]-cache-line-size and
- * [di]-cache-block-size properties. According to the PowerPC
- * Processor binding, -line-size should be provided if it
- * differs from the cache block size (that which is operated
- * on by cache instructions), so we look for -line-size first.
- * See cache_get_line_size(). */
-
- const char *line_size_props[2];
- const char *nr_sets_prop;
-};
-
-/* These are used to index the cache_type_info array. */
-#define CACHE_TYPE_UNIFIED 0 /* cache-size, cache-block-size, etc. */
-#define CACHE_TYPE_UNIFIED_D 1 /* d-cache-size, d-cache-block-size, etc */
-#define CACHE_TYPE_INSTRUCTION 2
-#define CACHE_TYPE_DATA 3
-
-static const struct cache_type_info cache_type_info[] = {
- {
- /* Embedded systems that use cache-size, cache-block-size,
- * etc. for the Unified (typically L2) cache. */
- .name = "Unified",
- .size_prop = "cache-size",
- .line_size_props = { "cache-line-size",
- "cache-block-size", },
- .nr_sets_prop = "cache-sets",
- },
- {
- /* PowerPC Processor binding says the [di]-cache-*
- * must be equal on unified caches, so just use
- * d-cache properties. */
- .name = "Unified",
- .size_prop = "d-cache-size",
- .line_size_props = { "d-cache-line-size",
- "d-cache-block-size", },
- .nr_sets_prop = "d-cache-sets",
- },
- {
- .name = "Instruction",
- .size_prop = "i-cache-size",
- .line_size_props = { "i-cache-line-size",
- "i-cache-block-size", },
- .nr_sets_prop = "i-cache-sets",
- },
- {
- .name = "Data",
- .size_prop = "d-cache-size",
- .line_size_props = { "d-cache-line-size",
- "d-cache-block-size", },
- .nr_sets_prop = "d-cache-sets",
- },
-};
-
-/* Cache object: each instance of this corresponds to a distinct cache
- * in the system. There are separate objects for Harvard caches: one
- * each for instruction and data, and each refers to the same OF node.
- * The refcount of the OF node is elevated for the lifetime of the
- * cache object. A cache object is released when its shared_cpu_map
- * is cleared (see cache_cpu_clear).
- *
- * A cache object is on two lists: an unsorted global list
- * (cache_list) of cache objects; and a singly-linked list
- * representing the local cache hierarchy, which is ordered by level
- * (e.g. L1d -> L1i -> L2 -> L3).
- */
-struct cache {
- struct device_node *ofnode; /* OF node for this cache, may be cpu */
- struct cpumask shared_cpu_map; /* online CPUs using this cache */
- int type; /* split cache disambiguation */
- int level; /* level not explicit in device tree */
- int group_id; /* id of the group of threads that share this cache */
- struct list_head list; /* global list of cache objects */
- struct cache *next_local; /* next cache of >= level */
-};
-
-static DEFINE_PER_CPU(struct cache_dir *, cache_dir_pcpu);
-
-/* traversal/modification of this list occurs only at cpu hotplug time;
- * access is serialized by cpu hotplug locking
- */
-static LIST_HEAD(cache_list);
-
-static struct cache_index_dir *kobj_to_cache_index_dir(struct kobject *k)
-{
- return container_of(k, struct cache_index_dir, kobj);
-}
-
-static const char *cache_type_string(const struct cache *cache)
-{
- return cache_type_info[cache->type].name;
-}
-
-static void cache_init(struct cache *cache, int type, int level,
- struct device_node *ofnode, int group_id)
-{
- cache->type = type;
- cache->level = level;
- cache->ofnode = of_node_get(ofnode);
- cache->group_id = group_id;
- INIT_LIST_HEAD(&cache->list);
- list_add(&cache->list, &cache_list);
-}
-
-static struct cache *new_cache(int type, int level,
- struct device_node *ofnode, int group_id)
-{
- struct cache *cache;
-
- cache = kzalloc_obj(*cache);
- if (cache)
- cache_init(cache, type, level, ofnode, group_id);
-
- return cache;
-}
-
-static void release_cache_debugcheck(struct cache *cache)
-{
- struct cache *iter;
-
- list_for_each_entry(iter, &cache_list, list)
- WARN_ONCE(iter->next_local == cache,
- "cache for %pOFP(%s) refers to cache for %pOFP(%s)\n",
- iter->ofnode,
- cache_type_string(iter),
- cache->ofnode,
- cache_type_string(cache));
-}
-
-static void release_cache(struct cache *cache)
-{
- if (!cache)
