Inline bitmaps

Handy for zero-copy RDMA tests (removes 4-byte s/g entries)

src/msgr_receive.cpp

@@ -207,20 +207,26 @@ void osd_messenger_t::handle_op_hdr(osd_client_t *cl)
     else if (cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE ||
         cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE)
     {
-        if (cur_op->req.sec_rw.attr_len > 0)
+        if (cur_op->req.sec_rw.bitmap_len > 0)
         {
-            if (cur_op->req.sec_rw.attr_len > sizeof(unsigned))
+            if (cur_op->req.sec_rw.bitmap_len > sizeof(void*))
-                cur_op->bitmap = cur_op->rmw_buf = malloc_or_die(cur_op->req.sec_rw.attr_len);
+                cur_op->bitmap = cur_op->rmw_buf = malloc_or_die(cur_op->req.sec_rw.bitmap_len);
             else
                 cur_op->bitmap = &cur_op->bmp_data;
-            cl->recv_list.push_back(cur_op->bitmap, cur_op->req.sec_rw.attr_len);
+            if (cur_op->req.sec_rw.bitmap_len <= 8)
+                memcpy(cur_op->bitmap, &cur_op->req.sec_rw.bitmap, cur_op->req.sec_rw.bitmap_len);
+            else
+            {
+                cl->recv_list.push_back(cur_op->bitmap, cur_op->req.sec_rw.bitmap_len);
+                cl->read_remaining += cur_op->req.sec_rw.bitmap_len;
+            }
         }
         if (cur_op->req.sec_rw.len > 0)
         {
             cur_op->buf = memalign_or_die(MEM_ALIGNMENT, cur_op->req.sec_rw.len);
             cl->recv_list.push_back(cur_op->buf, cur_op->req.sec_rw.len);
+            cl->read_remaining += cur_op->req.sec_rw.len;
         }
-        cl->read_remaining = cur_op->req.sec_rw.len + cur_op->req.sec_rw.attr_len;
     }
     else if (cur_op->req.hdr.opcode == OSD_OP_SEC_STABILIZE ||
         cur_op->req.hdr.opcode == OSD_OP_SEC_ROLLBACK)
@@ -295,7 +301,7 @@ bool osd_messenger_t::handle_reply_hdr(osd_client_t *cl)
     if (op->reply.hdr.opcode == OSD_OP_SEC_READ || op->reply.hdr.opcode == OSD_OP_READ)
     {
         // Read data. In this case we assume that the buffer is preallocated by the caller (!)
-        unsigned bmp_len = (op->reply.hdr.opcode == OSD_OP_SEC_READ ? op->reply.sec_rw.attr_len : op->reply.rw.bitmap_len);
+        unsigned bmp_len = (op->reply.hdr.opcode == OSD_OP_SEC_READ ? op->reply.sec_rw.bitmap_len : op->reply.rw.bitmap_len);
         unsigned expected_size = (op->reply.hdr.opcode == OSD_OP_SEC_READ ? op->req.sec_rw.len : op->req.rw.len);
         if (op->reply.hdr.retval >= 0 && (op->reply.hdr.retval != expected_size || bmp_len > op->bitmap_len))
         {
@@ -309,14 +315,24 @@ bool osd_messenger_t::handle_reply_hdr(osd_client_t *cl)
         if (op->reply.hdr.retval >= 0 && bmp_len > 0)
         {
             assert(op->bitmap);
-            cl->recv_list.push_back(op->bitmap, bmp_len);
+            if (bmp_len <= 8)
+            {
+                memcpy(op->bitmap, (op->reply.hdr.opcode == OSD_OP_SEC_READ
+                    ? &op->reply.sec_rw.bitmap
+                    : &op->reply.rw.bitmap), bmp_len);
+            }
+            else
+            {
+                cl->recv_list.push_back(op->bitmap, bmp_len);
+                cl->read_remaining += bmp_len;
+            }
         }
         if (op->reply.hdr.retval > 0)
         {
             assert(op->iov.count > 0);
             cl->recv_list.append(op->iov);
+            cl->read_remaining += op->reply.hdr.retval;
         }
-        cl->read_remaining = op->reply.hdr.retval + bmp_len;
         if (cl->read_remaining == 0)
         {
             goto reuse;

src/msgr_send.cpp

@@ -50,23 +50,37 @@ void osd_messenger_t::outbox_push(osd_op_t *cur_op)
     // Bitmap
     if (cur_op->op_type == OSD_OP_IN &&
         cur_op->req.hdr.opcode == OSD_OP_SEC_READ &&
-        cur_op->reply.sec_rw.attr_len > 0)
+        cur_op->reply.sec_rw.bitmap_len > 0)
     {
-        to_send_list.push_back((iovec){
+        if (cur_op->reply.sec_rw.bitmap_len <= 8)
-            .iov_base = cur_op->bitmap,
+        {
-            .iov_len = cur_op->reply.sec_rw.attr_len,
+            memcpy(&cur_op->reply.sec_rw.bitmap, cur_op->bitmap, cur_op->reply.sec_rw.bitmap_len);
-        });
+        }
-        to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
+        else
+        {
+            to_send_list.push_back((iovec){
+                .iov_base = cur_op->bitmap,
+                .iov_len = cur_op->reply.sec_rw.bitmap_len,
+            });
+            to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
+        }
     }
     else if (cur_op->op_type == OSD_OP_OUT &&
         (cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE || cur_op->req.hdr.opcode == OSD_OP_SEC_WRITE_STABLE) &&
-        cur_op->req.sec_rw.attr_len > 0)
+        cur_op->req.sec_rw.bitmap_len > 0)
     {
-        to_send_list.push_back((iovec){
+        if (cur_op->req.sec_rw.bitmap_len <= 8)
-            .iov_base = cur_op->bitmap,
+        {
-            .iov_len = cur_op->req.sec_rw.attr_len,
+            memcpy(&cur_op->req.sec_rw.bitmap, cur_op->bitmap, cur_op->req.sec_rw.bitmap_len);
-        });
+        }
-        to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
+        else
+        {
+            to_send_list.push_back((iovec){
+                .iov_base = cur_op->bitmap,
+                .iov_len = cur_op->req.sec_rw.attr_len,
+            });
+            to_outbox.push_back((msgr_sendp_t){ .op = cur_op, .flags = 0 });
+        }
     }
     // Operation data
     if ((cur_op->op_type == OSD_OP_IN

