diff --git a/osd.cpp b/osd.cpp
index d925aadff..c39918fea 100644
--- a/osd.cpp
+++ b/osd.cpp
@@ -332,11 +332,11 @@ void osd_t::exec_op(osd_op_t *cur_op)
     }
     else if (cur_op->req.hdr.opcode == OSD_OP_READ)
     {
-        exec_primary_read(cur_op);
+        continue_primary_read(cur_op);
     }
     else if (cur_op->req.hdr.opcode == OSD_OP_WRITE)
     {
-        exec_primary_write(cur_op);
+        continue_primary_write(cur_op);
     }
     else
     {
diff --git a/osd.h b/osd.h
index a8e91532e..264b3dff0 100644
--- a/osd.h
+++ b/osd.h
@@ -236,12 +236,12 @@ class osd_t
 
     // primary ops
     bool prepare_primary_rw(osd_op_t *cur_op);
-    void exec_primary_read(osd_op_t *cur_op);
-    void exec_primary_write(osd_op_t *cur_op);
+    void continue_primary_read(osd_op_t *cur_op);
+    void continue_primary_write(osd_op_t *cur_op);
     void exec_primary_sync(osd_op_t *cur_op);
     void finish_primary_op(osd_op_t *cur_op, int retval);
-    void handle_primary_read_subop(osd_op_t *cur_op, int ok);
-    void submit_read_subops(int read_pg_size, const uint64_t* osd_set, osd_op_t *cur_op);
+    void handle_primary_subop(osd_op_t *cur_op, int ok, uint64_t version);
+    void submit_primary_subops(int submit_type, int read_pg_size, const uint64_t* osd_set, osd_op_t *cur_op);
 public:
     osd_t(blockstore_config_t & config, blockstore_t *bs, ring_loop_t *ringloop);
     ~osd_t();
diff --git a/osd_primary.cpp b/osd_primary.cpp
index 799780888..912549f11 100644
--- a/osd_primary.cpp
+++ b/osd_primary.cpp
@@ -1,6 +1,10 @@
 #include "osd.h"
 #include "osd_rmw.h"
 
+#define SUBMIT_READ 0
+#define SUBMIT_RMW_READ 1
+#define SUBMIT_WRITE 2
+
 // read: read directly or read paired stripe(s), reconstruct, return
 // write: read paired stripe(s), modify, write
 //
@@ -12,14 +16,15 @@
 
 struct osd_primary_op_data_t
 {
+    int st = 0;
     pg_num_t pg_num;
     object_id oid;
     uint64_t target_ver;
+    uint64_t fact_ver = 0;
     int n_subops = 0, done = 0, errors = 0;
     int degraded = 0, pg_size, pg_minsize;
     osd_rmw_stripe_t *stripes;
     osd_op_t *subops = NULL;
-    bool should_read_version = false;
 };
 
 void osd_t::finish_primary_op(osd_op_t *cur_op, int retval)
@@ -71,127 +76,126 @@ bool osd_t::prepare_primary_rw(osd_op_t *cur_op)
     return true;
 }
 