- return;
-
- pr_debug("freeing L%d %s cache for %pOFP\n", cache->level,
- cache_type_string(cache), cache->ofnode);
-
- release_cache_debugcheck(cache);
- list_del(&cache->list);
- of_node_put(cache->ofnode);
- kfree(cache);
-}
-
-static void cache_cpu_set(struct cache *cache, int cpu)
-{
- struct cache *next = cache;
-
- while (next) {
- WARN_ONCE(cpumask_test_cpu(cpu, &next->shared_cpu_map),
- "CPU %i already accounted in %pOFP(%s)\n",
- cpu, next->ofnode,
- cache_type_string(next));
- cpumask_set_cpu(cpu, &next->shared_cpu_map);
- next = next->next_local;
- }
-}
-
-static int cache_size(const struct cache *cache, unsigned int *ret)
-{
- const char *propname;
- const __be32 *cache_size;
-
- propname = cache_type_info[cache->type].size_prop;
-
- cache_size = of_get_property(cache->ofnode, propname, NULL);
- if (!cache_size)
- return -ENODEV;
-
- *ret = of_read_number(cache_size, 1);
- return 0;
-}
-
-static int cache_size_kb(const struct cache *cache, unsigned int *ret)
-{
- unsigned int size;
-
- if (cache_size(cache, &size))
- return -ENODEV;
-
- *ret = size / 1024;
- return 0;
-}
-
-/* not cache_line_size() because that's a macro in include/linux/cache.h */
-static int cache_get_line_size(const struct cache *cache, unsigned int *ret)
-{
- const __be32 *line_size;
- int i, lim;
-
- lim = ARRAY_SIZE(cache_type_info[cache->type].line_size_props);
-
- for (i = 0; i < lim; i++) {
- const char *propname;
-
- propname = cache_type_info[cache->type].line_size_props[i];
- line_size = of_get_property(cache->ofnode, propname, NULL);
- if (line_size)
- break;
- }
-
- if (!line_size)
- return -ENODEV;
-
- *ret = of_read_number(line_size, 1);
- return 0;
-}
-
-static int cache_nr_sets(const struct cache *cache, unsigned int *ret)
-{
- const char *propname;
- const __be32 *nr_sets;
-
- propname = cache_type_info[cache->type].nr_sets_prop;
-
- nr_sets = of_get_property(cache->ofnode, propname, NULL);
- if (!nr_sets)
- return -ENODEV;
-
- *ret = of_read_number(nr_sets, 1);
- return 0;
-}
-
-static int cache_associativity(const struct cache *cache, unsigned int *ret)
-{
- unsigned int line_size;
- unsigned int nr_sets;
- unsigned int size;
-
- if (cache_nr_sets(cache, &nr_sets))
- goto err;
-
- /* If the cache is fully associative, there is no need to
- * check the other properties.
- */
- if (nr_sets == 1) {
- *ret = 0;
- return 0;
- }
-
- if (cache_get_line_size(cache, &line_size))
- goto err;
- if (cache_size(cache, &size))
- goto err;
-
- if (!(nr_sets > 0 && size > 0 && line_size > 0))
- goto err;
-
- *ret = (size / nr_sets) / line_size;
- return 0;
-err:
- return -ENODEV;
-}
-
-/* helper for dealing with split caches */
-static struct cache *cache_find_first_sibling(struct cache *cache)
-{
- struct cache *iter;
-
- if (cache->type == CACHE_TYPE_UNIFIED ||
- cache->type == CACHE_TYPE_UNIFIED_D)
- return cache;
-
- list_for_each_entry(iter, &cache_list, list)
- if (iter->ofnode == cache->ofnode &&
- iter->group_id == cache->group_id &&
- iter->next_local == cache)
- return iter;
-
- return cache;
-}
-
-/* return the first cache on a local list matching node and thread-group id */
-static struct cache *cache_lookup_by_node_group(const struct device_node *node,
- int group_id)
-{
- struct cache *cache = NULL;
- struct cache *iter;
-
- list_for_each_entry(iter, &cache_list, list) {
- if (iter->ofnode != node ||
- iter->group_id != group_id)
- continue;
- cache = cache_find_first_sibling(iter);
- break;
- }
-
- return cache;
-}
-
-static bool cache_node_is_unified(const struct device_node *np)
-{
- return of_get_property(np, "cache-unified", NULL);
-}
-
/*
- * Unified caches can have two different sets of tags. Most embedded
- * use cache-size, etc. for the unified cache size, but open firmware systems
- * use d-cache-size, etc. Check on initialization for which type we have, and
- * return the appropriate structure type. Assume it's embedded if it isn't
- * open firmware. If it's yet a 3rd type, then there will be missing entries
- * in /sys/devices/system/cpu/cpu0/cache/index2/, and this code will need
- * to be extended further.
