[PATCH 1/2] erofs: add support for the full decompressed lengthy
Gao Xiang
hsiangkao at linux.alibaba.com
Tue Aug 17 10:54:14 AEST 2021
On Mon, Aug 16, 2021 at 11:38:03PM +0800, Chao Yu wrote:
> On 2021/8/13 13:29, Gao Xiang wrote:
> > Previously, there is no need to get the full decompressed length since
> > EROFS supports partial decompression. However for some other cases
> > such as fiemap, the full decompressed length is necessary for iomap to
> > make it work properly.
> >
> > This patch adds a way to get the full decompressed length. Note that
> > it takes more metadata overhead and it'd be avoided if possible in the
> > performance sensitive scenario.
> >
> > Signed-off-by: Gao Xiang <hsiangkao at linux.alibaba.com>
> > ---
> > fs/erofs/internal.h | 5 +++
> > fs/erofs/zmap.c | 93 +++++++++++++++++++++++++++++++++++++++++----
> > 2 files changed, 90 insertions(+), 8 deletions(-)
> >
> > diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
> > index 25b094085ca6..2a05b09e1c06 100644
> > --- a/fs/erofs/internal.h
> > +++ b/fs/erofs/internal.h
> > @@ -356,6 +356,11 @@ struct erofs_map_blocks {
> > /* Flags used by erofs_map_blocks_flatmode() */
> > #define EROFS_GET_BLOCKS_RAW 0x0001
> > +/*
> > + * Used to get the exact decompressed length, e.g. fiemap (consider lookback
> > + * approach instead if possible since it's quite metadata expensive.)
> > + */
> > +#define EROFS_GET_BLOCKS_FIEMAP 0x0002
> > /* zmap.c */
> > #ifdef CONFIG_EROFS_FS_ZIP
> > diff --git a/fs/erofs/zmap.c b/fs/erofs/zmap.c
> > index f68aea4baed7..12256ef12819 100644
> > --- a/fs/erofs/zmap.c
> > +++ b/fs/erofs/zmap.c
> > @@ -212,9 +212,32 @@ static unsigned int decode_compactedbits(unsigned int lobits,
> > return lo;
> > }
> > +static int get_compacted_la_distance(unsigned int lclusterbits,
> > + unsigned int encodebits,
> > + unsigned int vcnt, u8 *in, int i)
> > +{
> > + const unsigned int lomask = (1 << lclusterbits) - 1;
> > + unsigned int lo, d1 = 0;
> > + u8 type;
> > +
> > + for (; i < vcnt; ++i) {
>
> for (di = 0; i < vcnt; ++i, ++d1)
Hi Chao,
Thanks for the review
this could cause lo potential uninitialized warning (actually it's
unpossible). So I resent [PATCH v1.1] as
https://lore.kernel.org/linux-erofs/20210814152727.78451-1-hsiangkao@linux.alibaba.com/
Please kindly help check out...
>
> > + lo = decode_compactedbits(lclusterbits, lomask,
> > + in, encodebits * i, &type);
> > +
> > + if (type != Z_EROFS_VLE_CLUSTER_TYPE_NONHEAD)
> > + return d1;
>
> [1]
>
> > + ++d1;
> > + }
> > +
> > + /* vcnt - 1 (Z_EROFS_VLE_CLUSTER_TYPE_NONHEAD) item */
> > + if (!(lo & Z_EROFS_VLE_DI_D0_CBLKCNT))
> > + d1 += lo - 1;
> > + return d1;
> > +}
> > +
> > static int unpack_compacted_index(struct z_erofs_maprecorder *m,
> > unsigned int amortizedshift,
> > - unsigned int eofs)
> > + unsigned int eofs, bool lookahead)
> > {
> > struct erofs_inode *const vi = EROFS_I(m->inode);
> > const unsigned int lclusterbits = vi->z_logical_clusterbits;
> > @@ -243,6 +266,11 @@ static int unpack_compacted_index(struct z_erofs_maprecorder *m,
> > m->type = type;
> > if (type == Z_EROFS_VLE_CLUSTER_TYPE_NONHEAD) {
> > m->clusterofs = 1 << lclusterbits;
> > +
> > + /* figure out lookahead_distance: delta[1] if needed */
> > + if (lookahead)
> > + m->delta[1] = get_compacted_la_distance(lclusterbits,
> > + encodebits, vcnt, in, i);
> > if (lo & Z_EROFS_VLE_DI_D0_CBLKCNT) {
> > if (!big_pcluster) {
> > DBG_BUGON(1);
> > @@ -313,7 +341,7 @@ static int unpack_compacted_index(struct z_erofs_maprecorder *m,
> > }
> > static int compacted_load_cluster_from_disk(struct z_erofs_maprecorder *m,
> > - unsigned long lcn)
> > + unsigned long lcn, bool lookahead)
> > {
> > struct inode *const inode = m->inode;
> > struct erofs_inode *const vi = EROFS_I(inode);
> > @@ -364,11 +392,12 @@ static int compacted_load_cluster_from_disk(struct z_erofs_maprecorder *m,
> > err = z_erofs_reload_indexes(m, erofs_blknr(pos));
> > if (err)
> > return err;
> > - return unpack_compacted_index(m, amortizedshift, erofs_blkoff(pos));
> > + return unpack_compacted_index(m, amortizedshift, erofs_blkoff(pos),
> > + lookahead);
> > }
> > static int z_erofs_load_cluster_from_disk(struct z_erofs_maprecorder *m,
> > - unsigned int lcn)
> > + unsigned int lcn, bool lookahead)
