NFSv4.1: Deal with wraparound issues when updating the layout stateid

[deliverable/linux.git] / fs / nfs / pnfs.c

diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index bdd93b969050fe03d745566a3829749a04f7d33a..f1387e87513ff619bcea36117fc29795106ec559 100644 (file)
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -200,8 +200,7 @@ static struct pnfs_layout_hdr *
 pnfs_alloc_layout_hdr(struct inode *ino, gfp_t gfp_flags)
 {
        struct pnfs_layoutdriver_type *ld = NFS_SERVER(ino)->pnfs_curr_ld;
-       return ld->alloc_layout_hdr ? ld->alloc_layout_hdr(ino, gfp_flags) :
-               kzalloc(sizeof(struct pnfs_layout_hdr), gfp_flags);
+       return ld->alloc_layout_hdr(ino, gfp_flags);
 }
 
 static void
@@ -218,7 +217,7 @@ pnfs_free_layout_hdr(struct pnfs_layout_hdr *lo)
                spin_unlock(&clp->cl_lock);
        }
        put_rpccred(lo->plh_lc_cred);
-       return ld->alloc_layout_hdr ? ld->free_layout_hdr(lo) : kfree(lo);
+       return ld->free_layout_hdr(lo);
 }
 
 static void
@@ -333,7 +332,7 @@ pnfs_layout_remove_lseg(struct pnfs_layout_hdr *lo,
        /* Matched by pnfs_get_layout_hdr in pnfs_layout_insert_lseg */
        atomic_dec(&lo->plh_refcount);
        if (list_empty(&lo->plh_segs))
-               set_bit(NFS_LAYOUT_DESTROYED, &lo->plh_flags);
+               clear_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags);
        rpc_wake_up(&NFS_SERVER(inode)->roc_rpcwaitq);
 }
 
@@ -464,10 +463,8 @@ pnfs_mark_matching_lsegs_invalid(struct pnfs_layout_hdr *lo,
 
        dprintk("%s:Begin lo %p\n", __func__, lo);
 
-       if (list_empty(&lo->plh_segs)) {
-               set_bit(NFS_LAYOUT_DESTROYED, &lo->plh_flags);
+       if (list_empty(&lo->plh_segs))
                return 0;
-       }
        list_for_each_entry_safe(lseg, next, &lo->plh_segs, pls_list)
                if (!recall_range ||
                    should_free_lseg(&lseg->pls_range, recall_range)) {
@@ -552,6 +549,15 @@ pnfs_destroy_all_layouts(struct nfs_client *clp)
        }
 }
 
+/*
+ * Compare 2 layout stateid sequence ids, to see which is newer,
+ * taking into account wraparound issues.
+ */
+static bool pnfs_seqid_is_newer(u32 s1, u32 s2)
+{
+       return (s32)s1 - (s32)s2 > 0;
+}
+
 /* update lo->plh_stateid with new if is more recent */
 void
 pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new,
@@ -561,7 +567,7 @@ pnfs_set_layout_stateid(struct pnfs_layout_hdr *lo, const nfs4_stateid *new,
 
        oldseq = be32_to_cpu(lo->plh_stateid.seqid);
        newseq = be32_to_cpu(new->seqid);
-       if ((int)(newseq - oldseq) > 0) {
+       if (list_empty(&lo->plh_segs) || pnfs_seqid_is_newer(newseq, oldseq)) {
                nfs4_stateid_copy(&lo->plh_stateid, new);
                if (update_barrier) {
                        u32 new_barrier = be32_to_cpu(new->seqid);
@@ -591,7 +597,6 @@ pnfs_layoutgets_blocked(struct pnfs_layout_hdr *lo, nfs4_stateid *stateid,
            (int)(lo->plh_barrier - be32_to_cpu(stateid->seqid)) >= 0)
                return true;
        return lo->plh_block_lgets ||
-               test_bit(NFS_LAYOUT_DESTROYED, &lo->plh_flags) ||
                test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags) ||
                (list_empty(&lo->plh_segs) &&
                 (atomic_read(&lo->plh_outstanding) > lget));
@@ -696,7 +701,7 @@ _pnfs_return_layout(struct inode *ino)
 
        spin_lock(&ino->i_lock);
        lo = nfsi->layout;
-       if (!lo || pnfs_test_layout_returned(lo)) {
+       if (!lo) {
                spin_unlock(&ino->i_lock);
                dprintk("NFS: %s no layout to return\n", __func__);
                goto out;
@@ -714,7 +719,6 @@ _pnfs_return_layout(struct inode *ino)
                goto out;
        }
        lo->plh_block_lgets++;
-       pnfs_mark_layout_returned(lo);
        spin_unlock(&ino->i_lock);
        pnfs_free_lseg_list(&tmp_list);
 
