summaryrefslogtreecommitdiffstats
path: root/fs/nfs/pnfs.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/nfs/pnfs.c')
-rw-r--r--fs/nfs/pnfs.c178
1 files changed, 166 insertions, 12 deletions
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index f38813a0a29..f57f5281a52 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -259,6 +259,7 @@ put_lseg(struct pnfs_layout_segment *lseg)
pnfs_free_lseg_list(&free_me);
}
}
+EXPORT_SYMBOL_GPL(put_lseg);
static bool
should_free_lseg(u32 lseg_iomode, u32 recall_iomode)
@@ -382,6 +383,7 @@ pnfs_destroy_all_layouts(struct nfs_client *clp)
plh_layouts);
dprintk("%s freeing layout for inode %lu\n", __func__,
lo->plh_inode->i_ino);
+ list_del_init(&lo->plh_layouts);
pnfs_destroy_layout(NFS_I(lo->plh_inode));
}
}
@@ -465,19 +467,38 @@ pnfs_choose_layoutget_stateid(nfs4_stateid *dst, struct pnfs_layout_hdr *lo,
static struct pnfs_layout_segment *
send_layoutget(struct pnfs_layout_hdr *lo,
struct nfs_open_context *ctx,
- u32 iomode)
+ u32 iomode,
+ gfp_t gfp_flags)
{
struct inode *ino = lo->plh_inode;
struct nfs_server *server = NFS_SERVER(ino);
struct nfs4_layoutget *lgp;
struct pnfs_layout_segment *lseg = NULL;
+ struct page **pages = NULL;
+ int i;
+ u32 max_resp_sz, max_pages;
dprintk("--> %s\n", __func__);
BUG_ON(ctx == NULL);
- lgp = kzalloc(sizeof(*lgp), GFP_KERNEL);
+ lgp = kzalloc(sizeof(*lgp), gfp_flags);
if (lgp == NULL)
return NULL;
+
+ /* allocate pages for xdr post processing */
+ max_resp_sz = server->nfs_client->cl_session->fc_attrs.max_resp_sz;
+ max_pages = max_resp_sz >> PAGE_SHIFT;
+
+ pages = kzalloc(max_pages * sizeof(struct page *), gfp_flags);
+ if (!pages)
+ goto out_err_free;
+
+ for (i = 0; i < max_pages; i++) {
+ pages[i] = alloc_page(gfp_flags);
+ if (!pages[i])
+ goto out_err_free;
+ }
+
lgp->args.minlength = NFS4_MAX_UINT64;
lgp->args.maxcount = PNFS_LAYOUT_MAXSIZE;
lgp->args.range.iomode = iomode;
@@ -486,7 +507,10 @@ send_layoutget(struct pnfs_layout_hdr *lo,
lgp->args.type = server->pnfs_curr_ld->id;
lgp->args.inode = ino;
lgp->args.ctx = get_nfs_open_context(ctx);
+ lgp->args.layout.pages = pages;
+ lgp->args.layout.pglen = max_pages * PAGE_SIZE;
lgp->lsegpp = &lseg;
+ lgp->gfp_flags = gfp_flags;
/* Synchronously retrieve layout information from server and
* store in lseg.
