rio/everywhere: add icreate/namelink fop

icreate creates inode, while namelink links the basename to it's
parent gfid.

For now mkdir is the primary user of these fops. Better distribution is
acheived by creating the inode on ,(say) mds1 and linking the basename to it's
parent gfid on mds2. The inode serves readdirp, stat etc.

More details about the fops are present at:
https://review.gluster.org/#/c/13395/3/design/DHT2/DHT2_Icreate_Namelink_Notes.md

This backport of three patches from experimental branch.
1- https://review.gluster.org/#/c/18085/
2- https://review.gluster.org/#/c/18086/
3- https://review.gluster.org/#/c/18094/

Updates gluster/glusterfs#243
Change-Id: I1bd3d5a441a3cfab1acfeb52f15c6c867d362592
Signed-off-by: Susant Palai <spalai@redhat.com>

14 files changed, 339 insertions, 1 deletions

diff --git a/libglusterfs/src/call-stub.c b/libglusterfs/src/call-stub.c
index 74a64294528..d5c4f1365e8 100644
--- a/libglusterfs/src/call-stub.c
+++ b/libglusterfs/src/call-stub.c
@@ -2009,6 +2009,126 @@ out:
         return stub;
 }
 
+call_stub_t *
+fop_icreate_stub (call_frame_t *frame, fop_icreate_t fn,
+                  loc_t *loc, mode_t mode, dict_t *xdata)
+{
+        call_stub_t *stub = NULL;
+
+        GF_VALIDATE_OR_GOTO ("call-stub", frame, out);
+        GF_VALIDATE_OR_GOTO ("call-stub", fn, out);
+
+        stub = stub_new (frame, 1, GF_FOP_ICREATE);
+        GF_VALIDATE_OR_GOTO ("call-stub", stub, out);
+
+        stub->fn.icreate = fn;
+
+        stub->args.mode = mode;
+        if (loc)
+                loc_copy (&stub->args.loc, loc);
+        if (xdata)
+                stub->args.xdata = dict_ref (xdata);
+
+ out:
+        return stub;
+}
+
+static void
+args_icreate_store_cbk (default_args_cbk_t *args,
+                        int32_t op_ret, int32_t op_errno,
+                        inode_t *inode, struct iatt *buf, dict_t *xdata)
+{
+        args->op_ret = op_ret;
+        args->op_errno = op_errno;
+        if (inode)
+                args->inode = inode_ref (inode);
+        if (buf)
+                args->stat = *buf;
+        if (xdata)
+                args->xdata = dict_ref (xdata);
+}
+
+call_stub_t *
+fop_icreate_cbk_stub (call_frame_t *frame,
+                      fop_icreate_cbk_t fn,
+                      int32_t op_ret, int32_t op_errno,
+                      inode_t *inode, struct iatt *buf, dict_t *xdata)
+{
+        call_stub_t *stub = NULL;
+
+        GF_VALIDATE_OR_GOTO ("call-stub", frame, out);
+
+        stub = stub_new (frame, 0, GF_FOP_ICREATE);
+        GF_VALIDATE_OR_GOTO ("call-stub", stub, out);
+
+        stub->fn_cbk.icreate = fn;
+        args_icreate_store_cbk (&stub->args_cbk,
+                                op_ret, op_errno, inode, buf, xdata);
+
+ out:
+        return stub;
+}
+
+call_stub_t *
+fop_namelink_stub (call_frame_t *frame,
+                   fop_namelink_t fn, loc_t *loc, dict_t *xdata)
+{
+        call_stub_t *stub = NULL;
+
+        GF_VALIDATE_OR_GOTO ("call-stub", frame, out);
+        GF_VALIDATE_OR_GOTO ("call-stub", fn, out);
+
+        stub = stub_new (frame, 1, GF_FOP_NAMELINK);
+        GF_VALIDATE_OR_GOTO ("call-stub", stub, out);
+
+        stub->fn.namelink = fn;
+
+        if (loc)
+                loc_copy (&stub->args.loc, loc);
+        if (xdata)
+                stub->args.xdata = dict_ref (xdata);
+
+ out:
+        return stub;
+}
+
+static void
+args_namelink_store_cbk (default_args_cbk_t *args,
+                         int32_t op_ret, int32_t op_errno,
+                         struct iatt *prebuf, struct iatt *postbuf, dict_t *xdata)
+{
+        args->op_ret = op_ret;
+        args->op_errno = op_errno;
+
+        if (prebuf)
+                args->prestat = *prebuf;
+        if (postbuf)
+                args->poststat = *postbuf;
+        if (xdata)
+                args->xdata = dict_ref (xdata);
+}
+
+call_stub_t *
+fop_namelink_cbk_stub (call_frame_t *frame,
+                       fop_namelink_cbk_t fn,
+                       int32_t op_ret, int32_t op_errno,
+                       struct iatt *prebuf, struct iatt *postbuf, dict_t *xdata)
+{
+        call_stub_t *stub = NULL;
+
+        GF_VALIDATE_OR_GOTO ("call-stub", frame, out);
+
+        stub = stub_new (frame, 0, GF_FOP_NAMELINK);
+        GF_VALIDATE_OR_GOTO ("call-stub", stub, out);
+
+        stub->fn_cbk.namelink = fn;
+        args_namelink_store_cbk (&stub->args_cbk,
+                                 op_ret, op_errno, prebuf, postbuf, xdata);
+
+ out:
+        return stub;
+}
+
 void
 call_resume_wind (call_stub_t *stub)
 {
diff --git a/libglusterfs/src/call-stub.h b/libglusterfs/src/call-stub.h
index 4d86a517537..e20f6a5a096 100644
--- a/libglusterfs/src/call-stub.h
+++ b/libglusterfs/src/call-stub.h
@@ -78,6 +78,8 @@ typedef struct _call_stub {
                 fop_getactivelk_t getactivelk;
 	        fop_setactivelk_t setactivelk;
                 fop_put_t put;
+                fop_icreate_t icreate;
+                fop_namelink_t namelink;
         } fn;
 
