Hi Andrew,
Can you please replace the patches on mm-unstable with this line,
it has bulk-allocator bug fixes and some design changes at the
request of Ying Huang. Full v2 notes are just before the test info.
= Cover Letter
Weighted interleave is a new interleave policy intended to make
use of heterogeneous memory environments appearing with CXL.
The existing interleave mechanism does an even round-robin
distribution of memory across all nodes in a nodemask, while
weighted interleave distributes memory across nodes according
to a provided weight. (Weight = # of page allocations per round)
Weighted interleave is intended to reduce average latency when
bandwidth is pressured - therefore increasing total throughput.
In other words: It allows greater use of the total available
bandwidth in a heterogeneous hardware environment (different
hardware provides different bandwidth capacity).
As bandwidth is pressured, latency increases - first linearly
and then exponentially. By keeping bandwidth usage distributed
according to available bandwidth, we therefore can reduce the
average latency of a cacheline fetch.
A good explanation of the bandwidth vs latency response curve:
https://mahmoudhatem.wordpress.com/2017/11/07/memory-bandwidth-vs-latency-response-curve/
From the article:
```
Constant region:
The latency response is fairly constant for the first 40%
of the sustained bandwidth.
Linear region:
In between 40% to 80% of the sustained bandwidth, the
latency response increases almost linearly with the bandwidth
demand of the system due to contention overhead by numerous
memory requests.
Exponential region:
Between 80% to 100% of the sustained bandwidth, the memory
latency is dominated by the contention latency which can be
as much as twice the idle latency or more.
Maximum sustained bandwidth :
Is 65% to 75% of the theoretical maximum bandwidth.
```
As a general rule of thumb:
* If bandwidth usage is low, latency does not increase. It is
optimal to place data in the nearest (lowest latency) device.
* If bandwidth usage is high, latency increases. It is optimal
to place data such that bandwidth use is optimized per-device.
This is the top line goal: Provide a user a mechanism to target using
the "maximum sustained bandwidth" of each hardware component in a
heterogenous memory system.
For example, the stream benchmark demonstrates that 1:1 (default)
interleave is actively harmful, while weighted interleave can be
beneficial. Default interleave distributes data such that too much
pressure is placed on devices with lower available bandwidth.
Stream Benchmark (High level results, 1 Socket + 1 CXL Device)
Default interleave : -78% (slower than DRAM)
Global weighting : -6% to +4% (workload dependant)
Targeted weights : +2.5% to +4% (consistently better than DRAM)
Global means the task-policy was set (set_mempolicy), while targeted
means VMA policies were set (mbind2). We see weighted interleave
is not always beneficial when applied globally, but is always
beneficial when applied to bandwidth-driving memory regions.
We implement sysfs entries for "system global" weights which can be
set by a daemon or administrator.
There are 3 patches in this set:
1) Implement system-global interleave weights as sysfs extension
in mm/mempolicy.c. These weights are RCU protected, and the
default/minimum weight is 1 for all nodes.
In future work, we intend to expose an interface for HMAT/CDAT
information to be used during boot to set reasonable system
default values based on the memory configuration of the system
discovered at boot or during device hotplug.
2) A mild refactor of some interleave-logic for re-use in the
new weighted interleave logic.
3) MPOL_WEIGHTED_INTERLEAVE extension for set_mempolicy/mbind
Included below are some performance and LTP test information,
and a sample numactl branch which can be used for testing.
= Performance summary =
(tests may have different configurations, see extended info below)
1) MLC (W2) : +38% over DRAM. +264% over default interleave.
MLC (W5) : +40% over DRAM. +226% over default interleave.
2) Stream : -6% to +4% over DRAM, +430% over default interleave.
3) XSBench : +19% over DRAM. +47% over default interleave.
= LTP Testing Summary =
existing mempolicy & mbind tests: pass
mempolicy & mbind + weighted interleave (global weights): pass
= version history
v2:
- MAJOR: Torture tested bulk allocator, fixed edge conditions
tracking the next me->il_node. Added documentation.
Prior version was stable, but the resulting me->il_node
could be wrong under certain circumstances.
- naming: iw_table_mtx -> iw_table_lock
- RCU: use synchronize+kfree and simplify the weight structure
- default: remove default table, since it's static for now
- sysfs setup: simplify setup, if table==NULL presume 1's
- node_store: only allocate (sizeof(u8) * nr_node_ids)
- allocators: update to deal with NULL table pointer
- read_once: __builtin_memcpy -> memcpy
- formatting
v1:
- RCU: This version protects the weight array with RCU.
- ktest fix: proper include (types.h) in uapi header
- doc: make mpol_params in docs reflect definition
- doc: mempolicy.c comments in MPOL_WEIGHTED_INTERLEAVE patch
- Dropped task-local weights and syscalls from the proposal
until affirmative use cases for task-local weights appear.
Link: https://lore.kernel.org/linux-mm/[email protected]/
=====================================================================
Performance tests - MLC
From - Ravi Jonnalagadda <[email protected]>
Hardware: Single-socket, multiple CXL memory expanders.
Workload: W2
Data Signature: 2:1 read:write
DRAM only bandwidth (GBps): 298.8
DRAM + CXL (default interleave) (GBps): 113.04
DRAM + CXL (weighted interleave)(GBps): 412.5
Gain over DRAM only: 1.38x
Gain over default interleave: 2.64x
Workload: W5
Data Signature: 1:1 read:write
DRAM only bandwidth (GBps): 273.2
DRAM + CXL (default interleave) (GBps): 117.23
DRAM + CXL (weighted interleave)(GBps): 382.7
Gain over DRAM only: 1.4x
Gain over default interleave: 2.26x
=====================================================================
Performance test - Stream
From - Gregory Price <[email protected]>
Hardware: Single socket, single CXL expander
numactl extension: https://github.com/gmprice/numactl/tree/weighted_interleave_master
Summary: 64 threads, ~18GB workload, 3GB per array, executed 100 times
Default interleave : -78% (slower than DRAM)
Global weighting : -6% to +4% (workload dependant)
mbind2 weights : +2.5% to +4% (consistently better than DRAM)
dram only:
numactl --cpunodebind=1 --membind=1 ./stream_c.exe --ntimes 100 --array-size 400M --malloc
Function Direction BestRateMBs AvgTime MinTime MaxTime
Copy: 0->0 200923.2 0.032662 0.031853 0.033301
Scale: 0->0 202123.0 0.032526 0.031664 0.032970
Add: 0->0 208873.2 0.047322 0.045961 0.047884
Triad: 0->0 208523.8 0.047262 0.046038 0.048414
CXL-only:
numactl --cpunodebind=1 -w --membind=2 ./stream_c.exe --ntimes 100 --array-size 400M --malloc
Copy: 0->0 22209.7 0.288661 0.288162 0.289342
Scale: 0->0 22288.2 0.287549 0.287147 0.288291
Add: 0->0 24419.1 0.393372 0.393135 0.393735
Triad: 0->0 24484.6 0.392337 0.392083 0.394331
Based on the above, the optimal weights are ~9:1
echo 9 > /sys/kernel/mm/mempolicy/weighted_interleave/node1
echo 1 > /sys/kernel/mm/mempolicy/weighted_interleave/node2
default interleave:
numactl --cpunodebind=1 --interleave=1,2 ./stream_c.exe --ntimes 100 --array-size 400M --malloc
Copy: 0->0 44666.2 0.143671 0.143285 0.144174
Scale: 0->0 44781.6 0.143256 0.142916 0.143713
Add: 0->0 48600.7 0.197719 0.197528 0.197858
Triad: 0->0 48727.5 0.197204 0.197014 0.197439
global weighted interleave:
numactl --cpunodebind=1 -w --interleave=1,2 ./stream_c.exe --ntimes 100 --array-size 400M --malloc
Copy: 0->0 190085.9 0.034289 0.033669 0.034645
Scale: 0->0 207677.4 0.031909 0.030817 0.033061
Add: 0->0 202036.8 0.048737 0.047516 0.053409
Triad: 0->0 217671.5 0.045819 0.044103 0.046755
targted regions w/ global weights (modified stream to mbind2 malloc'd regions))
numactl --cpunodebind=1 --membind=1 ./stream_c.exe -b --ntimes 100 --array-size 400M --malloc
Copy: 0->0 205827.0 0.031445 0.031094 0.031984
Scale: 0->0 208171.8 0.031320 0.030744 0.032505
Add: 0->0 217352.0 0.045087 0.044168 0.046515
Triad: 0->0 216884.8 0.045062 0.044263 0.046982
=====================================================================
Performance tests - XSBench
From - Hyeongtak Ji <[email protected]>
Hardware: Single socket, Single CXL memory Expander
NUMA node 0: 56 logical cores, 128 GB memory
NUMA node 2: 96 GB CXL memory
Threads: 56
Lookups: 170,000,000
Summary: +19% over DRAM. +47% over default interleave.
Performance tests - XSBench
1. dram only
$ numactl -m 0 ./XSBench -s XL –p 5000000
Runtime: 36.235 seconds
Lookups/s: 4,691,618
2. default interleave
$ numactl –i 0,2 ./XSBench –s XL –p 5000000
Runtime: 55.243 seconds
Lookups/s: 3,077,293
3. weighted interleave
numactl –w –i 0,2 ./XSBench –s XL –p 5000000
Runtime: 29.262 seconds
Lookups/s: 5,809,513
=====================================================================
LTP Tests: https://github.com/gmprice/ltp/tree/mempolicy2
= Existing tests
set_mempolicy, get_mempolicy, mbind
MPOL_WEIGHTED_INTERLEAVE added manually to test basic functionality
but did not adjust tests for weighting. Basically the weights were
set to 1, which is the default, and it should behavior like standard
MPOL_INTERLEAVE if logic is correct.