-void osd_t::exec_primary_read(osd_op_t *cur_op)
+void osd_t::continue_primary_read(osd_op_t *cur_op)
 {
-    if (!prepare_primary_rw(cur_op))
+    if (!cur_op->op_data && !prepare_primary_rw(cur_op))
     {
         return;
     }
-    osd_primary_op_data_t *op_data = (osd_primary_op_data_t*)cur_op->op_data;
-    auto & pg = pgs[op_data->pg_num];
-    for (int role = 0; role < pg.pg_minsize; role++)
+    osd_primary_op_data_t *op_data = cur_op->op_data;
+    if (op_data->st == 1)      goto resume_1;
+    else if (op_data->st == 2) goto resume_2;
     {
-        op_data->stripes[role].read_start = op_data->stripes[role].req_start;
-        op_data->stripes[role].read_end = op_data->stripes[role].req_end;
-    }
-    // Determine version
-    auto vo_it = pg.ver_override.find(op_data->oid);
-    op_data->target_ver = vo_it != pg.ver_override.end() ? vo_it->second : UINT64_MAX;
-    if (pg.state == PG_ACTIVE)
-    {
-        // Fast happy-path
-        cur_op->buf = alloc_read_buffer(op_data->stripes, pg.pg_minsize, 0);
-        submit_read_subops(pg.pg_minsize, pg.cur_set.data(), cur_op);
-        cur_op->send_list.push_back(cur_op->buf, cur_op->req.rw.len);
-    }
-    else
-    {
-        // PG may be degraded or have misplaced objects
-        auto st_it = pg.obj_states.find(op_data->oid);
-        uint64_t* cur_set = (st_it != pg.obj_states.end()
-            ? st_it->second->read_target.data()
-            : pg.cur_set.data());
-        if (extend_missing_stripes(op_data->stripes, cur_set, pg.pg_minsize, pg.pg_size) < 0)
+        auto & pg = pgs[op_data->pg_num];
+        for (int role = 0; role < pg.pg_minsize; role++)
         {
-            free(op_data);
-            finish_primary_op(cur_op, -EIO);
-            return;
+            op_data->stripes[role].read_start = op_data->stripes[role].req_start;
+            op_data->stripes[role].read_end = op_data->stripes[role].req_end;
         }
-        // Submit reads
-        cur_op->buf = alloc_read_buffer(op_data->stripes, pg.pg_size, 0);
-        submit_read_subops(pg.pg_size, cur_set, cur_op);
-        op_data->pg_minsize = pg.pg_minsize;
-        op_data->pg_size = pg.pg_size;
-        op_data->degraded = 1;
-    }
-}
-
-void osd_t::handle_primary_read_subop(osd_op_t *cur_op, int ok)
-{
-    osd_primary_op_data_t *op_data = (osd_primary_op_data_t*)cur_op->op_data;
-    if (!ok)
-        op_data->errors++;
-    else
-        op_data->done++;
-    if ((op_data->errors + op_data->done) >= op_data->n_subops)
-    {
-        delete[] op_data->subops;
-        op_data->subops = NULL;
-        if (op_data->errors > 0)
+        // Determine version
+        auto vo_it = pg.ver_override.find(op_data->oid);
+        op_data->target_ver = vo_it != pg.ver_override.end() ? vo_it->second : UINT64_MAX;
+        if (pg.state == PG_ACTIVE)
         {
-            free(op_data);
-            cur_op->op_data = NULL;
-            finish_primary_op(cur_op, -EIO);
-            return;
+            // Fast happy-path
+            cur_op->buf = alloc_read_buffer(op_data->stripes, pg.pg_minsize, 0);
