copy_file_range support in GlusterFS

    * libglusterfs changes to add new fop

    * Fuse changes:
      - Changes in fuse bridge xlator to receive and send responses

    * posix changes to perform the op on the backend filesystem

    * protocol and rpc changes for sending and receiving the fop

    * gfapi changes for performing the fop

    * tools: glfs-copy-file-range tool for testing copy_file_range fop

      - Although, copy_file_range support has been added to the upstream
	    fuse kernel module, no release has been made yet of a kernel
        which contains the support. It is expected to come in the
        upcoming release of linux-4.20

        So, as of now, executing copy_file_range fop on a fused based
        filesystem results in fuse kernel module sending read on the
	    source fd and write on the destination fd.

	    Therefore a small gfapi based tool has been written to be able
        test the copy_file_range fop. This tool is similar (in functionality)
	    to the example program given in copy_file_range man page.

	    So, running regular copy_file_range on a fuse mount point and
	    running gfapi based glfs-copy-file-range tool gives some idea about
	    how fast, the copy_file_range (or reflink) can be.

	    On the local machine this was the result obtained.

	    mount -t glusterfs workstation:new /mnt/glusterfs
	    [root@workstation ~]# cd /mnt/glusterfs/
	    [root@workstation glusterfs]# ls
	    file
	    [root@workstation glusterfs]# cd
	    [root@workstation ~]# time /tmp/a.out /mnt/glusterfs/file /mnt/glusterfs/new
	    real  0m6.495s
	    user  0m0.000s
	    sys   0m1.439s
	    [root@workstation ~]# time glfs-copy-file-range $(hostname) new /tmp/glfs.log /file /rrr
	    OPEN_SRC: opening /file is success
	    OPEN_DST: opening /rrr is success
	    FSTAT_SRC: fstat on /rrr is success
	    copy_file_range successful

        real  0m0.309s
        user  0m0.039s
        sys   0m0.017s

        This tool needs following arguments
         1) hostname
         2) volume name
         3) log file path
         4) source file path (relative to the gluster volume root)
         5) destination file path (relative to the gluster volume root)

        "glfs-copy-file-range <hostname> <volume> <log file path> <source> <destination>"

      - Added a testcase as well to run glfs-copy-file-range tool

    * io-stats changes to capture the fop for profiling

    * NOTE:

      - Added conditional check to see whether the copy_file_range syscall
        is available or not. If not, then return ENOSYS.

      - Added conditional check for kernel minor version in fuse_kernel.h
        and fuse-bridge while referring to copy_file_range. And the kernel
        minor version is kept as it is. i.e. 24. Increment it in future
        when there is a kernel release which contains the support for
        copy_file_range fop in fuse kernel module.

    * The document which contains a writeup on this enhancement can be found at
      https://docs.google.com/document/d/1BSILbXr_knynNwxSyyu503JoTz5QFM_4suNIh2WwrSc/edit

Change-Id: I280069c814dd21ce6ec3be00a884fc24ab692367
updates: #536
Signed-off-by: Raghavendra Bhat <raghavendra@redhat.com>

6 files changed, 265 insertions, 0 deletions

diff --git a/xlators/protocol/client/src/client-common.c b/xlators/protocol/client/src/client-common.c
index 7708c820918..64db98d661b 100644
--- a/xlators/protocol/client/src/client-common.c
+++ b/xlators/protocol/client/src/client-common.c
@@ -2556,6 +2556,38 @@ out:
 }
 
 int
+client_pre_copy_file_range_v2(xlator_t *this, gfx_copy_file_range_req *req,
+                              fd_t *fd_in, off64_t off_in, fd_t *fd_out,
+                              off64_t off_out, size_t size, int32_t flags,
+                              dict_t **xdata)
+{
+    int64_t remote_fd_in = -1;
+    int64_t remote_fd_out = -1;
+    int op_errno = ESTALE;
+
+    CLIENT_GET_REMOTE_FD(this, fd_in, FALLBACK_TO_ANON_FD, remote_fd_in,
+                         op_errno, out);
+
+    CLIENT_GET_REMOTE_FD(this, fd_out, FALLBACK_TO_ANON_FD, remote_fd_out,
+                         op_errno, out);
+    req->size = size;
+    req->off_in = off_in;
+    req->off_out = off_out;
+    req->fd_in = remote_fd_in;
+    req->fd_out = remote_fd_out;
+    req->flag = flags;
+
+    memcpy(req->gfid1, fd_in->inode->gfid, 16);
+    memcpy(req->gfid2, fd_out->inode->gfid, 16);
+
+    dict_to_xdr(*xdata, &req->xdata);
+
+    return 0;
+out:
+    return -op_errno;
+}
+
+int
 client_pre_statfs_v2(xlator_t *this, gfx_statfs_req *req, loc_t *loc,
                      dict_t *xdata)
 {
diff --git a/xlators/protocol/client/src/client-common.h b/xlators/protocol/client/src/client-common.h
index 5214eae128e..a2043d8742a 100644
--- a/xlators/protocol/client/src/client-common.h
+++ b/xlators/protocol/client/src/client-common.h
@@ -621,4 +621,10 @@ client_post_rename_v2(xlator_t *this, gfx_rename_rsp *rsp, struct iatt *stbuf,
                       struct iatt *prenewparent, struct iatt *postnewparent,
                       dict_t **xdata);
 
