Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / fs / nfs / objlayout / objlayout.c
diff --git a/kernel/fs/nfs/objlayout/objlayout.c b/kernel/fs/nfs/objlayout/objlayout.c
new file mode 100644 (file)
index 0000000..919efd4
--- /dev/null
@@ -0,0 +1,706 @@
+/*
+ *  pNFS Objects layout driver high level definitions
+ *
+ *  Copyright (C) 2007 Panasas Inc. [year of first publication]
+ *  All rights reserved.
+ *
+ *  Benny Halevy <bhalevy@panasas.com>
+ *  Boaz Harrosh <ooo@electrozaur.com>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License version 2
+ *  See the file COPYING included with this distribution for more details.
+ *
+ *  Redistribution and use in source and binary forms, with or without
+ *  modification, are permitted provided that the following conditions
+ *  are met:
+ *
+ *  1. Redistributions of source code must retain the above copyright
+ *     notice, this list of conditions and the following disclaimer.
+ *  2. Redistributions in binary form must reproduce the above copyright
+ *     notice, this list of conditions and the following disclaimer in the
+ *     documentation and/or other materials provided with the distribution.
+ *  3. Neither the name of the Panasas company nor the names of its
+ *     contributors may be used to endorse or promote products derived
+ *     from this software without specific prior written permission.
+ *
+ *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <linux/kmod.h>
+#include <linux/moduleparam.h>
+#include <linux/ratelimit.h>
+#include <scsi/osd_initiator.h>
+#include "objlayout.h"
+
+#define NFSDBG_FACILITY         NFSDBG_PNFS_LD
+/*
+ * Create a objlayout layout structure for the given inode and return it.
+ */
+struct pnfs_layout_hdr *
+objlayout_alloc_layout_hdr(struct inode *inode, gfp_t gfp_flags)
+{
+       struct objlayout *objlay;
+
+       objlay = kzalloc(sizeof(struct objlayout), gfp_flags);
+       if (!objlay)
+               return NULL;
+       spin_lock_init(&objlay->lock);
+       INIT_LIST_HEAD(&objlay->err_list);
+       dprintk("%s: Return %p\n", __func__, objlay);
+       return &objlay->pnfs_layout;
+}
+
+/*
+ * Free an objlayout layout structure
+ */
+void
+objlayout_free_layout_hdr(struct pnfs_layout_hdr *lo)
+{
+       struct objlayout *objlay = OBJLAYOUT(lo);
+
+       dprintk("%s: objlay %p\n", __func__, objlay);
+
+       WARN_ON(!list_empty(&objlay->err_list));
+       kfree(objlay);
+}
+
+/*
+ * Unmarshall layout and store it in pnfslay.
+ */
+struct pnfs_layout_segment *
+objlayout_alloc_lseg(struct pnfs_layout_hdr *pnfslay,
+                    struct nfs4_layoutget_res *lgr,
+                    gfp_t gfp_flags)
+{
+       int status = -ENOMEM;
+       struct xdr_stream stream;
+       struct xdr_buf buf = {
+               .pages =  lgr->layoutp->pages,
+               .page_len =  lgr->layoutp->len,
+               .buflen =  lgr->layoutp->len,
+               .len = lgr->layoutp->len,
+       };
+       struct page *scratch;
+       struct pnfs_layout_segment *lseg;
+
+       dprintk("%s: Begin pnfslay %p\n", __func__, pnfslay);
+
+       scratch = alloc_page(gfp_flags);
+       if (!scratch)
+               goto err_nofree;
+
+       xdr_init_decode(&stream, &buf, NULL);
+       xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
+
+       status = objio_alloc_lseg(&lseg, pnfslay, &lgr->range, &stream, gfp_flags);
+       if (unlikely(status)) {
+               dprintk("%s: objio_alloc_lseg Return err %d\n", __func__,
+                       status);
+               goto err;
+       }
+
+       __free_page(scratch);
+
+       dprintk("%s: Return %p\n", __func__, lseg);
+       return lseg;
+
+err:
+       __free_page(scratch);
+err_nofree:
+       dprintk("%s: Err Return=>%d\n", __func__, status);
+       return ERR_PTR(status);
+}
+
+/*
+ * Free a layout segement
+ */
+void
+objlayout_free_lseg(struct pnfs_layout_segment *lseg)
+{
+       dprintk("%s: freeing layout segment %p\n", __func__, lseg);
+
+       if (unlikely(!lseg))
+               return;
+
+       objio_free_lseg(lseg);
+}
+
+/*
+ * I/O Operations
+ */
+static inline u64
+end_offset(u64 start, u64 len)
+{
+       u64 end;
+
+       end = start + len;
+       return end >= start ? end : NFS4_MAX_UINT64;
+}
+
+static void _fix_verify_io_params(struct pnfs_layout_segment *lseg,
+                          struct page ***p_pages, unsigned *p_pgbase,
+                          u64 offset, unsigned long count)
+{
+       u64 lseg_end_offset;
+
+       BUG_ON(offset < lseg->pls_range.offset);
+       lseg_end_offset = end_offset(lseg->pls_range.offset,
+                                    lseg->pls_range.length);
+       BUG_ON(offset >= lseg_end_offset);
+       WARN_ON(offset + count > lseg_end_offset);
+
+       if (*p_pgbase > PAGE_SIZE) {
+               dprintk("%s: pgbase(0x%x) > PAGE_SIZE\n", __func__, *p_pgbase);
+               *p_pages += *p_pgbase >> PAGE_SHIFT;
+               *p_pgbase &= ~PAGE_MASK;
+       }
+}
+
+/*
+ * I/O done common code
+ */
+static void
+objlayout_iodone(struct objlayout_io_res *oir)
+{
+       if (likely(oir->status >= 0)) {
+               objio_free_result(oir);
+       } else {
+               struct objlayout *objlay = oir->objlay;
+
+               spin_lock(&objlay->lock);
+               objlay->delta_space_valid = OBJ_DSU_INVALID;
+               list_add(&objlay->err_list, &oir->err_list);
+               spin_unlock(&objlay->lock);
+       }
+}
+
+/*
+ * objlayout_io_set_result - Set an osd_error code on a specific osd comp.
+ *
+ * The @index component IO failed (error returned from target). Register
+ * the error for later reporting at layout-return.
+ */
+void
+objlayout_io_set_result(struct objlayout_io_res *oir, unsigned index,
+                       struct pnfs_osd_objid *pooid, int osd_error,
+                       u64 offset, u64 length, bool is_write)
+{
+       struct pnfs_osd_ioerr *ioerr = &oir->ioerrs[index];
+
+       BUG_ON(index >= oir->num_comps);
+       if (osd_error) {
+               ioerr->oer_component = *pooid;
+               ioerr->oer_comp_offset = offset;
+               ioerr->oer_comp_length = length;
+               ioerr->oer_iswrite = is_write;
+               ioerr->oer_errno = osd_error;
+
+               dprintk("%s: err[%d]: errno=%d is_write=%d dev(%llx:%llx) "
+                       "par=0x%llx obj=0x%llx offset=0x%llx length=0x%llx\n",
+                       __func__, index, ioerr->oer_errno,
+                       ioerr->oer_iswrite,
+                       _DEVID_LO(&ioerr->oer_component.oid_device_id),
+                       _DEVID_HI(&ioerr->oer_component.oid_device_id),
+                       ioerr->oer_component.oid_partition_id,
+                       ioerr->oer_component.oid_object_id,
+                       ioerr->oer_comp_offset,
+                       ioerr->oer_comp_length);
+       } else {
+               /* User need not call if no error is reported */
+               ioerr->oer_errno = 0;
+       }
+}
+
+/* Function scheduled on rpc workqueue to call ->nfs_readlist_complete().
+ * This is because the osd completion is called with ints-off from
+ * the block layer
+ */
+static void _rpc_read_complete(struct work_struct *work)
+{
+       struct rpc_task *task;
+       struct nfs_pgio_header *hdr;
+
+       dprintk("%s enter\n", __func__);
+       task = container_of(work, struct rpc_task, u.tk_work);
+       hdr = container_of(task, struct nfs_pgio_header, task);
+
+       pnfs_ld_read_done(hdr);
+}
+
+void
+objlayout_read_done(struct objlayout_io_res *oir, ssize_t status, bool sync)
+{
+       struct nfs_pgio_header *hdr = oir->rpcdata;
+
+       oir->status = hdr->task.tk_status = status;
+       if (status >= 0)
+               hdr->res.count = status;
+       else
+               hdr->pnfs_error = status;
+       objlayout_iodone(oir);
+       /* must not use oir after this point */
+
+       dprintk("%s: Return status=%zd eof=%d sync=%d\n", __func__,
+               status, hdr->res.eof, sync);
+
+       if (sync)
+               pnfs_ld_read_done(hdr);
+       else {
+               INIT_WORK(&hdr->task.u.tk_work, _rpc_read_complete);
+               schedule_work(&hdr->task.u.tk_work);
+       }
+}
+
+/*
+ * Perform sync or async reads.
+ */
+enum pnfs_try_status
+objlayout_read_pagelist(struct nfs_pgio_header *hdr)
+{
+       struct inode *inode = hdr->inode;
+       loff_t offset = hdr->args.offset;
+       size_t count = hdr->args.count;
+       int err;
+       loff_t eof;
+
+       eof = i_size_read(inode);
+       if (unlikely(offset + count > eof)) {
+               if (offset >= eof) {
+                       err = 0;
+                       hdr->res.count = 0;
+                       hdr->res.eof = 1;
+                       /*FIXME: do we need to call pnfs_ld_read_done() */
+                       goto out;
+               }
+               count = eof - offset;
+       }
+
+       hdr->res.eof = (offset + count) >= eof;
+       _fix_verify_io_params(hdr->lseg, &hdr->args.pages,
+                             &hdr->args.pgbase,
+                             hdr->args.offset, hdr->args.count);
+
+       dprintk("%s: inode(%lx) offset 0x%llx count 0x%Zx eof=%d\n",
+               __func__, inode->i_ino, offset, count, hdr->res.eof);
+
+       err = objio_read_pagelist(hdr);
+ out:
+       if (unlikely(err)) {
+               hdr->pnfs_error = err;
+               dprintk("%s: Returned Error %d\n", __func__, err);
+               return PNFS_NOT_ATTEMPTED;
+       }
+       return PNFS_ATTEMPTED;
+}
+
+/* Function scheduled on rpc workqueue to call ->nfs_writelist_complete().
+ * This is because the osd completion is called with ints-off from
+ * the block layer
+ */
+static void _rpc_write_complete(struct work_struct *work)
+{
+       struct rpc_task *task;
+       struct nfs_pgio_header *hdr;
+
+       dprintk("%s enter\n", __func__);
+       task = container_of(work, struct rpc_task, u.tk_work);
+       hdr = container_of(task, struct nfs_pgio_header, task);
+
+       pnfs_ld_write_done(hdr);
+}
+
+void
+objlayout_write_done(struct objlayout_io_res *oir, ssize_t status, bool sync)
+{
+       struct nfs_pgio_header *hdr = oir->rpcdata;
+
+       oir->status = hdr->task.tk_status = status;
+       if (status >= 0) {
+               hdr->res.count = status;
+               hdr->verf.committed = oir->committed;
+       } else {
+               hdr->pnfs_error = status;
+       }
+       objlayout_iodone(oir);
+       /* must not use oir after this point */
+
+       dprintk("%s: Return status %zd committed %d sync=%d\n", __func__,
+               status, hdr->verf.committed, sync);
+
+       if (sync)
+               pnfs_ld_write_done(hdr);
+       else {
+               INIT_WORK(&hdr->task.u.tk_work, _rpc_write_complete);
+               schedule_work(&hdr->task.u.tk_work);
+       }
+}
+
+/*
+ * Perform sync or async writes.
