2009-11-03 04:29:05

by Rusty Russell

[permalink] [raw]
Subject: [PATCH 14/14] cpumask: don't recommend set_cpus_allowed hack in Documentation/cpu-hotplug.txt

Signed-off-by: Rusty Russell <[email protected]>
Cc: Gautham R Shenoy <[email protected]>
Cc: Ashok Raj <[email protected]>
---
Documentation/cpu-hotplug.txt | 49 ++++++++++++++----------------------------
1 file changed, 17 insertions(+), 32 deletions(-)

diff --git a/Documentation/cpu-hotplug.txt b/Documentation/cpu-hotplug.txt
--- a/Documentation/cpu-hotplug.txt
+++ b/Documentation/cpu-hotplug.txt
@@ -309,41 +309,26 @@ A: The following are what is required fo

Q: I need to ensure that a particular cpu is not removed when there is some
work specific to this cpu is in progress.
-A: First switch the current thread context to preferred cpu
+A: There are two ways. If your code can be run in interrupt context, use
+ smp_call_function_single(), otherwise use work_on_cpu(). Note that
+ work_on_cpu() can fail due to out of memory:

int my_func_on_cpu(int cpu)
{
- cpumask_t saved_mask, new_mask = CPU_MASK_NONE;
- int curr_cpu, err = 0;
-
- saved_mask = current->cpus_allowed;
- cpu_set(cpu, new_mask);
- err = set_cpus_allowed(current, new_mask);
-
- if (err)
- return err;
-
- /*
- * If we got scheduled out just after the return from
- * set_cpus_allowed() before running the work, this ensures
- * we stay locked.
- */
- curr_cpu = get_cpu();
-
- if (curr_cpu != cpu) {
- err = -EAGAIN;
- goto ret;
- } else {
- /*
- * Do work : But cant sleep, since get_cpu() disables preempt
- */
- }
- ret:
- put_cpu();
- set_cpus_allowed(current, saved_mask);
- return err;
- }
-
+ int err;
+ cpu_hotplug_lock();
+ if (!cpu_online(cpu))
+ err = -EINVAL;
+ else
+#if NEEDS_BLOCKING
+ err = work_on_cpu(cpu, __my_func_on_cpu, NULL);
+#else
+ smp_call_function_single(cpu, __my_func_on_cpu, &err,
+ true);
+#endif
+ cpu_hotplug_unlock();
+ return err;
+ }

Q: How do we determine how many CPUs are available for hotplug.
A: There is no clear spec defined way from ACPI that can give us that


Subject: Re: [PATCH 14/14] cpumask: don't recommend set_cpus_allowed hack in Documentation/cpu-hotplug.txt

Hi Rusty,

On Tue, Nov 03, 2009 at 02:59:06PM +1030, Rusty Russell wrote:
> Signed-off-by: Rusty Russell <[email protected]>
> Cc: Gautham R Shenoy <[email protected]>
> Cc: Ashok Raj <[email protected]>
> ---
> Documentation/cpu-hotplug.txt | 49 ++++++++++++++----------------------------
> 1 file changed, 17 insertions(+), 32 deletions(-)
>
> diff --git a/Documentation/cpu-hotplug.txt b/Documentation/cpu-hotplug.txt
> --- a/Documentation/cpu-hotplug.txt
> +++ b/Documentation/cpu-hotplug.txt
> @@ -309,41 +309,26 @@ A: The following are what is required fo
>
> Q: I need to ensure that a particular cpu is not removed when there is some
> work specific to this cpu is in progress.
> -A: First switch the current thread context to preferred cpu
> +A: There are two ways. If your code can be run in interrupt context, use
> + smp_call_function_single(), otherwise use work_on_cpu(). Note that
> + work_on_cpu() can fail due to out of memory:
>
> int my_func_on_cpu(int cpu)
> {
> + int err;
> + cpu_hotplug_lock();
^^^^^^^^^^^^^^^^^^^
This needs to be get_online_cpus();
> + if (!cpu_online(cpu))
> + err = -EINVAL;
> + else
> +#if NEEDS_BLOCKING
> + err = work_on_cpu(cpu, __my_func_on_cpu, NULL);
> +#else
> + smp_call_function_single(cpu, __my_func_on_cpu, &err,
> + true);
> +#endif
> + cpu_hotplug_unlock();
^^^^^^^^^^^^^^^^^^^^^
And this one should be put_online_cpus();

> + return err;
> + }
>
> Q: How do we determine how many CPUs are available for hotplug.
> A: There is no clear spec defined way from ACPI that can give us that

--
Thanks and Regards
gautham

2009-11-04 08:02:34

by Rusty Russell

[permalink] [raw]
Subject: Re: [PATCH 14/14] cpumask: don't recommend set_cpus_allowed hack in Documentation/cpu-hotplug.txt

On Tue, 3 Nov 2009 04:27:23 pm Gautham R Shenoy wrote:
> Hi Rusty,
>
> On Tue, Nov 03, 2009 at 02:59:06PM +1030, Rusty Russell wrote:
...
> > + int err;
> > + cpu_hotplug_lock();
> ^^^^^^^^^^^^^^^^^^^
> This needs to be get_online_cpus();

Thanks, both fixed!

Rusty.