nfsd4: cl_count is unused
[safe/jmp/linux-2.6] / fs / exofs / ios.c
index 4372542..5293bc4 100644 (file)
  */
 
 #include <scsi/scsi_device.h>
-#include <scsi/osd_sense.h>
+#include <asm/div64.h>
 
 #include "exofs.h"
 
-int exofs_check_ok_resid(struct osd_request *or, u64 *in_resid, u64 *out_resid)
-{
-       struct osd_sense_info osi;
-       int ret = osd_req_decode_sense(or, &osi);
-
-       if (ret) { /* translate to Linux codes */
-               if (osi.additional_code == scsi_invalid_field_in_cdb) {
-                       if (osi.cdb_field_offset == OSD_CFO_STARTING_BYTE)
-                               ret = -EFAULT;
-                       if (osi.cdb_field_offset == OSD_CFO_OBJECT_ID)
-                               ret = -ENOENT;
-                       else
-                               ret = -EINVAL;
-               } else if (osi.additional_code == osd_quota_error)
-                       ret = -ENOSPC;
-               else
-                       ret = -EIO;
-       }
-
-       /* FIXME: should be include in osd_sense_info */
-       if (in_resid)
-               *in_resid = or->in.req ? or->in.req->resid_len : 0;
-
-       if (out_resid)
-               *out_resid = or->out.req ? or->out.req->resid_len : 0;
-
-       return ret;
-}
+#define EXOFS_DBGMSG2(M...) do {} while (0)
+/* #define EXOFS_DBGMSG2 EXOFS_DBGMSG */
 
 void exofs_make_credential(u8 cred_a[OSD_CAP_LEN], const struct osd_obj_id *obj)
 {
        osd_sec_init_nosec_doall_caps(cred_a, obj, false, true);
 }
 
-/*
- * Perform a synchronous OSD operation.
- */
-int exofs_sync_op(struct osd_request *or, int timeout, uint8_t *credential)
+int exofs_read_kern(struct osd_dev *od, u8 *cred, struct osd_obj_id *obj,
+                   u64 offset, void *p, unsigned length)
 {
+       struct osd_request *or = osd_start_request(od, GFP_KERNEL);
+/*     struct osd_sense_info osi = {.key = 0};*/
        int ret;
 
-       or->timeout = timeout;
-       ret = osd_finalize_request(or, 0, credential, NULL);
-       if (ret) {
+       if (unlikely(!or)) {
+               EXOFS_DBGMSG("%s: osd_start_request failed.\n", __func__);
+               return -ENOMEM;
+       }
+       ret = osd_req_read_kern(or, obj, offset, p, length);
+       if (unlikely(ret)) {
+               EXOFS_DBGMSG("%s: osd_req_read_kern failed.\n", __func__);
+               goto out;
+       }
+
+       ret = osd_finalize_request(or, 0, cred, NULL);
+       if (unlikely(ret)) {
                EXOFS_DBGMSG("Faild to osd_finalize_request() => %d\n", ret);
-               return ret;
+               goto out;
        }
 
        ret = osd_execute_request(or);
-
-       if (ret)
+       if (unlikely(ret))
                EXOFS_DBGMSG("osd_execute_request() => %d\n", ret);
        /* osd_req_decode_sense(or, ret); */
+
+out:
+       osd_end_request(or);
        return ret;
 }
 