 	union {
@@ -131,6 +133,8 @@ typedef struct _call_stub {
                 fop_getactivelk_cbk_t getactivelk;
                 fop_setactivelk_cbk_t setactivelk;
                 fop_put_cbk_t put;
+                fop_icreate_cbk_t icreate;
+                fop_namelink_cbk_t namelink;
 	} fn_cbk;
 
         default_args_t args;
@@ -776,6 +780,26 @@ fop_put_cbk_stub (call_frame_t *frame, fop_put_cbk_t fn, int32_t op_ret,
                   struct iatt *preparent, struct iatt *postparent,
                   dict_t *xdata);
 
+call_stub_t *
+fop_icreate_stub (call_frame_t *frame, fop_icreate_t fn,
+                  loc_t *loc, mode_t mode, dict_t *xdata);
+
+call_stub_t *
+fop_namelink_stub (call_frame_t *frame,
+                   fop_namelink_t fn, loc_t *loc, dict_t *xdata);
+
+call_stub_t *
+fop_icreate_cbk_stub (call_frame_t *frame,
+                      fop_icreate_cbk_t fn,
+                      int32_t op_ret, int32_t op_errno,
+                      inode_t *inode, struct iatt *buf, dict_t *xdata);
+
+call_stub_t *
+fop_namelink_cbk_stub (call_frame_t *frame,
+                       fop_namelink_cbk_t fn,
+                       int32_t op_ret, int32_t op_errno,
+                       struct iatt *prebuf, struct iatt *postbuf, dict_t *xdata);
+
 void call_resume (call_stub_t *stub);
 void call_resume_keep_stub (call_stub_t *stub);
 void call_stub_destroy (call_stub_t *stub);
diff --git a/libglusterfs/src/common-utils.c b/libglusterfs/src/common-utils.c
index 68d2e243f58..ad0359d7cb2 100644
--- a/libglusterfs/src/common-utils.c
+++ b/libglusterfs/src/common-utils.c
@@ -4591,6 +4591,8 @@ fop_enum_to_pri_string (glusterfs_fop_t fop)
         case GF_FOP_READDIRP:
         case GF_FOP_GETACTIVELK:
         case GF_FOP_SETACTIVELK:
+        case GF_FOP_ICREATE:
+        case GF_FOP_NAMELINK:
                 return "HIGH";
 
         case GF_FOP_CREATE:
@@ -4940,4 +4942,3 @@ get_struct_variable (int mem_num, gf_gsync_status_t *sts_val)
 out:
         return NULL;
 }
-
diff --git a/libglusterfs/src/default-args.c b/libglusterfs/src/default-args.c
index 9ac0f400493..20cd05e68cb 100644
--- a/libglusterfs/src/default-args.c
+++ b/libglusterfs/src/default-args.c
@@ -1580,6 +1580,29 @@ args_lease_cbk_store (default_args_cbk_t *args,
                 args->xdata = dict_ref (xdata);
 }
 
+int
+args_icreate_store (default_args_t *args,
+                    loc_t *loc, mode_t mode, dict_t *xdata)
+{
+        loc_copy (&args->loc, loc);
+        args->mode = mode;
+
+        if (xdata)
+                args->xdata = dict_ref (xdata);
+        return 0;
+}
+
+int
+args_namelink_store (default_args_t *args,
+                     loc_t *loc, dict_t *xdata)
+{
+        loc_copy (&args->loc, loc);
+
+        if (xdata)
+                args->xdata = dict_ref (xdata);
+        return 0;
+}
+
 void
 args_cbk_wipe (default_args_cbk_t *args_cbk)
 {
diff --git a/libglusterfs/src/default-args.h b/libglusterfs/src/default-args.h
index f97d8830de9..7d46fefa663 100644
--- a/libglusterfs/src/default-args.h
+++ b/libglusterfs/src/default-args.h
@@ -488,6 +488,14 @@ args_getactivelk_cbk_store (default_args_cbk_t *args,
 int
 args_setactivelk_store (default_args_t *args, loc_t *loc,
                           lock_migration_info_t *locklist, dict_t *xdata);
+
+int
+args_icreate_store (default_args_t *args,
+                  loc_t *loc, mode_t mode, dict_t *xdata);
+
+int
+args_namelink_store (default_args_t *args, loc_t *loc, dict_t *xdata);
+
 void
 args_cbk_init (default_args_cbk_t *args_cbk);
 #endif /* _DEFAULT_ARGS_H */
diff --git a/libglusterfs/src/defaults-tmpl.c b/libglusterfs/src/defaults-tmpl.c
index 21b99d9566c..b9f6274800e 100644
--- a/libglusterfs/src/defaults-tmpl.c
+++ b/libglusterfs/src/defaults-tmpl.c
@@ -82,6 +82,8 @@ struct xlator_fops _default_fops = {
         .getactivelk = default_getactivelk,
         .setactivelk = default_setactivelk,
         .put = default_put,
+        .icreate = default_icreate,
+        .namelink = default_namelink,
 };
 struct xlator_fops *default_fops = &_default_fops;
 
diff --git a/libglusterfs/src/defaults.h b/libglusterfs/src/defaults.h
index c202dbe4941..5e56cc2c9f9 100644
--- a/libglusterfs/src/defaults.h
+++ b/libglusterfs/src/defaults.h
@@ -359,6 +359,12 @@ default_put (call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
              mode_t umask, uint32_t flags, struct iovec *vector, int32_t count,
              off_t off, struct iobref *iobref, dict_t *xattr, dict_t *xdata);
 