+ */
+enum pnfs_try_status
+objlayout_write_pagelist(struct nfs_pgio_header *hdr, int how)
+{
+       int err;
+
+       _fix_verify_io_params(hdr->lseg, &hdr->args.pages,
+                             &hdr->args.pgbase,
+                             hdr->args.offset, hdr->args.count);
+
+       err = objio_write_pagelist(hdr, how);
+       if (unlikely(err)) {
+               hdr->pnfs_error = err;
+               dprintk("%s: Returned Error %d\n", __func__, err);
+               return PNFS_NOT_ATTEMPTED;
+       }
+       return PNFS_ATTEMPTED;
+}
+
+void
+objlayout_encode_layoutcommit(struct pnfs_layout_hdr *pnfslay,
+                             struct xdr_stream *xdr,
+                             const struct nfs4_layoutcommit_args *args)
+{
+       struct objlayout *objlay = OBJLAYOUT(pnfslay);
+       struct pnfs_osd_layoutupdate lou;
+       __be32 *start;
+
+       dprintk("%s: Begin\n", __func__);
+
+       spin_lock(&objlay->lock);
+       lou.dsu_valid = (objlay->delta_space_valid == OBJ_DSU_VALID);
+       lou.dsu_delta = objlay->delta_space_used;
+       objlay->delta_space_used = 0;
+       objlay->delta_space_valid = OBJ_DSU_INIT;
+       lou.olu_ioerr_flag = !list_empty(&objlay->err_list);
+       spin_unlock(&objlay->lock);
+
+       start = xdr_reserve_space(xdr, 4);
+
+       BUG_ON(pnfs_osd_xdr_encode_layoutupdate(xdr, &lou));
+
+       *start = cpu_to_be32((xdr->p - start - 1) * 4);
+
+       dprintk("%s: Return delta_space_used %lld err %d\n", __func__,
+               lou.dsu_delta, lou.olu_ioerr_flag);
+}
+
+static int
+err_prio(u32 oer_errno)
+{
+       switch (oer_errno) {
+       case 0:
+               return 0;
+
+       case PNFS_OSD_ERR_RESOURCE:
+               return OSD_ERR_PRI_RESOURCE;
+       case PNFS_OSD_ERR_BAD_CRED:
+               return OSD_ERR_PRI_BAD_CRED;
+       case PNFS_OSD_ERR_NO_ACCESS:
+               return OSD_ERR_PRI_NO_ACCESS;
+       case PNFS_OSD_ERR_UNREACHABLE:
+               return OSD_ERR_PRI_UNREACHABLE;
+       case PNFS_OSD_ERR_NOT_FOUND:
+               return OSD_ERR_PRI_NOT_FOUND;
+       case PNFS_OSD_ERR_NO_SPACE:
+               return OSD_ERR_PRI_NO_SPACE;
+       default:
+               WARN_ON(1);
+               /* fallthrough */
+       case PNFS_OSD_ERR_EIO:
+               return OSD_ERR_PRI_EIO;
+       }
+}
+
+static void
+merge_ioerr(struct pnfs_osd_ioerr *dest_err,
+           const struct pnfs_osd_ioerr *src_err)
+{
+       u64 dest_end, src_end;
+
+       if (!dest_err->oer_errno) {
+               *dest_err = *src_err;
+               /* accumulated device must be blank */
+               memset(&dest_err->oer_component.oid_device_id, 0,
+                       sizeof(dest_err->oer_component.oid_device_id));
+
+               return;
+       }
+
+       if (dest_err->oer_component.oid_partition_id !=
+                               src_err->oer_component.oid_partition_id)
+               dest_err->oer_component.oid_partition_id = 0;
+
+       if (dest_err->oer_component.oid_object_id !=
+                               src_err->oer_component.oid_object_id)
+               dest_err->oer_component.oid_object_id = 0;
+
+       if (dest_err->oer_comp_offset > src_err->oer_comp_offset)
+               dest_err->oer_comp_offset = src_err->oer_comp_offset;
+
+       dest_end = end_offset(dest_err->oer_comp_offset,
+                             dest_err->oer_comp_length);
+       src_end =  end_offset(src_err->oer_comp_offset,
+                             src_err->oer_comp_length);
+       if (dest_end < src_end)
+               dest_end = src_end;
+
+       dest_err->oer_comp_length = dest_end - dest_err->oer_comp_offset;
+
+       if ((src_err->oer_iswrite == dest_err->oer_iswrite) &&
+           (err_prio(src_err->oer_errno) > err_prio(dest_err->oer_errno))) {
+                       dest_err->oer_errno = src_err->oer_errno;
+       } else if (src_err->oer_iswrite) {
+               dest_err->oer_iswrite = true;
+               dest_err->oer_errno = src_err->oer_errno;