+ * A firmware CPU node can represent multiple hardware cache groups.
+ * Use the first logical CPU in the corresponding thread-group mask as
+ * the identifier of that cache instance.
*/
-static int cache_is_unified_d(const struct device_node *np)
-{
- return of_get_property(np,
- cache_type_info[CACHE_TYPE_UNIFIED_D].size_prop, NULL) ?
- CACHE_TYPE_UNIFIED_D : CACHE_TYPE_UNIFIED;
-}
-
-static struct cache *cache_do_one_devnode_unified(struct device_node *node, int group_id,
- int level)
-{
- pr_debug("creating L%d ucache for %pOFP\n", level, node);
-
- return new_cache(cache_is_unified_d(node), level, node, group_id);
-}
-
-static struct cache *cache_do_one_devnode_split(struct device_node *node, int group_id,
- int level)
-{
- struct cache *dcache, *icache;
-
- pr_debug("creating L%d dcache and icache for %pOFP\n", level,
- node);
-
- dcache = new_cache(CACHE_TYPE_DATA, level, node, group_id);
- icache = new_cache(CACHE_TYPE_INSTRUCTION, level, node, group_id);
-
- if (!dcache || !icache)
- goto err;
-
- dcache->next_local = icache;
-
- return dcache;
-err:
- release_cache(dcache);
- release_cache(icache);
- return NULL;
-}
-
-static struct cache *cache_do_one_devnode(struct device_node *node, int group_id, int level)
-{
- struct cache *cache;
-
- if (cache_node_is_unified(node))
- cache = cache_do_one_devnode_unified(node, group_id, level);
- else
- cache = cache_do_one_devnode_split(node, group_id, level);
-
- return cache;
-}
-
-static struct cache *cache_lookup_or_instantiate(struct device_node *node,
- int group_id,
- int level)
-{
- struct cache *cache;
-
- cache = cache_lookup_by_node_group(node, group_id);
-
- WARN_ONCE(cache && cache->level != level,
- "cache level mismatch on lookup (got %d, expected %d)\n",
- cache->level, level);
-
- if (!cache)
- cache = cache_do_one_devnode(node, group_id, level);
-
- return cache;
-}
-
-static void link_cache_lists(struct cache *smaller, struct cache *bigger)
-{
- while (smaller->next_local) {
- if (smaller->next_local == bigger)
- return; /* already linked */
- smaller = smaller->next_local;
- }
-
- smaller->next_local = bigger;
-
- /*
- * The cache->next_local list sorts by level ascending:
- * L1d -> L1i -> L2 -> L3 ...
- */
- WARN_ONCE((smaller->level == 1 && bigger->level > 2) ||
- (smaller->level > 1 && bigger->level != smaller->level + 1),
- "linking L%i cache %pOFP to L%i cache %pOFP; skipped a level?\n",
- smaller->level, smaller->ofnode, bigger->level, bigger->ofnode);
-}
-
-static void do_subsidiary_caches_debugcheck(struct cache *cache)
+int cacheinfo_get_arch_id(unsigned int cpu, unsigned int level,
+ unsigned int *id)
{
- WARN_ONCE(cache->level != 1,
- "instantiating cache chain from L%d %s cache for "
- "%pOFP instead of an L1\n", cache->level,
- cache_type_string(cache), cache->ofnode);
- WARN_ONCE(!of_node_is_type(cache->ofnode, "cpu"),
- "instantiating cache chain from node %pOFP of type '%s' "
- "instead of a cpu node\n", cache->ofnode,
- of_node_get_device_type(cache->ofnode));
-}
+ const struct cpumask *mask = NULL;
-/*
- * If sub-groups of threads in a core containing @cpu_id share the
- * L@level-cache (information obtained via "ibm,thread-groups"
- * device-tree property), then we identify the group by the first
- * thread-sibling in the group. We define this to be the group-id.
- *
- * In the absence of any thread-group information for L@level-cache,
- * this function returns -1.
