diff options
| author | Adam Li <adamli@os.amperecomputing.com> | 2024-02-26 05:24:52 +0300 |
|---|---|---|
| committer | Greg Kroah-Hartman <gregkh@linuxfoundation.org> | 2024-05-02 17:35:17 +0300 |
| commit | e830c804e26733fb09219cb9b92f167715c3b8a0 (patch) | |
| tree | 8418cd66a0c7c0b5772a243f87a15af19c78593a /Documentation | |
| parent | 997c7abd551007b25607d28c242be05923c640ad (diff) | |
| download | linux-e830c804e26733fb09219cb9b92f167715c3b8a0.tar.xz | |
net: make SK_MEMORY_PCPU_RESERV tunable
[ Upstream commit 12a686c2e761f1f1f6e6e2117a9ab9c6de2ac8a7 ]
This patch adds /proc/sys/net/core/mem_pcpu_rsv sysctl file,
to make SK_MEMORY_PCPU_RESERV tunable.
Commit 3cd3399dd7a8 ("net: implement per-cpu reserves for
memory_allocated") introduced per-cpu forward alloc cache:
"Implement a per-cpu cache of +1/-1 MB, to reduce number
of changes to sk->sk_prot->memory_allocated, which
would otherwise be cause of false sharing."
sk_prot->memory_allocated points to global atomic variable:
atomic_long_t tcp_memory_allocated ____cacheline_aligned_in_smp;
If increasing the per-cpu cache size from 1MB to e.g. 16MB,
changes to sk->sk_prot->memory_allocated can be further reduced.
Performance may be improved on system with many cores.
Signed-off-by: Adam Li <adamli@os.amperecomputing.com>
Reviewed-by: Christoph Lameter (Ampere) <cl@linux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stable-dep-of: 3584718cf2ec ("net: fix sk_memory_allocated_{add|sub} vs softirqs")
Signed-off-by: Sasha Levin <sashal@kernel.org>
Diffstat (limited to 'Documentation')
| -rw-r--r-- | Documentation/admin-guide/sysctl/net.rst | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/Documentation/admin-guide/sysctl/net.rst b/Documentation/admin-guide/sysctl/net.rst index 396091651955..7250c0542828 100644 --- a/Documentation/admin-guide/sysctl/net.rst +++ b/Documentation/admin-guide/sysctl/net.rst @@ -206,6 +206,11 @@ Will increase power usage. Default: 0 (off) +mem_pcpu_rsv +------------ + +Per-cpu reserved forward alloc cache size in page units. Default 1MB per CPU. + rmem_default ------------ |