+            submit_primary_subops(SUBMIT_READ, pg.pg_minsize, pg.cur_set.data(), cur_op);
+            cur_op->send_list.push_back(cur_op->buf, cur_op->req.rw.len);
+            op_data->st = 1;
         }
-        if (op_data->degraded)
+        else
         {
-            // Reconstruct missing stripes
-            // FIXME: Always EC(k+1) by now. Add different coding schemes
-            osd_rmw_stripe_t *stripes = op_data->stripes;
-            for (int role = 0; role < op_data->pg_minsize; role++)
+            // PG may be degraded or have misplaced objects
+            auto st_it = pg.obj_states.find(op_data->oid);
+            uint64_t* cur_set = (st_it != pg.obj_states.end()
+                ? st_it->second->read_target.data()
+                : pg.cur_set.data());
+            if (extend_missing_stripes(op_data->stripes, cur_set, pg.pg_minsize, pg.pg_size) < 0)
             {
-                if (stripes[role].read_end != 0 && stripes[role].missing)
-                {
-                    reconstruct_stripe(stripes, op_data->pg_size, role);
-                }
-                if (stripes[role].req_end != 0)
-                {
-                    // Send buffer in parts to avoid copying
-                    cur_op->send_list.push_back(
-                        stripes[role].read_buf + (stripes[role].read_start - stripes[role].req_start), stripes[role].req_end
-                    );
-                }
+                free(op_data);
+                finish_primary_op(cur_op, -EIO);
+                return;
             }
+            // Submit reads
+            op_data->pg_minsize = pg.pg_minsize;
+            op_data->pg_size = pg.pg_size;
+            op_data->degraded = 1;
+            cur_op->buf = alloc_read_buffer(op_data->stripes, pg.pg_size, 0);
+            submit_primary_subops(SUBMIT_READ, pg.pg_size, cur_set, cur_op);
+            op_data->st = 1;
         }
+    }
+resume_1:
+    return;
+resume_2:
+    if (op_data->errors > 0)
+    {
         free(op_data);
         cur_op->op_data = NULL;
-        finish_primary_op(cur_op, cur_op->req.rw.len);
+        finish_primary_op(cur_op, -EIO);
+        return;
     }
+    if (op_data->degraded)
+    {
+        // Reconstruct missing stripes
+        // FIXME: Always EC(k+1) by now. Add different coding schemes
+        osd_rmw_stripe_t *stripes = op_data->stripes;
+        for (int role = 0; role < op_data->pg_minsize; role++)
+        {
+            if (stripes[role].read_end != 0 && stripes[role].missing)
+            {
+                reconstruct_stripe(stripes, op_data->pg_size, role);
+            }
+            if (stripes[role].req_end != 0)
+            {
+                // Send buffer in parts to avoid copying
+                cur_op->send_list.push_back(
+                    stripes[role].read_buf + (stripes[role].req_start - stripes[role].read_start),
+                    stripes[role].req_end - stripes[role].req_start
+                );
+            }
+        }
+    }
+    free(op_data);
+    cur_op->op_data = NULL;
+    finish_primary_op(cur_op, cur_op->req.rw.len);
 }
 
