[ICX] [SPR] [ipc/msg] performance: Mitigate the lock contention with percpu counter

This is a public version of https://bugs.launchpad.net/bugs/2034980

Backport: ipc/msg: mitigate the lock contention with percpu counter (merged upstream in 6.1) to jammy

[Impact]

The msg_bytes and msg_hdrs atomic counters are frequently updated when IPC msg queue is in heavy use, causing heavy
cache bounce and overhead. Change them to percpu_counter greatly improve the performance. Since there is one percpu
struct per namespace, additional memory cost is minimal. Reading of the count done in msgctl call, which is infrequent.
So the need to sum up the counts in each CPU is infrequent.

[Fix]

Backport:
  72d1e611082e ipc/msg: mitigate the lock contention with percpu counter

For clean backport/build, those are also required:

5d0ce3595ab75 percpu: add percpu_counter_add_local and percpu_counter_sub_local
38cd5b12b7854 ipc: Remove extra braces
0889f44e28103 ipc: Check permissions for checkpoint_restart sysctls at open time
dd141a4955d5e ipc: Remove extra1 field abuse to pass ipc namespace
def7343ff03bb ipc: Use the same namespace to modify and validate
1f5c135ee509e ipc: Store ipc sysctls in the ipc namespace
dc55e35f9e810 ipc: Store mqueue sysctls in the ipc namespace
0e9beb8a96f21 ipc/ipc_sysctl.c: remove fallback for !CONFIG_PROC_SYSCTL
5563cabdde7ee ipc: check checkpoint_restore_ns_capable() to modify C/R proc files

[Test Plan]

Test as the original patch, with pts/stress-ng message passing
and compare performance.

[Where problems could occur]

Performance regression in IPC communication/workload.