[PATCH 0/3] libceph: update rados.h and related code

[PATCH 0/3] libceph: update rados.h and related code

[Alex Elder]

The ceph kernel and user space code bases have a few header files
in common, but because they're maintained separately they sometimes
diverge.

This series updates "include/linux/ceph/rados.h" to nearly match
its user space counterpart in "src/include/rados.h".

There are still a few things that should be reconciled by updating
that user space code; that work is covered by:
    http://tracker.ceph.com/issues/4167

					-Alex

[PATCH 1/3] libceph: update rados.h
[PATCH 2/3] libceph: add ceph_osd_state_name()
[PATCH 3/3] libceph: update ceph_osd_op_name()

[PATCH 1/3] libceph: update rados.h

[Alex Elder]

Update most of "include/linux/ceph/rados.h" to match its user space
counterpart in "src/include/rados.h" in the ceph tree.

Almost everything that has changed is either:
    - added or revised comments
    - added definitions (therefore no real effect on existing code)
    - defining the same value a different way (e.g., "1 << 0" vs "1")

The only exceptions are:
    - The declaration of ceph_osd_state_name() was excluded; that
      will be inserted in the next patch.
    - ceph_osd_op_mode_read() and ceph_osd_op_mode_modify() are
      defined differently, but they were never used in the kernel
    - CEPH_OSD_FLAG_PEERSTAT is now CEPH_OSD_FLAG_PEERSTAT_OLD, but
      that was never used in the kernel

Anything that was present in this file but not in its user space
counterpart was left intact here.  I left the definitions of
EOLDSNAPC and EBLACKLISTED using numerical values here; I'm
not sure the right way to go with those.

This and the next two commits resolve:
    http://tracker.ceph.com/issues/4164

Signed-off-by: Alex Elder <elder@inktank.com>
---
 include/linux/ceph/rados.h |   91
++++++++++++++++++++++++++++++++++----------
 1 file changed, 71 insertions(+), 20 deletions(-)

diff --git a/include/linux/ceph/rados.h b/include/linux/ceph/rados.h
index 2c04afe..9c3b4aa 100644
--- a/include/linux/ceph/rados.h
+++ b/include/linux/ceph/rados.h
@@ -145,8 +145,10 @@ struct ceph_eversion {
  */

 /* status bits */
-#define CEPH_OSD_EXISTS 1
-#define CEPH_OSD_UP     2
+#define CEPH_OSD_EXISTS  (1<<0)
+#define CEPH_OSD_UP      (1<<1)
+#define CEPH_OSD_AUTOOUT (1<<2)  /* osd was automatically marked out */
+#define CEPH_OSD_NEW     (1<<3)  /* osd is new, never marked in */

 /* osd weights.  fixed point value: 0x10000 == 1.0 ("in"), 0 == "out" */
 #define CEPH_OSD_IN  0x10000
@@ -161,9 +163,25 @@ struct ceph_eversion {
 #define CEPH_OSDMAP_PAUSERD  (1<<2)  /* pause all reads */
 #define CEPH_OSDMAP_PAUSEWR  (1<<3)  /* pause all writes */
 #define CEPH_OSDMAP_PAUSEREC (1<<4)  /* pause recovery */
+#define CEPH_OSDMAP_NOUP     (1<<5)  /* block osd boot */
+#define CEPH_OSDMAP_NODOWN   (1<<6)  /* block osd mark-down/failure */
+#define CEPH_OSDMAP_NOOUT    (1<<7)  /* block osd auto mark-out */
+#define CEPH_OSDMAP_NOIN     (1<<8)  /* block osd auto mark-in */
+#define CEPH_OSDMAP_NOBACKFILL (1<<9) /* block osd backfill */
+#define CEPH_OSDMAP_NORECOVER (1<<10) /* block osd recovery and backfill */
+
+/*
+ * The error code to return when an OSD can't handle a write
+ * because it is too large.
+ */
+#define OSD_WRITETOOBIG EMSGSIZE