+int
+client_pre_copy_file_range_v2(xlator_t *this, gfx_copy_file_range_req *req,
+                              fd_t *fd_in, off64_t off_in, fd_t *fd_out,
+                              off64_t off_out, size_t size, int32_t flags,
+                              dict_t **xdata);
+
 #endif /* __CLIENT_COMMON_H__ */
diff --git a/xlators/protocol/client/src/client-helpers.c b/xlators/protocol/client/src/client-helpers.c
index 849fdfca0bc..55e87b3c370 100644
--- a/xlators/protocol/client/src/client-helpers.c
+++ b/xlators/protocol/client/src/client-helpers.c
@@ -2459,6 +2459,20 @@ client_handle_fop_requirements_v2(
                 lease, this, &this_req->compound_req_v2_u.compound_lease_req,
                 op_errno, out, &args->loc, &args->lease, args->xdata);
             break;
+        case GF_FOP_COPY_FILE_RANGE:
+            /*
+             * Not going to handle the copy_file_range fop in compound
+             * operation. This is because, compound operation is going
+             * to be removed. In fact, AFR one of the heavy consumer of
+             * compound operations has stopped using that.
+             * https://github.com/gluster/glusterfs/issues/414
+             * Therefore, sending ENOTSUP error for this fop coming as
+             * comound request. Though, there was no need of handling
+             * "case GF_FOP_COPY_FILE_RANGE" technically, this comment
+             * under the label of GF_FOP_COPY_FILE_RANGE will help in
+             * understanding that this fop does not handle the compund
+             * request and why.
+             */
         default:
             return ENOTSUP;
     }
@@ -2631,6 +2645,14 @@ compound_request_cleanup_v2(gfx_compound_req *req)
             case GF_FOP_SEEK:
                 CLIENT4_COMPOUND_FOP_CLEANUP(curr_req, seek);
                 break;
+            case GF_FOP_COPY_FILE_RANGE:
+                /*
+                 * This fop is not handled in compund operations.
+                 * Check the comment added under this fop's section
+                 * in the compound_request_cleanup_v2. Therefore
+                 * keeping this label only as a placeholder with
+                 * a message that, this fop is not handled.
+                 */
             default:
                 break;
         }
@@ -3004,6 +3026,12 @@ client_process_response_v2(call_frame_t *frame, xlator_t *this,
                                   &this_args_cbk->lease, xdata);
             break;
         }
+        case GF_FOP_COPY_FILE_RANGE:
+            /*
+             * Not handling this fop. Returning ENOTSUP. Check
+             * the comment added for this fop in the function
+             * client_handle_fop_requirements_v2.
+             */
         default:
             return -ENOTSUP;
     }
diff --git a/xlators/protocol/client/src/client-rpc-fops_v2.c b/xlators/protocol/client/src/client-rpc-fops_v2.c
index ca180c1db4b..8f3ee41e5c5 100644
--- a/xlators/protocol/client/src/client-rpc-fops_v2.c
+++ b/xlators/protocol/client/src/client-rpc-fops_v2.c
@@ -2833,6 +2833,72 @@ out:
     return 0;
 }
 