+int exofs_get_io_state(struct exofs_layout *layout,
+                      struct exofs_io_state **pios)
+{
+       struct exofs_io_state *ios;
+
+       /*TODO: Maybe use kmem_cach per sbi of size
+        * exofs_io_state_size(layout->s_numdevs)
+        */
+       ios = kzalloc(exofs_io_state_size(layout->s_numdevs), GFP_KERNEL);
+       if (unlikely(!ios)) {
+               EXOFS_DBGMSG("Faild kzalloc bytes=%d\n",
+                            exofs_io_state_size(layout->s_numdevs));
+               *pios = NULL;
+               return -ENOMEM;
+       }
+
+       ios->layout = layout;
+       ios->obj.partition = layout->s_pid;
+       *pios = ios;
+       return 0;
+}
+
+void exofs_put_io_state(struct exofs_io_state *ios)
+{
+       if (ios) {
+               unsigned i;
+
+               for (i = 0; i < ios->numdevs; i++) {
+                       struct exofs_per_dev_state *per_dev = &ios->per_dev[i];
+
+                       if (per_dev->or)
+                               osd_end_request(per_dev->or);
+                       if (per_dev->bio)
+                               bio_put(per_dev->bio);
+               }
+
+               kfree(ios);
+       }
+}
+
+unsigned exofs_layout_od_id(struct exofs_layout *layout,
+                           osd_id obj_no, unsigned layout_index)
+{
+/*     switch (layout->lay_func) {
+       case LAYOUT_MOVING_WINDOW:
+       {*/
+               unsigned dev_mod = obj_no;
+
+               return (layout_index + dev_mod * layout->mirrors_p1) %
+                                                             layout->s_numdevs;
+/*     }
+       case LAYOUT_FUNC_IMPLICT:
+               return layout->devs[layout_index];
+       }*/
+}
+
+static inline struct osd_dev *exofs_ios_od(struct exofs_io_state *ios,
+                                          unsigned layout_index)
+{
+       return ios->layout->s_ods[
+               exofs_layout_od_id(ios->layout, ios->obj.id, layout_index)];
+}
+
+static void _sync_done(struct exofs_io_state *ios, void *p)
+{
+       struct completion *waiting = p;
+
+       complete(waiting);
+}
+
+static void _last_io(struct kref *kref)
+{
+       struct exofs_io_state *ios = container_of(
+                                       kref, struct exofs_io_state, kref);
+
+       ios->done(ios, ios->private);
+}
+
+static void _done_io(struct osd_request *or, void *p)
+{
+       struct exofs_io_state *ios = p;
+
+       kref_put(&ios->kref, _last_io);
+}
+
+static int exofs_io_execute(struct exofs_io_state *ios)
+{
+       DECLARE_COMPLETION_ONSTACK(wait);
+       bool sync = (ios->done == NULL);
+       int i, ret;
+
+       if (sync) {
+               ios->done = _sync_done;
+               ios->private = &wait;
+       }
+
+       for (i = 0; i < ios->numdevs; i++) {
+               struct osd_request *or = ios->per_dev[i].or;
+               if (unlikely(!or))
+                       continue;
+
+               ret = osd_finalize_request(or, 0, ios->cred, NULL);
+               if (unlikely(ret)) {
+                       EXOFS_DBGMSG("Faild to osd_finalize_request() => %d\n",
+                                    ret);
+                       return ret;
+               }
+       }
+
+       kref_init(&ios->kref);
+
+       for (i = 0; i < ios->numdevs; i++) {
+               struct osd_request *or = ios->per_dev[i].or;
+               if (unlikely(!or))
+                       continue;
+
+               kref_get(&ios->kref);
+               osd_execute_request_async(or, _done_io, ios);
+       }
+
+       kref_put(&ios->kref, _last_io);
+       ret = 0;
+
+       if (sync) {
+               wait_for_completion(&wait);
+               ret = exofs_check_io(ios, NULL);
+       }
+       return ret;
+}
+
+static void _clear_bio(struct bio *bio)
+{
+       struct bio_vec *bv;
+       unsigned i;
+
+       __bio_for_each_segment(bv, bio, i, 0) {
+               unsigned this_count = bv->bv_len;
+
+               if (likely(PAGE_SIZE == this_count))
+                       clear_highpage(bv->bv_page);
+               else
+                       zero_user(bv->bv_page, bv->bv_offset, this_count);
+       }
+}
+
+int exofs_check_io(struct exofs_io_state *ios, u64 *resid)
+{
+       enum osd_err_priority acumulated_osd_err = 0;
+       int acumulated_lin_err = 0;
+       int i;
+
+       for (i = 0; i < ios->numdevs; i++) {
+               struct osd_sense_info osi;
+               struct osd_request *or = ios->per_dev[i].or;
+               int ret;
+
+               if (unlikely(!or))
+                       continue;
+
+               ret = osd_req_decode_sense(or, &osi);
+               if (likely(!ret))
+                       continue;
+
+               if (OSD_ERR_PRI_CLEAR_PAGES == osi.osd_err_pri) {
+                       /* start read offset passed endof file */
+                       _clear_bio(ios->per_dev[i].bio);
+                       EXOFS_DBGMSG("start read offset passed end of file "
+                               "offset=0x%llx, length=0x%llx\n",
+                               _LLU(ios->per_dev[i].offset),
+                               _LLU(ios->per_dev[i].length));
+
+                       continue; /* we recovered */
+               }
+
+               if (osi.osd_err_pri >= acumulated_osd_err) {
+                       acumulated_osd_err = osi.osd_err_pri;
+                       acumulated_lin_err = ret;
+               }
+       }
+
+       /* TODO: raid specific residual calculations */
+       if (resid) {
+               if (likely(!acumulated_lin_err))
+                       *resid = 0;
+               else
+                       *resid = ios->length;
+       }
+
+       return acumulated_lin_err;
+}
+
 /*
- * Perform an asynchronous OSD operation.
+ * L - logical offset into the file
+ *
+ * U - The number of bytes in a stripe within a group
+ *
+ *     U = stripe_unit * group_width
+ *
+ * T - The number of bytes striped within a group of component objects
+ *     (before advancing to the next group)
+ *
+ *     T = stripe_unit * group_width * group_depth
+ *
+ * S - The number of bytes striped across all component objects
+ *     before the pattern repeats
+ *
+ *     S = stripe_unit * group_width * group_depth * group_count
+ *
+ * M - The "major" (i.e., across all components) stripe number
+ *
+ *     M = L / S
+ *
+ * G - Counts the groups from the beginning of the major stripe
+ *
+ *     G = (L - (M * S)) / T   [or (L % S) / T]
+ *
+ * H - The byte offset within the group
+ *
+ *     H = (L - (M * S)) % T   [or (L % S) % T]