 /*
  * osd ops
+ *
+ * WARNING: do not use these op codes directly.  Use the helpers
+ * defined below instead.  In certain cases, op code behavior was
+ * redefined, resulting in special-cases in the helpers.
  */
 #define CEPH_OSD_OP_MODE       0xf000
 #define CEPH_OSD_OP_MODE_RD    0x1000
@@ -177,6 +195,7 @@ struct ceph_eversion {
 #define CEPH_OSD_OP_TYPE_ATTR  0x0300
 #define CEPH_OSD_OP_TYPE_EXEC  0x0400
 #define CEPH_OSD_OP_TYPE_PG    0x0500
+#define CEPH_OSD_OP_TYPE_MULTI 0x0600 /* multiobject */

 enum {
 	/** data **/
@@ -217,6 +236,23 @@ enum {

 	CEPH_OSD_OP_WATCH   = CEPH_OSD_OP_MODE_WR | CEPH_OSD_OP_TYPE_DATA | 15,

+	/* omap */
+	CEPH_OSD_OP_OMAPGETKEYS   = CEPH_OSD_OP_MODE_RD |
CEPH_OSD_OP_TYPE_DATA | 17,
+	CEPH_OSD_OP_OMAPGETVALS   = CEPH_OSD_OP_MODE_RD |
CEPH_OSD_OP_TYPE_DATA | 18,
+	CEPH_OSD_OP_OMAPGETHEADER = CEPH_OSD_OP_MODE_RD |
CEPH_OSD_OP_TYPE_DATA | 19,
+	CEPH_OSD_OP_OMAPGETVALSBYKEYS  =
+	  CEPH_OSD_OP_MODE_RD | CEPH_OSD_OP_TYPE_DATA | 20,
+	CEPH_OSD_OP_OMAPSETVALS   = CEPH_OSD_OP_MODE_WR |
CEPH_OSD_OP_TYPE_DATA | 21,
+	CEPH_OSD_OP_OMAPSETHEADER = CEPH_OSD_OP_MODE_WR |
CEPH_OSD_OP_TYPE_DATA | 22,
+	CEPH_OSD_OP_OMAPCLEAR     = CEPH_OSD_OP_MODE_WR |
CEPH_OSD_OP_TYPE_DATA | 23,
+	CEPH_OSD_OP_OMAPRMKEYS    = CEPH_OSD_OP_MODE_WR |
CEPH_OSD_OP_TYPE_DATA | 24,
+	CEPH_OSD_OP_OMAP_CMP      = CEPH_OSD_OP_MODE_RD |
CEPH_OSD_OP_TYPE_DATA | 25,
+
+	/** multi **/
+	CEPH_OSD_OP_CLONERANGE = CEPH_OSD_OP_MODE_WR | CEPH_OSD_OP_TYPE_MULTI | 1,
+	CEPH_OSD_OP_ASSERT_SRC_VERSION = CEPH_OSD_OP_MODE_RD |
CEPH_OSD_OP_TYPE_MULTI | 2,
+	CEPH_OSD_OP_SRC_CMPXATTR = CEPH_OSD_OP_MODE_RD |
CEPH_OSD_OP_TYPE_MULTI | 3,
+
 	/** attrs **/
 	/* read */
 	CEPH_OSD_OP_GETXATTR  = CEPH_OSD_OP_MODE_RD | CEPH_OSD_OP_TYPE_ATTR | 1,
@@ -238,6 +274,7 @@ enum {
 	CEPH_OSD_OP_SCRUB_RESERVE   = CEPH_OSD_OP_MODE_SUB | 6,
 	CEPH_OSD_OP_SCRUB_UNRESERVE = CEPH_OSD_OP_MODE_SUB | 7,
 	CEPH_OSD_OP_SCRUB_STOP      = CEPH_OSD_OP_MODE_SUB | 8,
+	CEPH_OSD_OP_SCRUB_MAP     = CEPH_OSD_OP_MODE_SUB | 9,

 	/** lock **/
 	CEPH_OSD_OP_WRLOCK    = CEPH_OSD_OP_MODE_WR | CEPH_OSD_OP_TYPE_LOCK | 1,
@@ -248,10 +285,12 @@ enum {
 	CEPH_OSD_OP_DNLOCK    = CEPH_OSD_OP_MODE_WR | CEPH_OSD_OP_TYPE_LOCK | 6,

 	/** exec **/
+	/* note: the RD bit here is wrong; see special-case below in helper */
 	CEPH_OSD_OP_CALL    = CEPH_OSD_OP_MODE_RD | CEPH_OSD_OP_TYPE_EXEC | 1,