== set_mempolicy01 : passed 18, failed 0
== set_mempolicy02 : passed 10, failed 0
== set_mempolicy03 : passed 64, failed 0
== set_mempolicy04 : passed 32, failed 0
== set_mempolicy05 - n/a on non-x86
== set_mempolicy06 : passed 10, failed 0
this is set_mempolicy02 + MPOL_WEIGHTED_INTERLEAVE
== set_mempolicy07 : passed 32, failed 0
set_mempolicy04 + MPOL_WEIGHTED_INTERLEAVE
== get_mempolicy01 : passed 12, failed 0
change: added MPOL_WEIGHTED_INTERLEAVE
== get_mempolicy02 : passed 2, failed 0
== mbind01 : passed 15, failed 0
added MPOL_WEIGHTED_INTERLEAVE
== mbind02 : passed 4, failed 0
added MPOL_WEIGHTED_INTERLEAVE
== mbind03 : passed 16, failed 0
added MPOL_WEIGHTED_INTERLEAVE
== mbind04 : passed 48, failed 0
added MPOL_WEIGHTED_INTERLEAVE
=====================================================================
numactl (set_mempolicy) w/ global weighting test
numactl fork: https://github.com/gmprice/numactl/tree/weighted_interleave_master
command: numactl -w --interleave=0,1 ./eatmem
result (weights 1:1):
0176a000 weighted interleave:0-1 heap anon=65793 dirty=65793 active=0 N0=32897 N1=32896 kernelpagesize_kB=4
7fceeb9ff000 weighted interleave:0-1 anon=65537 dirty=65537 active=0 N0=32768 N1=32769 kernelpagesize_kB=4
50% distribution is correct
result (weights 5:1):
01b14000 weighted interleave:0-1 heap anon=65793 dirty=65793 active=0 N0=54828 N1=10965 kernelpagesize_kB=4
7f47a1dff000 weighted interleave:0-1 anon=65537 dirty=65537 active=0 N0=54614 N1=10923 kernelpagesize_kB=4
16.666% distribution is correct
result (weights 1:5):
01f07000 weighted interleave:0-1 heap anon=65793 dirty=65793 active=0 N0=10966 N1=54827 kernelpagesize_kB=4
7f17b1dff000 weighted interleave:0-1 anon=65537 dirty=65537 active=0 N0=10923 N1=54614 kernelpagesize_kB=4
16.666% distribution is correct
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
int main (void)
{
char* mem = malloc(1024*1024*256);
memset(mem, 1, 1024*1024*256);
for (int i = 0; i < ((1024*1024*256)/4096); i++)
{
mem = malloc(4096);
mem[0] = 1;
}
printf("done\n");
getchar();
return 0;
}
=====================================================================
Suggested-by: Gregory Price <[email protected]>
Suggested-by: Johannes Weiner <[email protected]>
Suggested-by: Hasan Al Maruf <[email protected]>
Suggested-by: Hao Wang <[email protected]>
Suggested-by: Ying Huang <[email protected]>
Suggested-by: Dan Williams <[email protected]>
Suggested-by: Michal Hocko <[email protected]>
Suggested-by: Zhongkun He <[email protected]>
Suggested-by: Frank van der Linden <[email protected]>
Suggested-by: John Groves <[email protected]>
Suggested-by: Vinicius Tavares Petrucci <[email protected]>
Suggested-by: Srinivasulu Thanneeru <[email protected]>
Suggested-by: Ravi Jonnalagadda <[email protected]>
Suggested-by: Jonathan Cameron <[email protected]>
Suggested-by: Hyeongtak Ji <[email protected]>
Suggested-by: Andi Kleen <[email protected]>
Signed-off-by: Gregory Price <[email protected]>
Gregory Price (2):
mm/mempolicy: refactor a read-once mechanism into a function for
re-use
mm/mempolicy: introduce MPOL_WEIGHTED_INTERLEAVE for weighted
interleaving
Rakie Kim (1):
mm/mempolicy: implement the sysfs-based weighted_interleave interface
.../ABI/testing/sysfs-kernel-mm-mempolicy | 4 +
...fs-kernel-mm-mempolicy-weighted-interleave | 26 +
.../admin-guide/mm/numa_memory_policy.rst | 9 +
include/linux/mempolicy.h | 5 +
include/uapi/linux/mempolicy.h | 1 +
mm/mempolicy.c | 491 +++++++++++++++++-
6 files changed, 523 insertions(+), 13 deletions(-)
create mode 100644 Documentation/ABI/testing/sysfs-kernel-mm-mempolicy
create mode 100644 Documentation/ABI/testing/sysfs-kernel-mm-mempolicy-weighted-interleave
--
2.39.1
When a system has multiple NUMA nodes and it becomes bandwidth hungry,
using the current MPOL_INTERLEAVE could be an wise option.
However, if those NUMA nodes consist of different types of memory such
as socket-attached DRAM and CXL/PCIe attached DRAM, the round-robin
based interleave policy does not optimally distribute data to make use
of their different bandwidth characteristics.
Instead, interleave is more effective when the allocation policy follows
each NUMA nodes' bandwidth weight rather than a simple 1:1 distribution.
This patch introduces a new memory policy, MPOL_WEIGHTED_INTERLEAVE,
enabling weighted interleave between NUMA nodes. Weighted interleave
allows for proportional distribution of memory across multiple numa
nodes, preferably apportioned to match the bandwidth of each node.
For example, if a system has 1 CPU node (0), and 2 memory nodes (0,1),
with bandwidth of (100GB/s, 50GB/s) respectively, the appropriate
weight distribution is (2:1).
Weights for each node can be assigned via the new sysfs extension:
/sys/kernel/mm/mempolicy/weighted_interleave/
For now, the default value of all nodes will be `1`, which matches
the behavior of standard 1:1 round-robin interleave. An extension
will be added in the future to allow default values to be registered
at kernel and device bringup time.
The policy allocates a number of pages equal to the set weights. For
example, if the weights are (2,1), then 2 pages will be allocated on
node0 for every 1 page allocated on node1.
The new flag MPOL_WEIGHTED_INTERLEAVE can be used in set_mempolicy(2)
and mbind(2).
There are 3 integration points:
weighted_interleave_nodes:
Counts the number of allocations as they occur, and applies the
weight for the current node. When the weight reaches 0, switch
to the next node.
weighted_interleave_nid:
Gets the total weight of the nodemask as well as each individual
node weight, then calculates the node based on the given index.
bulk_array_weighted_interleave:
Gets the total weight of the nodemask as well as each individual
node weight, then calculates the number of "interleave rounds" as
well as any delta ("partial round"). Calculates the number of
pages for each node and allocates them.
If a node was scheduled for interleave via interleave_nodes, the
current weight (pol->cur_weight) will be allocated first, before
the remaining bulk calculation is done.
One piece of complexity is the interaction between a recent refactor
which split the logic to acquire the "ilx" (interleave index) of an
allocation and the actually application of the interleave. The
calculation of the `interleave index` is done by `get_vma_policy()`,
while the actual selection of the node will be later appliex by the
relevant weighted_interleave function.
Suggested-by: Hasan Al Maruf <[email protected]>
Signed-off-by: Gregory Price <[email protected]>
Co-developed-by: Rakie Kim <[email protected]>
Signed-off-by: Rakie Kim <[email protected]>
Co-developed-by: Honggyu Kim <[email protected]>
Signed-off-by: Honggyu Kim <[email protected]>
Co-developed-by: Hyeongtak Ji <[email protected]>
Signed-off-by: Hyeongtak Ji <[email protected]>
Co-developed-by: Srinivasulu Thanneeru <[email protected]>
Signed-off-by: Srinivasulu Thanneeru <[email protected]>
Co-developed-by: Ravi Jonnalagadda <[email protected]>
Signed-off-by: Ravi Jonnalagadda <[email protected]>
---
.../admin-guide/mm/numa_memory_policy.rst | 9 +
include/linux/mempolicy.h | 5 +
include/uapi/linux/mempolicy.h | 1 +
mm/mempolicy.c | 234 +++++++++++++++++-
4 files changed, 246 insertions(+), 3 deletions(-)
diff --git a/Documentation/admin-guide/mm/numa_memory_policy.rst b/Documentation/admin-guide/mm/numa_memory_policy.rst
index eca38fa81e0f..a70f20ce1ffb 100644
--- a/Documentation/admin-guide/mm/numa_memory_policy.rst
+++ b/Documentation/admin-guide/mm/numa_memory_policy.rst
@@ -250,6 +250,15 @@ MPOL_PREFERRED_MANY
can fall back to all existing numa nodes. This is effectively
MPOL_PREFERRED allowed for a mask rather than a single node.
+MPOL_WEIGHTED_INTERLEAVE
+ This mode operates the same as MPOL_INTERLEAVE, except that
+ interleaving behavior is executed based on weights set in
+ /sys/kernel/mm/mempolicy/weighted_interleave/
+
+ Weighted interleave allocates pages on nodes according to a
+ weight. For example if nodes [0,1] are weighted [5,2], 5 pages
+ will be allocated on node0 for every 2 pages allocated on node1.
+
NUMA memory policy supports the following optional mode flags:
MPOL_F_STATIC_NODES
diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
index 931b118336f4..c1a083eb0dd5 100644
--- a/include/linux/mempolicy.h
+++ b/include/linux/mempolicy.h
@@ -54,6 +54,11 @@ struct mempolicy {
nodemask_t cpuset_mems_allowed; /* relative to these nodes */
nodemask_t user_nodemask; /* nodemask passed by user */
} w;
+
+ /* Weighted interleave settings */
+ struct {
+ u8 cur_weight;
+ } wil;
};
/*
diff --git a/include/uapi/linux/mempolicy.h b/include/uapi/linux/mempolicy.h
index a8963f7ef4c2..1f9bb10d1a47 100644
--- a/include/uapi/linux/mempolicy.h
+++ b/include/uapi/linux/mempolicy.h
@@ -23,6 +23,7 @@ enum {
MPOL_INTERLEAVE,
MPOL_LOCAL,
MPOL_PREFERRED_MANY,
+ MPOL_WEIGHTED_INTERLEAVE,
MPOL_MAX, /* always last member of enum */
};
diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index 427bddf115df..aa3b2389d3e0 100644
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -19,6 +19,13 @@
* for anonymous memory. For process policy an process counter
* is used.
*
+ * weighted interleave
+ * Allocate memory interleaved over a set of nodes based on
+ * a set of weights (per-node), with normal fallback if it
+ * fails. Otherwise operates the same as interleave.
+ * Example: nodeset(0,1) & weights (2,1) - 2 pages allocated
+ * on node 0 for every 1 page allocated on node 1.
+ *
* bind Only allocate memory on a specific set of nodes,
* no fallback.