-void osd_t::submit_read_subops(int read_pg_size, const uint64_t* osd_set, osd_op_t *cur_op)
+void osd_t::submit_primary_subops(int submit_type, int pg_size, const uint64_t* osd_set, osd_op_t *cur_op)
 {
-    osd_primary_op_data_t *op_data = (osd_primary_op_data_t*)cur_op->op_data;
+    bool w = submit_type == SUBMIT_WRITE;
+    osd_primary_op_data_t *op_data = cur_op->op_data;
     osd_rmw_stripe_t *stripes = op_data->stripes;
     // Allocate subops
-    int n_subops = 0, force_read = -1;
-    for (int role = 0; role < read_pg_size; role++)
+    int n_subops = 0, zero_read = -1;
+    for (int role = 0; role < pg_size; role++)
     {
-        if (osd_set[role] == this->osd_num || osd_set[role] != 0 && force_read == -1)
+        if (osd_set[role] == this->osd_num || osd_set[role] != 0 && zero_read == -1)
         {
-            force_read = role;
+            zero_read = role;
         }
-        if (osd_set[role] != 0 && stripes[role].read_end != 0)
+        if (osd_set[role] != 0 &&
+            (w ? stripes[role].write_end : stripes[role].read_end) != 0)
         {
             n_subops++;
         }
     }
-    if (!n_subops && op_data->should_read_version)
+    if (!n_subops && submit_type == SUBMIT_RMW_READ)
     {
         n_subops = 1;
     }
     else
     {
-        force_read = -1;
+        zero_read = -1;
     }
     osd_op_t *subops = new osd_op_t[n_subops];
+    op_data->done = op_data->errors = 0;
     op_data->n_subops = n_subops;
     op_data->subops = subops;
     int subop = 0;
-    for (int role = 0; role < read_pg_size; role++)
+    for (int role = 0; role < pg_size; role++)
     {
-        if (stripes[role].read_end == 0 && force_read != role)
+        if ((submit_type == SUBMIT_WRITE ? stripes[role].write_end : stripes[role].read_end) == 0 && zero_read != role)
         {
             continue;
         }
@@ -201,19 +205,19 @@ void osd_t::submit_read_subops(int read_pg_size, const uint64_t* osd_set, osd_op
             if (role_osd_num == this->osd_num)
             {
                 subops[subop].bs_op = new blockstore_op_t({
-                    .opcode = BS_OP_READ,
+                    .opcode = (uint64_t)(w ? BS_OP_WRITE : BS_OP_READ),
                     .callback = [cur_op, this](blockstore_op_t *subop)
                     {
-                        handle_primary_read_subop(cur_op, subop->retval == subop->len);
+                        handle_primary_subop(cur_op, subop->retval == subop->len, subop->version);
                     },
                     .oid = {
                         .inode = op_data->oid.inode,
                         .stripe = op_data->oid.stripe | role,
                     },
-                    .version = op_data->target_ver,
-                    .offset = stripes[role].read_start,
-                    .len = stripes[role].read_end - stripes[role].read_start,
-                    .buf = stripes[role].read_buf,
+                    .version = w ? 0 : (submit_type == SUBMIT_RMW_READ ? UINT64_MAX : op_data->target_ver),
+                    .offset = w ? stripes[role].write_start : stripes[role].read_start,
+                    .len = w ? stripes[role].write_end - stripes[role].write_start : stripes[role].read_end - stripes[role].read_start,
+                    .buf = w ? stripes[role].write_buf : stripes[role].read_buf,
                 });
                 bs->enqueue_op(subops[subop].bs_op);
             }
@@ -225,22 +229,22 @@ void osd_t::submit_read_subops(int read_pg_size, const uint64_t* osd_set, osd_op
                     .header = {
                         .magic = SECONDARY_OSD_OP_MAGIC,
                         .id = this->next_subop_id++,
-                        .opcode = OSD_OP_SECONDARY_READ,
+                        .opcode = (uint64_t)(w ? OSD_OP_SECONDARY_WRITE : OSD_OP_SECONDARY_READ),
                     },
                     .oid = {
                         .inode = op_data->oid.inode,
                         .stripe = op_data->oid.stripe | role,
                     },
-                    .version = op_data->target_ver,
-                    .offset = stripes[role].read_start,
-                    .len = stripes[role].read_end - stripes[role].read_start,
+                    .version = w ? 0 : (submit_type == SUBMIT_RMW_READ ? UINT64_MAX : op_data->target_ver),
+                    .offset = w ? stripes[role].write_start : stripes[role].read_start,
+                    .len = w ? stripes[role].write_end - stripes[role].write_start : stripes[role].read_end - stripes[role].read_start,
                 };
-                subops[subop].buf = stripes[role].read_buf;
+                subops[subop].buf = w ? stripes[role].write_buf : stripes[role].read_buf;
                 subops[subop].callback = [cur_op, this](osd_op_t *subop)
                 {
                     // so it doesn't get freed
                     subop->buf = NULL;
-                    handle_primary_read_subop(cur_op, subop->reply.hdr.retval == subop->req.sec_rw.len);
+                    handle_primary_subop(cur_op, subop->reply.hdr.retval == subop->req.sec_rw.len, subop->reply.sec_rw.version);
                 };
                 outbox_push(clients[subops[subop].peer_fd], &subops[subop]);
             }
@@ -249,47 +253,98 @@ void osd_t::submit_read_subops(int read_pg_size, const uint64_t* osd_set, osd_op
     }
 }
 