- */
-static int get_group_id(unsigned int cpu_id, int level)
-{
if (has_big_cores && level == 1)
- return cpumask_first(per_cpu(thread_group_l1_cache_map,
- cpu_id));
+ mask = per_cpu(thread_group_l1_cache_map, cpu);
else if (thread_group_shares_l2 && level == 2)
- return cpumask_first(per_cpu(thread_group_l2_cache_map,
- cpu_id));
+ mask = per_cpu(thread_group_l2_cache_map, cpu);
else if (thread_group_shares_l3 && level == 3)
- return cpumask_first(per_cpu(thread_group_l3_cache_map,
- cpu_id));
- return -1;
-}
-
-static void do_subsidiary_caches(struct cache *cache, unsigned int cpu_id)
-{
- struct device_node *subcache_node;
- int level = cache->level;
+ mask = per_cpu(thread_group_l3_cache_map, cpu);
- do_subsidiary_caches_debugcheck(cache);
+ if (!mask || cpumask_empty(mask))
+ return -ENOENT;
- while ((subcache_node = of_find_next_cache_node(cache->ofnode))) {
- struct cache *subcache;
- int group_id;
-
- level++;
- group_id = get_group_id(cpu_id, level);
- subcache = cache_lookup_or_instantiate(subcache_node, group_id, level);
- of_node_put(subcache_node);
- if (!subcache)
- break;
-
- link_cache_lists(cache, subcache);
- cache = subcache;
- }
-}
+ *id = cpumask_first(mask);
+ if (*id >= nr_cpu_ids)
+ return -ENOENT;
-static struct cache *cache_chain_instantiate(unsigned int cpu_id)
-{
- struct device_node *cpu_node;
- struct cache *cpu_cache = NULL;
- int group_id;
-
- pr_debug("creating cache object(s) for CPU %i\n", cpu_id);
-
- cpu_node = of_get_cpu_node(cpu_id, NULL);
- WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
- if (!cpu_node)
- goto out;
-
- group_id = get_group_id(cpu_id, 1);
-
- cpu_cache = cache_lookup_or_instantiate(cpu_node, group_id, 1);
- if (!cpu_cache)
- goto out;
-
- do_subsidiary_caches(cpu_cache, cpu_id);
-
- cache_cpu_set(cpu_cache, cpu_id);
-out:
- of_node_put(cpu_node);
-
- return cpu_cache;
-}
-
-static struct cache_dir *cacheinfo_create_cache_dir(unsigned int cpu_id)
-{
- struct cache_dir *cache_dir;
- struct device *dev;
- struct kobject *kobj = NULL;
-
- dev = get_cpu_device(cpu_id);
- WARN_ONCE(!dev, "no dev for CPU %i\n", cpu_id);
- if (!dev)
- goto err;
-
- kobj = kobject_create_and_add("cache", &dev->kobj);
- if (!kobj)
- goto err;
-
- cache_dir = kzalloc_obj(*cache_dir);
- if (!cache_dir)
- goto err;
-
- cache_dir->kobj = kobj;
-
- WARN_ON_ONCE(per_cpu(cache_dir_pcpu, cpu_id) != NULL);
-
- per_cpu(cache_dir_pcpu, cpu_id) = cache_dir;
-
- return cache_dir;
-err:
- kobject_put(kobj);
- return NULL;
-}
-
-static void cache_index_release(struct kobject *kobj)
-{
- struct cache_index_dir *index;
-
- index = kobj_to_cache_index_dir(kobj);
-
- pr_debug("freeing index directory for L%d %s cache\n",
- index->cache->level, cache_type_string(index->cache));
-
- kfree(index);
-}
-
-static ssize_t cache_index_show(struct kobject *k, struct attribute *attr, char *buf)
-{
- struct kobj_attribute *kobj_attr;
-
- kobj_attr = container_of(attr, struct kobj_attribute, attr);
-
- return kobj_attr->show(k, kobj_attr, buf);
-}
-
-static struct cache *index_kobj_to_cache(struct kobject *k)
-{
- struct cache_index_dir *index;
-
- index = kobj_to_cache_index_dir(k);
-
- return index->cache;
-}
-
-static ssize_t size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
- unsigned int size_kb;
- struct cache *cache;
-
- cache = index_kobj_to_cache(k);
-
- if (cache_size_kb(cache, &size_kb))
- return -ENODEV;
-
- return sysfs_emit(buf, "%uK\n", size_kb);
-}
-
-static struct kobj_attribute cache_size_attr =
- __ATTR(size, 0444, size_show, NULL);
-
-
-static ssize_t line_size_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
- unsigned int line_size;
- struct cache *cache;
-
- cache = index_kobj_to_cache(k);
-
- if (cache_get_line_size(cache, &line_size))
- return -ENODEV;
-
- return sysfs_emit(buf, "%u\n", line_size);
-}
-
-static struct kobj_attribute cache_line_size_attr =
- __ATTR(coherency_line_size, 0444, line_size_show, NULL);
-
-static ssize_t nr_sets_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
- unsigned int nr_sets;
- struct cache *cache;
-
- cache = index_kobj_to_cache(k);
-
- if (cache_nr_sets(cache, &nr_sets))