* FIXME: memory is allocated starting with the first node
@@ -313,6 +320,7 @@ static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
policy->mode = mode;
policy->flags = flags;
policy->home_node = NUMA_NO_NODE;
+ policy->wil.cur_weight = 0;
return policy;
}
@@ -425,6 +433,10 @@ static const struct mempolicy_operations mpol_ops[MPOL_MAX] = {
.create = mpol_new_nodemask,
.rebind = mpol_rebind_preferred,
},
+ [MPOL_WEIGHTED_INTERLEAVE] = {
+ .create = mpol_new_nodemask,
+ .rebind = mpol_rebind_nodemask,
+ },
};
static bool migrate_folio_add(struct folio *folio, struct list_head *foliolist,
@@ -846,7 +858,8 @@ static long do_set_mempolicy(unsigned short mode, unsigned short flags,
old = current->mempolicy;
current->mempolicy = new;
- if (new && new->mode == MPOL_INTERLEAVE)
+ if (new && (new->mode == MPOL_INTERLEAVE ||
+ new->mode == MPOL_WEIGHTED_INTERLEAVE))
current->il_prev = MAX_NUMNODES-1;
task_unlock(current);
mpol_put(old);
@@ -872,6 +885,7 @@ static void get_policy_nodemask(struct mempolicy *pol, nodemask_t *nodes)
case MPOL_INTERLEAVE:
case MPOL_PREFERRED:
case MPOL_PREFERRED_MANY:
+ case MPOL_WEIGHTED_INTERLEAVE:
*nodes = pol->nodes;
break;
case MPOL_LOCAL:
@@ -956,6 +970,13 @@ static long do_get_mempolicy(int *policy, nodemask_t *nmask,
} else if (pol == current->mempolicy &&
pol->mode == MPOL_INTERLEAVE) {
*policy = next_node_in(current->il_prev, pol->nodes);
+ } else if (pol == current->mempolicy &&
+ (pol->mode == MPOL_WEIGHTED_INTERLEAVE)) {
+ if (pol->wil.cur_weight)
+ *policy = current->il_prev;
+ else
+ *policy = next_node_in(current->il_prev,
+ pol->nodes);
} else {
err = -EINVAL;
goto out;
@@ -1785,7 +1806,8 @@ struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
pol = __get_vma_policy(vma, addr, ilx);
if (!pol)
pol = get_task_policy(current);
- if (pol->mode == MPOL_INTERLEAVE) {
+ if (pol->mode == MPOL_INTERLEAVE ||
+ pol->mode == MPOL_WEIGHTED_INTERLEAVE) {
*ilx += vma->vm_pgoff >> order;
*ilx += (addr - vma->vm_start) >> (PAGE_SHIFT + order);
}
@@ -1835,6 +1857,28 @@ bool apply_policy_zone(struct mempolicy *policy, enum zone_type zone)
return zone >= dynamic_policy_zone;
}
+static unsigned int weighted_interleave_nodes(struct mempolicy *policy)
+{
+ unsigned int next;
+ struct task_struct *me = current;
+ u8 __rcu *table;
+
+ next = next_node_in(me->il_prev, policy->nodes);
+ if (next == MAX_NUMNODES)
+ return next;
+
+ rcu_read_lock();
+ table = rcu_dereference(iw_table);
+ if (!policy->wil.cur_weight)
+ policy->wil.cur_weight = table ? table[next] : 1;
+ rcu_read_unlock();
+
+ policy->wil.cur_weight--;
+ if (!policy->wil.cur_weight)
+ me->il_prev = next;
+ return next;
+}
+
/* Do dynamic interleaving for a process */
static unsigned int interleave_nodes(struct mempolicy *policy)
{
@@ -1869,6 +1913,9 @@ unsigned int mempolicy_slab_node(void)
case MPOL_INTERLEAVE:
return interleave_nodes(policy);
+ case MPOL_WEIGHTED_INTERLEAVE:
+ return weighted_interleave_nodes(policy);
+
case MPOL_BIND:
case MPOL_PREFERRED_MANY:
{
@@ -1907,6 +1954,39 @@ static unsigned int read_once_policy_nodemask(struct mempolicy *pol,
return nodes_weight(*mask);
}
+static unsigned int weighted_interleave_nid(struct mempolicy *pol, pgoff_t ilx)
+{
+ nodemask_t nodemask;
+ unsigned int target, nr_nodes;
+ u8 __rcu *table;
+ unsigned int weight_total = 0;
+ u8 weight;
+ int nid;
+
+ nr_nodes = read_once_policy_nodemask(pol, &nodemask);
+ if (!nr_nodes)
+ return numa_node_id();
+
+ rcu_read_lock();
+ table = rcu_dereference(iw_table);
+ /* calculate the total weight */
+ for_each_node_mask(nid, nodemask)
+ weight_total += table ? table[nid] : 1;
+
+ /* Calculate the node offset based on totals */
+ target = ilx % weight_total;
+ nid = first_node(nodemask);
+ while (target) {
+ weight = table ? table[nid] : 1;
+ if (target < weight)
+ break;
+ target -= weight;
+ nid = next_node_in(nid, nodemask);
+ }
+ rcu_read_unlock();
+ return nid;
+}
+
/*
* Do static interleaving for interleave index @ilx. Returns the ilx'th
* node in pol->nodes (starting from ilx=0), wrapping around if ilx
@@ -1967,6 +2047,11 @@ static nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *pol,
*nid = (ilx == NO_INTERLEAVE_INDEX) ?
interleave_nodes(pol) : interleave_nid(pol, ilx);
break;
+ case MPOL_WEIGHTED_INTERLEAVE:
+ *nid = (ilx == NO_INTERLEAVE_INDEX) ?
+ weighted_interleave_nodes(pol) :
+ weighted_interleave_nid(pol, ilx);
+ break;
}
return nodemask;
@@ -2028,6 +2113,7 @@ bool init_nodemask_of_mempolicy(nodemask_t *mask)
case MPOL_PREFERRED_MANY:
case MPOL_BIND:
case MPOL_INTERLEAVE:
+ case MPOL_WEIGHTED_INTERLEAVE:
*mask = mempolicy->nodes;
break;
@@ -2127,7 +2213,8 @@ struct page *alloc_pages_mpol(gfp_t gfp, unsigned int order,
* If the policy is interleave or does not allow the current
* node in its nodemask, we allocate the standard way.
*/
- if (pol->mode != MPOL_INTERLEAVE &&
+ if ((pol->mode != MPOL_INTERLEAVE &&
+ pol->mode != MPOL_WEIGHTED_INTERLEAVE) &&
(!nodemask || node_isset(nid, *nodemask))) {
/*
* First, try to allocate THP only on local node, but
@@ -2263,6 +2350,135 @@ static unsigned long alloc_pages_bulk_array_interleave(gfp_t gfp,
return total_allocated;
}
+static unsigned long alloc_pages_bulk_array_weighted_interleave(gfp_t gfp,
+ struct mempolicy *pol, unsigned long nr_pages,
+ struct page **page_array)
+{
+ struct task_struct *me = current;
+ unsigned long total_allocated = 0;
+ unsigned long nr_allocated;
+ unsigned long rounds;
+ unsigned long node_pages, delta;
+ u8 weight, resume_weight;
+ u8 __rcu *table;
+ u8 *weights;
+ unsigned int weight_total = 0;
+ unsigned long rem_pages = nr_pages;
+ nodemask_t nodes;
+ int nnodes, node, weight_nodes, resume_node;
+ int prev_node = NUMA_NO_NODE;
+ bool delta_depleted = false;
+ int i;
+
+ if (!nr_pages)
+ return 0;
+
+ nnodes = read_once_policy_nodemask(pol, &nodes);
+ if (!nnodes)
+ return 0;
+
+ /* Continue allocating from most recent node and adjust the nr_pages */
+ if (pol->wil.cur_weight) {
+ node = next_node_in(me->il_prev, nodes);
+ node_pages = pol->wil.cur_weight;
+ if (node_pages > rem_pages)
+ node_pages = rem_pages;
+ nr_allocated = __alloc_pages_bulk(gfp, node, NULL, node_pages,
+ NULL, page_array);
+ page_array += nr_allocated;
+ total_allocated += nr_allocated;
+ /* if that's all the pages, no need to interleave */
+ if (rem_pages <= pol->wil.cur_weight) {
+ pol->wil.cur_weight -= rem_pages;
+ return total_allocated;
+ }
+ /* Otherwise we adjust nr_pages down, and continue from there */
+ rem_pages -= pol->wil.cur_weight;
+ pol->wil.cur_weight = 0;
+ prev_node = node;
+ }
+
+ /* fetch the weights for this operation and calculate total weight */
+ weights = kmalloc(nnodes, GFP_KERNEL);
+ if (!weights)
+ return total_allocated;
+
+ rcu_read_lock();
+ table = rcu_dereference(iw_table);
+ weight_nodes = 0;
+ while (weight_nodes < nnodes) {
+ node = next_node_in(prev_node, nodes);
+ weight = table ? table[node] : 1;
+ weights[weight_nodes++] = weight;
+ weight_total += weight;
+ }
+ rcu_read_unlock();
+
+ /*
+ * Now we can continue allocating from 0 instead of an offset
+ * We calculate the number of rounds and any partial rounds so
+ * that we minimize the number of calls to __alloc_pages_bulk
+ * This requires us to track which node we should resume from.
+ *
+ * if (rounds > 0) and (delta == 0), resume_node will always be
+ * the current me->il_prev
+ *
+ * if (delta > 0) and delta is depleted exactly on a node-weight
+ * boundary, resume node will be the node last allocated from when
+ * delta reached 0.
+ *
+ * if (delta > 0) and delta is not depleted on a node-weight boundary,
+ * resume node will be the node prior to the node last allocated from.