@@ -723,9 +727,9 @@ _pnfs_return_layout(struct inode *ino)
        lrp = kzalloc(sizeof(*lrp), GFP_KERNEL);
        if (unlikely(lrp == NULL)) {
                status = -ENOMEM;
-               pnfs_layout_io_set_failed(lo, IOMODE_RW);
-               pnfs_layout_io_set_failed(lo, IOMODE_READ);
-               pnfs_clear_layout_returned(lo);
+               spin_lock(&ino->i_lock);
+               lo->plh_block_lgets--;
+               spin_unlock(&ino->i_lock);
                pnfs_put_layout_hdr(lo);
                goto out;
        }
@@ -859,7 +863,6 @@ pnfs_layout_insert_lseg(struct pnfs_layout_hdr *lo,
 
        dprintk("%s:Begin\n", __func__);
 
-       assert_spin_locked(&lo->plh_inode->i_lock);
        list_for_each_entry(lp, &lo->plh_segs, pls_list) {
                if (cmp_layout(&lseg->pls_range, &lp->pls_range) > 0)
                        continue;
@@ -913,21 +916,19 @@ pnfs_find_alloc_layout(struct inode *ino,
 
        dprintk("%s Begin ino=%p layout=%p\n", __func__, ino, nfsi->layout);
 
-       assert_spin_locked(&ino->i_lock);
-       if (nfsi->layout) {
-               if (test_bit(NFS_LAYOUT_DESTROYED, &nfsi->layout->plh_flags))
-                       return NULL;
-               pnfs_get_layout_hdr(nfsi->layout);
-               return nfsi->layout;
-       }
+       if (nfsi->layout != NULL)
+               goto out_existing;
        spin_unlock(&ino->i_lock);
        new = alloc_init_layout_hdr(ino, ctx, gfp_flags);
        spin_lock(&ino->i_lock);
 
-       if (likely(nfsi->layout == NULL))       /* Won the race? */
+       if (likely(nfsi->layout == NULL)) {     /* Won the race? */
                nfsi->layout = new;
-       else
-               pnfs_free_layout_hdr(new);
+               return new;
+       }
+       pnfs_free_layout_hdr(new);
+out_existing:
+       pnfs_get_layout_hdr(nfsi->layout);
        return nfsi->layout;
 }
 
@@ -970,7 +971,6 @@ pnfs_find_lseg(struct pnfs_layout_hdr *lo,
 
        dprintk("%s:Begin\n", __func__);
 
-       assert_spin_locked(&lo->plh_inode->i_lock);
        list_for_each_entry(lseg, &lo->plh_segs, pls_list) {
                if (test_bit(NFS_LSEG_VALID, &lseg->pls_flags) &&
                    is_matching_lseg(&lseg->pls_range, range)) {
@@ -1120,9 +1120,6 @@ pnfs_update_layout(struct inode *ino,
        if (list_empty(&lo->plh_segs))
                first = true;
 
-       /* Enable LAYOUTRETURNs */
-       pnfs_clear_layout_returned(lo);
-
        spin_unlock(&ino->i_lock);
        if (first) {
                /* The lo must be on the clp list if there is any
@@ -1193,6 +1190,10 @@ pnfs_layout_process(struct nfs4_layoutget *lgp)
                dprintk("%s forget reply due to state\n", __func__);
                goto out_forget_reply;
        }
+
+       /* Done processing layoutget. Set the layout stateid */
+       pnfs_set_layout_stateid(lo, &res->stateid, false);
+
        init_lseg(lo, lseg);
        lseg->pls_range = res->range;
        pnfs_get_lseg(lseg);
@@ -1203,8 +1204,6 @@ pnfs_layout_process(struct nfs4_layoutget *lgp)
                set_bit(NFS_LAYOUT_ROC, &lo->plh_flags);
        }
 
-       /* Done processing layoutget. Set the layout stateid */
-       pnfs_set_layout_stateid(lo, &res->stateid, false);
        spin_unlock(&ino->i_lock);
        return lseg;
 out: