Skip to contentSkip to conversation

xfsprogs: v4 inode type in directory

PATCH Superseded Mark Tinguely 2 participants 7 messages Oct 16, 2013 – Oct 17, 2013

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

cover xfsprogs: v4 inode type in directory

Here are the patches that enable the inode in the directory feature in v4 superblocks.

patch 1: add the entries to xfs_sb.h (sync with kernel)
patch 2: add the XFS_FSOP_GEOM_FLAGS_FTYPE to xfs_fs.h (sync with kernel)
add the entry to repair so that xfs_info reports the feature
patch 3: add the feature to mkfs.xfs and manual page.
note: this new feature is ignored for superblock v5
automatically turns on this feature.

--Mark.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#2 lore raw

PATCH 2/3 xfsprog: add dirent filetype information for xfs_info

Make xfs_info aware of the directory inode type by using the XFS_FSOP_GEOM_FLAGS_FTYPE set in the kernel.

---
changed the display field to "ftype", same as the mkfs.xfs entry that made it.

growfs/xfs_growfs.c | 14 +++++++++-----

 growfs/xfs_growfs.c |   14 +++++++++-----
 include/xfs_fs.h    |    3 ++-
 2 files changed, 11 insertions(+), 6 deletions(-)
growfs/xfs_growfs.c +9 -5
Index: b/growfs/xfs_growfs.c
===================================================================
@@ -55,7 +55,8 @@ report_info(
 	int		attrversion,
 	int		projid32bit,
 	int		crcs_enabled,
-	int		cimode)
+	int		cimode,
+	int		ftype_enabled)
 {
 	printf(_(
 	    "meta-data=%-22s isize=%-6u agcount=%u, agsize=%u blks\n"
@@ -63,7 +64,7 @@ report_info(
 	    "         =%-22s crc=%u\n"
 	    "data     =%-22s bsize=%-6u blocks=%llu, imaxpct=%u\n"
 	    "         =%-22s sunit=%-6u swidth=%u blks\n"
-	    "naming   =version %-14u bsize=%-6u ascii-ci=%d\n"
+	    "naming   =version %-14u bsize=%-6u ascii-ci=%d ftype=%d\n"
 	    "log      =%-22s bsize=%-6u blocks=%u, version=%u\n"
 	    "         =%-22s sectsz=%-5u sunit=%u blks, lazy-count=%u\n"
 	    "realtime =%-22s extsz=%-6u blocks=%llu, rtextents=%llu\n"),
@@ -74,7 +75,7 @@ report_info(
 		"", geo.blocksize, (unsigned long long)geo.datablocks,
 			geo.imaxpct,
 		"", geo.sunit, geo.swidth,
-  		dirversion, geo.dirblocksize, cimode,
+  		dirversion, geo.dirblocksize, cimode, ftype_enabled,
 		isint ? _("internal") : logname ? logname : _("external"),
 			geo.blocksize, geo.logblocks, logversion,
 		"", geo.logsectsize, geo.logsunit / geo.blocksize, lazycount,
@@ -121,6 +122,7 @@ main(int argc, char **argv)
 	libxfs_init_t		xi;	/* libxfs structure */
 	int			projid32bit;
 	int			crcs_enabled;
+	int			ftype_enabled = 0;
 
 	progname = basename(argv[0]);
 	setlocale(LC_ALL, "");
@@ -242,10 +244,12 @@ main(int argc, char **argv)
 	ci = geo.flags & XFS_FSOP_GEOM_FLAGS_DIRV2CI ? 1 : 0;
 	projid32bit = geo.flags & XFS_FSOP_GEOM_FLAGS_PROJID32 ? 1 : 0;
 	crcs_enabled = geo.flags & XFS_FSOP_GEOM_FLAGS_V5SB ? 1 : 0;
+	ftype_enabled = geo.flags & XFS_FSOP_GEOM_FLAGS_FTYPE ? 1 : 0;
 	if (nflag) {
 		report_info(geo, datadev, isint, logdev, rtdev,
 				lazycount, dirversion, logversion,
-				attrversion, projid32bit, crcs_enabled, ci);
+				attrversion, projid32bit, crcs_enabled, ci,
+				ftype_enabled);
 		exit(0);
 	}
 