@@ -496,7 +520,26 @@ send_layoutget(struct pnfs_layout_hdr *lo,
/* remember that LAYOUTGET failed and suspend trying */
set_bit(lo_fail_bit(iomode), &lo->plh_flags);
}
+
+ /* free xdr pages */
+ for (i = 0; i < max_pages; i++)
+ __free_page(pages[i]);
+ kfree(pages);
+
return lseg;
+
+out_err_free:
+ /* free any allocated xdr pages, lgp as it's not used */
+ if (pages) {
+ for (i = 0; i < max_pages; i++) {
+ if (!pages[i])
+ break;
+ __free_page(pages[i]);
+ }
+ kfree(pages);
+ }
+ kfree(lgp);
+ return NULL;
}
bool pnfs_roc(struct inode *ino)
@@ -625,11 +668,11 @@ pnfs_insert_layout(struct pnfs_layout_hdr *lo,
}
static struct pnfs_layout_hdr *
-alloc_init_layout_hdr(struct inode *ino)
+alloc_init_layout_hdr(struct inode *ino, gfp_t gfp_flags)
{
struct pnfs_layout_hdr *lo;
- lo = kzalloc(sizeof(struct pnfs_layout_hdr), GFP_KERNEL);
+ lo = kzalloc(sizeof(struct pnfs_layout_hdr), gfp_flags);
if (!lo)
return NULL;
atomic_set(&lo->plh_refcount, 1);
@@ -641,7 +684,7 @@ alloc_init_layout_hdr(struct inode *ino)
}
static struct pnfs_layout_hdr *
-pnfs_find_alloc_layout(struct inode *ino)
+pnfs_find_alloc_layout(struct inode *ino, gfp_t gfp_flags)
{
struct nfs_inode *nfsi = NFS_I(ino);
struct pnfs_layout_hdr *new = NULL;
@@ -656,7 +699,7 @@ pnfs_find_alloc_layout(struct inode *ino)
return nfsi->layout;
}
spin_unlock(&ino->i_lock);
- new = alloc_init_layout_hdr(ino);
+ new = alloc_init_layout_hdr(ino, gfp_flags);
spin_lock(&ino->i_lock);
if (likely(nfsi->layout == NULL)) /* Won the race? */
@@ -716,7 +759,8 @@ pnfs_find_lseg(struct pnfs_layout_hdr *lo, u32 iomode)
struct pnfs_layout_segment *
pnfs_update_layout(struct inode *ino,
struct nfs_open_context *ctx,
- enum pnfs_iomode iomode)
+ enum pnfs_iomode iomode,
+ gfp_t gfp_flags)
{
struct nfs_inode *nfsi = NFS_I(ino);
struct nfs_client *clp = NFS_SERVER(ino)->nfs_client;
@@ -727,7 +771,7 @@ pnfs_update_layout(struct inode *ino,
if (!pnfs_enabled_sb(NFS_SERVER(ino)))
return NULL;
spin_lock(&ino->i_lock);
- lo = pnfs_find_alloc_layout(ino);
+ lo = pnfs_find_alloc_layout(ino, gfp_flags);
if (lo == NULL) {
dprintk("%s ERROR: can't get pnfs_layout_hdr\n", __func__);
goto out_unlock;
@@ -767,7 +811,7 @@ pnfs_update_layout(struct inode *ino,
spin_unlock(&clp->cl_lock);
}
- lseg = send_layoutget(lo, ctx, iomode);
+ lseg = send_layoutget(lo, ctx, iomode, gfp_flags);
if (!lseg && first) {
spin_lock(&clp->cl_lock);
list_del_init(&lo->plh_layouts);
@@ -806,7 +850,7 @@ pnfs_layout_process(struct nfs4_layoutget *lgp)
goto out;
}
/* Inject layout blob into I/O device driver */
- lseg = NFS_SERVER(ino)->pnfs_curr_ld->alloc_lseg(lo, res);
+ lseg = NFS_SERVER(ino)->pnfs_curr_ld->alloc_lseg(lo, res, lgp->gfp_flags);
if (!lseg || IS_ERR(lseg)) {
if (!lseg)
status = -ENOMEM;
@@ -859,7 +903,8 @@ static int pnfs_read_pg_test(struct nfs_pageio_descriptor *pgio,
/* This is first coelesce call for a series of nfs_pages */
pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
prev->wb_context,
- IOMODE_READ);
+ IOMODE_READ,
+ GFP_KERNEL);
}
return NFS_SERVER(pgio->pg_inode)->pnfs_curr_ld->pg_test(pgio, prev, req);
}
@@ -881,7 +926,8 @@ static int pnfs_write_pg_test(struct nfs_pageio_descriptor *pgio,
/* This is first coelesce call for a series of nfs_pages */
pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
prev->wb_context,
- IOMODE_RW);
+ IOMODE_RW,
+ GFP_NOFS);
}
return NFS_SERVER(pgio->pg_inode)->pnfs_curr_ld->pg_test(pgio, prev, req);
}
@@ -945,3 +991,111 @@ pnfs_try_to_read_data(struct nfs_read_data *rdata,
dprintk("%s End (trypnfs:%d)\n", __func__, trypnfs);
return trypnfs;
}
+
+/*
+ * Currently there is only one (whole file) write lseg.