+ *
+ * N - The "minor" (i.e., across the group) stripe number
+ *
+ *     N = H / U
+ *
+ * C - The component index coresponding to L
+ *
+ *     C = (H - (N * U)) / stripe_unit + G * group_width
+ *     [or (L % U) / stripe_unit + G * group_width]
+ *
+ * O - The component offset coresponding to L
+ *
+ *     O = L % stripe_unit + N * stripe_unit + M * group_depth * stripe_unit
  */
-int exofs_async_op(struct osd_request *or, osd_req_done_fn *async_done,
-                  void *caller_context, u8 *cred)
+struct _striping_info {
+       u64 obj_offset;
+       u64 group_length;
+       u64 total_group_length;
+       u64 Major;
+       unsigned dev;
+       unsigned unit_off;
+};
+
+static void _calc_stripe_info(struct exofs_io_state *ios, u64 file_offset,
+                             struct _striping_info *si)
+{
+       u32     stripe_unit = ios->layout->stripe_unit;
+       u32     group_width = ios->layout->group_width;
+       u64     group_depth = ios->layout->group_depth;
+
+       u32     U = stripe_unit * group_width;
+       u64     T = U * group_depth;
+       u64     S = T * ios->layout->group_count;
+       u64     M = div64_u64(file_offset, S);
+
+       /*
+       G = (L - (M * S)) / T
+       H = (L - (M * S)) % T
+       */
+       u64     LmodS = file_offset - M * S;
+       u32     G = div64_u64(LmodS, T);
+       u64     H = LmodS - G * T;
+
+       u32     N = div_u64(H, U);
+
+       /* "H - (N * U)" is just "H % U" so it's bound to u32 */
+       si->dev = (u32)(H - (N * U)) / stripe_unit + G * group_width;
+       si->dev *= ios->layout->mirrors_p1;
+
+       div_u64_rem(file_offset, stripe_unit, &si->unit_off);
+
+       si->obj_offset = si->unit_off + (N * stripe_unit) +
+                                 (M * group_depth * stripe_unit);
+
+       si->group_length = T - H;
+       si->total_group_length = T;
+       si->Major = M;
+}
+
+static int _add_stripe_unit(struct exofs_io_state *ios,  unsigned *cur_pg,
+               unsigned pgbase, struct exofs_per_dev_state *per_dev,
+               int cur_len)
+{
+       unsigned pg = *cur_pg;
+       struct request_queue *q =
+                       osd_request_queue(exofs_ios_od(ios, per_dev->dev));
+
+       per_dev->length += cur_len;
+
+       if (per_dev->bio == NULL) {
+               unsigned pages_in_stripe = ios->layout->group_width *
+                                       (ios->layout->stripe_unit / PAGE_SIZE);
+               unsigned bio_size = (ios->nr_pages + pages_in_stripe) /
+                                               ios->layout->group_width;
+
+               per_dev->bio = bio_kmalloc(GFP_KERNEL, bio_size);
+               if (unlikely(!per_dev->bio)) {
+                       EXOFS_DBGMSG("Faild to allocate BIO size=%u\n",
+                                    bio_size);
+                       return -ENOMEM;
+               }
+       }
+
+       while (cur_len > 0) {
+               unsigned pglen = min_t(unsigned, PAGE_SIZE - pgbase, cur_len);
+               unsigned added_len;
+
+               BUG_ON(ios->nr_pages <= pg);
+               cur_len -= pglen;
+
+               added_len = bio_add_pc_page(q, per_dev->bio, ios->pages[pg],
+                                           pglen, pgbase);
+               if (unlikely(pglen != added_len))
+                       return -ENOMEM;
+               pgbase = 0;
+               ++pg;
+       }
+       BUG_ON(cur_len);
+
+       *cur_pg = pg;
+       return 0;
+}
+
+static int _prepare_one_group(struct exofs_io_state *ios, u64 length,
+                             struct _striping_info *si, unsigned first_comp)
+{
+       unsigned stripe_unit = ios->layout->stripe_unit;
+       unsigned mirrors_p1 = ios->layout->mirrors_p1;
+       unsigned devs_in_group = ios->layout->group_width * mirrors_p1;
+       unsigned dev = si->dev;
+       unsigned first_dev = dev - (dev % devs_in_group);
+       unsigned comp = first_comp + (dev - first_dev);
+       unsigned max_comp = ios->numdevs ? ios->numdevs - mirrors_p1 : 0;
+       unsigned cur_pg = ios->pages_consumed;
+       int ret = 0;
+
+       while (length) {
+               struct exofs_per_dev_state *per_dev = &ios->per_dev[comp];
+               unsigned cur_len, page_off = 0;
+
+               if (!per_dev->length) {
+                       per_dev->dev = dev;
+                       if (dev < si->dev) {
+                               per_dev->offset = si->obj_offset + stripe_unit -
+                                                                  si->unit_off;
+                               cur_len = stripe_unit;
+                       } else if (dev == si->dev) {
+                               per_dev->offset = si->obj_offset;
+                               cur_len = stripe_unit - si->unit_off;
+                               page_off = si->unit_off & ~PAGE_MASK;
+                               BUG_ON(page_off && (page_off != ios->pgbase));
+                       } else { /* dev > si->dev */
+                               per_dev->offset = si->obj_offset - si->unit_off;
+                               cur_len = stripe_unit;
+                       }
+
+                       if (max_comp < comp)
+                               max_comp = comp;
+
+                       dev += mirrors_p1;
+                       dev = (dev % devs_in_group) + first_dev;
+               } else {
+                       cur_len = stripe_unit;
+               }
+               if (cur_len >= length)
+                       cur_len = length;
+
+               ret = _add_stripe_unit(ios, &cur_pg, page_off , per_dev,
+                                      cur_len);
+               if (unlikely(ret))
+                       goto out;
+
+               comp += mirrors_p1;
+               comp = (comp % devs_in_group) + first_comp;
+
+               length -= cur_len;
+       }
+out:
+       ios->numdevs = max_comp + mirrors_p1;
+       ios->pages_consumed = cur_pg;
+       return ret;
+}
+
+static int _prepare_for_striping(struct exofs_io_state *ios)
+{
+       u64 length = ios->length;
+       struct _striping_info si;
+       unsigned devs_in_group = ios->layout->group_width *
+                                ios->layout->mirrors_p1;
+       unsigned first_comp = 0;
+       int ret = 0;
+
+       _calc_stripe_info(ios, ios->offset, &si);
+
+       if (!ios->pages) {
+               if (ios->kern_buff) {
+                       struct exofs_per_dev_state *per_dev = &ios->per_dev[0];
+
+                       per_dev->offset = si.obj_offset;