src/osd_ops.h

@@ -35,7 +35,7 @@
 #define MEM_ALIGNMENT               512
 #endif
 #define OSD_RW_MAX                  64*1024*1024
-#define OSD_PROTOCOL_VERSION        1
+#define OSD_PROTOCOL_VERSION        2
 
 // common request and reply headers
 struct __attribute__((__packed__)) osd_op_header_t
@@ -74,8 +74,10 @@ struct __attribute__((__packed__)) osd_op_sec_rw_t
     // length
     uint32_t len;
     // bitmap/attribute length - bitmap comes after header, but before data
-    uint32_t attr_len;
+    uint32_t bitmap_len;
     uint32_t pad0;
+    // inline bitmap (when it's no longer than 8 bytes)
+    uint64_t bitmap;
 };
 
 struct __attribute__((__packed__)) osd_reply_sec_rw_t
@@ -84,8 +86,10 @@ struct __attribute__((__packed__)) osd_reply_sec_rw_t
     // for reads and writes: assigned or read version number
     uint64_t version;
     // for reads: bitmap/attribute length (just to double-check)
-    uint32_t attr_len;
+    uint32_t bitmap_len;
     uint32_t pad0;
+    // inline bitmap (when it's no longer than 8 bytes)
+    uint64_t bitmap;
 };
 
 // delete object on the secondary OSD
@@ -199,6 +203,8 @@ struct __attribute__((__packed__)) osd_reply_rw_t
     // for reads: bitmap length
     uint32_t bitmap_len;
     uint32_t pad0;
+    // inline bitmap (when it's no longer than 8 bytes)
+    uint64_t bitmap;
 };
 
 // sync to the primary OSD

src/osd_primary.cpp

@@ -235,7 +235,10 @@ resume_2:
         {
             reconstruct_stripes_jerasure(stripes, op_data->pg_size, op_data->pg_data_size, clean_entry_bitmap_size);
         }
-        cur_op->iov.push_back(op_data->stripes[0].bmp_buf, cur_op->reply.rw.bitmap_len);
+        if (cur_op->reply.rw.bitmap_len <= 8)
+            memcpy(&cur_op->reply.rw.bitmap, op_data->stripes[0].bmp_buf, cur_op->reply.rw.bitmap_len);
+        else
+            cur_op->iov.push_back(op_data->stripes[0].bmp_buf, cur_op->reply.rw.bitmap_len);
         for (int role = 0; role < op_data->pg_size; role++)
         {
             if (stripes[role].req_end != 0)
@@ -250,7 +253,10 @@ resume_2:
     }
     else
     {
-        cur_op->iov.push_back(op_data->stripes[0].bmp_buf, cur_op->reply.rw.bitmap_len);
+        if (cur_op->reply.rw.bitmap_len <= 8)
+            memcpy(&cur_op->reply.rw.bitmap, op_data->stripes[0].bmp_buf, cur_op->reply.rw.bitmap_len);
+        else
+            cur_op->iov.push_back(op_data->stripes[0].bmp_buf, cur_op->reply.rw.bitmap_len);
         cur_op->iov.push_back(cur_op->buf, cur_op->req.rw.len);
     }
     finish_op(cur_op, cur_op->req.rw.len);

src/osd_primary_subops.cpp

@@ -200,7 +200,7 @@ int osd_t::submit_primary_subop_batch(int submit_type, inode_t inode, uint64_t o
                     .version = op_version,
                     .offset = wr ? stripes[stripe_num].write_start : stripes[stripe_num].read_start,
                     .len = wr ? stripes[stripe_num].write_end - stripes[stripe_num].write_start : stripes[stripe_num].read_end - stripes[stripe_num].read_start,
-                    .attr_len = wr ? clean_entry_bitmap_size : 0,
+                    .bitmap_len = wr ? clean_entry_bitmap_size : 0,
                 };
 #ifdef OSD_DEBUG
                 printf(

src/osd_secondary.cpp

@@ -20,9 +20,9 @@ void osd_t::secondary_op_callback(osd_op_t *op)
     if (op->req.hdr.opcode == OSD_OP_SEC_READ)
     {
         if (op->bs_op->retval >= 0)
-            op->reply.sec_rw.attr_len = clean_entry_bitmap_size;
+            op->reply.sec_rw.bitmap_len = clean_entry_bitmap_size;
         else
-            op->reply.sec_rw.attr_len = 0;
+            op->reply.sec_rw.bitmap_len = 0;
         if (op->bs_op->retval > 0)
             op->iov.push_back(op->buf, op->bs_op->retval);
     }
@@ -81,7 +81,7 @@ void osd_t::exec_secondary(osd_op_t *cur_op)
         if (cur_op->req.hdr.opcode == OSD_OP_SEC_READ)
         {
             // Allocate memory for the read operation
-            if (clean_entry_bitmap_size > sizeof(unsigned))
+            if (clean_entry_bitmap_size > sizeof(void*))
                 cur_op->bitmap = cur_op->rmw_buf = malloc_or_die(clean_entry_bitmap_size);
             else
                 cur_op->bitmap = &cur_op->bmp_data;