I tested dm-writecache performance on a machine with Optane nvdimm and it
turned out that for larger writes, cached stores + cache flushing perform
better than non-temporal stores. This is the throughput of dm-writecache
measured with this command:
dd if=/dev/zero of=/dev/mapper/wc bs=64 oflag=dire
Hi
I performed some microbenchmarks on a system with real Optane-based nvdimm
and I found out that the fastest method how to write to persistent memory
is to fill a cacheline with 8 8-byte writes and then issue clwb or
clflushopt on the cacheline. With this method, we can achieve 1.6 GB/s
thro
The use of mddev_unlock is out-sync. Some paths have
mddev_unlock after mddev_lock_nointr while some do not.
Since these paths are similar in semantic, there might
be a missing unlock bug.
Fix this by adding mddev_unlock in 3 paths.
Signed-off-by: Dinghao Liu
---
drivers/md/dm-raid.c | 9 +
On Tue, 24 Mar 2020 15:11:49 -0400
Mike Snitzer wrote:
> On Tue, Mar 24 2020 at 2:59pm -0400,
> Lukas Straub wrote:
>
> > On Tue, 24 Mar 2020 13:18:22 -0400
> > Mike Snitzer wrote:
> >
> > > On Thu, Feb 27 2020 at 8:26am -0500,
> > > Lukas Straub wrote:
> > >
> > > > If a full metadat
On Thu, Mar 26, 2020 at 11:31:45PM +, Alasdair G Kergon wrote:
> I'm experimenting with ways of recording changes to the udev database so
> you can look back at the history of the storage stack on a particular
> machine. This is still a work-in-progress, but it's reached a point
> where I'd li
On Thu, Mar 26, 2020 at 11:31:45PM +, Alasdair G Kergon wrote:
> I'm experimenting with ways of recording changes to the udev database so
> you can look back at the history of the storage stack on a particular
> machine. This is still a work-in-progress, but it's reached a point
> where I'd li
On Fri, Mar 27, 2020 at 08:53:29PM +, Alasdair G Kergon wrote:
> On Fri, Mar 27, 2020 at 03:50:06PM -0400, Brian McCullough wrote:
> > In your instructions, you say to put the "working" script in /usr/sbin,
> > while in the udev rule, it is in /sbin/
>
> In Fedora they are the same.
Ah. So t
On 03/26/2020 07:46 AM, Christoph Hellwig wrote:
> On Thu, Mar 26, 2020 at 10:34:42AM -0400, Martin K. Petersen wrote:
>>> I just worry about the proliferation of identical merging and
>>> splitting code throughout the block stack as we add additional
>>> single-range, no payload operations (Verify