@@ -282,7 +286,7 @@ main(int argc, char **argv)
 
 	report_info(geo, datadev, isint, logdev, rtdev,
 			lazycount, dirversion, logversion,
-			attrversion, projid32bit, crcs_enabled, ci);
+			attrversion, projid32bit, crcs_enabled, ci, ftype_enabled);
 
 	ddsize = xi.dsize;
 	dlsize = ( xi.logBBsize? xi.logBBsize :
include/xfs_fs.h +2 -1
Index: b/include/xfs_fs.h
===================================================================
@@ -233,10 +233,11 @@ typedef struct xfs_fsop_resblks {
 #define XFS_FSOP_GEOM_FLAGS_LOGV2	0x0100	/* log format version 2	*/
 #define XFS_FSOP_GEOM_FLAGS_SECTOR	0x0200	/* sector sizes >1BB	*/
 #define XFS_FSOP_GEOM_FLAGS_ATTR2	0x0400	/* inline attributes rework */
-#define XFS_FSOP_GEOM_FLAGS_PROJID32	0x0800  /* 32-bit project IDs	*/
+#define XFS_FSOP_GEOM_FLAGS_PROJID32	0x0800	/* 32-bit project IDs	*/
 #define XFS_FSOP_GEOM_FLAGS_DIRV2CI	0x1000	/* ASCII only CI names	*/
 #define XFS_FSOP_GEOM_FLAGS_LAZYSB	0x4000	/* lazy superblock counters */
 #define XFS_FSOP_GEOM_FLAGS_V5SB	0x8000	/* version 5 superblock */
+#define XFS_FSOP_GEOM_FLAGS_FTYPE	0x10000	/* inode directory types */
 
 
 /*

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#3 lore raw

PATCH 3/3 xfsprog: add mkfs.xfs sb v4 support for dirent filetype field

In reply to Mark Tinguely

Add directory inode type feature to mkfs.xfs and its manual page.

In sb v4, "mkfs.xfs -n ftype=1" turns on the feature. The feature is still automatically turned on for sb v5. Reject the "ftype=1" request if used with the "crc=1" setting.

---
updated the manual entry.
changed the variables in xfs_mkfs.c
call usage() if the crc and ftype are used together.
update the sb_features_incompat entry for sb v5 and feature bit for v4

man/man8/mkfs.xfs.8 |    7 +++++++
 man/man8/mkfs.xfs.8 |   15 +++++++++++++++
 mkfs/xfs_mkfs.c     |   42 +++++++++++++++++++++++++++++++-----------
 mkfs/xfs_mkfs.h     |    4 +++-
 3 files changed, 49 insertions(+), 12 deletions(-)
man/man8/mkfs.xfs.8 +15 -0
Index: b/man/man8/mkfs.xfs.8
===================================================================
@@ -517,6 +517,21 @@ option enables ASCII only case-insensiti
 are stored in directories using the case they were created with.
 .IP
 Note: Version 1 directories are not supported.
+.TP
+.BI ftype= value
+This feature introduced in Linux 3.12 allows the inode type to
+be stored in the directory structure so that readdir(3) and getdents(2)
+do not need to look up the inode to determine the inode type.
+
+The
+.I value
+is either 0 or 1, with 1 signifiying that filetype information
+will be stored in the directory structure. The default value is 0.
+
+When CRCs are enabled via "mkfs.xfs -m crc=1", the ftype functionality is
+always enabled. This feature can not be turned off for such filesystem
+configurations.
+.IP
 .RE
 .TP
 .BI \-p " protofile"
mkfs/xfs_mkfs.c +31 -11

115 lines not on this page · raw message

mkfs/xfs_mkfs.h +3 -1
Index: b/mkfs/xfs_mkfs.h
===================================================================
@@ -36,12 +36,14 @@
 	        XFS_DFL_SB_VERSION_BITS |                               \
 	0 ) : XFS_SB_VERSION_1 )
 
-#define XFS_SB_VERSION2_MKFS(crc, lazycount, attr2, projid32bit, parent) (\
+#define XFS_SB_VERSION2_MKFS(crc, lazycount, attr2, projid32bit, parent, \
+			     ftype) (\
 	((lazycount) ? XFS_SB_VERSION2_LAZYSBCOUNTBIT : 0) |		\
 	((attr2) ? XFS_SB_VERSION2_ATTR2BIT : 0) |			\
 	((projid32bit) ? XFS_SB_VERSION2_PROJID32BIT : 0) |		\
 	((parent) ? XFS_SB_VERSION2_PARENTBIT : 0) |			\
 	((crc) ? XFS_SB_VERSION2_CRCBIT : 0) |				\
+	((ftype) ? XFS_SB_VERSION2_FTYPE : 0) |				\
 	0 )
 
 #define	XFS_DFL_BLOCKSIZE_LOG	12		/* 4096 byte blocks */

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#1 lore raw

PATCH 1/3 xfsprog: add xfs sb v4 support for dirent filetype field

In reply to Mark Tinguely

Add xfsprog superblock v4 support for the inode type information in the xfs directory feature in xfs_sb.h.

This support adds a feature bit for version 4 superblocks and leaves the original superblock 5 incompatibility bit.