- return -ENODEV;
-
- return sysfs_emit(buf, "%u\n", nr_sets);
-}
-
-static struct kobj_attribute cache_nr_sets_attr =
- __ATTR(number_of_sets, 0444, nr_sets_show, NULL);
-
-static ssize_t associativity_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
- unsigned int associativity;
- struct cache *cache;
-
- cache = index_kobj_to_cache(k);
-
- if (cache_associativity(cache, &associativity))
- return -ENODEV;
-
- return sysfs_emit(buf, "%u\n", associativity);
-}
-
-static struct kobj_attribute cache_assoc_attr =
- __ATTR(ways_of_associativity, 0444, associativity_show, NULL);
-
-static ssize_t type_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
- struct cache *cache;
-
- cache = index_kobj_to_cache(k);
-
- return sysfs_emit(buf, "%s\n", cache_type_string(cache));
-}
-
-static struct kobj_attribute cache_type_attr =
- __ATTR(type, 0444, type_show, NULL);
-
-static ssize_t level_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
- struct cache_index_dir *index;
- struct cache *cache;
-
- index = kobj_to_cache_index_dir(k);
- cache = index->cache;
-
- return sysfs_emit(buf, "%d\n", cache->level);
-}
-
-static struct kobj_attribute cache_level_attr =
- __ATTR(level, 0444, level_show, NULL);
-
-static ssize_t
-show_shared_cpumap(struct kobject *k, struct kobj_attribute *attr, char *buf, bool list)
-{
- struct cache_index_dir *index;
- struct cache *cache;
- const struct cpumask *mask;
-
- index = kobj_to_cache_index_dir(k);
- cache = index->cache;
-
- mask = &cache->shared_cpu_map;
-
- return sysfs_emit(buf, list ? "%*pbl\n" : "%*pb\n",
- cpumask_pr_args(mask));
-}
-
-static ssize_t shared_cpu_map_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
-{
- return show_shared_cpumap(k, attr, buf, false);
+ return 0;
}
-static ssize_t shared_cpu_list_show(struct kobject *k, struct kobj_attribute *attr, char *buf)
+static void powerpc_cache_leaf_init(struct cacheinfo *leaf,
+ unsigned int level,
+ enum cache_type type)
{
- return show_shared_cpumap(k, attr, buf, true);
+ leaf->level = level;
+ leaf->type = type;
}
-static struct kobj_attribute cache_shared_cpu_map_attr =
- __ATTR(shared_cpu_map, 0444, shared_cpu_map_show, NULL);
-
-static struct kobj_attribute cache_shared_cpu_list_attr =
- __ATTR(shared_cpu_list, 0444, shared_cpu_list_show, NULL);
-
-/* Attributes which should always be created -- the kobject/sysfs core
- * does this automatically via kobj_type->default_groups. This is the
- * minimum data required to uniquely identify a cache.
- */
-static struct attribute *cache_index_default_attrs[] = {
- &cache_type_attr.attr,
- &cache_level_attr.attr,
- &cache_shared_cpu_map_attr.attr,
- &cache_shared_cpu_list_attr.attr,
- NULL,
-};
-ATTRIBUTE_GROUPS(cache_index_default);
-
-/* Attributes which should be created if the cache device node has the
- * right properties -- see cacheinfo_create_index_opt_attrs
+/*
+ * A node carrying cache-unified represents one unified cache leaf.
+ * Other nodes represent separate data and instruction cache leaves.
+ *
+ * Cache property availability does not determine whether a leaf
+ * exists. Missing optional properties are handled by generic
+ * cacheinfo.
*/
-static struct kobj_attribute *cache_index_opt_attrs[] = {
- &cache_size_attr,
- &cache_line_size_attr,
- &cache_nr_sets_attr,
- &cache_assoc_attr,
-};
-
-static const struct sysfs_ops cache_index_ops = {
- .show = cache_index_show,
-};
-
-static const struct kobj_type cache_index_type = {
- .release = cache_index_release,
- .sysfs_ops = &cache_index_ops,
- .default_groups = cache_index_default_groups,
-};
-
-static void cacheinfo_create_index_opt_attrs(struct cache_index_dir *dir)
+static unsigned int powerpc_cache_leaves(struct device_node *np)
{
- const char *cache_type;
- struct cache *cache;
- char *buf;
- int i;
-
- buf = kmalloc(PAGE_SIZE, GFP_KERNEL);
- if (!buf)
- return;
-
- cache = dir->cache;
- cache_type = cache_type_string(cache);
-
- /* We don't want to create an attribute that can't provide a
- * meaningful value. Check the return value of each optional
- * attribute's ->show method before registering the
- * attribute.
