backporting bcache-testing to kernel 3.4

backporting bcache-testing to kernel 3.4

[Dongsu Park]

Hi,

has anyone succeeded backporting bcache(-testing) branch to kernel 3.4?

I tried to get it working, patching commits from bcache-testing branch
as well as block patches by Kent, on top of Kernel 3.4.23.

* writeback mode
* backing device: MD-RAID0, or a HDD
* cache device: loopback device on tmpfs (for test)
* testing command:
   dd if=/dev/zero of=/dev/bcache0 bs=4K count=1M oflag=sync

Everything works fine, except for a problem of sync operation.
Every write request is being immediately flushed into backing device.
That's actually what I want to avoid in writeback caches.
Observing block I/Os with iostat, it's weird that no data is actually
written to backing device, despite of high IOPs. e.g.:

Device:  rrqm/s   wrqm/s  r/s     w/s    rMB/s    wMB/s
avgrq-sz avgqu-sz   await r_await w_await svctm  %util
vdb      0.00     0.00    0.00 2550.00    0.00     0.00
0.00     2.05       0.81  0.00    0.81    0.26   67.00

If I test bcache-testing branch based on kernel 3.9-rc3, which means
without backporting, then it works well without such massive bogus syncs.
bcache-3.2 is also fine. So I suppose there might be any change on the
block layer between 3.4 and 3.9, making bcache work in such a strange way.

Any idea?

Dongsu

Re: backporting bcache-testing to kernel 3.4

[Kent Overstreet]

This sounds like the bug I _think_ I just fixed - Do you suppose you
could try again with the current bcache branch?

On Mon, Apr 22, 2013 at 12:21 PM, Dongsu Park
<dongsu.park-EIkl63zCoXaH+58JC4qpiA@public.gmane.org> wrote:
> Hi,
>
> has anyone succeeded backporting bcache(-testing) branch to kernel 3.4?
>
> I tried to get it working, patching commits from bcache-testing branch
> as well as block patches by Kent, on top of Kernel 3.4.23.
>
> * writeback mode
> * backing device: MD-RAID0, or a HDD
> * cache device: loopback device on tmpfs (for test)
> * testing command:
>    dd if=/dev/zero of=/dev/bcache0 bs=4K count=1M oflag=sync
>
> Everything works fine, except for a problem of sync operation.
> Every write request is being immediately flushed into backing device.
> That's actually what I want to avoid in writeback caches.
> Observing block I/Os with iostat, it's weird that no data is actually
> written to backing device, despite of high IOPs. e.g.:
>
> Device:  rrqm/s   wrqm/s  r/s     w/s    rMB/s    wMB/s
> avgrq-sz avgqu-sz   await r_await w_await svctm  %util
> vdb      0.00     0.00    0.00 2550.00    0.00     0.00
> 0.00     2.05       0.81  0.00    0.81    0.26   67.00
>
> If I test bcache-testing branch based on kernel 3.9-rc3, which means
> without backporting, then it works well without such massive bogus syncs.
> bcache-3.2 is also fine. So I suppose there might be any change on the
> block layer between 3.4 and 3.9, making bcache work in such a strange way.
>
> Any idea?
>
> Dongsu
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html