+ *
+ * (rounds == 0) and (delta == 0) is not possible (earlier exit)
+ */
+ rounds = rem_pages / weight_total;
+ delta = rem_pages % weight_total;
+ /* If no delta, we'll resume from current prev_node and first weight */
+ for (i = 0; i < nnodes; i++) {
+ node = next_node_in(prev_node, nodes);
+ weight = weights[i];
+ node_pages = weight * rounds;
+ /* If a delta exists, add this node's portion of the delta */
+ if (delta >= weight) {
+ node_pages += weight;
+ delta -= weight;
+ resume_node = node;
+ resume_weight = i < (nnodes - 1) ? weights[i+1] :
+ weights[0];
+ /* stop tracking iff (delta == weight) */
+ delta_depleted = !delta;
+ } else if (delta) { /* <= weight */
+ /* if delta depleted, resume from this node */
+ node_pages += delta;
+ delta = 0;
+ resume_node = prev_node;
+ resume_weight = weight - (node_pages % weight);
+ delta_depleted = true; /* stop tracking */
+ } else if (!delta_depleted) {
+ /* if there was no delta, track last allocated node */
+ resume_node = node;
+ resume_weight = i < (nnodes - 1) ? weights[i+1] :
+ weights[0];
+ }
+ /* node_pages can be 0 if an allocation fails and rounds == 0 */
+ if (!node_pages)
+ break;
+ nr_allocated = __alloc_pages_bulk(gfp, node, NULL, node_pages,
+ NULL, page_array);
+ page_array += nr_allocated;
+ total_allocated += nr_allocated;
+ if (total_allocated == nr_pages)
+ break;
+ prev_node = node;
+ }
+ /* resume allocating from the calculated node and weight */
+ me->il_prev = resume_node;
+ pol->wil.cur_weight = resume_weight;
+ kfree(weights);
+ return total_allocated;
+}
+
static unsigned long alloc_pages_bulk_array_preferred_many(gfp_t gfp, int nid,
struct mempolicy *pol, unsigned long nr_pages,
struct page **page_array)
@@ -2303,6 +2519,10 @@ unsigned long alloc_pages_bulk_array_mempolicy(gfp_t gfp,
return alloc_pages_bulk_array_interleave(gfp, pol,
nr_pages, page_array);
+ if (pol->mode == MPOL_WEIGHTED_INTERLEAVE)
+ return alloc_pages_bulk_array_weighted_interleave(
+ gfp, pol, nr_pages, page_array);
+
if (pol->mode == MPOL_PREFERRED_MANY)
return alloc_pages_bulk_array_preferred_many(gfp,
numa_node_id(), pol, nr_pages, page_array);
@@ -2378,6 +2598,7 @@ bool __mpol_equal(struct mempolicy *a, struct mempolicy *b)
case MPOL_INTERLEAVE:
case MPOL_PREFERRED:
case MPOL_PREFERRED_MANY:
+ case MPOL_WEIGHTED_INTERLEAVE:
return !!nodes_equal(a->nodes, b->nodes);
case MPOL_LOCAL:
return true;
@@ -2514,6 +2735,10 @@ int mpol_misplaced(struct folio *folio, struct vm_area_struct *vma,
polnid = interleave_nid(pol, ilx);
break;
+ case MPOL_WEIGHTED_INTERLEAVE:
+ polnid = weighted_interleave_nid(pol, ilx);
+ break;
+
case MPOL_PREFERRED:
if (node_isset(curnid, pol->nodes))
goto out;
@@ -2888,6 +3113,7 @@ static const char * const policy_modes[] =
[MPOL_PREFERRED] = "prefer",
[MPOL_BIND] = "bind",
[MPOL_INTERLEAVE] = "interleave",
+ [MPOL_WEIGHTED_INTERLEAVE] = "weighted interleave",
[MPOL_LOCAL] = "local",
[MPOL_PREFERRED_MANY] = "prefer (many)",
};
@@ -2947,6 +3173,7 @@ int mpol_parse_str(char *str, struct mempolicy **mpol)
}
break;
case MPOL_INTERLEAVE:
+ case MPOL_WEIGHTED_INTERLEAVE:
/*
* Default to online nodes with memory if no nodelist
*/
@@ -3057,6 +3284,7 @@ void mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
case MPOL_PREFERRED_MANY:
case MPOL_BIND:
case MPOL_INTERLEAVE:
+ case MPOL_WEIGHTED_INTERLEAVE:
nodes = pol->nodes;
break;
default:
--
2.39.1
Gregory Price <[email protected]> writes:
> When a system has multiple NUMA nodes and it becomes bandwidth hungry,
> using the current MPOL_INTERLEAVE could be an wise option.
>
> However, if those NUMA nodes consist of different types of memory such
> as socket-attached DRAM and CXL/PCIe attached DRAM, the round-robin
> based interleave policy does not optimally distribute data to make use
> of their different bandwidth characteristics.
>
> Instead, interleave is more effective when the allocation policy follows
> each NUMA nodes' bandwidth weight rather than a simple 1:1 distribution.
>
> This patch introduces a new memory policy, MPOL_WEIGHTED_INTERLEAVE,
> enabling weighted interleave between NUMA nodes. Weighted interleave
> allows for proportional distribution of memory across multiple numa
> nodes, preferably apportioned to match the bandwidth of each node.
>
> For example, if a system has 1 CPU node (0), and 2 memory nodes (0,1),
> with bandwidth of (100GB/s, 50GB/s) respectively, the appropriate
> weight distribution is (2:1).
>
> Weights for each node can be assigned via the new sysfs extension:
> /sys/kernel/mm/mempolicy/weighted_interleave/
>
> For now, the default value of all nodes will be `1`, which matches
> the behavior of standard 1:1 round-robin interleave. An extension
> will be added in the future to allow default values to be registered
> at kernel and device bringup time.
>
> The policy allocates a number of pages equal to the set weights. For
> example, if the weights are (2,1), then 2 pages will be allocated on
> node0 for every 1 page allocated on node1.
>
> The new flag MPOL_WEIGHTED_INTERLEAVE can be used in set_mempolicy(2)
> and mbind(2).
>
> There are 3 integration points:
>
> weighted_interleave_nodes:
> Counts the number of allocations as they occur, and applies the
> weight for the current node. When the weight reaches 0, switch
> to the next node.
>
> weighted_interleave_nid:
> Gets the total weight of the nodemask as well as each individual
> node weight, then calculates the node based on the given index.
>
> bulk_array_weighted_interleave:
> Gets the total weight of the nodemask as well as each individual
> node weight, then calculates the number of "interleave rounds" as
> well as any delta ("partial round"). Calculates the number of
> pages for each node and allocates them.
>
> If a node was scheduled for interleave via interleave_nodes, the
> current weight (pol->cur_weight) will be allocated first, before
> the remaining bulk calculation is done.
>
> One piece of complexity is the interaction between a recent refactor
> which split the logic to acquire the "ilx" (interleave index) of an
> allocation and the actually application of the interleave. The
> calculation of the `interleave index` is done by `get_vma_policy()`,
> while the actual selection of the node will be later appliex by the
> relevant weighted_interleave function.
>
> Suggested-by: Hasan Al Maruf <[email protected]>
> Signed-off-by: Gregory Price <[email protected]>
> Co-developed-by: Rakie Kim <[email protected]>
> Signed-off-by: Rakie Kim <[email protected]>
> Co-developed-by: Honggyu Kim <[email protected]>
> Signed-off-by: Honggyu Kim <[email protected]>
> Co-developed-by: Hyeongtak Ji <[email protected]>
> Signed-off-by: Hyeongtak Ji <[email protected]>
> Co-developed-by: Srinivasulu Thanneeru <[email protected]>
> Signed-off-by: Srinivasulu Thanneeru <[email protected]>
> Co-developed-by: Ravi Jonnalagadda <[email protected]>
> Signed-off-by: Ravi Jonnalagadda <[email protected]>
> ---
> .../admin-guide/mm/numa_memory_policy.rst | 9 +
> include/linux/mempolicy.h | 5 +
> include/uapi/linux/mempolicy.h | 1 +
> mm/mempolicy.c | 234 +++++++++++++++++-
> 4 files changed, 246 insertions(+), 3 deletions(-)
>
> diff --git a/Documentation/admin-guide/mm/numa_memory_policy.rst b/Documentation/admin-guide/mm/numa_memory_policy.rst
> index eca38fa81e0f..a70f20ce1ffb 100644
> --- a/Documentation/admin-guide/mm/numa_memory_policy.rst
> +++ b/Documentation/admin-guide/mm/numa_memory_policy.rst
> @@ -250,6 +250,15 @@ MPOL_PREFERRED_MANY
> can fall back to all existing numa nodes. This is effectively
> MPOL_PREFERRED allowed for a mask rather than a single node.
>
> +MPOL_WEIGHTED_INTERLEAVE
> + This mode operates the same as MPOL_INTERLEAVE, except that
> + interleaving behavior is executed based on weights set in
> + /sys/kernel/mm/mempolicy/weighted_interleave/
> +
> + Weighted interleave allocates pages on nodes according to a
> + weight. For example if nodes [0,1] are weighted [5,2], 5 pages
> + will be allocated on node0 for every 2 pages allocated on node1.
> +
> NUMA memory policy supports the following optional mode flags:
>
> MPOL_F_STATIC_NODES
> diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
> index 931b118336f4..c1a083eb0dd5 100644
> --- a/include/linux/mempolicy.h
> +++ b/include/linux/mempolicy.h
> @@ -54,6 +54,11 @@ struct mempolicy {
> nodemask_t cpuset_mems_allowed; /* relative to these nodes */
> nodemask_t user_nodemask; /* nodemask passed by user */
> } w;
> +
> + /* Weighted interleave settings */
> + struct {
> + u8 cur_weight;
> + } wil;
> };
>
> /*
> diff --git a/include/uapi/linux/mempolicy.h b/include/uapi/linux/mempolicy.h
> index a8963f7ef4c2..1f9bb10d1a47 100644
> --- a/include/uapi/linux/mempolicy.h
> +++ b/include/uapi/linux/mempolicy.h
> @@ -23,6 +23,7 @@ enum {
> MPOL_INTERLEAVE,
> MPOL_LOCAL,
> MPOL_PREFERRED_MANY,
> + MPOL_WEIGHTED_INTERLEAVE,
> MPOL_MAX, /* always last member of enum */
> };
>
> diff --git a/mm/mempolicy.c b/mm/mempolicy.c
> index 427bddf115df..aa3b2389d3e0 100644
> --- a/mm/mempolicy.c
> +++ b/mm/mempolicy.c
> @@ -19,6 +19,13 @@
> * for anonymous memory. For process policy an process counter
> * is used.
> *
> + * weighted interleave
> + * Allocate memory interleaved over a set of nodes based on
> + * a set of weights (per-node), with normal fallback if it
> + * fails. Otherwise operates the same as interleave.
> + * Example: nodeset(0,1) & weights (2,1) - 2 pages allocated
> + * on node 0 for every 1 page allocated on node 1.
> + *
> * bind Only allocate memory on a specific set of nodes,
> * no fallback.