+                       per_dev->dev = si.dev;
+
+                       /* no cross device without page array */
+                       BUG_ON((ios->layout->group_width > 1) &&
+                              (si.unit_off + ios->length >
+                               ios->layout->stripe_unit));
+               }
+               ios->numdevs = ios->layout->mirrors_p1;
+               return 0;
+       }
+
+       while (length) {
+               if (length < si.group_length)
+                       si.group_length = length;
+
+               ret = _prepare_one_group(ios, si.group_length, &si, first_comp);
+               if (unlikely(ret))
+                       goto out;
+
+               length -= si.group_length;
+
+               si.group_length = si.total_group_length;
+               si.unit_off = 0;
+               ++si.Major;
+               si.obj_offset = si.Major * ios->layout->stripe_unit *
+                                               ios->layout->group_depth;
+
+               si.dev = (si.dev - (si.dev % devs_in_group)) + devs_in_group;
+               si.dev %= ios->layout->s_numdevs;
+
+               first_comp += devs_in_group;
+               first_comp %= ios->layout->s_numdevs;
+       }
+
+out:
+       return ret;
+}
+
+int exofs_sbi_create(struct exofs_io_state *ios)
 {
+       int i, ret;
+
+       for (i = 0; i < ios->layout->s_numdevs; i++) {
+               struct osd_request *or;
+
+               or = osd_start_request(exofs_ios_od(ios, i), GFP_KERNEL);
+               if (unlikely(!or)) {
+                       EXOFS_ERR("%s: osd_start_request failed\n", __func__);
+                       ret = -ENOMEM;
+                       goto out;
+               }
+               ios->per_dev[i].or = or;
+               ios->numdevs++;
+
+               osd_req_create_object(or, &ios->obj);
+       }
+       ret = exofs_io_execute(ios);
+
+out:
+       return ret;
+}
+
+int exofs_sbi_remove(struct exofs_io_state *ios)
+{
+       int i, ret;
+
+       for (i = 0; i < ios->layout->s_numdevs; i++) {
+               struct osd_request *or;
+
+               or = osd_start_request(exofs_ios_od(ios, i), GFP_KERNEL);
+               if (unlikely(!or)) {
+                       EXOFS_ERR("%s: osd_start_request failed\n", __func__);
+                       ret = -ENOMEM;
+                       goto out;
+               }
+               ios->per_dev[i].or = or;
+               ios->numdevs++;
+
+               osd_req_remove_object(or, &ios->obj);
+       }
+       ret = exofs_io_execute(ios);
+
+out:
+       return ret;
+}
+
+static int _sbi_write_mirror(struct exofs_io_state *ios, int cur_comp)
+{
+       struct exofs_per_dev_state *master_dev = &ios->per_dev[cur_comp];
+       unsigned dev = ios->per_dev[cur_comp].dev;
+       unsigned last_comp = cur_comp + ios->layout->mirrors_p1;
+       int ret = 0;
+
+       if (ios->pages && !master_dev->length)
+               return 0; /* Just an empty slot */
+
+       for (; cur_comp < last_comp; ++cur_comp, ++dev) {
+               struct exofs_per_dev_state *per_dev = &ios->per_dev[cur_comp];
+               struct osd_request *or;
+
+               or = osd_start_request(exofs_ios_od(ios, dev), GFP_KERNEL);
+               if (unlikely(!or)) {
+                       EXOFS_ERR("%s: osd_start_request failed\n", __func__);
+                       ret = -ENOMEM;
+                       goto out;
+               }
+               per_dev->or = or;
+               per_dev->offset = master_dev->offset;
+
+               if (ios->pages) {
+                       struct bio *bio;
+
+                       if (per_dev != master_dev) {
+                               bio = bio_kmalloc(GFP_KERNEL,
+                                                 master_dev->bio->bi_max_vecs);
+                               if (unlikely(!bio)) {
+                                       EXOFS_DBGMSG(
+                                             "Faild to allocate BIO size=%u\n",
+                                             master_dev->bio->bi_max_vecs);
+                                       ret = -ENOMEM;
+                                       goto out;
+                               }
+
+                               __bio_clone(bio, master_dev->bio);
+                               bio->bi_bdev = NULL;
+                               bio->bi_next = NULL;
+                               per_dev->length = master_dev->length;
+                               per_dev->bio =  bio;
+                               per_dev->dev = dev;
+                       } else {
+                               bio = master_dev->bio;
+                               /* FIXME: bio_set_dir() */
+                               bio->bi_rw |= (1 << BIO_RW);
+                       }
+
+                       osd_req_write(or, &ios->obj, per_dev->offset, bio,
+                                     per_dev->length);
+                       EXOFS_DBGMSG("write(0x%llx) offset=0x%llx "
+                                     "length=0x%llx dev=%d\n",
+                                    _LLU(ios->obj.id), _LLU(per_dev->offset),
+                                    _LLU(per_dev->length), dev);
+               } else if (ios->kern_buff) {
+                       ret = osd_req_write_kern(or, &ios->obj, per_dev->offset,
+                                          ios->kern_buff, ios->length);
+                       if (unlikely(ret))
+                               goto out;
+                       EXOFS_DBGMSG2("write_kern(0x%llx) offset=0x%llx "
+                                     "length=0x%llx dev=%d\n",
+                                    _LLU(ios->obj.id), _LLU(per_dev->offset),
+                                    _LLU(ios->length), dev);
+               } else {
+                       osd_req_set_attributes(or, &ios->obj);
+                       EXOFS_DBGMSG2("obj(0x%llx) set_attributes=%d dev=%d\n",
+                                    _LLU(ios->obj.id), ios->out_attr_len, dev);
+               }
+
+               if (ios->out_attr)
+                       osd_req_add_set_attr_list(or, ios->out_attr,
+                                                 ios->out_attr_len);
+
+               if (ios->in_attr)
+                       osd_req_add_get_attr_list(or, ios->in_attr,
+                                                 ios->in_attr_len);
+       }
+
+out:
+       return ret;
+}
+
+int exofs_sbi_write(struct exofs_io_state *ios)
+{
+       int i;
        int ret;
 