+ */
+static struct pnfs_layout_segment *pnfs_list_write_lseg(struct inode *inode)
+{
+ struct pnfs_layout_segment *lseg, *rv = NULL;
+
+ list_for_each_entry(lseg, &NFS_I(inode)->layout->plh_segs, pls_list)
+ if (lseg->pls_range.iomode == IOMODE_RW)
+ rv = lseg;
+ return rv;
+}
+
+void
+pnfs_set_layoutcommit(struct nfs_write_data *wdata)
+{
+ struct nfs_inode *nfsi = NFS_I(wdata->inode);
+ loff_t end_pos = wdata->args.offset + wdata->res.count;
+ bool mark_as_dirty = false;
+
+ spin_lock(&nfsi->vfs_inode.i_lock);
+ if (!test_and_set_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) {
+ /* references matched in nfs4_layoutcommit_release */
+ get_lseg(wdata->lseg);
+ wdata->lseg->pls_lc_cred =
+ get_rpccred(wdata->args.context->state->owner->so_cred);
+ mark_as_dirty = true;
+ dprintk("%s: Set layoutcommit for inode %lu ",
+ __func__, wdata->inode->i_ino);
+ }
+ if (end_pos > wdata->lseg->pls_end_pos)
+ wdata->lseg->pls_end_pos = end_pos;
+ spin_unlock(&nfsi->vfs_inode.i_lock);
+
+ /* if pnfs_layoutcommit_inode() runs between inode locks, the next one
+ * will be a noop because NFS_INO_LAYOUTCOMMIT will not be set */
+ if (mark_as_dirty)
+ mark_inode_dirty_sync(wdata->inode);
+}
+EXPORT_SYMBOL_GPL(pnfs_set_layoutcommit);
+
+/*
+ * For the LAYOUT4_NFSV4_1_FILES layout type, NFS_DATA_SYNC WRITEs and
+ * NFS_UNSTABLE WRITEs with a COMMIT to data servers must store enough
+ * data to disk to allow the server to recover the data if it crashes.
+ * LAYOUTCOMMIT is only needed when the NFL4_UFLG_COMMIT_THRU_MDS flag
+ * is off, and a COMMIT is sent to a data server, or
+ * if WRITEs to a data server return NFS_DATA_SYNC.
+ */
+int
+pnfs_layoutcommit_inode(struct inode *inode, bool sync)
+{
+ struct nfs4_layoutcommit_data *data;
+ struct nfs_inode *nfsi = NFS_I(inode);
+ struct pnfs_layout_segment *lseg;
+ struct rpc_cred *cred;
+ loff_t end_pos;
+ int status = 0;
+
+ dprintk("--> %s inode %lu\n", __func__, inode->i_ino);
+
+ if (!test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags))
+ return 0;
+
+ /* Note kzalloc ensures data->res.seq_res.sr_slot == NULL */
+ data = kzalloc(sizeof(*data), GFP_NOFS);
+ if (!data) {
+ mark_inode_dirty_sync(inode);
+ status = -ENOMEM;
+ goto out;
+ }
+
+ spin_lock(&inode->i_lock);
+ if (!test_and_clear_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) {
+ spin_unlock(&inode->i_lock);
+ kfree(data);
+ goto out;
+ }
+ /*
+ * Currently only one (whole file) write lseg which is referenced
+ * in pnfs_set_layoutcommit and will be found.
+ */
+ lseg = pnfs_list_write_lseg(inode);
+
+ end_pos = lseg->pls_end_pos;
+ cred = lseg->pls_lc_cred;
+ lseg->pls_end_pos = 0;
+ lseg->pls_lc_cred = NULL;
+
+ memcpy(&data->args.stateid.data, nfsi->layout->plh_stateid.data,
+ sizeof(nfsi->layout->plh_stateid.data));
+ spin_unlock(&inode->i_lock);
+
+ data->args.inode = inode;
+ data->lseg = lseg;
+ data->cred = cred;
+ nfs_fattr_init(&data->fattr);
+ data->args.bitmask = NFS_SERVER(inode)->cache_consistency_bitmask;
+ data->res.fattr = &data->fattr;
+ data->args.lastbytewritten = end_pos - 1;
+ data->res.server = NFS_SERVER(inode);
+
+ status = nfs4_proc_layoutcommit(data, sync);
+out:
+ dprintk("<-- %s status %d\n", __func__, status);
+ return status;
+}