+int32_t default_icreate (call_frame_t *frame, xlator_t *this,
+                         loc_t *loc, mode_t mode, dict_t *xdata);
+
+int32_t default_namelink (call_frame_t *frame,
+                          xlator_t *this, loc_t *loc, dict_t *xdata);
+
 /* Resume */
 int32_t default_getspec_resume (call_frame_t *frame,
                                 xlator_t *this,
@@ -912,6 +918,14 @@ default_put_cbk_resume (call_frame_t *frame, void *cookie, xlator_t *this,
                         struct iatt *buf, struct iatt *preparent,
                         struct iatt *postparent, dict_t *xdata);
 
+int32_t
+default_icreate_resume (call_frame_t *frame, xlator_t *this,
+                        loc_t *loc, mode_t mode, dict_t *xdata);
+
+int32_t
+default_namelink_resume (call_frame_t *frame,
+                         xlator_t *this, loc_t *loc, dict_t *xdata);
+
 /* _CBK */
 int32_t
 default_lookup_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
@@ -1166,6 +1180,18 @@ default_put_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
                  struct iatt *postparent, dict_t *xdata);
 
 int32_t
+default_icreate_cbk (call_frame_t *frame,
+                     void *cookie, xlator_t *this,
+                     int32_t op_ret, int32_t op_errno,
+                     inode_t *inode, struct iatt *buf, dict_t *xdata);
+
+int32_t
+default_namelink_cbk (call_frame_t *frame, void *cookie,
+                      xlator_t *this, int32_t op_ret,
+                      int32_t op_errno, struct iatt *prebuf,
+                      struct iatt *postbuf, dict_t *xdata);
+
+int32_t
 default_lookup_failure_cbk (call_frame_t *frame, int32_t op_errno);
 
 int32_t
@@ -1321,6 +1347,12 @@ int32_t
 default_put_failure_cbk (call_frame_t *frame, int32_t op_errno);
 
 int32_t
+default_icreate_failure_cbk (call_frame_t *frame, int32_t op_errno);
+
+int32_t
+default_namelink_failure_cbk (call_frame_t *frame, int32_t op_errno);
+
+int32_t
 default_mem_acct_init (xlator_t *this);
 
 void
diff --git a/libglusterfs/src/generator.py b/libglusterfs/src/generator.py
index d9262e473b7..7fa5f433501 100755
--- a/libglusterfs/src/generator.py
+++ b/libglusterfs/src/generator.py
@@ -575,6 +575,23 @@ ops['put'] = (
         ('cbk-arg',     'xdata',                'dict_t *'),
 )
 
+ops['icreate'] = (
+        ('fop-arg',     'loc',                   'loc_t *'),
+        ('fop-arg',     'mode',                  'mode_t'),
+        ('fop-arg',     'xdata',                 'dict_t *'),
+        ('cbk-arg',     'inode',                 'inode_t *'),
+        ('cbk-arg',     'buf',                   'struct iatt *'),
+        ('cbk-arg',     'xdata',                 'dict_t *'),
+)
+
+ops['namelink'] = (
+        ('fop-arg',     'loc',                   'loc_t *'),
+        ('fop-arg',     'xdata',                 'dict_t *'),
+        ('cbk-arg',     'prebuf',                'struct iatt *'),
+        ('cbk-arg',     'postbuf',               'struct iatt *'),
+        ('cbk-arg',     'xdata',                 'dict_t *'),
+)
+
 #####################################################################
 xlator_cbks['forget'] = (
         ('fn-arg',      'this',        'xlator_t *'),
diff --git a/libglusterfs/src/globals.c b/libglusterfs/src/globals.c
index 1c574463123..6bed1b546ee 100644
--- a/libglusterfs/src/globals.c
+++ b/libglusterfs/src/globals.c
@@ -75,6 +75,8 @@ const char *gf_fop_list[GF_FOP_MAXVALUE] = {
         [GF_FOP_GETACTIVELK] = "GETACTIVELK",
         [GF_FOP_SETACTIVELK] = "SETACTIVELK",
         [GF_FOP_PUT]         = "PUT",
+        [GF_FOP_ICREATE]     = "ICREATE",
+        [GF_FOP_NAMELINK]    = "NAMELINK",
 };
 