+       }
+}
+
+static void
+encode_accumulated_error(struct objlayout *objlay, __be32 *p)
+{
+       struct objlayout_io_res *oir, *tmp;
+       struct pnfs_osd_ioerr accumulated_err = {.oer_errno = 0};
+
+       list_for_each_entry_safe(oir, tmp, &objlay->err_list, err_list) {
+               unsigned i;
+
+               for (i = 0; i < oir->num_comps; i++) {
+                       struct pnfs_osd_ioerr *ioerr = &oir->ioerrs[i];
+
+                       if (!ioerr->oer_errno)
+                               continue;
+
+                       printk(KERN_ERR "NFS: %s: err[%d]: errno=%d "
+                               "is_write=%d dev(%llx:%llx) par=0x%llx "
+                               "obj=0x%llx offset=0x%llx length=0x%llx\n",
+                               __func__, i, ioerr->oer_errno,
+                               ioerr->oer_iswrite,
+                               _DEVID_LO(&ioerr->oer_component.oid_device_id),
+                               _DEVID_HI(&ioerr->oer_component.oid_device_id),
+                               ioerr->oer_component.oid_partition_id,
+                               ioerr->oer_component.oid_object_id,
+                               ioerr->oer_comp_offset,
+                               ioerr->oer_comp_length);
+
+                       merge_ioerr(&accumulated_err, ioerr);
+               }
+               list_del(&oir->err_list);
+               objio_free_result(oir);
+       }
+
+       pnfs_osd_xdr_encode_ioerr(p, &accumulated_err);
+}
+
+void
+objlayout_encode_layoutreturn(struct pnfs_layout_hdr *pnfslay,
+                             struct xdr_stream *xdr,
+                             const struct nfs4_layoutreturn_args *args)
+{
+       struct objlayout *objlay = OBJLAYOUT(pnfslay);
+       struct objlayout_io_res *oir, *tmp;
+       __be32 *start;
+
+       dprintk("%s: Begin\n", __func__);
+       start = xdr_reserve_space(xdr, 4);
+       BUG_ON(!start);
+
+       spin_lock(&objlay->lock);
+
+       list_for_each_entry_safe(oir, tmp, &objlay->err_list, err_list) {
+               __be32 *last_xdr = NULL, *p;
+               unsigned i;
+               int res = 0;
+
+               for (i = 0; i < oir->num_comps; i++) {
+                       struct pnfs_osd_ioerr *ioerr = &oir->ioerrs[i];
+
+                       if (!ioerr->oer_errno)
+                               continue;
+
+                       dprintk("%s: err[%d]: errno=%d is_write=%d "
+                               "dev(%llx:%llx) par=0x%llx obj=0x%llx "
+                               "offset=0x%llx length=0x%llx\n",
+                               __func__, i, ioerr->oer_errno,
+                               ioerr->oer_iswrite,
+                               _DEVID_LO(&ioerr->oer_component.oid_device_id),
+                               _DEVID_HI(&ioerr->oer_component.oid_device_id),
+                               ioerr->oer_component.oid_partition_id,
+                               ioerr->oer_component.oid_object_id,
+                               ioerr->oer_comp_offset,
+                               ioerr->oer_comp_length);
+
+                       p = pnfs_osd_xdr_ioerr_reserve_space(xdr);
+                       if (unlikely(!p)) {
+                               res = -E2BIG;
+                               break; /* accumulated_error */
+                       }
+
+                       last_xdr = p;
+                       pnfs_osd_xdr_encode_ioerr(p, &oir->ioerrs[i]);
+               }
+
+               /* TODO: use xdr_write_pages */
+               if (unlikely(res)) {
+                       /* no space for even one error descriptor */
+                       BUG_ON(!last_xdr);
+
+                       /* we've encountered a situation with lots and lots of
+                        * errors and no space to encode them all. Use the last
+                        * available slot to report the union of all the
+                        * remaining errors.
