pnfs/flexfiles: retry getting layout segment for reads

JIRA: https://issues.redhat.com/browse/RHEL-87767

commit eb3fabd
Author: Mike Snitzer <snitzer@kernel.org>
Date:   Thu Jan 16 20:05:39 2025 -0500

    pnfs/flexfiles: retry getting layout segment for reads

    If ff_layout_pg_get_read()'s attempt to get a layout segment results
    in -EAGAIN have ff_layout_pg_init_read() retry it after sleeping.

    If "softerr" mount is used, use 'io_maxretrans' to limit the number of
    attempts to get a layout segment.

    This fixes a long-standing issue of O_DIRECT reads failing with
    -EAGAIN (11) when using flexfiles Client Side Mirroring (CSM).

    Cc: stable@vger.kernel.org
    Signed-off-by: Mike Snitzer <snitzer@kernel.org>
    Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>

Author: bcodding-rh

fs/nfs/flexfilelayout/flexfilelayout.c

@@ -831,6 +831,9 @@ ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio,
 	struct nfs4_pnfs_ds *ds;
 	u32 ds_idx;
 
+	if (NFS_SERVER(pgio->pg_inode)->flags &
+			(NFS_MOUNT_SOFT|NFS_MOUNT_SOFTERR))
+		pgio->pg_maxretrans = io_maxretrans;
 retry:
 	pnfs_generic_pg_check_layout(pgio, req);
 	/* Use full layout for now */
@@ -844,6 +847,8 @@ ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio,
 		if (!pgio->pg_lseg)
 			goto out_nolseg;
 	}
+	/* Reset wb_nio, since getting layout segment was successful */
+	req->wb_nio = 0;
 
 	ds = ff_layout_get_ds_for_read(pgio, &ds_idx);
 	if (!ds) {
@@ -860,14 +865,24 @@ ff_layout_pg_init_read(struct nfs_pageio_descriptor *pgio,
 	pgm->pg_bsize = mirror->mirror_ds->ds_versions[0].rsize;
 
 	pgio->pg_mirror_idx = ds_idx;
-
-	if (NFS_SERVER(pgio->pg_inode)->flags &
-			(NFS_MOUNT_SOFT|NFS_MOUNT_SOFTERR))
-		pgio->pg_maxretrans = io_maxretrans;
 	return;
 out_nolseg:
-	if (pgio->pg_error < 0)
-		return;
+	if (pgio->pg_error < 0) {
+		if (pgio->pg_error != -EAGAIN)
+			return;
+		/* Retry getting layout segment if lower layer returned -EAGAIN */
+		if (pgio->pg_maxretrans && req->wb_nio++ > pgio->pg_maxretrans) {
+			if (NFS_SERVER(pgio->pg_inode)->flags & NFS_MOUNT_SOFTERR)
+				pgio->pg_error = -ETIMEDOUT;
+			else
+				pgio->pg_error = -EIO;
+			return;
+		}
+		pgio->pg_error = 0;
+		/* Sleep for 1 second before retrying */
+		ssleep(1);
+		goto retry;
+	}
 out_mds:
 	trace_pnfs_mds_fallback_pg_init_read(pgio->pg_inode,
 			0, NFS4_MAX_UINT64, IOMODE_READ,