On 08/28/2010 03:46 AM, Alex Elder wrote:
> On Thu, 2010-06-17 at 16:53 +0800, Tao Ma wrote:
>> > Hi Dave,
>> > On 06/14/2010 08:29 PM, Dave Chinner wrote:
>>> > > I just had a thought - if you want to avoid holes being reported to
>>> > > fiemap, then add a BMV_IF_NO_HOLES flag to xfs_getbmap() and skip
>>> > > holes in the mappin gloop when this flag is set. That will make
>>> > > fiemap fill in the full number of extents without hacking the
>>> > > extent count...
>> > Here is the updated one. I have used BVM_IF_NO_HOLES in xfs_getbmap
>> > to skip increasing index 'cur_ext'. It is a bit ugly, see my commit
>> > log. I guess maybe we can add another flag in xfs_bmapi so that it
>> > don't even give us the holes?
> Dave said he would commit this but it hasn't been done
> yet, so there's still time to comment!
yeah, it is never late. ;)
>
> . . .
>
>> > diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
>> > index 99587de..d49107d 100644
>> > --- a/fs/xfs/xfs_bmap.c
>> > +++ b/fs/xfs/xfs_bmap.c
>> > @@ -5744,12 +5744,24 @@ xfs_getbmap(
>> > map[i].br_startblock))
>> > goto out_free_map;
>> >
>> > - nexleft--;
>> > bmv->bmv_offset =
>> > out[cur_ext].bmv_offset +
>> > out[cur_ext].bmv_length;
>> > bmv->bmv_length =
>> > max_t(__int64_t, 0, bmvend -
>> > bmv->bmv_offset);
>> > +
>> > + /*
>> > + * In case we don't want to return the hole,
>> > + * don't increase cur_ext so that we can reuse
>> > + * it in the next loop.
>> > + */
>> > + if ((iflags& BMV_IF_NO_HOLES)&&
>> > + out[cur_ext].bmv_block == -1LL) {
> I personally would prefer this to use:
> if ((iflags& BMV_IF_NO_HOLES)&&
> map[i].br_startblock == HOLESTARTBLOCK) {
>
> I think this is more obvious that this "extent" represents
> a hole (as opposed to bmv_block == -1, which is how
> xfs_getbmapx_fix_eof_hole() encodes this).
>
> Otherwise looks good.
OK, I have updated the patch and attached it. So could you please send
it to Linus?
Regards,
Tao
>From 09a3cbaa456b35ed071cdd08a8b3c4c129baf451 Mon Sep 17 00:00:00 2001
From: Tao Ma <tao.ma@xxxxxxxxxx>
Date: Mon, 30 Aug 2010 10:04:51 +0800
Subject: [PATCH v4] xfs: Make fiemap works with sparse file.
In xfs_vn_fiemap, we set bvm_count to fi_extent_max + 1 and want
to return fi_extent_max extents, but actually it won't work for
a sparse file. The reason is that in xfs_getbmap we will
calculate holes and set it in 'out', while out is malloced by
bmv_count(fi_extent_max+1) which didn't consider holes. So in the
worst case, if 'out' vector looks like
[hole, extent, hole, extent, hole, ... hole, extent, hole],
we will only return half of fi_extent_max extents.
This patch add a new parameter BMV_IF_NO_HOLES for bvm_iflags.
So with this flags, we don't use our 'out' in xfs_getbmap for
a hole. The solution is a bit ugly by just don't increasing
index of 'out' vector. I felt that it is not easy to skip it
at the very beginning since we have the complicated check and
some function like xfs_getbmapx_fix_eof_hole to adjust 'out'.
Cc: Dave Chinner <david@xxxxxxxxxxxxx>
Reviewed-by: Alex Elder <aelder@xxxxxxx>
Signed-off-by: Tao Ma <tao.ma@xxxxxxxxxx>
---
fs/xfs/linux-2.6/xfs_iops.c | 2 +-
fs/xfs/xfs_bmap.c | 14 +++++++++++++-
fs/xfs/xfs_fs.h | 4 +++-
3 files changed, 17 insertions(+), 3 deletions(-)
diff --git a/fs/xfs/linux-2.6/xfs_iops.c b/fs/xfs/linux-2.6/xfs_iops.c
index 68be25d..b1fc2a6 100644
--- a/fs/xfs/linux-2.6/xfs_iops.c
+++ b/fs/xfs/linux-2.6/xfs_iops.c
@@ -664,7 +664,7 @@ xfs_vn_fiemap(
fieinfo->fi_extents_max + 1;
bm.bmv_count = min_t(__s32, bm.bmv_count,
(PAGE_SIZE * 16 / sizeof(struct getbmapx)));
- bm.bmv_iflags = BMV_IF_PREALLOC;
+ bm.bmv_iflags = BMV_IF_PREALLOC | BMV_IF_NO_HOLES;
if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR)
bm.bmv_iflags |= BMV_IF_ATTRFORK;
if (!(fieinfo->fi_flags & FIEMAP_FLAG_SYNC))
diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c
index 23f14e5..f90dadd 100644
--- a/fs/xfs/xfs_bmap.c
+++ b/fs/xfs/xfs_bmap.c
@@ -5533,12 +5533,24 @@ xfs_getbmap(
map[i].br_startblock))
goto out_free_map;
- nexleft--;
bmv->bmv_offset =
out[cur_ext].bmv_offset +
out[cur_ext].bmv_length;
bmv->bmv_length =
max_t(__int64_t, 0, bmvend - bmv->bmv_offset);
+
+ /*
+ * In case we don't want to return the hole,
+ * don't increase cur_ext so that we can reuse
+ * it in the next loop.
+ */
+ if ((iflags & BMV_IF_NO_HOLES) &&
+ map[i].br_startblock == HOLESTARTBLOCK) {
+ memset(&out[cur_ext], 0, sizeof(out[cur_ext]));
+ continue;
+ }
+
+ nexleft--;
bmv->bmv_entries++;
cur_ext++;
}
diff --git a/fs/xfs/xfs_fs.h b/fs/xfs/xfs_fs.h
index 7cf7220..87c2e9d 100644
--- a/fs/xfs/xfs_fs.h
+++ b/fs/xfs/xfs_fs.h
@@ -114,8 +114,10 @@ struct getbmapx {
#define BMV_IF_NO_DMAPI_READ 0x2 /* Do not generate DMAPI read event */
#define BMV_IF_PREALLOC 0x4 /* rtn status BMV_OF_PREALLOC
if req */
#define BMV_IF_DELALLOC 0x8 /* rtn status BMV_OF_DELALLOC
if req */
+#define BMV_IF_NO_HOLES 0x10 /* Do not return holes */
#define BMV_IF_VALID \
- (BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC|BMV_IF_DELALLOC)
+ (BMV_IF_ATTRFORK|BMV_IF_NO_DMAPI_READ|BMV_IF_PREALLOC| \
+ BMV_IF_DELALLOC|BMV_IF_NO_HOLES)
/* bmv_oflags values - returned for each non-header segment */
#define BMV_OF_PREALLOC 0x1 /* segment = unwritten
pre-allocation */
--
1.7.1.571.gba4d01
|