On 02/24/2016 03:23 PM, Waiman Long wrote:
On 02/24/2016 03:28 AM, Jan Kara wrote:
On Tue 23-02-16 14:04:32, Waiman Long wrote:
When many threads are trying to add or delete inode to or from
a superblock's s_inodes list, spinlock contention on the list can
become a performance bottleneck.
This
On 02/25/2016 03:06 AM, Ingo Molnar wrote:
* Jan Kara wrote:
With an exit microbenchmark that creates a large number of threads,
attachs many inodes to them and then exits. The runtimes of that
microbenchmark with 1000 threads before and after the patch on a 4-socket
Intel E7-4820 v3 system (4
* Jan Kara wrote:
> > > > With an exit microbenchmark that creates a large number of threads,
> > > > attachs many inodes to them and then exits. The runtimes of that
> > > > microbenchmark with 1000 threads before and after the patch on a
> > > > 4-socket
> > > > Intel E7-4820 v3 system (40
On 02/24/2016 03:28 AM, Jan Kara wrote:
On Tue 23-02-16 14:04:32, Waiman Long wrote:
When many threads are trying to add or delete inode to or from
a superblock's s_inodes list, spinlock contention on the list can
become a performance bottleneck.
This patch changes the s_inodes field to become
On Wed 24-02-16 09:36:30, Ingo Molnar wrote:
>
> * Jan Kara wrote:
>
> > On Tue 23-02-16 14:04:32, Waiman Long wrote:
> > > When many threads are trying to add or delete inode to or from
> > > a superblock's s_inodes list, spinlock contention on the list can
> > > become a performance bottleneck
* Jan Kara wrote:
> On Tue 23-02-16 14:04:32, Waiman Long wrote:
> > When many threads are trying to add or delete inode to or from
> > a superblock's s_inodes list, spinlock contention on the list can
> > become a performance bottleneck.
> >
> > This patch changes the s_inodes field to become
On Tue 23-02-16 14:04:32, Waiman Long wrote:
> When many threads are trying to add or delete inode to or from
> a superblock's s_inodes list, spinlock contention on the list can
> become a performance bottleneck.
>
> This patch changes the s_inodes field to become a per-cpu list with
> per-cpu spi
When many threads are trying to add or delete inode to or from
a superblock's s_inodes list, spinlock contention on the list can
become a performance bottleneck.
This patch changes the s_inodes field to become a per-cpu list with
per-cpu spinlocks. As a result, the following superblock inode list
8 matches
Mail list logo