 	/** pg **/
 	CEPH_OSD_OP_PGLS      = CEPH_OSD_OP_MODE_RD | CEPH_OSD_OP_TYPE_PG | 1,
+	CEPH_OSD_OP_PGLS_FILTER = CEPH_OSD_OP_MODE_RD | CEPH_OSD_OP_TYPE_PG | 2,
 };

 static inline int ceph_osd_op_type_lock(int op)
@@ -274,6 +313,10 @@ static inline int ceph_osd_op_type_pg(int op)
 {
 	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_PG;
 }
+static inline int ceph_osd_op_type_multi(int op)
+{
+	return (op & CEPH_OSD_OP_TYPE) == CEPH_OSD_OP_TYPE_MULTI;
+}

 static inline int ceph_osd_op_mode_subop(int op)
 {
@@ -281,11 +324,12 @@ static inline int ceph_osd_op_mode_subop(int op)
 }
 static inline int ceph_osd_op_mode_read(int op)
 {
-	return (op & CEPH_OSD_OP_MODE) == CEPH_OSD_OP_MODE_RD;
+	return (op & CEPH_OSD_OP_MODE_RD) &&
+		op != CEPH_OSD_OP_CALL;
 }
 static inline int ceph_osd_op_mode_modify(int op)
 {
-	return (op & CEPH_OSD_OP_MODE) == CEPH_OSD_OP_MODE_WR;
+	return op & CEPH_OSD_OP_MODE_WR;
 }

 /*
@@ -294,34 +338,38 @@ static inline int ceph_osd_op_mode_modify(int op)
  */
 #define CEPH_OSD_TMAP_HDR 'h'
 #define CEPH_OSD_TMAP_SET 's'
+#define CEPH_OSD_TMAP_CREATE 'c' /* create key */
 #define CEPH_OSD_TMAP_RM  'r'
+#define CEPH_OSD_TMAP_RMSLOPPY 'R'

 extern const char *ceph_osd_op_name(int op);

-
 /*
  * osd op flags
  *
  * An op may be READ, WRITE, or READ|WRITE.
  */
 enum {
-	CEPH_OSD_FLAG_ACK = 1,          /* want (or is) "ack" ack */
-	CEPH_OSD_FLAG_ONNVRAM = 2,      /* want (or is) "onnvram" ack */
-	CEPH_OSD_FLAG_ONDISK = 4,       /* want (or is) "ondisk" ack */
-	CEPH_OSD_FLAG_RETRY = 8,        /* resend attempt */
-	CEPH_OSD_FLAG_READ = 16,        /* op may read */
-	CEPH_OSD_FLAG_WRITE = 32,       /* op may write */
-	CEPH_OSD_FLAG_ORDERSNAP = 64,   /* EOLDSNAP if snapc is out of order */
-	CEPH_OSD_FLAG_PEERSTAT = 128,   /* msg includes osd_peer_stat */
-	CEPH_OSD_FLAG_BALANCE_READS = 256,
-	CEPH_OSD_FLAG_PARALLELEXEC = 512, /* execute op in parallel */
-	CEPH_OSD_FLAG_PGOP = 1024,      /* pg op, no object */
-	CEPH_OSD_FLAG_EXEC = 2048,      /* op may exec */
-	CEPH_OSD_FLAG_EXEC_PUBLIC = 4096, /* op may exec (public) */
+	CEPH_OSD_FLAG_ACK =            0x0001,  /* want (or is) "ack" ack */
+	CEPH_OSD_FLAG_ONNVRAM =        0x0002,  /* want (or is) "onnvram" ack */
+	CEPH_OSD_FLAG_ONDISK =         0x0004,  /* want (or is) "ondisk" ack */
+	CEPH_OSD_FLAG_RETRY =          0x0008,  /* resend attempt */
+	CEPH_OSD_FLAG_READ =           0x0010,  /* op may read */
+	CEPH_OSD_FLAG_WRITE =          0x0020,  /* op may write */
+	CEPH_OSD_FLAG_ORDERSNAP =      0x0040,  /* EOLDSNAP if snapc is out of
order */
+	CEPH_OSD_FLAG_PEERSTAT_OLD =   0x0080,  /* DEPRECATED msg includes
osd_peer_stat */
+	CEPH_OSD_FLAG_BALANCE_READS =  0x0100,
+	CEPH_OSD_FLAG_PARALLELEXEC =   0x0200,  /* execute op in parallel */
+	CEPH_OSD_FLAG_PGOP =           0x0400,  /* pg op, no object */
+	CEPH_OSD_FLAG_EXEC =           0x0800,  /* op may exec */
+	CEPH_OSD_FLAG_EXEC_PUBLIC =    0x1000,  /* DEPRECATED op may exec
(public) */
+	CEPH_OSD_FLAG_LOCALIZE_READS = 0x2000,  /* read from nearby replica,
if any */
+	CEPH_OSD_FLAG_RWORDERED =      0x4000,  /* order wrt concurrent reads */
 };