- */
- for (i = 0; i < ARRAY_SIZE(cache_index_opt_attrs); i++) {
- struct kobj_attribute *attr;
- ssize_t rc;
-
- attr = cache_index_opt_attrs[i];
-
- rc = attr->show(&dir->kobj, attr, buf);
- if (rc <= 0) {
- pr_debug("not creating %s attribute for "
- "%pOFP(%s) (rc = %zd)\n",
- attr->attr.name, cache->ofnode,
- cache_type, rc);
- continue;
- }
- if (sysfs_create_file(&dir->kobj, &attr->attr))
- pr_debug("could not create %s attribute for %pOFP(%s)\n",
- attr->attr.name, cache->ofnode, cache_type);
- }
-
- kfree(buf);
+ return of_property_read_bool(np, "cache-unified") ? 1 : 2;
}
-static void cacheinfo_create_index_dir(struct cache *cache, int index,
- struct cache_dir *cache_dir)
+int init_cache_level(unsigned int cpu)
{
- struct cache_index_dir *index_dir;
- int rc;
+ struct cpu_cacheinfo *ci = get_cpu_cacheinfo(cpu);
+ struct device_node *next;
+ struct device_node *np;
+ unsigned int leaves = 0;
+ unsigned int levels = 0;
- index_dir = kzalloc_obj(*index_dir);
- if (!index_dir)
- return;
+ np = of_cpu_device_node_get(cpu);
+ if (!np)
+ return -ENOENT;
- index_dir->cache = cache;
-
- rc = kobject_init_and_add(&index_dir->kobj, &cache_index_type,
- cache_dir->kobj, "index%d", index);
- if (rc) {
- kobject_put(&index_dir->kobj);
- return;
- }
+ for (;;) {
+ leaves += powerpc_cache_leaves(np);
+ levels++;
- index_dir->next = cache_dir->index;
- cache_dir->index = index_dir;
+ next = of_find_next_cache_node(np);
+ of_node_put(np);
+ np = next;
- cacheinfo_create_index_opt_attrs(index_dir);
-}
-
-static void cacheinfo_sysfs_populate(unsigned int cpu_id,
- struct cache *cache_list)
-{
- struct cache_dir *cache_dir;
- struct cache *cache;
- int index = 0;
-
- cache_dir = cacheinfo_create_cache_dir(cpu_id);
- if (!cache_dir)
- return;
-
- cache = cache_list;
- while (cache) {
- cacheinfo_create_index_dir(cache, index, cache_dir);
- index++;
- cache = cache->next_local;
+ if (!np)
+ break;
}
-}
-void cacheinfo_cpu_online(unsigned int cpu_id)
-{
- struct cache *cache;
+ if (!leaves)
+ return -ENOENT;
- cache = cache_chain_instantiate(cpu_id);
- if (!cache)
- return;
+ ci->num_levels = levels;
+ ci->num_leaves = leaves;
- cacheinfo_sysfs_populate(cpu_id, cache);
-}
-
-/* functions needed to remove cache entry for cpu offline or suspend/resume */
-
-#if (defined(CONFIG_PPC_PSERIES) && defined(CONFIG_SUSPEND)) || \
- defined(CONFIG_HOTPLUG_CPU)
-
-static struct cache *cache_lookup_by_cpu(unsigned int cpu_id)
-{
- struct device_node *cpu_node;
- struct cache *cache;
- int group_id;
-
- cpu_node = of_get_cpu_node(cpu_id, NULL);
- WARN_ONCE(!cpu_node, "no OF node found for CPU %i\n", cpu_id);
- if (!cpu_node)
- return NULL;
-
- group_id = get_group_id(cpu_id, 1);
- cache = cache_lookup_by_node_group(cpu_node, group_id);
- of_node_put(cpu_node);
-
- return cache;
-}
-
-static void remove_index_dirs(struct cache_dir *cache_dir)
-{
- struct cache_index_dir *index;
-
- index = cache_dir->index;
-
- while (index) {
- struct cache_index_dir *next;
-
- next = index->next;
- kobject_put(&index->kobj);
- index = next;
- }
+ return 0;
}
-static void remove_cache_dir(struct cache_dir *cache_dir)
+int populate_cache_leaves(unsigned int cpu)
{
- remove_index_dirs(cache_dir);
+ struct cpu_cacheinfo *ci = get_cpu_cacheinfo(cpu);
+ struct cacheinfo *leaf = ci->info_list;
+ struct device_node *next;
+ struct device_node *np;
+ unsigned int index = 0;
+ unsigned int level = 1;