 const char *gf_upcall_list[GF_UPCALL_FLAGS_MAXVALUE] = {
diff --git a/libglusterfs/src/glusterfs.h b/libglusterfs/src/glusterfs.h
index ad8e396fde5..aa28b80222e 100644
--- a/libglusterfs/src/glusterfs.h
+++ b/libglusterfs/src/glusterfs.h
@@ -306,6 +306,7 @@ const char *fop_enum_to_pri_string (glusterfs_fop_t fop);
 #define GF_SET_OVERWRITE      0x2 /* Overwrite with the buf given */
 #define GF_SET_DIR_ONLY       0x4
 #define GF_SET_EPOCH_TIME     0x8 /* used by afr dir lookup selfheal */
+#define GF_AUXILLARY_PARGFID  0xd /* RIO dummy parent gfid */
 
 /* key value which quick read uses to get small files in lookup cbk */
 #define GF_CONTENT_KEY "glusterfs.content"
diff --git a/libglusterfs/src/syncop.c b/libglusterfs/src/syncop.c
index a4882f5eef2..58c606b238f 100644
--- a/libglusterfs/src/syncop.c
+++ b/libglusterfs/src/syncop.c
@@ -3251,3 +3251,79 @@ syncop_setactivelk (xlator_t *subvol, loc_t *loc,
         return args.op_ret;
 
 }
+
+int
+syncop_icreate_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
+                   int32_t op_ret, int32_t op_errno, inode_t *inode,
+                   struct iatt *buf, dict_t *xdata)
+{
+        struct syncargs *args = NULL;
+
+        args = cookie;
+
+        args->op_ret   = op_ret;
+        args->op_errno = op_errno;
+        if (xdata)
+                args->xdata  = dict_ref (xdata);
+
+	if (buf)
+		args->iatt1 = *buf;
+
+        __wake (args);
+
+        return 0;
+}
+
+int
+syncop_namelink_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
+                   int32_t op_ret, int32_t op_errno, struct iatt *prebuf,
+                   struct iatt *postbuf, dict_t *xdata)
+{
+        struct syncargs *args = NULL;
+
+        args = cookie;
+
+        args->op_ret   = op_ret;
+        args->op_errno = op_errno;
+
+        if (xdata)
+                args->xdata  = dict_ref (xdata);
+
+        __wake (args);
+
+        return 0;
+}
+
+int
+syncop_icreate (xlator_t *subvol, loc_t *loc, mode_t mode, dict_t *xdata)
+{
+        struct syncargs args = {0, };
+
+        SYNCOP (subvol, (&args), syncop_icreate_cbk, subvol->fops->icreate,
+                loc, mode, xdata);
+
+        if (xdata)
+                xdata = args.xdata;
+        else if (args.xdata)
+                dict_unref (args.xdata);
+
+        errno = args.op_errno;
+        return args.op_ret;
+}
+
+int
+syncop_namelink (xlator_t *subvol, loc_t *loc, dict_t *xdata)
+{
+        struct syncargs args = {0, };
+
+        SYNCOP (subvol, (&args), syncop_namelink_cbk, subvol->fops->namelink,
+                loc, xdata);
+
+        if (xdata)
+                xdata = args.xdata;
+        else if (args.xdata)
+                dict_unref (args.xdata);
+
+        errno = args.op_errno;
+        return args.op_ret;
+}
diff --git a/libglusterfs/src/syncop.h b/libglusterfs/src/syncop.h
index 8f84cb7c4a4..3b2556535b7 100644
--- a/libglusterfs/src/syncop.h
+++ b/libglusterfs/src/syncop.h
@@ -569,4 +569,14 @@ syncop_put (xlator_t *subvol, loc_t *loc, mode_t mode, mode_t umask,
             off_t offset, struct iobref *iobref, dict_t *xattr,
             struct iatt *iatt, dict_t *xdata_in, dict_t **xdata_out);
 