+int
+client4_0_copy_file_range_cbk(struct rpc_req *req, struct iovec *iov, int count,
+                              void *myframe)
+{
+    gfx_common_3iatt_rsp rsp = {
+        0,
+    };
+    call_frame_t *frame = NULL;
+    struct iatt stbuf = {
+        0,
+    };
+    struct iatt prestat = {
+        0,
+    };
+    struct iatt poststat = {
+        0,
+    };
+    int ret = 0;
+    xlator_t *this = NULL;
+    dict_t *xdata = NULL;
+    clnt_local_t *local = NULL;
+
+    this = THIS;
+
+    frame = myframe;
+    local = frame->local;
+
+    if (-1 == req->rpc_status) {
+        rsp.op_ret = -1;
+        rsp.op_errno = ENOTCONN;
+        goto out;
+    }
+
+    ret = xdr_to_generic(*iov, &rsp, (xdrproc_t)xdr_gfx_common_3iatt_rsp);
+    if (ret < 0) {
+        gf_msg(this->name, GF_LOG_ERROR, EINVAL, PC_MSG_XDR_DECODING_FAILED,
+               "XDR decoding failed");
+        rsp.op_ret = -1;
+        rsp.op_errno = EINVAL;
+        goto out;
+    }
+
+    ret = client_post_common_3iatt(this, &rsp, &stbuf, &prestat, &poststat,
+                                   &xdata);
+    if (ret < 0)
+        goto out;
+out:
+    if (rsp.op_ret == -1) {
+        gf_msg(this->name, GF_LOG_WARNING, gf_error_to_errno(rsp.op_errno),
+               PC_MSG_REMOTE_OP_FAILED, "remote operation failed");
+    } else if (rsp.op_ret >= 0) {
+        if (local->attempt_reopen)
+            client_attempt_reopen(local->fd, this);
+        if (local->attempt_reopen_out)
+            client_attempt_reopen(local->fd_out, this);
+    }
+    CLIENT_STACK_UNWIND(copy_file_range, frame, rsp.op_ret,
+                        gf_error_to_errno(rsp.op_errno), &stbuf, &prestat,
+                        &poststat, xdata);
+
+    if (xdata)
+        dict_unref(xdata);
+
+    return 0;
+}
+
 int32_t
 client4_0_releasedir(call_frame_t *frame, xlator_t *this, void *data)
 {
@@ -5846,6 +5912,80 @@ unwind:
 }
 
 int32_t
+client4_0_copy_file_range(call_frame_t *frame, xlator_t *this, void *data)
+{
+    clnt_args_t *args = NULL;
+    clnt_conf_t *conf = NULL;
+    clnt_local_t *local = NULL;
+    gfx_copy_file_range_req req = {
+        {
+            0,
+        },
+    };
+    int op_errno = ESTALE;
+    int ret = 0;
+
+    if (!frame || !this || !data)
+        goto unwind;
+
+    args = data;
+    conf = this->private;
+
+    ret = client_pre_copy_file_range_v2(this, &req, args->fd, args->off_in,
+                                        args->fd_out, args->off_out, args->size,
+                                        args->flags, &args->xdata);
+
+    if (ret) {
+        op_errno = -ret;
+        goto unwind;
+    }
+
+    ret = client_fd_fop_prepare_local(frame, args->fd, req.fd_in);
+    if (ret) {
+        op_errno = -ret;
+        goto unwind;
+    }
+
+    /*
+     * Since frame->local is allocated in above function call
+     * itself, better to use it (with the assumption that it
+     * has been allocated) directly instead of again calling
+     * client_fd_fop_prepare_local or modifying it, as doing
+     * so requires changes in other places as well.
+     */
+
+    local = frame->local;
+    local->fd_out = fd_ref(args->fd_out);
+    local->attempt_reopen_out = client_is_reopen_needed(args->fd_out, this,
+                                                        req.fd_out);
+
+    ret = client_submit_request(
+        this, &req, frame, conf->fops, GFS3_OP_COPY_FILE_RANGE,
+        client4_0_copy_file_range_cbk, NULL, NULL, 0, NULL, 0, NULL,
+        (xdrproc_t)xdr_gfx_copy_file_range_req);
+    if (ret) {
+        /*
+         * If the lower layers fail to submit a request, they'll also
+         * do the unwind for us (see rpc_clnt_submit), so don't unwind
+         * here in such cases.
+         */
+        gf_msg(this->name, GF_LOG_WARNING, 0, PC_MSG_FOP_SEND_FAILED,
+               "failed to send the fop");
+    }
+
+    GF_FREE(req.xdata.pairs.pairs_val);
+
+    return 0;
+
+unwind:
+    CLIENT_STACK_UNWIND(copy_file_range, frame, -1, op_errno, NULL, NULL, NULL,
+                        NULL);
+    GF_FREE(req.xdata.pairs.pairs_val);
+
+    return 0;
+}
+
+int32_t
 client4_0_fsetattr(call_frame_t *frame, xlator_t *this, void *data)
 {
     clnt_args_t *args = NULL;
@@ -6257,6 +6397,7 @@ rpc_clnt_procedure_t clnt4_0_fop_actors[GF_FOP_MAXVALUE] = {
     [GF_FOP_COMPOUND] = {"COMPOUND", client4_0_compound},
     [GF_FOP_ICREATE] = {"ICREATE", client4_0_icreate},
     [GF_FOP_NAMELINK] = {"NAMELINK", client4_0_namelink},
+    [GF_FOP_COPY_FILE_RANGE] = {"COPY-FILE-RANGE", client4_0_copy_file_range},
 };
 