> * FIXME: memory is allocated starting with the first node
> @@ -313,6 +320,7 @@ static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
> policy->mode = mode;
> policy->flags = flags;
> policy->home_node = NUMA_NO_NODE;
> + policy->wil.cur_weight = 0;
>
> return policy;
> }
> @@ -425,6 +433,10 @@ static const struct mempolicy_operations mpol_ops[MPOL_MAX] = {
> .create = mpol_new_nodemask,
> .rebind = mpol_rebind_preferred,
> },
> + [MPOL_WEIGHTED_INTERLEAVE] = {
> + .create = mpol_new_nodemask,
> + .rebind = mpol_rebind_nodemask,
> + },
> };
>
> static bool migrate_folio_add(struct folio *folio, struct list_head *foliolist,
> @@ -846,7 +858,8 @@ static long do_set_mempolicy(unsigned short mode, unsigned short flags,
>
> old = current->mempolicy;
> current->mempolicy = new;
> - if (new && new->mode == MPOL_INTERLEAVE)
> + if (new && (new->mode == MPOL_INTERLEAVE ||
> + new->mode == MPOL_WEIGHTED_INTERLEAVE))
> current->il_prev = MAX_NUMNODES-1;
> task_unlock(current);
> mpol_put(old);
> @@ -872,6 +885,7 @@ static void get_policy_nodemask(struct mempolicy *pol, nodemask_t *nodes)
> case MPOL_INTERLEAVE:
> case MPOL_PREFERRED:
> case MPOL_PREFERRED_MANY:
> + case MPOL_WEIGHTED_INTERLEAVE:
> *nodes = pol->nodes;
> break;
> case MPOL_LOCAL:
> @@ -956,6 +970,13 @@ static long do_get_mempolicy(int *policy, nodemask_t *nmask,
> } else if (pol == current->mempolicy &&
> pol->mode == MPOL_INTERLEAVE) {
> *policy = next_node_in(current->il_prev, pol->nodes);
> + } else if (pol == current->mempolicy &&
> + (pol->mode == MPOL_WEIGHTED_INTERLEAVE)) {
> + if (pol->wil.cur_weight)
> + *policy = current->il_prev;
> + else
> + *policy = next_node_in(current->il_prev,
> + pol->nodes);
> } else {
> err = -EINVAL;
> goto out;
> @@ -1785,7 +1806,8 @@ struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
> pol = __get_vma_policy(vma, addr, ilx);
> if (!pol)
> pol = get_task_policy(current);
> - if (pol->mode == MPOL_INTERLEAVE) {
> + if (pol->mode == MPOL_INTERLEAVE ||
> + pol->mode == MPOL_WEIGHTED_INTERLEAVE) {
Should change the comments above get_vma_policy() definition too.
> *ilx += vma->vm_pgoff >> order;
> *ilx += (addr - vma->vm_start) >> (PAGE_SHIFT + order);
> }
> @@ -1835,6 +1857,28 @@ bool apply_policy_zone(struct mempolicy *policy, enum zone_type zone)
> return zone >= dynamic_policy_zone;
> }
>
> +static unsigned int weighted_interleave_nodes(struct mempolicy *policy)
> +{
> + unsigned int next;
> + struct task_struct *me = current;
> + u8 __rcu *table;
> +
> + next = next_node_in(me->il_prev, policy->nodes);
> + if (next == MAX_NUMNODES)
> + return next;
> +
> + rcu_read_lock();
> + table = rcu_dereference(iw_table);
> + if (!policy->wil.cur_weight)
> + policy->wil.cur_weight = table ? table[next] : 1;
> + rcu_read_unlock();
> +
> + policy->wil.cur_weight--;
> + if (!policy->wil.cur_weight)
> + me->il_prev = next;
> + return next;
> +}
> +
> /* Do dynamic interleaving for a process */
> static unsigned int interleave_nodes(struct mempolicy *policy)
> {
> @@ -1869,6 +1913,9 @@ unsigned int mempolicy_slab_node(void)
> case MPOL_INTERLEAVE:
> return interleave_nodes(policy);
>
> + case MPOL_WEIGHTED_INTERLEAVE:
> + return weighted_interleave_nodes(policy);
> +
> case MPOL_BIND:
> case MPOL_PREFERRED_MANY:
> {
> @@ -1907,6 +1954,39 @@ static unsigned int read_once_policy_nodemask(struct mempolicy *pol,
> return nodes_weight(*mask);
> }
>
> +static unsigned int weighted_interleave_nid(struct mempolicy *pol, pgoff_t ilx)
> +{
> + nodemask_t nodemask;
> + unsigned int target, nr_nodes;
> + u8 __rcu *table;
> + unsigned int weight_total = 0;
> + u8 weight;
> + int nid;
> +
> + nr_nodes = read_once_policy_nodemask(pol, &nodemask);
> + if (!nr_nodes)
> + return numa_node_id();
> +
> + rcu_read_lock();
> + table = rcu_dereference(iw_table);
> + /* calculate the total weight */
> + for_each_node_mask(nid, nodemask)
> + weight_total += table ? table[nid] : 1;
> +
> + /* Calculate the node offset based on totals */
> + target = ilx % weight_total;
> + nid = first_node(nodemask);
> + while (target) {
> + weight = table ? table[nid] : 1;
> + if (target < weight)
> + break;
> + target -= weight;
> + nid = next_node_in(nid, nodemask);
> + }
> + rcu_read_unlock();
> + return nid;
> +}
> +
> /*
> * Do static interleaving for interleave index @ilx. Returns the ilx'th
> * node in pol->nodes (starting from ilx=0), wrapping around if ilx
> @@ -1967,6 +2047,11 @@ static nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *pol,
> *nid = (ilx == NO_INTERLEAVE_INDEX) ?
> interleave_nodes(pol) : interleave_nid(pol, ilx);
> break;
> + case MPOL_WEIGHTED_INTERLEAVE:
> + *nid = (ilx == NO_INTERLEAVE_INDEX) ?
> + weighted_interleave_nodes(pol) :
> + weighted_interleave_nid(pol, ilx);
> + break;
> }
>
> return nodemask;
> @@ -2028,6 +2113,7 @@ bool init_nodemask_of_mempolicy(nodemask_t *mask)
> case MPOL_PREFERRED_MANY:
> case MPOL_BIND:
> case MPOL_INTERLEAVE:
> + case MPOL_WEIGHTED_INTERLEAVE:
> *mask = mempolicy->nodes;
> break;
>
> @@ -2127,7 +2213,8 @@ struct page *alloc_pages_mpol(gfp_t gfp, unsigned int order,
> * If the policy is interleave or does not allow the current
> * node in its nodemask, we allocate the standard way.
> */
> - if (pol->mode != MPOL_INTERLEAVE &&
> + if ((pol->mode != MPOL_INTERLEAVE &&
> + pol->mode != MPOL_WEIGHTED_INTERLEAVE) &&
> (!nodemask || node_isset(nid, *nodemask))) {
> /*
> * First, try to allocate THP only on local node, but
> @@ -2263,6 +2350,135 @@ static unsigned long alloc_pages_bulk_array_interleave(gfp_t gfp,
> return total_allocated;
> }
>
> +static unsigned long alloc_pages_bulk_array_weighted_interleave(gfp_t gfp,
> + struct mempolicy *pol, unsigned long nr_pages,
> + struct page **page_array)
> +{
> + struct task_struct *me = current;
> + unsigned long total_allocated = 0;
> + unsigned long nr_allocated;
> + unsigned long rounds;
> + unsigned long node_pages, delta;
> + u8 weight, resume_weight;
> + u8 __rcu *table;
> + u8 *weights;
> + unsigned int weight_total = 0;
> + unsigned long rem_pages = nr_pages;
> + nodemask_t nodes;
> + int nnodes, node, weight_nodes, resume_node;
> + int prev_node = NUMA_NO_NODE;
It appears that we should initialize prev_node with me->il_prev?
Details are as below.
> + bool delta_depleted = false;
> + int i;
> +
> + if (!nr_pages)
> + return 0;
> +
> + nnodes = read_once_policy_nodemask(pol, &nodes);
> + if (!nnodes)
> + return 0;
> +
> + /* Continue allocating from most recent node and adjust the nr_pages */
> + if (pol->wil.cur_weight) {
> + node = next_node_in(me->il_prev, nodes);
> + node_pages = pol->wil.cur_weight;
> + if (node_pages > rem_pages)
> + node_pages = rem_pages;
> + nr_allocated = __alloc_pages_bulk(gfp, node, NULL, node_pages,
> + NULL, page_array);
> + page_array += nr_allocated;
> + total_allocated += nr_allocated;
> + /* if that's all the pages, no need to interleave */
> + if (rem_pages <= pol->wil.cur_weight) {
> + pol->wil.cur_weight -= rem_pages;
If "pol->wil.cur_weight == 0" here, we need to change me->il_prev?
> + return total_allocated;
> + }
> + /* Otherwise we adjust nr_pages down, and continue from there */
> + rem_pages -= pol->wil.cur_weight;
> + pol->wil.cur_weight = 0;
> + prev_node = node;
> + }
else {
prev_node = me->il_prev;
}
> +
> + /* fetch the weights for this operation and calculate total weight */
> + weights = kmalloc(nnodes, GFP_KERNEL);
> + if (!weights)
> + return total_allocated;
> +
> + rcu_read_lock();
> + table = rcu_dereference(iw_table);
> + weight_nodes = 0;
We can replace "weight_nodes" with "i" and use a "for" loop?
> + while (weight_nodes < nnodes) {
> + node = next_node_in(prev_node, nodes);
IIUC, "node" will not change in the loop, so all "weight" below will be
the same value. To keep it simple, I think we can just copy weights
from the global iw_table and consider the default value?
> + weight = table ? table[node] : 1;
> + weights[weight_nodes++] = weight;
> + weight_total += weight;
> + }
> + rcu_read_unlock();
> +
> + /*
> + * Now we can continue allocating from 0 instead of an offset
> + * We calculate the number of rounds and any partial rounds so
> + * that we minimize the number of calls to __alloc_pages_bulk
> + * This requires us to track which node we should resume from.
> + *
> + * if (rounds > 0) and (delta == 0), resume_node will always be
> + * the current me->il_prev
> + *
> + * if (delta > 0) and delta is depleted exactly on a node-weight
> + * boundary, resume node will be the node last allocated from when
> + * delta reached 0.