-       ret = osd_finalize_request(or, 0, cred, NULL);
-       if (ret) {
-               EXOFS_DBGMSG("Faild to osd_finalize_request() => %d\n", ret);
+       ret = _prepare_for_striping(ios);
+       if (unlikely(ret))
                return ret;
+
+       for (i = 0; i < ios->numdevs; i += ios->layout->mirrors_p1) {
+               ret = _sbi_write_mirror(ios, i);
+               if (unlikely(ret))
+                       return ret;
        }
 
-       ret = osd_execute_request_async(or, async_done, caller_context);
+       ret = exofs_io_execute(ios);
+       return ret;
+}
+
+static int _sbi_read_mirror(struct exofs_io_state *ios, unsigned cur_comp)
+{
+       struct osd_request *or;
+       struct exofs_per_dev_state *per_dev = &ios->per_dev[cur_comp];
+       unsigned first_dev = (unsigned)ios->obj.id;
+
+       if (ios->pages && !per_dev->length)
+               return 0; /* Just an empty slot */
+
+       first_dev = per_dev->dev + first_dev % ios->layout->mirrors_p1;
+       or = osd_start_request(exofs_ios_od(ios, first_dev), GFP_KERNEL);
+       if (unlikely(!or)) {
+               EXOFS_ERR("%s: osd_start_request failed\n", __func__);
+               return -ENOMEM;
+       }
+       per_dev->or = or;
 