- /* Remove cache dir from sysfs */
- kobject_del(cache_dir->kobj);
+ np = of_cpu_device_node_get(cpu);
+ if (!np)
+ return -ENOENT;
- kobject_put(cache_dir->kobj);
+ for (;;) {
+ bool unified = of_property_read_bool(np, "cache-unified");
- kfree(cache_dir);
-}
+ if (unified) {
+ if (index >= ci->num_leaves)
+ goto err;
-static void cache_cpu_clear(struct cache *cache, int cpu)
-{
- while (cache) {
- struct cache *next = cache->next_local;
+ powerpc_cache_leaf_init(&leaf[index++], level,
+ CACHE_TYPE_UNIFIED);
+ } else {
+ if (index + 1 >= ci->num_leaves)
+ goto err;
- WARN_ONCE(!cpumask_test_cpu(cpu, &cache->shared_cpu_map),
- "CPU %i not accounted in %pOFP(%s)\n",
- cpu, cache->ofnode,
- cache_type_string(cache));
+ powerpc_cache_leaf_init(&leaf[index++], level,
+ CACHE_TYPE_DATA);
+ powerpc_cache_leaf_init(&leaf[index++], level,
+ CACHE_TYPE_INST);
+ }
- cpumask_clear_cpu(cpu, &cache->shared_cpu_map);
+ next = of_find_next_cache_node(np);
+ of_node_put(np);
+ np = next;
- /* Release the cache object if all the cpus using it
- * are offline */
- if (cpumask_empty(&cache->shared_cpu_map))
- release_cache(cache);
+ if (!np)
+ break;
- cache = next;
+ level++;
}
-}
-
-void cacheinfo_cpu_offline(unsigned int cpu_id)
-{
- struct cache_dir *cache_dir;
- struct cache *cache;
-
- /* Prevent userspace from seeing inconsistent state - remove
- * the sysfs hierarchy first */
- cache_dir = per_cpu(cache_dir_pcpu, cpu_id);
-
- /* careful, sysfs population may have failed */
- if (cache_dir)
- remove_cache_dir(cache_dir);
-
- per_cpu(cache_dir_pcpu, cpu_id) = NULL;
-
- /* clear the CPU's bit in its cache chain, possibly freeing
- * cache objects */
- cache = cache_lookup_by_cpu(cpu_id);
- if (cache)
- cache_cpu_clear(cache, cpu_id);
-}
-void cacheinfo_teardown(void)
-{
- unsigned int cpu;
-
- lockdep_assert_cpus_held();
+ if (index != ci->num_leaves)
+ return -EINVAL;
- for_each_online_cpu(cpu)
- cacheinfo_cpu_offline(cpu);
-}
-
-void cacheinfo_rebuild(void)
-{
- unsigned int cpu;
-
- lockdep_assert_cpus_held();
+ return 0;
- for_each_online_cpu(cpu)
- cacheinfo_cpu_online(cpu);
+err:
+ of_node_put(np);
+ return -EINVAL;
}
-
-#endif /* (CONFIG_PPC_PSERIES && CONFIG_SUSPEND) || CONFIG_HOTPLUG_CPU */
diff --git a/arch/powerpc/kernel/cacheinfo.h b/arch/powerpc/kernel/cacheinfo.h
deleted file mode 100644
index 52bd3fc6642d..000000000000
--- a/arch/powerpc/kernel/cacheinfo.h
+++ /dev/null
@@ -1,13 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-#ifndef _PPC_CACHEINFO_H
-#define _PPC_CACHEINFO_H
-
-/* These are just hooks for sysfs.c to use. */
-extern void cacheinfo_cpu_online(unsigned int cpu_id);
-extern void cacheinfo_cpu_offline(unsigned int cpu_id);
-
-/* Allow migration/suspend to tear down and rebuild the hierarchy. */
-extern void cacheinfo_teardown(void);
-extern void cacheinfo_rebuild(void);
-
-#endif /* _PPC_CACHEINFO_H */
diff --git a/arch/powerpc/kernel/sysfs.c b/arch/powerpc/kernel/sysfs.c
index 2100c5e2c821..519e0a75f30c 100644
--- a/arch/powerpc/kernel/sysfs.c
+++ b/arch/powerpc/kernel/sysfs.c
@@ -24,7 +24,6 @@
#include <asm/idle.h>
#include <asm/svm.h>
-#include "cacheinfo.h"
#include "setup.h"
#ifdef CONFIG_PPC64
@@ -928,7 +927,6 @@ static int register_cpu_online(unsigned int cpu)
device_create_file(s, &dev_attr_altivec_idle_wait_time);
}
#endif
- cacheinfo_cpu_online(cpu);
return 0;
}
@@ -1024,7 +1022,6 @@ static int unregister_cpu_online(unsigned int cpu)
