On Tue, Jul 26, 2011 at 11:44:37AM +0800, Osier Yang wrote:
于 2011年07月26日 09:48, Dave Allan 写道:
>Fixes bz 639591
>---
> tools/virsh.pod | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
>diff --git a/tools/virsh.pod b/tools/virsh.pod
>index 5b7fa9c..dd3457c 100644
>--- a/tools/virsh.pod
>+++ b/tools/virsh.pod
>@@ -696,7 +696,7 @@ If I<--live> is specified, set scheduler information of a
running guest.
> If I<--config> is specified, affect the next boot of a persistent guest.
> If I<--current> is specified, affect the current guest state.
>
>-B<Note>: The cpu_shares parameter has a valid value range of 0-262144;
Negative
>+B<Note>: The cpu_shares parameter has a valid value range of 2-262144;
Negative
> values are wrapped to positive, and larger values are capped at the maximum.
> Therefore, -1 is a useful shorthand for 262144.
>
What Linda commented in the bug is wrong.
# echo 2048 > /sys/fs/cgroup/cpu/cpu.shares
-bash: echo: write error: Invalid argument
# echo 4 > /sys/fs/cgroup/cpu/cpu.shares
-bash: echo: write error: Invalid argument
The error is not from cgroup. Actually it doesn't report
any error and silently set cpu.shares to 2 if the value
one passed is lower than 2.
But I guess kernel will not think it as a bug, what we can do
is improving the doc. But I guess the improvement in this
patch doesn't tell the whole truth.
# virsh schedinfo test --set cpu_shares=0
Scheduler : posix
cpu_shares : 2
Only saying the valid range is 0-2621244 doesn't explain why we
still allows 0,
but no error, and the cpu.shares is silently set to "2".
Regards
Osier
Ok, then self-NAK until we get this sorted out.
Dave