> + *
> + * if (delta > 0) and delta is not depleted on a node-weight boundary,
> + * resume node will be the node prior to the node last allocated from.
> + *
> + * (rounds == 0) and (delta == 0) is not possible (earlier exit)
> + */
> + rounds = rem_pages / weight_total;
> + delta = rem_pages % weight_total;
> + /* If no delta, we'll resume from current prev_node and first weight */
> + for (i = 0; i < nnodes; i++) {
> + node = next_node_in(prev_node, nodes);
> + weight = weights[i];
> + node_pages = weight * rounds;
> + /* If a delta exists, add this node's portion of the delta */
> + if (delta >= weight) {
> + node_pages += weight;
> + delta -= weight;
> + resume_node = node;
> + resume_weight = i < (nnodes - 1) ? weights[i+1] :
> + weights[0];
> + /* stop tracking iff (delta == weight) */
> + delta_depleted = !delta;
> + } else if (delta) { /* <= weight */
The comment is unnecessary and wrong.
> + /* if delta depleted, resume from this node */
> + node_pages += delta;
> + delta = 0;
> + resume_node = prev_node;
> + resume_weight = weight - (node_pages % weight);
> + delta_depleted = true; /* stop tracking */
> + } else if (!delta_depleted) {
> + /* if there was no delta, track last allocated node */
> + resume_node = node;
> + resume_weight = i < (nnodes - 1) ? weights[i+1] :
> + weights[0];
> + }
Can the above code be simplified as something like below?
resume_node = prev_node;
resume_weight = 0;
for (...) {
...
if (delta > weight) {
node_pages += weight;
delta -= weight;
} else if (delta) {
node_pages += delta;
/* if delta depleted, resume from this node */
if (delta < weight) {
resume_node = prev_node;
resume_weight = weight - delta;
} else {
resume_node = node;
}
delta = 0;
}
...
}
--
Best Regards,
Huang, Ying
> + /* node_pages can be 0 if an allocation fails and rounds == 0 */
> + if (!node_pages)
> + break;
> + nr_allocated = __alloc_pages_bulk(gfp, node, NULL, node_pages,
> + NULL, page_array);
> + page_array += nr_allocated;
> + total_allocated += nr_allocated;
> + if (total_allocated == nr_pages)
> + break;
> + prev_node = node;
> + }
> + /* resume allocating from the calculated node and weight */
> + me->il_prev = resume_node;
> + pol->wil.cur_weight = resume_weight;
> + kfree(weights);
> + return total_allocated;
> +}
> +
> static unsigned long alloc_pages_bulk_array_preferred_many(gfp_t gfp, int nid,
> struct mempolicy *pol, unsigned long nr_pages,
> struct page **page_array)
> @@ -2303,6 +2519,10 @@ unsigned long alloc_pages_bulk_array_mempolicy(gfp_t gfp,
> return alloc_pages_bulk_array_interleave(gfp, pol,
> nr_pages, page_array);
>
> + if (pol->mode == MPOL_WEIGHTED_INTERLEAVE)
> + return alloc_pages_bulk_array_weighted_interleave(
> + gfp, pol, nr_pages, page_array);
> +
> if (pol->mode == MPOL_PREFERRED_MANY)
> return alloc_pages_bulk_array_preferred_many(gfp,
> numa_node_id(), pol, nr_pages, page_array);
> @@ -2378,6 +2598,7 @@ bool __mpol_equal(struct mempolicy *a, struct mempolicy *b)
> case MPOL_INTERLEAVE:
> case MPOL_PREFERRED:
> case MPOL_PREFERRED_MANY:
> + case MPOL_WEIGHTED_INTERLEAVE:
> return !!nodes_equal(a->nodes, b->nodes);
> case MPOL_LOCAL:
> return true;
> @@ -2514,6 +2735,10 @@ int mpol_misplaced(struct folio *folio, struct vm_area_struct *vma,
> polnid = interleave_nid(pol, ilx);
> break;
>
> + case MPOL_WEIGHTED_INTERLEAVE:
> + polnid = weighted_interleave_nid(pol, ilx);
> + break;
> +
> case MPOL_PREFERRED:
> if (node_isset(curnid, pol->nodes))
> goto out;
> @@ -2888,6 +3113,7 @@ static const char * const policy_modes[] =
> [MPOL_PREFERRED] = "prefer",
> [MPOL_BIND] = "bind",
> [MPOL_INTERLEAVE] = "interleave",
> + [MPOL_WEIGHTED_INTERLEAVE] = "weighted interleave",
> [MPOL_LOCAL] = "local",
> [MPOL_PREFERRED_MANY] = "prefer (many)",
> };
> @@ -2947,6 +3173,7 @@ int mpol_parse_str(char *str, struct mempolicy **mpol)
> }
> break;
> case MPOL_INTERLEAVE:
> + case MPOL_WEIGHTED_INTERLEAVE:
> /*
> * Default to online nodes with memory if no nodelist
> */
> @@ -3057,6 +3284,7 @@ void mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
> case MPOL_PREFERRED_MANY:
> case MPOL_BIND:
> case MPOL_INTERLEAVE:
> + case MPOL_WEIGHTED_INTERLEAVE:
> nodes = pol->nodes;
> break;
> default:
On Tue, Jan 23, 2024 at 11:02:03AM +0800, Huang, Ying wrote:
> Gregory Price <[email protected]> writes:
>
> > + int prev_node = NUMA_NO_NODE;
>
> It appears that we should initialize prev_node with me->il_prev?
> Details are as below.
>
yeah good catch, was a rebase error from my tested code, where this is
the case. patching now.
> > + if (rem_pages <= pol->wil.cur_weight) {
> > + pol->wil.cur_weight -= rem_pages;
>
> If "pol->wil.cur_weight == 0" here, we need to change me->il_prev?
>
you are right, and also need to fetch the next cur_weight. Seems I
missed this specific case in my tests. (had this tested with a single
node but not 2, so it looked right).
Added to my test suite.
> We can replace "weight_nodes" with "i" and use a "for" loop?
>
> > + while (weight_nodes < nnodes) {
> > + node = next_node_in(prev_node, nodes);
>
> IIUC, "node" will not change in the loop, so all "weight" below will be
> the same value. To keep it simple, I think we can just copy weights
> from the global iw_table and consider the default value?
>
another rebase error here from my tested code, this should have been
node = prev_node;
while (...)
node = next_node_in(node, nodes);
I can change it to a for loop as suggested, but for more info on why I
did it this way, see the chunk below
> > + } else if (!delta_depleted) {
> > + /* if there was no delta, track last allocated node */
> > + resume_node = node;
> > + resume_weight = i < (nnodes - 1) ? weights[i+1] :
> > + weights[0];
^ this line acquires the weight of the *NEXT* node
another chunk prior to this does the same
thing. I suppose i can use next_node_in()
instead and just copy the entire weigh array
though, if that is preferable.
> > + }
>
> Can the above code be simplified as something like below?
>
> resume_node = prev_node;
> resume_weight = 0;
> for (...) {
> ...
> if (delta > weight) {
> node_pages += weight;
> delta -= weight;
> } else if (delta) {
> node_pages += delta;
> /* if delta depleted, resume from this node */
> if (delta < weight) {
> resume_node = prev_node;
> resume_weight = weight - delta;
> } else {
> resume_node = node;
> }
> delta = 0;
> }
> ...
> }
>
I'll take another look at it, but this logic is annoying because of the
corner case: me->il_prev can be NUMA_NO_NODE or an actual numa node.
If it's NUMA_NO_NODE, then the logic you have above will say "the next
node has no remaining weights assigned" and skip it on the next call to
weighted_interleave_nid or weighted_interleave_nodes.
This is incorrect - we want the weight of the first node to be
resume_weight, which is what this chunk does:
if (delta >= weight) {
/* if delta == weight, get next node weight */
resume_weight = i < (nnodes - 1) ? weights[i+1] : weights[0];
else if (delta) { /* delta < weight */
/* there's a remaining weight, use the that for resume weight */
resume_weight = weight - (node_pages % weight);
} else if (!delta_depleted) {
/* there was never a delta, track the last node and get the weight
* of the node AFTER that node, that's the resume weight */
resume_weight = i < (nnodes - 1) ? weights[i+1] : weights[0];
}
If il_prev is an actual node, and delta == 0, we want to return with
(il_prev = prev_node) but with the weight set to the weight of the
first node we're about to allocate from.
This is the reason for the annoying logic here: We have to come out of
this loop with the actual node and the actual weight.
I'll try to clean it up further and get my test suite to pass.
~Gregory
On Mon, Jan 22, 2024 at 11:54:34PM -0500, Gregory Price wrote:
> >
> > Can the above code be simplified as something like below?
> >
> > resume_node = prev_node;
--- resume_weight = 0;
+++ resume_weight = weights[node];
> > for (...) {
> > ...
> > }
> >
>
> I'll take another look at it, but this logic is annoying because of the
> corner case: me->il_prev can be NUMA_NO_NODE or an actual numa node.
>
After a quick look, as long as no one objects to (me->il_prev) remaining
NUMA_NO_NODE while having a weight assigned to pol->wil.cur_weight, then
this looks like it can be simplified as above.
I don't think it's harmful, but i'll have to take a quick look at what
happens on rebind to make sure we don't have a stale weight.
~Gregory
Gregory Price <[email protected]> writes:
> On Tue, Jan 23, 2024 at 11:02:03AM +0800, Huang, Ying wrote:
>> Gregory Price <[email protected]> writes:
>>
>> > + int prev_node = NUMA_NO_NODE;
>>
>> It appears that we should initialize prev_node with me->il_prev?
>> Details are as below.
>>
>
> yeah good catch, was a rebase error from my tested code, where this is
> the case. patching now.
>
>> > + if (rem_pages <= pol->wil.cur_weight) {
>> > + pol->wil.cur_weight -= rem_pages;
>>
>> If "pol->wil.cur_weight == 0" here, we need to change me->il_prev?
>>
> you are right, and also need to fetch the next cur_weight. Seems I
> missed this specific case in my tests. (had this tested with a single
> node but not 2, so it looked right).
>
> Added to my test suite.
>
>> We can replace "weight_nodes" with "i" and use a "for" loop?
>>
>> > + while (weight_nodes < nnodes) {
>> > + node = next_node_in(prev_node, nodes);
>>
>> IIUC, "node" will not change in the loop, so all "weight" below will be
>> the same value. To keep it simple, I think we can just copy weights
>> from the global iw_table and consider the default value?