-       if (ret)
-               EXOFS_DBGMSG("osd_execute_request_async() => %d\n", ret);
+       if (ios->pages) {
+               osd_req_read(or, &ios->obj, per_dev->offset,
+                               per_dev->bio, per_dev->length);
+               EXOFS_DBGMSG("read(0x%llx) offset=0x%llx length=0x%llx"
+                            " dev=%d\n", _LLU(ios->obj.id),
+                            _LLU(per_dev->offset), _LLU(per_dev->length),
+                            first_dev);
+       } else if (ios->kern_buff) {
+               int ret = osd_req_read_kern(or, &ios->obj, per_dev->offset,
+                                           ios->kern_buff, ios->length);
+               EXOFS_DBGMSG2("read_kern(0x%llx) offset=0x%llx "
+                             "length=0x%llx dev=%d ret=>%d\n",
+                             _LLU(ios->obj.id), _LLU(per_dev->offset),
+                             _LLU(ios->length), first_dev, ret);
+               if (unlikely(ret))
+                       return ret;
+       } else {
+               osd_req_get_attributes(or, &ios->obj);
+               EXOFS_DBGMSG2("obj(0x%llx) get_attributes=%d dev=%d\n",
+                             _LLU(ios->obj.id), ios->in_attr_len, first_dev);
+       }
+       if (ios->out_attr)
+               osd_req_add_set_attr_list(or, ios->out_attr, ios->out_attr_len);
+
+       if (ios->in_attr)
+               osd_req_add_get_attr_list(or, ios->in_attr, ios->in_attr_len);
+
+       return 0;
+}
+
+int exofs_sbi_read(struct exofs_io_state *ios)
+{
+       int i;
+       int ret;
+
+       ret = _prepare_for_striping(ios);
+       if (unlikely(ret))
+               return ret;
+
+       for (i = 0; i < ios->numdevs; i += ios->layout->mirrors_p1) {
+               ret = _sbi_read_mirror(ios, i);
+               if (unlikely(ret))
+                       return ret;
+       }
+
+       ret = exofs_io_execute(ios);
        return ret;
 }
 
