On Wed, Jul 30, 2014 at 04:34:54PM +0200, Jean-Pierre André wrote:
> Hi,
> 
> I cannot really check the logic for the maximum trimming size
> which can be accepted by the lower layers.
> 
> I am checking on a ntfs loop device over ext3, and I see the
> trimmings truncated to a variable length within the same
> device. I even have seen different maximum lengths when
> repeating the same test on a different (but identical) file.
> 
> I guess this is related to the fragmentation of the underlying
> file system (ext3), with trimming truncated to ends of
> fragments. The loop device apparently does not reflect the
> limitation imposed by the underlying file system (it probably
> cannot, anyway, if this is variable).
> 
> Now, a loop device is not a meaningful target for ntfs-3g.
> Are you confident your length limitation would be fine
> on a SSD ?

loop has a discard_max_bytes of 4G, so the code to limit the size of
discards in ntfs-3g is never invoked.  As you say the effective amount
that can be discarded probably also depends on the ext3 block size.
Also on bugs/races in the kernel, since the trimming code in the
kernel is new and not much tested -- we have found many problems with
it when writing virt-sparsify.

You can emulate smaller discard sizes by inserting:

  *discard_granularity = 256 * 4096;

in fstrim_limits.

v5 will fix a few more bugs -- posting in a minute.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-builder quickly builds VMs from scratch
http://libguestfs.org/virt-builder.1.html

------------------------------------------------------------------------------
Infragistics Professional
Build stunning WinForms apps today!
Reboot your WinForms applications with our WinForms controls. 
Build a bridge from your legacy apps to the future.
http://pubads.g.doubleclick.net/gampad/clk?id=153845071&iu=/4140/ostg.clktrk
_______________________________________________
ntfs-3g-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ntfs-3g-devel

Reply via email to