> > {
> > const unsigned int datamode = EROFS_I(m->inode)->datalayout;
> > @@ -376,7 +405,7 @@ static int z_erofs_load_cluster_from_disk(struct z_erofs_maprecorder *m,
> > return legacy_load_cluster_from_disk(m, lcn);
> > if (datamode == EROFS_INODE_FLAT_COMPRESSION)
> > - return compacted_load_cluster_from_disk(m, lcn);
> > + return compacted_load_cluster_from_disk(m, lcn, lookahead);
> > return -EINVAL;
> > }
> > @@ -399,7 +428,7 @@ static int z_erofs_extent_lookback(struct z_erofs_maprecorder *m,
> > /* load extent head logical cluster if needed */
> > lcn -= lookback_distance;
> > - err = z_erofs_load_cluster_from_disk(m, lcn);
> > + err = z_erofs_load_cluster_from_disk(m, lcn, false);
> > if (err)
> > return err;
> > @@ -450,7 +479,7 @@ static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
> > if (m->compressedlcs)
> > goto out;
> > - err = z_erofs_load_cluster_from_disk(m, lcn);
> > + err = z_erofs_load_cluster_from_disk(m, lcn, false);
> > if (err)
> > return err;
> > @@ -498,6 +527,48 @@ static int z_erofs_get_extent_compressedlen(struct z_erofs_maprecorder *m,
> > return -EFSCORRUPTED;
> > }
> > +static int z_erofs_get_extent_decompressedlen(struct z_erofs_maprecorder *m)
> > +{
> > + struct inode *inode = m->inode;
> > + struct erofs_inode *vi = EROFS_I(inode);
> > + struct erofs_map_blocks *map = m->map;
> > + unsigned int lclusterbits = vi->z_logical_clusterbits;
> > + u64 lcn = m->lcn, headlcn = map->m_la >> lclusterbits;
> > + int err;
> > +
> > + do {
> > + /* handle the last EOF pcluster (no next HEAD lcluster) */
> > + if ((lcn << lclusterbits) >= inode->i_size) {
> > + map->m_llen = inode->i_size - map->m_la;
> > + return 0;
> > + }
> > +
> > + err = z_erofs_load_cluster_from_disk(m, lcn, true);
> > + if (err)
> > + return err;
> > +
> > + if (m->type == Z_EROFS_VLE_CLUSTER_TYPE_NONHEAD) {
> > + if (!m->delta[1])
> > + DBG_BUGON(m->clusterofs != 1 << lclusterbits);
>
> DBG_BUGON(!m->delta[1] &&
> m->clusterofs != 1 << lclusterbits);
>
Ok, will update.
> > + } else if (m->type == Z_EROFS_VLE_CLUSTER_TYPE_PLAIN ||
> > + m->type == Z_EROFS_VLE_CLUSTER_TYPE_HEAD) {
> > + /* go on until the next HEAD lcluster */
> > + if (lcn != headlcn)
> > + break;
> > + m->delta[1] = 1;
>
> If I didn't miss anything, return value [1] of get_compacted_la_distance()
> won't be used anyway here? right?
These are two different things, first, return value [1] is used to calculate
length to the next lcluster, e.g.
NONHEAD NONHEAD HEAD
^
m->lcn
so here in the first loop, delta[1] = 2, and m->type == NONHEAD,
the second loop, m->type == HEAD, lcn != headlcn, so we are done.
Yet here the logic is instead
if m->lcn == m->headlcn, so m->lcn points to this lcluster HEAD, but we
need to find the next lcluster, e.g.
HEAD NONHEAD NONHEAD NONHEAD HEAD
^ ^
m->lcn, m->headlcn we need to get this
so we lcn == headlcn, we need to go on and get the next HEAD lcluster
one.
It will use since m->lcn could be originally equal to headlcn (point to
this HEAD lcluster)
Thanks,
Gao Xiang
>
> Thanks,
>
> > + } else {
> > + erofs_err(inode->i_sb, "unknown type %u @ lcn %llu of nid %llu",
> > + m->type, lcn, vi->nid);
> > + DBG_BUGON(1);
> > + return -EOPNOTSUPP;
> > + }
> > + lcn += m->delta[1];
> > + } while (m->delta[1]);
> > +
> > + map->m_llen = (lcn << lclusterbits) + m->clusterofs - map->m_la;
> > + return 0;
> > +}
> > +
> > int z_erofs_map_blocks_iter(struct inode *inode,
> > struct erofs_map_blocks *map,
> > int flags)
> > @@ -531,7 +602,7 @@ int z_erofs_map_blocks_iter(struct inode *inode,
> > initial_lcn = ofs >> lclusterbits;
> > endoff = ofs & ((1 << lclusterbits) - 1);
> > - err = z_erofs_load_cluster_from_disk(&m, initial_lcn);
> > + err = z_erofs_load_cluster_from_disk(&m, initial_lcn, false);
> > if (err)
> > goto unmap_out;
> > @@ -581,6 +652,12 @@ int z_erofs_map_blocks_iter(struct inode *inode,
> > err = z_erofs_get_extent_compressedlen(&m, initial_lcn);
> > if (err)
> > goto out;
> > +
> > + if (flags & EROFS_GET_BLOCKS_FIEMAP) {
> > + err = z_erofs_get_extent_decompressedlen(&m);
> > + if (!err)
> > + map->m_flags |= EROFS_MAP_FULL_MAPPED;
> > + }
> > unmap_out:
> > if (m.kaddr)
> > kunmap_atomic(m.kaddr);
> >
More information about the Linux-erofs
mailing list