 rpc_clnt_prog_t clnt4_0_fop_prog = {
diff --git a/xlators/protocol/client/src/client.c b/xlators/protocol/client/src/client.c
index 38723b43b45..c8e84f6e1b7 100644
--- a/xlators/protocol/client/src/client.c
+++ b/xlators/protocol/client/src/client.c
@@ -1129,6 +1129,41 @@ out:
     return 0;
 }
 
+int32_t
+client_copy_file_range(call_frame_t *frame, xlator_t *this, fd_t *fd_in,
+                       off_t off_in, fd_t *fd_out, off_t off_out, size_t len,
+                       uint32_t flags, dict_t *xdata)
+{
+    int ret = -1;
+    clnt_conf_t *conf = NULL;
+    rpc_clnt_procedure_t *proc = NULL;
+    clnt_args_t args = {
+        0,
+    };
+
+    conf = this->private;
+    if (!conf || !conf->fops)
+        goto out;
+
+    args.fd = fd_in;
+    args.fd_out = fd_out;
+    args.offset = off_in;
+    args.off_out = off_out;
+    args.size = len;
+    args.flags = flags;
+    args.xdata = xdata;
+
+    proc = &conf->fops->proctable[GF_FOP_COPY_FILE_RANGE];
+    if (proc->fn)
+        ret = proc->fn(frame, this, &args);
+out:
+    if (ret)
+        STACK_UNWIND_STRICT(copy_file_range, frame, -1, ENOTCONN, NULL, NULL,
+                            NULL, NULL);
+
+    return 0;
+}
+
 static gf_boolean_t
 is_client_rpc_init_command(dict_t *dict, xlator_t *this, char **value)
 {
@@ -2898,6 +2933,7 @@ struct xlator_fops fops = {
     .icreate = client_icreate,
     .namelink = client_namelink,
     .put = client_put,
+    .copy_file_range = client_copy_file_range,
 };
 
 struct xlator_dumpops dumpops = {
diff --git a/xlators/protocol/client/src/client.h b/xlators/protocol/client/src/client.h
index 5fc75a84628..71f84f3ca89 100644
--- a/xlators/protocol/client/src/client.h
+++ b/xlators/protocol/client/src/client.h
@@ -269,6 +269,7 @@ typedef struct client_local {
     loc_t loc;
     loc_t loc2;
     fd_t *fd;
+    fd_t *fd_out; /* used in copy_file_range */
     clnt_fd_ctx_t *fdctx;
     uint32_t flags;
     struct iobref *iobref;
@@ -280,6 +281,11 @@ typedef struct client_local {
     pthread_mutex_t mutex;
     char *name;
     gf_boolean_t attempt_reopen;
+    /*
+     * The below boolean variable is used
+     * only for copy_file_range fop
+     */
+    gf_boolean_t attempt_reopen_out;
     /* required for compound fops */
     compound_args_t *compound_args;
     unsigned int length; /* length of a compound fop */
@@ -289,7 +295,13 @@ typedef struct client_local {
 
 typedef struct client_args {
     loc_t *loc;
+    /*
+     * This is the source fd for copy_file_range and
+     * the default fd for any other fd based fop which
+     * requires only one fd (i.e. opetates on one fd)
+     */
     fd_t *fd;
+    fd_t *fd_out; /* this is the destination fd for copy_file_range */
     const char *linkname;
     struct iobref *iobref;
     struct iovec *vector;
@@ -301,7 +313,17 @@ typedef struct client_args {
     struct gf_flock *flock;
     const char *volume;
     const char *basename;
+
     off_t offset;
+    /*
+     * According to the man page of copy_file_range,
+     * the offsets for source and destination file
+     * are of type loff_t. But the type loff_t is
+     * linux specific and is actual a typedef of
+     * off64_t.
+     */
+    off64_t off_in;  /* used in copy_file_range for source fd */
+    off64_t off_out; /* used in copy_file_range for dst fd */
     int32_t mask;
     int32_t cmd;
     size_t size;