 enum {
 	CEPH_OSD_OP_FLAG_EXCL = 1,      /* EXCL object create */
+	CEPH_OSD_OP_FLAG_FAILOK = 2,    /* continue despite failure */
 };

 #define EOLDSNAPC    ERESTART  /* ORDERSNAP flag set; writer has old
snapc*/
@@ -381,7 +429,11 @@ struct ceph_osd_op {
 			__le64 ver;
 			__u8 flag;	/* 0 = unwatch, 1 = watch */
 		} __attribute__ ((packed)) watch;
-};
+		struct {
+			__le64 offset, length;
+			__le64 src_offset;
+		} __attribute__ ((packed)) clonerange;
+	};
 	__le32 payload_len;
 } __attribute__ ((packed));

@@ -424,5 +476,4 @@ struct ceph_osd_reply_head {
 } __attribute__ ((packed));


-
 #endif
-- 
1.7.9.5

[PATCH 2/3] libceph: add ceph_osd_state_name()

[Alex Elder]

Add the definition of ceph_osd_state_name(), to match its
counterpart in user space.

Signed-off-by: Alex Elder <elder@inktank.com>
---
 include/linux/ceph/rados.h |    2 ++
 net/ceph/ceph_strings.c    |   15 +++++++++++++++
 2 files changed, 17 insertions(+)

diff --git a/include/linux/ceph/rados.h b/include/linux/ceph/rados.h
index 9c3b4aa..b65182a 100644
--- a/include/linux/ceph/rados.h
+++ b/include/linux/ceph/rados.h
@@ -150,6 +150,8 @@ struct ceph_eversion {
 #define CEPH_OSD_AUTOOUT (1<<2)  /* osd was automatically marked out */
 #define CEPH_OSD_NEW     (1<<3)  /* osd is new, never marked in */

+extern const char *ceph_osd_state_name(int s);
+
 /* osd weights.  fixed point value: 0x10000 == 1.0 ("in"), 0 == "out" */
 #define CEPH_OSD_IN  0x10000
 #define CEPH_OSD_OUT 0
diff --git a/net/ceph/ceph_strings.c b/net/ceph/ceph_strings.c
index 3fbda04..833075c 100644
--- a/net/ceph/ceph_strings.c
+++ b/net/ceph/ceph_strings.c
@@ -68,6 +68,21 @@ const char *ceph_osd_op_name(int op)
 	return "???";
 }

+const char *ceph_osd_state_name(int s)
+{
+	switch (s) {
+	case CEPH_OSD_EXISTS:
+		return "exists";
+	case CEPH_OSD_UP:
+		return "up";
+	case CEPH_OSD_AUTOOUT:
+		return "autoout";
+	case CEPH_OSD_NEW:
+		return "new";
+	default:
+		return "???";
+	}
+}

 const char *ceph_pool_op_name(int op)
 {
-- 
1.7.9.5

[PATCH 3/3] libceph: update ceph_osd_op_name()

[Alex Elder]

Update ceph_osd_op_name() to include the newly-added definitions in
"rados.h", and to match its counterpart in the user space code.