-void osd_t::exec_primary_write(osd_op_t *cur_op)
+void osd_t::handle_primary_subop(osd_op_t *cur_op, int ok, uint64_t version)
 {
-    if (!prepare_primary_rw(cur_op))
+    osd_primary_op_data_t *op_data = cur_op->op_data;
+    op_data->fact_ver = version;
+    if (!ok)
+    {
+        op_data->errors++;
+    }
+    else
+    {
+        op_data->done++;
+    }
+    if ((op_data->errors + op_data->done) >= op_data->n_subops)
+    {
+        delete[] op_data->subops;
+        op_data->subops = NULL;
+        op_data->st++;
+        if (cur_op->req.hdr.opcode == OSD_OP_READ)
+        {
+            continue_primary_read(cur_op);
+        }
+        else
+        {
+            continue_primary_write(cur_op);
+        }
+    }
+}
+
+void osd_t::continue_primary_write(osd_op_t *cur_op)
+{
+    if (!cur_op->op_data && !prepare_primary_rw(cur_op))
     {
         return;
     }
-    osd_primary_op_data_t *op_data = (osd_primary_op_data_t*)cur_op->op_data;
+    osd_primary_op_data_t *op_data = cur_op->op_data;
+    // FIXME: Handle operation cancel
     auto & pg = pgs[op_data->pg_num];
+    if (op_data->st == 1)      goto resume_1;
+    else if (op_data->st == 2) goto resume_2;
+    else if (op_data->st == 3) goto resume_3;
+    else if (op_data->st == 4) goto resume_4;
+    else if (op_data->st == 5) goto resume_5;
     // Check if actions are pending for this object
-    auto act_it = pg.obj_stab_actions.lower_bound((obj_piece_id_t){
-        .oid = op_data->oid,
-        .osd_num = 0,
-    });
-    if (act_it != pg.obj_stab_actions.end() &&
-        act_it->first.oid.inode == op_data->oid.inode &&
-        (act_it->first.oid.stripe & ~STRIPE_MASK) == op_data->oid.stripe)
     {
-        // FIXME postpone the request until actions are done
-        free(op_data);
-        finish_primary_op(cur_op, -EIO);
-        return;
+        auto act_it = pg.obj_stab_actions.lower_bound((obj_piece_id_t){
+            .oid = op_data->oid,
+            .osd_num = 0,
+        });
+        if (act_it != pg.obj_stab_actions.end() &&
+            act_it->first.oid.inode == op_data->oid.inode &&
+            (act_it->first.oid.stripe & ~STRIPE_MASK) == op_data->oid.stripe)
+        {
+            // FIXME postpone the request until actions are done
+            free(op_data);
+            finish_primary_op(cur_op, -EIO);
+            return;
+        }
     }
+resume_1:
     // Check if there are other write requests to the same object
-    
+    {
+        auto vo_it = pg.ver_override.find(op_data->oid);
+        if (vo_it != pg.ver_override.end())
+        {
+            op_data->st = 1;
+            //pg.write_queue.push_back(cur_op);
+            return;
+        }
+    }
     // Determine blocks to read
     cur_op->rmw_buf = calc_rmw_reads(cur_op->buf, op_data->stripes, pg.cur_set.data(), pg.pg_size, pg.pg_minsize, pg.pg_cursize);
-    op_data->should_read_version = true;
     // Read required blocks
-    submit_read_subops(pg.pg_size, pg.cur_set.data(), cur_op);
-    // ->>>>> Continue from the callback
+    submit_primary_subops(SUBMIT_RMW_READ, pg.pg_size, pg.cur_set.data(), cur_op);
+    op_data->st = 2;
+resume_2:
+    return;
+resume_3:
+    // Save version override
+    pg.ver_override[op_data->oid] = op_data->fact_ver;
     // Calculate parity
     calc_rmw_parity(op_data->stripes, op_data->pg_size);
-    // Save version override if degraded
-    
     // Send writes
-    
-    // ->>>>> Continue from the callback
+    submit_primary_subops(SUBMIT_WRITE, pg.pg_size, pg.cur_set.data(), cur_op);
+    op_data->st = 4;
+resume_4:
+    return;
+resume_5:
     // Remember version as unstable
     
     // Remove version override if degraded
     
+    finish_primary_op(cur_op, cur_op->req.rw.len);
 }
 
 void osd_t::exec_primary_sync(osd_op_t *cur_op)