NFSv4.1: Ensure that layoutget uses the correct gfp modes

Currently, writebacks may end up recursing back into the filesystem due to
GFP_KERNEL direct reclaims in the pnfs subsystem.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

1 files changed, 19 insertions, 14 deletions

diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 65455f58b109..f57f5281a520 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -467,7 +467,8 @@ 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);
@@ -480,7 +481,7 @@ send_layoutget(struct pnfs_layout_hdr *lo,
         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;
 
@@ -488,12 +489,12 @@ send_layoutget(struct pnfs_layout_hdr *lo,
         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_KERNEL);
+        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_KERNEL);
+                pages[i] = alloc_page(gfp_flags);
                 if (!pages[i])
                         goto out_err_free;
         }
@@ -509,6 +510,7 @@ send_layoutget(struct pnfs_layout_hdr *lo,
         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.
@@ -666,11 +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);
@@ -682,7 +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;
@@ -697,7 +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? */
@@ -757,7 +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;
@@ -768,7 +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;
@@ -808,7 +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);
@@ -847,7 +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;
@@ -900,7 +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);
 }
@@ -922,7 +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);
 }