On 29/05/17(Mon) 14:30, Stefan Fritsch wrote:
> On Mon, 29 May 2017, Martin Pieuchot wrote:
>
> > A complete diff would be easier to review.
>
> ok, here it comes.
>
> The original commit message was this:
>
> Implement VFS read clustering for MSDOSFS.
>
> The logic used in msdosfs_bmap() to loop calling pcbmap() comes from
> FreeBSD and is not really efficient but it is good enough since it is
> only called when generating I/O.
>
> With this diff I get a 100% improvement when reading big files from a
> crappy USB stick.
>
> With this and bread_cluster(9) modified to not re-fetch B_CACHED
> buffers,
> reading large contiguous files with chunk sizes of MAXPHYS is almost as
> fast as physio(9) on the same device.
>
> For a 'real world' example, when copying music files from a USB stick I
> see a speed jump from 15MB/s on -current to 24Mb/s with this diff.
>
> While here rename some 'lbn' variables into 'cn' to better reflect what
> we're dealing with.
>
> Tested by Mathieu, with support from deraadt@
ok mpi@
Now I understood why I got confused between block & cluster, the current
msdosfs_bmap() function has a bug. ``ap->a_bn'' is representing a
cluster, not a block. Diff below fixes that, for the archives.
Index: msdosfs/msdosfs_vnops.c
===================================================================
RCS file: /cvs/src/sys/msdosfs/msdosfs_vnops.c,v
retrieving revision 1.113
diff -u -p -r1.113 msdosfs_vnops.c
--- msdosfs/msdosfs_vnops.c 30 Aug 2016 19:47:23 -0000 1.113
+++ msdosfs/msdosfs_vnops.c 29 May 2017 13:24:11 -0000
@@ -1765,7 +1765,6 @@ msdosfs_bmap(void *v)
{
struct vop_bmap_args *ap = v;
struct denode *dep = VTODE(ap->a_vp);
- struct msdosfsmount *pmp = dep->de_pmp;
if (ap->a_vpp != NULL)
*ap->a_vpp = dep->de_devvp;
@@ -1777,7 +1776,7 @@ msdosfs_bmap(void *v)
*/
*ap->a_runp = 0;
}
- return (pcbmap(dep, de_bn2cn(pmp, ap->a_bn), ap->a_bnp, 0, 0));
+ return (pcbmap(dep, ap->a_bn, ap->a_bnp, 0, 0));
}
int