+                        */
+                       encode_accumulated_error(objlay, last_xdr);
+                       goto loop_done;
+               }
+               list_del(&oir->err_list);
+               objio_free_result(oir);
+       }
+loop_done:
+       spin_unlock(&objlay->lock);
+
+       *start = cpu_to_be32((xdr->p - start - 1) * 4);
+       dprintk("%s: Return\n", __func__);
+}
+
+enum {
+       OBJLAYOUT_MAX_URI_LEN = 256, OBJLAYOUT_MAX_OSDNAME_LEN = 64,
+       OBJLAYOUT_MAX_SYSID_HEX_LEN = OSD_SYSTEMID_LEN * 2 + 1,
+       OSD_LOGIN_UPCALL_PATHLEN  = 256
+};
+
+static char osd_login_prog[OSD_LOGIN_UPCALL_PATHLEN] = "/sbin/osd_login";
+
+module_param_string(osd_login_prog, osd_login_prog, sizeof(osd_login_prog),
+                   0600);
+MODULE_PARM_DESC(osd_login_prog, "Path to the osd_login upcall program");
+
+struct __auto_login {
+       char uri[OBJLAYOUT_MAX_URI_LEN];
+       char osdname[OBJLAYOUT_MAX_OSDNAME_LEN];
+       char systemid_hex[OBJLAYOUT_MAX_SYSID_HEX_LEN];
+};
+
+static int __objlayout_upcall(struct __auto_login *login)
+{
+       static char *envp[] = { "HOME=/",
+               "TERM=linux",
+               "PATH=/sbin:/usr/sbin:/bin:/usr/bin",
+               NULL
+       };
+       char *argv[8];
+       int ret;
+
+       if (unlikely(!osd_login_prog[0])) {
+               dprintk("%s: osd_login_prog is disabled\n", __func__);
+               return -EACCES;
+       }
+
+       dprintk("%s uri: %s\n", __func__, login->uri);
+       dprintk("%s osdname %s\n", __func__, login->osdname);
+       dprintk("%s systemid_hex %s\n", __func__, login->systemid_hex);
+
+       argv[0] = (char *)osd_login_prog;
+       argv[1] = "-u";
+       argv[2] = login->uri;
+       argv[3] = "-o";
+       argv[4] = login->osdname;
+       argv[5] = "-s";
+       argv[6] = login->systemid_hex;
+       argv[7] = NULL;
+
+       ret = call_usermodehelper(argv[0], argv, envp, UMH_WAIT_PROC);
+       /*
+        * Disable the upcall mechanism if we're getting an ENOENT or
+        * EACCES error. The admin can re-enable it on the fly by using
+        * sysfs to set the objlayoutdriver.osd_login_prog module parameter once
+        * the problem has been fixed.
+        */
+       if (ret == -ENOENT || ret == -EACCES) {
+               printk(KERN_ERR "PNFS-OBJ: %s was not found please set "
+                       "objlayoutdriver.osd_login_prog kernel parameter!\n",
+                       osd_login_prog);
+               osd_login_prog[0] = '\0';
+       }
+       dprintk("%s %s return value: %d\n", __func__, osd_login_prog, ret);
+
+       return ret;
+}
+
+/* Assume dest is all zeros */
+static void __copy_nfsS_and_zero_terminate(struct nfs4_string s,
+                                          char *dest, int max_len,
+                                          const char *var_name)
+{
+       if (!s.len)
+               return;
+
+       if (s.len >= max_len) {
+               pr_warn_ratelimited(
+                       "objlayout_autologin: %s: s.len(%d) >= max_len(%d)",
+                       var_name, s.len, max_len);
+               s.len = max_len - 1; /* space for null terminator */
+       }
+
+       memcpy(dest, s.data, s.len);
+}
+
+/* Assume sysid is all zeros */
+static void _sysid_2_hex(struct nfs4_string s,
+                 char sysid[OBJLAYOUT_MAX_SYSID_HEX_LEN])
+{
+       int i;
+       char *cur;
+
+       if (!s.len)
+               return;
+
+       if (s.len != OSD_SYSTEMID_LEN) {
+               pr_warn_ratelimited(
+                   "objlayout_autologin: systemid_len(%d) != OSD_SYSTEMID_LEN",
+                   s.len);
+               if (s.len > OSD_SYSTEMID_LEN)
+                       s.len = OSD_SYSTEMID_LEN;
+       }
+
+       cur = sysid;
+       for (i = 0; i < s.len; i++)
+               cur = hex_byte_pack(cur, s.data[i]);
+}
+
+int objlayout_autologin(struct pnfs_osd_deviceaddr *deviceaddr)
+{
+       int rc;
+       struct __auto_login login;
+
+       if (!deviceaddr->oda_targetaddr.ota_netaddr.r_addr.len)
+               return -ENODEV;
+
+       memset(&login, 0, sizeof(login));
+       __copy_nfsS_and_zero_terminate(
+               deviceaddr->oda_targetaddr.ota_netaddr.r_addr,
+               login.uri, sizeof(login.uri), "URI");
+
+       __copy_nfsS_and_zero_terminate(
+               deviceaddr->oda_osdname,
+               login.osdname, sizeof(login.osdname), "OSDNAME");
+
+       _sysid_2_hex(deviceaddr->oda_systemid, login.systemid_hex);
+
+       rc = __objlayout_upcall(&login);
+       if (rc > 0) /* script returns positive values */
+               rc = -ENODEV;
+
+       return rc;
+}