-int extract_attr_from_req(struct osd_request *or, struct osd_attr *attr)
+int extract_attr_from_ios(struct exofs_io_state *ios, struct osd_attr *attr)
 {
        struct osd_attr cur_attr = {.attr_page = 0}; /* start with zeros */
        void *iter = NULL;
@@ -112,7 +728,8 @@ int extract_attr_from_req(struct osd_request *or, struct osd_attr *attr)
 
        do {
                nelem = 1;
-               osd_req_decode_get_attr_list(or, &cur_attr, &nelem, &iter);
+               osd_req_decode_get_attr_list(ios->per_dev[0].or,
+                                            &cur_attr, &nelem, &iter);
                if ((cur_attr.attr_page == attr->attr_page) &&
                    (cur_attr.attr_id == attr->attr_id)) {
                        attr->len = cur_attr.len;
@@ -123,3 +740,83 @@ int extract_attr_from_req(struct osd_request *or, struct osd_attr *attr)
 
        return -EIO;
 }
+
+static int _truncate_mirrors(struct exofs_io_state *ios, unsigned cur_comp,
+                            struct osd_attr *attr)
+{
+       int last_comp = cur_comp + ios->layout->mirrors_p1;
+
+       for (; cur_comp < last_comp; ++cur_comp) {
+               struct exofs_per_dev_state *per_dev = &ios->per_dev[cur_comp];
+               struct osd_request *or;
+
+               or = osd_start_request(exofs_ios_od(ios, cur_comp), GFP_KERNEL);
+               if (unlikely(!or)) {
+                       EXOFS_ERR("%s: osd_start_request failed\n", __func__);
+                       return -ENOMEM;
+               }
+               per_dev->or = or;
+
+               osd_req_set_attributes(or, &ios->obj);
+               osd_req_add_set_attr_list(or, attr, 1);
+       }
+
+       return 0;
+}
+
+int exofs_oi_truncate(struct exofs_i_info *oi, u64 size)
+{
+       struct exofs_sb_info *sbi = oi->vfs_inode.i_sb->s_fs_info;
+       struct exofs_io_state *ios;
+       struct exofs_trunc_attr {
+               struct osd_attr attr;
+               __be64 newsize;
+       } *size_attrs;
+       struct _striping_info si;
+       int i, ret;
+
+       ret = exofs_get_io_state(&sbi->layout, &ios);
+       if (unlikely(ret))
+               return ret;
+
+       size_attrs = kcalloc(ios->layout->group_width, sizeof(*size_attrs),
+                            GFP_KERNEL);
+       if (unlikely(!size_attrs)) {
+               ret = -ENOMEM;
+               goto out;
+       }
+
+       ios->obj.id = exofs_oi_objno(oi);
+       ios->cred = oi->i_cred;
+
+       ios->numdevs = ios->layout->s_numdevs;
+       _calc_stripe_info(ios, size, &si);
+
+       for (i = 0; i < ios->layout->group_width; ++i) {
+               struct exofs_trunc_attr *size_attr = &size_attrs[i];
+               u64 obj_size;
+
+               if (i < si.dev)
+                       obj_size = si.obj_offset +
+                                       ios->layout->stripe_unit - si.unit_off;
+               else if (i == si.dev)
+                       obj_size = si.obj_offset;
+               else /* i > si.dev */
+                       obj_size = si.obj_offset - si.unit_off;
+
+               size_attr->newsize = cpu_to_be64(obj_size);
+               size_attr->attr = g_attr_logical_length;
+               size_attr->attr.val_ptr = &size_attr->newsize;
+
+               ret = _truncate_mirrors(ios, i * ios->layout->mirrors_p1,
+                                       &size_attr->attr);
+               if (unlikely(ret))
+                       goto out;
+       }
+       ret = exofs_io_execute(ios);
+
+out:
+       kfree(size_attrs);
+       exofs_put_io_state(ios);
+       return ret;
+}