>>
>
> another rebase error here from my tested code, this should have been
> node = prev_node;
> while (...)
> node = next_node_in(node, nodes);
>
> I can change it to a for loop as suggested, but for more info on why I
> did it this way, see the chunk below
>
>> > + } else if (!delta_depleted) {
>> > + /* if there was no delta, track last allocated node */
>> > + resume_node = node;
>> > + resume_weight = i < (nnodes - 1) ? weights[i+1] :
>> > + weights[0];
> ^ this line acquires the weight of the *NEXT* node
> another chunk prior to this does the same
> thing. I suppose i can use next_node_in()
> instead and just copy the entire weigh array
> though, if that is preferable.
Yes. I think copy the entire weight array make code logic simpler.
--
Best Regards,
Huang, Ying
Gregory Price <[email protected]> writes:
> On Mon, Jan 22, 2024 at 11:54:34PM -0500, Gregory Price wrote:
>> >
>> > Can the above code be simplified as something like below?
>> >
>> > resume_node = prev_node;
> --- resume_weight = 0;
> +++ resume_weight = weights[node];
>> > for (...) {
>> > ...
>> > }
>> >
>>
>> I'll take another look at it, but this logic is annoying because of the
>> corner case: me->il_prev can be NUMA_NO_NODE or an actual numa node.
>>
>
> After a quick look, as long as no one objects to (me->il_prev) remaining
> NUMA_NO_NODE
MAX_NUMNODES-1 ?
> while having a weight assigned to pol->wil.cur_weight,
I think that it is OK.
And, IIUC, pol->wil.cur_weight can be 0, as in
weighted_interleave_nodes(), if it's 0, it will be assigned to default
weight for the node.
> then
> this looks like it can be simplified as above.
>
> I don't think it's harmful, but i'll have to take a quick look at what
> happens on rebind to make sure we don't have a stale weight.
Make sense.
--
Best Regards,
Huang, Ying
Gregory Price <[email protected]> writes:
> When a system has multiple NUMA nodes and it becomes bandwidth hungry,
> using the current MPOL_INTERLEAVE could be an wise option.
>
> However, if those NUMA nodes consist of different types of memory such
> as socket-attached DRAM and CXL/PCIe attached DRAM, the round-robin
> based interleave policy does not optimally distribute data to make use
> of their different bandwidth characteristics.
>
> Instead, interleave is more effective when the allocation policy follows
> each NUMA nodes' bandwidth weight rather than a simple 1:1 distribution.
>
> This patch introduces a new memory policy, MPOL_WEIGHTED_INTERLEAVE,
> enabling weighted interleave between NUMA nodes. Weighted interleave
> allows for proportional distribution of memory across multiple numa
> nodes, preferably apportioned to match the bandwidth of each node.
>
> For example, if a system has 1 CPU node (0), and 2 memory nodes (0,1),
> with bandwidth of (100GB/s, 50GB/s) respectively, the appropriate
> weight distribution is (2:1).
>
> Weights for each node can be assigned via the new sysfs extension:
> /sys/kernel/mm/mempolicy/weighted_interleave/
>
> For now, the default value of all nodes will be `1`, which matches
> the behavior of standard 1:1 round-robin interleave. An extension
> will be added in the future to allow default values to be registered
> at kernel and device bringup time.
>
> The policy allocates a number of pages equal to the set weights. For
> example, if the weights are (2,1), then 2 pages will be allocated on
> node0 for every 1 page allocated on node1.
>
> The new flag MPOL_WEIGHTED_INTERLEAVE can be used in set_mempolicy(2)
> and mbind(2).
>
> There are 3 integration points:
>
> weighted_interleave_nodes:
> Counts the number of allocations as they occur, and applies the
> weight for the current node. When the weight reaches 0, switch
> to the next node.
>
> weighted_interleave_nid:
> Gets the total weight of the nodemask as well as each individual
> node weight, then calculates the node based on the given index.
>
> bulk_array_weighted_interleave:
> Gets the total weight of the nodemask as well as each individual
> node weight, then calculates the number of "interleave rounds" as
> well as any delta ("partial round"). Calculates the number of
> pages for each node and allocates them.
>
> If a node was scheduled for interleave via interleave_nodes, the
> current weight (pol->cur_weight) will be allocated first, before
> the remaining bulk calculation is done.
>
> One piece of complexity is the interaction between a recent refactor
> which split the logic to acquire the "ilx" (interleave index) of an
> allocation and the actually application of the interleave. The
> calculation of the `interleave index` is done by `get_vma_policy()`,
> while the actual selection of the node will be later appliex by the
> relevant weighted_interleave function.
>
> Suggested-by: Hasan Al Maruf <[email protected]>
> Signed-off-by: Gregory Price <[email protected]>
> Co-developed-by: Rakie Kim <[email protected]>
> Signed-off-by: Rakie Kim <[email protected]>
> Co-developed-by: Honggyu Kim <[email protected]>
> Signed-off-by: Honggyu Kim <[email protected]>
> Co-developed-by: Hyeongtak Ji <[email protected]>
> Signed-off-by: Hyeongtak Ji <[email protected]>
> Co-developed-by: Srinivasulu Thanneeru <[email protected]>
> Signed-off-by: Srinivasulu Thanneeru <[email protected]>
> Co-developed-by: Ravi Jonnalagadda <[email protected]>
> Signed-off-by: Ravi Jonnalagadda <[email protected]>
> ---
> .../admin-guide/mm/numa_memory_policy.rst | 9 +
> include/linux/mempolicy.h | 5 +
> include/uapi/linux/mempolicy.h | 1 +
> mm/mempolicy.c | 234 +++++++++++++++++-
> 4 files changed, 246 insertions(+), 3 deletions(-)
>
> diff --git a/Documentation/admin-guide/mm/numa_memory_policy.rst b/Documentation/admin-guide/mm/numa_memory_policy.rst
> index eca38fa81e0f..a70f20ce1ffb 100644
> --- a/Documentation/admin-guide/mm/numa_memory_policy.rst
> +++ b/Documentation/admin-guide/mm/numa_memory_policy.rst
> @@ -250,6 +250,15 @@ MPOL_PREFERRED_MANY
> can fall back to all existing numa nodes. This is effectively
> MPOL_PREFERRED allowed for a mask rather than a single node.
>
> +MPOL_WEIGHTED_INTERLEAVE
> + This mode operates the same as MPOL_INTERLEAVE, except that
> + interleaving behavior is executed based on weights set in
> + /sys/kernel/mm/mempolicy/weighted_interleave/
> +
> + Weighted interleave allocates pages on nodes according to a
> + weight. For example if nodes [0,1] are weighted [5,2], 5 pages
> + will be allocated on node0 for every 2 pages allocated on node1.
> +
> NUMA memory policy supports the following optional mode flags:
>
> MPOL_F_STATIC_NODES
> diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
> index 931b118336f4..c1a083eb0dd5 100644
> --- a/include/linux/mempolicy.h
> +++ b/include/linux/mempolicy.h
> @@ -54,6 +54,11 @@ struct mempolicy {
> nodemask_t cpuset_mems_allowed; /* relative to these nodes */
> nodemask_t user_nodemask; /* nodemask passed by user */
> } w;
> +
> + /* Weighted interleave settings */
> + struct {
> + u8 cur_weight;
> + } wil;
> };
>
> /*
> diff --git a/include/uapi/linux/mempolicy.h b/include/uapi/linux/mempolicy.h
> index a8963f7ef4c2..1f9bb10d1a47 100644
> --- a/include/uapi/linux/mempolicy.h
> +++ b/include/uapi/linux/mempolicy.h
> @@ -23,6 +23,7 @@ enum {
> MPOL_INTERLEAVE,
> MPOL_LOCAL,
> MPOL_PREFERRED_MANY,
> + MPOL_WEIGHTED_INTERLEAVE,
> MPOL_MAX, /* always last member of enum */
> };
>
> diff --git a/mm/mempolicy.c b/mm/mempolicy.c
> index 427bddf115df..aa3b2389d3e0 100644
> --- a/mm/mempolicy.c
> +++ b/mm/mempolicy.c
> @@ -19,6 +19,13 @@
> * for anonymous memory. For process policy an process counter
> * is used.
> *
> + * weighted interleave
> + * Allocate memory interleaved over a set of nodes based on
> + * a set of weights (per-node), with normal fallback if it
> + * fails. Otherwise operates the same as interleave.
> + * Example: nodeset(0,1) & weights (2,1) - 2 pages allocated
> + * on node 0 for every 1 page allocated on node 1.
> + *
> * bind Only allocate memory on a specific set of nodes,
> * no fallback.
> * FIXME: memory is allocated starting with the first node
> @@ -313,6 +320,7 @@ static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
> policy->mode = mode;
> policy->flags = flags;
> policy->home_node = NUMA_NO_NODE;
> + policy->wil.cur_weight = 0;
>
> return policy;
> }
> @@ -425,6 +433,10 @@ static const struct mempolicy_operations mpol_ops[MPOL_MAX] = {
> .create = mpol_new_nodemask,
> .rebind = mpol_rebind_preferred,
> },
> + [MPOL_WEIGHTED_INTERLEAVE] = {
> + .create = mpol_new_nodemask,
> + .rebind = mpol_rebind_nodemask,
> + },
> };
>
> static bool migrate_folio_add(struct folio *folio, struct list_head *foliolist,
> @@ -846,7 +858,8 @@ static long do_set_mempolicy(unsigned short mode, unsigned short flags,
>
> old = current->mempolicy;
> current->mempolicy = new;
> - if (new && new->mode == MPOL_INTERLEAVE)
> + if (new && (new->mode == MPOL_INTERLEAVE ||
> + new->mode == MPOL_WEIGHTED_INTERLEAVE))
> current->il_prev = MAX_NUMNODES-1;
> task_unlock(current);
> mpol_put(old);
> @@ -872,6 +885,7 @@ static void get_policy_nodemask(struct mempolicy *pol, nodemask_t *nodes)
> case MPOL_INTERLEAVE:
> case MPOL_PREFERRED:
> case MPOL_PREFERRED_MANY:
> + case MPOL_WEIGHTED_INTERLEAVE:
> *nodes = pol->nodes;
> break;
> case MPOL_LOCAL:
> @@ -956,6 +970,13 @@ static long do_get_mempolicy(int *policy, nodemask_t *nmask,
> } else if (pol == current->mempolicy &&
> pol->mode == MPOL_INTERLEAVE) {
> *policy = next_node_in(current->il_prev, pol->nodes);
> + } else if (pol == current->mempolicy &&
> + (pol->mode == MPOL_WEIGHTED_INTERLEAVE)) {
> + if (pol->wil.cur_weight)
> + *policy = current->il_prev;
> + else
> + *policy = next_node_in(current->il_prev,
> + pol->nodes);
Per my understanding, we should always use "*policy = next_node_in()"
here, as in weighted_interleave_nodes().