device_remove_file(s, &dev_attr_altivec_idle_wait_time);
}
#endif
- cacheinfo_cpu_offline(cpu);
of_node_put(s->of_node);
s->of_node = NULL;
return 0;
diff --git a/arch/powerpc/platforms/pseries/mobility.c b/arch/powerpc/platforms/pseries/mobility.c
index b5c2abd12432..8b7f0ce107a6 100644
--- a/arch/powerpc/platforms/pseries/mobility.c
+++ b/arch/powerpc/platforms/pseries/mobility.c
@@ -9,6 +9,7 @@
#define pr_fmt(fmt) "mobility: " fmt
+#include <linux/cacheinfo.h>
#include <linux/cpu.h>
#include <linux/kernel.h>
#include <linux/kobject.h>
@@ -29,7 +30,6 @@
#include "pseries.h"
#include "vas.h" /* vas_migration_handler() */
#include "papr-hvpipe.h" /* hvpipe_migration_handler() */
-#include "../../kernel/cacheinfo.h"
static struct kobject *mobility_kobj;
--
2.54.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 5/5] sched/topology: Remove stale comment around llc_bytes on powerpc
2026-11-18 16:36 [PATCH 0/5] cacheinfo, powerpc: Move powerpc to generic cacheinfo Shrikanth Hegde
` (3 preceding siblings ...)
2026-11-18 16:36 ` [PATCH 4/5] powerpc/cacheinfo: Move to generic cacheinfo Shrikanth Hegde
@ 2026-11-18 16:36 ` Shrikanth Hegde
4 siblings, 0 replies; 6+ messages in thread
From: Shrikanth Hegde @ 2026-11-18 16:36 UTC (permalink / raw)
To: linuxppc-dev, linux-kernel, driver-core
Cc: sshegde, christophe.leroy, maddy, gregkh, rafael, dakr, johan,
sudeep.holla, ritesh.list, tim.c.chen, yu.c.chen
PowerPC now populates generic cacheinfo, allowing
get_effective_llc_bytes() to obtain the LLC size.
Remove the comment stating that get_effective_llc_bytes() always
returns zero on PowerPC.
Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
---
kernel/sched/topology.c | 7 -------
1 file changed, 7 deletions(-)
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
index 99878da03bce..6d4e4a32bc4e 100644
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -1001,13 +1001,6 @@ void sched_cache_active_set(void)
* Update the bottom sched_domain's llc_bytes for @cpus sharing a physical
* LLC. Called from cacheinfo_cpu_online() or
* cacheinfo_cpu_pre_down() with cpu hotplug lock held.
- *
- * Note: get_effective_llc_bytes() returns 0 on PowerPC.
- * thus cache aware scheduling is disabled on PowerPC for
- * now. PowerPC does not use the generic cacheinfo framework --
- * it has its own cacheinfo with a separate struct cache hierarchy
- * and does not populates the per-CPU struct cpu_cacheinfo array
- * that get_cpu_cacheinfo_llc() reads.
*/
void sched_update_llc_bytes(const struct cpumask *cpus)
{
--
2.54.0
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2026-11-18 16:37 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-18 16:36 [PATCH 0/5] cacheinfo, powerpc: Move powerpc to generic cacheinfo Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 1/5] cacheinfo: Allow architectures to provide custom cache IDs Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 2/5] cacheinfo: Support unified caches described by d-cache properties Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 3/5] cacheinfo: Add teardown and rebuild helpers Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 4/5] powerpc/cacheinfo: Move to generic cacheinfo Shrikanth Hegde
2026-11-18 16:36 ` [PATCH 5/5] sched/topology: Remove stale comment around llc_bytes on powerpc Shrikanth Hegde
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®