Re: [PATCH RT] drop migrate_disabled section in _cpu_down

From
Steven Rostedt <rostedt@goodmis.org>
Date
2014-03-24 14:20:33
Also in
lkml
Reply to
Nicholas Mc Guire, 2014-03-24
On Mon, 24 Mar 2014 13:20:25 +0100
Nicholas Mc Guire [off-list ref] wrote:
HI !


Minor cleanup: drop unnecessary migrate_disable in _cpu_down

in kernel/cpu.c:_cpu_down
    ...
    cpumask_andnot(cpumask, cpu_online_mask, cpumask_of(cpu));
    set_cpus_allowed_ptr(current, cpumask);
    ...
    migrate_disable();
    mycpu = smp_processor_id();
    if (mycpu == cpu) {
            printk(KERN_ERR "Yuck! Still on unplug CPU\n!");
            migrate_enable();
            err = -EBUSY;
            goto restore_cpus;
    }
    migrate_enable();

If we got migrated off "cpu" then we can not get migrate
back as the cpus_allowed was updated. so if smp_processor_id(); returns
anything != cpu we are fine even if migrated off to a third cpu between
the mycpu = smp_processor_id() and the if (mycpu == cpu). No need for
migration protection here.

Further the migrate disable is effectively only needed for the condition
check in the if statement - so I think it could be replaced by:

    if (cpu==smp_processor_id()) {
            pr_err("Yuck! Still on unplug CPU\n!");
            err = -EBUSY;
            goto restore_cpus;
    }

even in the case that the cpus_allowed had not been updated.

Patch is against 3.12.13-rt21
This is checking for an anomaly, and it's in a very slow path. I feel
more comfortable with keeping the code as is, than to change it for a
micro optimization.

-- Steve

Signed-off-by: Nicholas Mc Guire <redacted>
Thread (2 messages) 2 messages, 2 authors, last reply 2014-03-24
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help