> } else {
> err = -EINVAL;
> goto out;
> @@ -1785,7 +1806,8 @@ struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
> pol = __get_vma_policy(vma, addr, ilx);
> if (!pol)
> pol = get_task_policy(current);
> - if (pol->mode == MPOL_INTERLEAVE) {
> + if (pol->mode == MPOL_INTERLEAVE ||
> + pol->mode == MPOL_WEIGHTED_INTERLEAVE) {
> *ilx += vma->vm_pgoff >> order;
> *ilx += (addr - vma->vm_start) >> (PAGE_SHIFT + order);
> }
> @@ -1835,6 +1857,28 @@ bool apply_policy_zone(struct mempolicy *policy, enum zone_type zone)
> return zone >= dynamic_policy_zone;
> }
>
> +static unsigned int weighted_interleave_nodes(struct mempolicy *policy)
> +{
> + unsigned int next;
> + struct task_struct *me = current;
> + u8 __rcu *table;
> +
> + next = next_node_in(me->il_prev, policy->nodes);
> + if (next == MAX_NUMNODES)
> + return next;
> +
> + rcu_read_lock();
> + table = rcu_dereference(iw_table);
> + if (!policy->wil.cur_weight)
> + policy->wil.cur_weight = table ? table[next] : 1;
> + rcu_read_unlock();
> +
> + policy->wil.cur_weight--;
> + if (!policy->wil.cur_weight)
> + me->il_prev = next;
> + return next;
> +}
> +
[snip]
--
Best Regards,
Huang, Ying
On Tue, Jan 23, 2024 at 04:35:19PM +0800, Huang, Ying wrote:
> Gregory Price <[email protected]> writes:
>
> > On Mon, Jan 22, 2024 at 11:54:34PM -0500, Gregory Price wrote:
> >> >
> >> > Can the above code be simplified as something like below?
> >> >
> >> > resume_node = prev_node;
> > --- resume_weight = 0;
> > +++ resume_weight = weights[node];
> >> > for (...) {
> >> > ...
> >> > }
> >> >
> >>
> >> I'll take another look at it, but this logic is annoying because of the
> >> corner case: me->il_prev can be NUMA_NO_NODE or an actual numa node.
> >>
> >
> > After a quick look, as long as no one objects to (me->il_prev) remaining
> > NUMA_NO_NODE
>
> MAX_NUMNODES-1 ?
>
When setting a new policy, the il_prev gets set to NUMA_NO_NODE. It's
not harmful and is just (-1), which is functionally the same as
(MAX_NUMNODES-1) for the purpose of iterating the nodemask with
next_node_in(). So it's fine to set (resume_node = me->il_prev)
as discussed.
I have a cleaned up function I'll push when i fix up a few other spots.
> > while having a weight assigned to pol->wil.cur_weight,
>
> I think that it is OK.
>
> And, IIUC, pol->wil.cur_weight can be 0, as in
> weighted_interleave_nodes(), if it's 0, it will be assigned to default
> weight for the node.
>
cur_weight is different than the global weights. cur_weight tells us
how many pages are remaining to allocate for the current node.
(cur_weight = 0) can happen in two scenarios:
- initial setting of mempolicy (NUMA_NO_NODE w/ cur_weight=0)
- weighted_interleave_nodes decrements it down to 0
Now that i'm looking at it - the second condition should not exist, and
we can eliminate it. The logic in weighted_interleave_nodes is actually
annoyingly unclear at the moment, so I'm going to re-factor it a bit to
be more explicit.
~Gregory
Gregory Price <[email protected]> writes:
> On Tue, Jan 23, 2024 at 04:35:19PM +0800, Huang, Ying wrote:
>> Gregory Price <[email protected]> writes:
>>
>> > On Mon, Jan 22, 2024 at 11:54:34PM -0500, Gregory Price wrote:
>> >> >
>> >> > Can the above code be simplified as something like below?
>> >> >
>> >> > resume_node = prev_node;
>> > --- resume_weight = 0;
>> > +++ resume_weight = weights[node];
>> >> > for (...) {
>> >> > ...
>> >> > }
>> >> >
>> >>
>> >> I'll take another look at it, but this logic is annoying because of the
>> >> corner case: me->il_prev can be NUMA_NO_NODE or an actual numa node.
>> >>
>> >
>> > After a quick look, as long as no one objects to (me->il_prev) remaining
>> > NUMA_NO_NODE
>>
>> MAX_NUMNODES-1 ?
>>
>
> When setting a new policy, the il_prev gets set to NUMA_NO_NODE. It's
IIUC, it is set to MAX_NUMNODES-1 as below,
@@ -846,7 +858,8 @@ static long do_set_mempolicy(unsigned short mode, unsigned short flags,
old = current->mempolicy;
current->mempolicy = new;
- if (new && new->mode == MPOL_INTERLEAVE)
+ if (new && (new->mode == MPOL_INTERLEAVE ||
+ new->mode == MPOL_WEIGHTED_INTERLEAVE))
current->il_prev = MAX_NUMNODES-1;
task_unlock(current);
mpol_put(old);
I don't think we need to change this.
> not harmful and is just (-1), which is functionally the same as
> (MAX_NUMNODES-1) for the purpose of iterating the nodemask with
> next_node_in(). So it's fine to set (resume_node = me->il_prev)
> as discussed.
>
> I have a cleaned up function I'll push when i fix up a few other spots.
>
>> > while having a weight assigned to pol->wil.cur_weight,
>>
>> I think that it is OK.
>>
>> And, IIUC, pol->wil.cur_weight can be 0, as in
>> weighted_interleave_nodes(), if it's 0, it will be assigned to default
>> weight for the node.
>>
>
> cur_weight is different than the global weights. cur_weight tells us
> how many pages are remaining to allocate for the current node.
>
> (cur_weight = 0) can happen in two scenarios:
> - initial setting of mempolicy (NUMA_NO_NODE w/ cur_weight=0)
> - weighted_interleave_nodes decrements it down to 0
>
> Now that i'm looking at it - the second condition should not exist, and
> we can eliminate it. The logic in weighted_interleave_nodes is actually
> annoyingly unclear at the moment, so I'm going to re-factor it a bit to
> be more explicit.
I am OK with either way. Just a reminder, the first condition may be
true in alloc_pages_bulk_array_weighted_interleave() and perhaps some
other places.
--
Best Regards,
Huang, Ying
On Wed, Jan 24, 2024 at 09:51:20AM +0800, Huang, Ying wrote:
> Gregory Price <[email protected]> writes:
>
> + if (new && (new->mode == MPOL_INTERLEAVE ||
> + new->mode == MPOL_WEIGHTED_INTERLEAVE))
> current->il_prev = MAX_NUMNODES-1;
> task_unlock(current);
> mpol_put(old);
>
> I don't think we need to change this.
>
Ah you're right it's set to MAX_NUMNODES-1 here, but NUMA_NO_NODE can be
passed in as an argument to alloc_pages_bulk_array_mempolicy, like here:
vm_area_alloc_pages()
if (IS_ENABLED(CONFIG_NUMA) && nid == NUMA_NO_NODE)
nr = alloc_pages_bulk_array_mempolicy(bulk_gfp,
nr_pages_request,
pages + nr_allocated);
> > (cur_weight = 0) can happen in two scenarios:
> > - initial setting of mempolicy (NUMA_NO_NODE w/ cur_weight=0)
> > - weighted_interleave_nodes decrements it down to 0
> >
> > Now that i'm looking at it - the second condition should not exist, and
> > we can eliminate it. The logic in weighted_interleave_nodes is actually
> > annoyingly unclear at the moment, so I'm going to re-factor it a bit to
> > be more explicit.
>
> I am OK with either way. Just a reminder, the first condition may be
> true in alloc_pages_bulk_array_weighted_interleave() and perhaps some
> other places.
>
Yeah, the bulk allocator handles it correctly, it's just a matter of
clarity for weighted_interleave_nodes.
What isn't necessarily handled correctly is the rebind code. Rebind due
to a cgroup/mems_allowed change can cause a stale weight to be carried.
Basically cur_weight is not cleared, but the node it applied to may no
longer be the next node when next_node_in() is called.
The race condition is 1) exceedingly rare, and 2) not necessarily harmful,
just inaccurate. The worst case scenario is that a node receives up to 255
additional allocations once after a rebind (but more likely 10-20).
I was considering forcing the interleave forward like this:
@@ -356,6 +361,10 @@ static void mpol_rebind_nodemask(struct mempolicy *pol, const nodemask_t *nodes)
tmp = *nodes;
pol->nodes = tmp;
+
+ /* Weighted interleave policies are forced forward to the next node */
+ if (pol->mode & MPOL_WEIGHTED_INTERLEAVE)
+ pol->wil.cur_weight = 0;
}
But this creates 2 race conditions when we read cur_weight and nodemask
in the allocator path.
Example 1:
1) bulk allocator READ_ONCE(mask), READ_ONCE(cur_weight)
2) rebind changes nodemask and { cur_weight = 0; }
3) bulk allocator sets pol->wil.cur_weight
In this scenario, resume_weight is stale coming out of bulk allocations
if the resume_node has been removed from the node mask.
Example 2:
1) rebind changes nodemask
2) bulk allocator READ_ONCE(mask), READ_ONCE(cur_weight)
3) rebind sets { cur_weight = 0; }
In this scenario, cur_weight is stale going into bulk allocations.
Neither of these can force a violation of mems_allowed, just a
mis-application of a weight.
I'll need to think on this a bit. We can either leave this as-is,
meaning the first allocation after a rebind may apply the wrong weight
to a node, or we can try to track the current-interleave-node and
validate next_node_in(mask) == current-interleave-node before leaving
the allocator path (this may also be just as racey).
turns out concurrent counting is still hard :]
~Gregory