Signed-off-by: Alex Elder <elder@inktank.com>
---
 net/ceph/ceph_strings.c |   24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/net/ceph/ceph_strings.c b/net/ceph/ceph_strings.c
index 833075c..1348df9 100644
--- a/net/ceph/ceph_strings.c
+++ b/net/ceph/ceph_strings.c
@@ -21,9 +21,15 @@ const char *ceph_osd_op_name(int op)
 	switch (op) {
 	case CEPH_OSD_OP_READ: return "read";
 	case CEPH_OSD_OP_STAT: return "stat";
+	case CEPH_OSD_OP_MAPEXT: return "mapext";
+	case CEPH_OSD_OP_SPARSE_READ: return "sparse-read";
+	case CEPH_OSD_OP_NOTIFY: return "notify";
+	case CEPH_OSD_OP_NOTIFY_ACK: return "notify-ack";
+	case CEPH_OSD_OP_ASSERT_VER: return "assert-version";

 	case CEPH_OSD_OP_MASKTRUNC: return "masktrunc";

+	case CEPH_OSD_OP_CREATE: return "create";
 	case CEPH_OSD_OP_WRITE: return "write";
 	case CEPH_OSD_OP_DELETE: return "delete";
 	case CEPH_OSD_OP_TRUNCATE: return "truncate";
@@ -39,6 +45,11 @@ const char *ceph_osd_op_name(int op)
 	case CEPH_OSD_OP_TMAPUP: return "tmapup";
 	case CEPH_OSD_OP_TMAPGET: return "tmapget";
 	case CEPH_OSD_OP_TMAPPUT: return "tmapput";
+	case CEPH_OSD_OP_WATCH: return "watch";
+
+	case CEPH_OSD_OP_CLONERANGE: return "clonerange";
+	case CEPH_OSD_OP_ASSERT_SRC_VERSION: return "assert-src-version";
+	case CEPH_OSD_OP_SRC_CMPXATTR: return "src-cmpxattr";

 	case CEPH_OSD_OP_GETXATTR: return "getxattr";
 	case CEPH_OSD_OP_GETXATTRS: return "getxattrs";
@@ -53,6 +64,10 @@ const char *ceph_osd_op_name(int op)
 	case CEPH_OSD_OP_BALANCEREADS: return "balance-reads";
 	case CEPH_OSD_OP_UNBALANCEREADS: return "unbalance-reads";
 	case CEPH_OSD_OP_SCRUB: return "scrub";
+	case CEPH_OSD_OP_SCRUB_RESERVE: return "scrub-reserve";
+	case CEPH_OSD_OP_SCRUB_UNRESERVE: return "scrub-unreserve";
+	case CEPH_OSD_OP_SCRUB_STOP: return "scrub-stop";
+	case CEPH_OSD_OP_SCRUB_MAP: return "scrub-map";

 	case CEPH_OSD_OP_WRLOCK: return "wrlock";
 	case CEPH_OSD_OP_WRUNLOCK: return "wrunlock";
@@ -64,6 +79,15 @@ const char *ceph_osd_op_name(int op)
 	case CEPH_OSD_OP_CALL: return "call";

 	case CEPH_OSD_OP_PGLS: return "pgls";
+	case CEPH_OSD_OP_PGLS_FILTER: return "pgls-filter";
+	case CEPH_OSD_OP_OMAPGETKEYS: return "omap-get-keys";
+	case CEPH_OSD_OP_OMAPGETVALS: return "omap-get-vals";
+	case CEPH_OSD_OP_OMAPGETHEADER: return "omap-get-header";
+	case CEPH_OSD_OP_OMAPGETVALSBYKEYS: return "omap-get-vals-by-keys";
+	case CEPH_OSD_OP_OMAPSETVALS: return "omap-set-vals";
+	case CEPH_OSD_OP_OMAPSETHEADER: return "omap-set-header";
+	case CEPH_OSD_OP_OMAPCLEAR: return "omap-clear";
+	case CEPH_OSD_OP_OMAPRMKEYS: return "omap-rm-keys";
 	}
 	return "???";
 }
-- 
1.7.9.5

Re: [PATCH 0/3] libceph: update rados.h and related code

[Josh Durgin]

On 02/16/2013 08:56 AM, Alex Elder wrote:
> The ceph kernel and user space code bases have a few header files
> in common, but because they're maintained separately they sometimes
> diverge.
>
> This series updates "include/linux/ceph/rados.h" to nearly match
> its user space counterpart in "src/include/rados.h".
>
> There are still a few things that should be reconciled by updating
> that user space code; that work is covered by:
>      http://tracker.ceph.com/issues/4167
>
> 					-Alex
>
> [PATCH 1/3] libceph: update rados.h
> [PATCH 2/3] libceph: add ceph_osd_state_name()
> [PATCH 3/3] libceph: update ceph_osd_op_name()

These look good.
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>