+int
+syncop_setactivelk_cbk (call_frame_t *frame, void *cookie, xlator_t *this,
+                        int32_t op_ret, int32_t op_errno, dict_t *xdata);
+
+int
+syncop_icreate (xlator_t *subvol, loc_t *loc, mode_t mode, dict_t *xdata_out);
+
+int
+syncop_namelink (xlator_t *subvol, loc_t *loc, dict_t *xdata_out);
+
 #endif /* _SYNCOP_H */
diff --git a/libglusterfs/src/xlator.c b/libglusterfs/src/xlator.c
index 677dd6ce80f..3e108312d60 100644
--- a/libglusterfs/src/xlator.c
+++ b/libglusterfs/src/xlator.c
@@ -106,6 +106,8 @@ fill_defaults (xlator_t *xl)
         SET_DEFAULT_FOP (put);
 
         SET_DEFAULT_FOP (getspec);
+        SET_DEFAULT_FOP (icreate);
+        SET_DEFAULT_FOP (namelink);
 
         if (!xl->cbks)
                 xl->cbks = &default_cbks;
diff --git a/libglusterfs/src/xlator.h b/libglusterfs/src/xlator.h
index a3115f09c52..81199bd1809 100644
--- a/libglusterfs/src/xlator.h
+++ b/libglusterfs/src/xlator.h
@@ -471,6 +471,16 @@ typedef int32_t (*fop_put_cbk_t) (call_frame_t *frame, void *cookie,
                                   struct iatt *buf, struct iatt *preparent,
                                   struct iatt *postparent, dict_t *xdata);
 
+typedef int32_t (*fop_icreate_cbk_t) (call_frame_t *frame, void *cookie,
+                                      xlator_t *this, int32_t op_ret,
+                                      int32_t op_errno, inode_t *inode,
+                                      struct iatt *buf, dict_t *xdata);
+
+typedef int32_t (*fop_namelink_cbk_t) (call_frame_t *frame, void *cookie,
+                                       xlator_t *this, int32_t op_ret,
+                                       int32_t op_errno, struct iatt *prebuf,
+                                       struct iatt *postbuf, dict_t *xdata);
+
 typedef int32_t (*fop_lookup_t) (call_frame_t *frame,
                                  xlator_t *this,
                                  loc_t *loc,
@@ -737,6 +747,12 @@ typedef int32_t (*fop_put_t) (call_frame_t *frame, xlator_t *this, loc_t *loc,
                               struct iobref *iobref, dict_t *xattr,
                               dict_t *xdata);
 
+typedef int32_t (*fop_icreate_t) (call_frame_t *frame, xlator_t *this,
+                                  loc_t *loc, mode_t mode, dict_t *xdata);
+
+typedef int32_t (*fop_namelink_t) (call_frame_t *frame, xlator_t *this,
+                                   loc_t *loc, dict_t *xdata);
+
 /* WARNING: make sure the list is in order with FOP definition in
    `rpc/xdr/src/glusterfs-fops.x`.
    If it is not in order, mainly the metrics related feature would be broken */
@@ -799,6 +815,8 @@ struct xlator_fops {
         fop_getactivelk_t    getactivelk;
         fop_setactivelk_t    setactivelk;
         fop_put_t            put;
+        fop_icreate_t        icreate;
+        fop_namelink_t       namelink;
 
         /* these entries are used for a typechecking hack in STACK_WIND _only_ */
         /* make sure to add _cbk variables only after defining regular fops as
@@ -861,6 +879,8 @@ struct xlator_fops {
         fop_getactivelk_cbk_t    getactivelk_cbk;
         fop_setactivelk_cbk_t    setactivelk_cbk;
         fop_put_cbk_t            put_cbk;
+        fop_icreate_cbk_t        icreate_cbk;
+        fop_namelink_cbk_t       namelink_cbk;
 };
 
 typedef int32_t (*cbk_forget_t) (xlator_t *this,