Date: Tue, 04 Jun 2024 08:29:18 +0200
From: Joel Granados via B4 Relay <devnull+j.granados.samsung.com@...nel.org>
To: Peter Zijlstra <peterz@...radead.org>, Ingo Molnar <mingo@...hat.com>, 
 Will Deacon <will@...nel.org>, Waiman Long <longman@...hat.com>, 
 Boqun Feng <boqun.feng@...il.com>, Suren Baghdasaryan <surenb@...gle.com>, 
 Kent Overstreet <kent.overstreet@...ux.dev>, 
 Andrew Morton <akpm@...ux-foundation.org>, 
 Luis Chamberlain <mcgrof@...nel.org>, Kees Cook <keescook@...omium.org>, 
 Joel Granados <j.granados@...sung.com>, 
 "David S. Miller" <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>, 
 Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>
Cc: linux-kernel@...r.kernel.org, linux-mm@...ck.org, 
 linux-fsdevel@...r.kernel.org, netdev@...r.kernel.org
Subject: [PATCH 0/8] sysctl: Remove sentinel check from sysctl internals

From: Joel Granados <j.granados@...sung.com>

What?
Remove the loop stopping criteria check for ->procname == NULL, the
array size calculation based on sentinels and the superfluous sentinels
created within the sysctl infrastructure. None are needed as they are
now solely based on ctl_table ARRAY_SIZE. Finally, sentinels that have
been added in recent releases (not present for the original patchsets)
were removed.

Why?
By removing the sysctl sentinel elements we avoid kernel bloat as
ctl_table arrays get moved out of kernel/sysctl.c into their own
respective subsystems. This move was started long ago to avoid merge
conflicts; the sentinel removal bit is to avoid bloating the kernel by
one element as arrays moved out. It includes work in /arch [1], /dirver
[2], fs/ [3], kernel [4], net/ [5], mm/ [6], security/ [6], io_uring [6]
and other misc directories [6]. It will reduce the overall build time
size of the kernel and run time memory bloat by about ~64 bytes per
declared ctl_table array (more info here [0]).

Testing:
* Ran sysctl selftests (./tools/testing/selftests/sysctl/sysctl.sh)
* Ran this through 0-day with no errors or warnings

Savings in vmlinux:
  A total of 64 bytes per sentinel is saved after removal. Here is the
  aggregated savings for all the removal patchsets ([1,2,3,4,5,6]) for
  the x86_64 arch (actual savings will depend on kernel conf):
 |------|---------|-----------|-------|-----------|--------|---------|----------------|
 |dir   | arch[1] | driver[2] | fs[3] | kernel[4] | net[5] | misc[6] | Total(approx.) |
 |------|---------|-----------|-------|-----------|--------|---------|----------------|
 |Bytes | 192     | 2432      | 1920  | 1984      | 3976   | 963     |    11467       |
 |------|---------|-----------|-------|-----------|--------|---------|----------------|

Savings in allocated memory:
  The estimated savings during boot for config [3] are 6272 bytes. See
  [7] for how to measure it.

Comments/feedback greatly appreciated

Best
Joel

[0] Links Related to the ctl_table sentinel removal:
    * Good summaries from Luis:
      https://lore.kernel.org/all/ZO5Yx5JFogGi%2FcBo@bombadil.infradead.org/
      https://lore.kernel.org/all/ZMFizKFkVxUFtSqa@bombadil.infradead.org/
    * Patches adjusting sysctl register calls:
      https://lore.kernel.org/all/20230302204612.782387-1-mcgrof@kernel.org/
      https://lore.kernel.org/all/20230302202826.776286-1-mcgrof@kernel.org/
    * Discussions about expectations and approach
      https://lore.kernel.org/all/20230321130908.6972-1-frank.li@vivo.com
      https://lore.kernel.org/all/20220220060626.15885-1-tangmeng@uniontech.com

[1] https://lore.kernel.org/20231002-jag-sysctl_remove_empty_elem_arch-v3-0-606da2840a7a@samsung.com
[2] https://lore.kernel.org/20231002-jag-sysctl_remove_empty_elem_drivers-v2-0-02dd0d46f71e@samsung.com
[3] https://lore.kernel.org/20231121-jag-sysctl_remove_empty_elem_fs-v2-0-39eab723a034@samsung.com
[4] https://lore.kernel.org/20240328-jag-sysctl_remove_empty_elem_kernel-v3-0-285d273912fe@samsung.com
[5] https://lore.kernel.org/20240501-jag-sysctl_remset_net-v6-0-370b702b6b4a@samsung.com
[6] https://lore.kernel.org/20240328-jag-sysctl_remset_misc-v1-0-47c1463b3af2@samsung.com

[7]
To measure the in memory savings apply this on top of this patchset.
"
diff --git i/fs/proc/proc_sysctl.c w/fs/proc/proc_sysctl.c
index a6aeaa928dd2..6ca5341bcddf 100644
--- i/fs/proc/proc_sysctl.c
+++ w/fs/proc/proc_sysctl.c
@@ -963,6 +963,7 @@ static struct ctl_dir *new_dir(struct ctl_table_set *set,
        table[0].procname = new_name;
        table[0].mode = S_IFDIR|S_IRUGO|S_IXUGO;
        init_header(&new->header, set->dir.header.root, set, node, table, 1);
+       printk("%ld sysctl saved mem kzalloc\n", sizeof(struct ctl_table));

        return new;
 }
@@ -1186,6 +1187,7 @@ static struct ctl_table_header *new_links(struct ctl_dir *dir, struct ctl_table_
                link_name += len;
                link++;
        }
+       printk("%ld sysctl saved mem kzalloc\n", sizeof(struct ctl_table));
        init_header(links, dir->header.root, dir->header.set, node, link_table,
                    head->ctl_table_size);
        links->nreg = head->ctl_table_size;
"
and then run the following bash script in the kernel:

accum=0
for n in $(dmesg | grep kzalloc | awk '{print $3}') ; do
    accum=$(calc "$accum + $n")
done
echo $accum

Signed-off-by: Joel Granados <j.granados@...sung.com>

---
Joel Granados (8):
      locking: Remove superfluous sentinel element from kern_lockdep_table
      mm profiling: Remove superfluous sentinel element from ctl_table
      sysctl: Remove check for sentinel element in ctl_table arrays
      sysctl: Replace nr_entries with ctl_table_size in new_links
      sysctl: Remove superfluous empty allocations from sysctl internals
      sysctl: Remove "child" sysctl code comments
      sysctl: Remove ctl_table sentinel code comments
      sysctl: Warn on an empty procname element

 fs/proc/proc_sysctl.c    | 50 +++++++++++++++++++++---------------------------
 kernel/locking/lockdep.c |  1 -
 lib/alloc_tag.c          |  1 -
 net/sysctl_net.c         | 11 ++---------
 4 files changed, 24 insertions(+), 39 deletions(-)
---
base-commit: 1613e604df0cd359cf2a7fbd9be7a0bcfacfabd0
change-id: 20240603-jag-sysctl_remset-4afb8c723003

Best regards,
-- 
Joel Granados <j.granados@...sung.com>

Powered by blists - more mailing lists