---
Same as the previous version. I did not change to:

	return XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
	       (xfs_sb_has_incompat_feature(sbp, XFS_SB_FEAT_INCOMPAT_FTYPE) ||
		((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) &&
		 (sbp->sb_features2 & XFS_SB_VERSION2_FTYPE)));

because xfs_sb_has_incompat_feature is not valid in superblock v4

include/xfs_sb.h | 20 ++++++++++++--------

 include/xfs_sb.h |   10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
include/xfs_sb.h +7 -3
Index: b/include/xfs_sb.h
===================================================================
@@ -84,11 +84,13 @@ struct xfs_trans;
 #define XFS_SB_VERSION2_PARENTBIT	0x00000010	/* parent pointers */
 #define XFS_SB_VERSION2_PROJID32BIT	0x00000080	/* 32 bit project id */
 #define XFS_SB_VERSION2_CRCBIT		0x00000100	/* metadata CRCs */
+#define XFS_SB_VERSION2_FTYPE		0x00000200	/* inode type in dir */
 
 #define	XFS_SB_VERSION2_OKREALFBITS	\
 	(XFS_SB_VERSION2_LAZYSBCOUNTBIT	| \
 	 XFS_SB_VERSION2_ATTR2BIT	| \
-	 XFS_SB_VERSION2_PROJID32BIT)
+	 XFS_SB_VERSION2_PROJID32BIT	| \
+	 XFS_SB_VERSION2_FTYPE)
 #define	XFS_SB_VERSION2_OKSASHFBITS	\
 	(0)
 #define XFS_SB_VERSION2_OKREALBITS	\
@@ -631,8 +633,10 @@ static inline int xfs_sb_version_has_pqu
 
 static inline int xfs_sb_version_hasftype(struct xfs_sb *sbp)
 {
-	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 &&
-		xfs_sb_has_incompat_feature(sbp, XFS_SB_FEAT_INCOMPAT_FTYPE);
+	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 &&
+		xfs_sb_has_incompat_feature(sbp, XFS_SB_FEAT_INCOMPAT_FTYPE)) ||
+	       (xfs_sb_version_hasmorebits(sbp) &&
+		 (sbp->sb_features2 & XFS_SB_VERSION2_FTYPE));
 }
 
 /*

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

#4 lore raw

Re: [PATCH 3/3] xfsprog: add mkfs.xfs sb v4 support for dirent filetype field

In reply to Mark Tinguely
Quoted 31 lines · Mark Tinguely

Add directory inode type feature to mkfs.xfs and its manual page.

In sb v4, "mkfs.xfs -n ftype=1" turns on the feature. The feature is still automatically turned on for sb v5. Reject the "ftype=1" request if used with the "crc=1" setting.

---
updated the manual entry.
changed the variables in xfs_mkfs.c
call usage() if the crc and ftype are used together.
update the sb_features_incompat entry for sb v5 and feature bit for v4

man/man8/mkfs.xfs.8 |    7 +++++++
 man/man8/mkfs.xfs.8 |   15 +++++++++++++++
 mkfs/xfs_mkfs.c     |   42 +++++++++++++++++++++++++++++++-----------
 mkfs/xfs_mkfs.h     |    4 +++-
 3 files changed, 49 insertions(+), 12 deletions(-)

Index: b/man/man8/mkfs.xfs.8
===================================================================

man/man8/mkfs.xfs.8 +4 -0
@@ -517,6 +517,21 @@ option enables ASCII only case-insensiti
 are stored in directories using the case they were created with.
 .IP
 Note: Version 1 directories are not supported.
+.TP
+.BI ftype= value
+This feature introduced in Linux 3.12 allows the inode type to
+be stored in the directory structure so that readdir(3) and getdents(2)

man/man8/mkfs.xfs.8 option enables ASCII only case-insensiti · line 533

Quoted 10 lines
man/man8/mkfs.xfs.8 +10 -0
+do not need to look up the inode to determine the inode type.
+
+The
+.I value
+is either 0 or 1, with 1 signifiying that filetype information
+will be stored in the directory structure. The default value is 0.
+
+When CRCs are enabled via "mkfs.xfs -m crc=1", the ftype functionality is
+always enabled. This feature can not be turned off for such filesystem
+configurations.

I don't think you need the "mkfs.xfs -m crc=1" in that paragraph. If you want to put the "-m crc=1" option in there, the way it is done in the rest of the man page is just as "-m crc=1" in bold type, not in quotes, and it doesn't need the mkfs.xfs prefix.

man/man8/mkfs.xfs.8 main( · line 1484

Quoted 6 lines
man/man8/mkfs.xfs.8 +2 -0
@@ -1475,6 +1480,8 @@ main(
 					if (c < 0 || c > 1)
 						illegal(value, "m crc");
 					crcs_enabled = c;
+					if (dirftype && crcs_enabled)
+						usage();

That doesn't work. dirftype will only be set if ftype is enabled. You should be checking nftype here. Also, consider this inconsistent parsing of teh same options:

$ sudo ~/packages/mkfs.xfs -f -n ftype=0 -m crc=1 /dev/vda
meta-data=/dev/vda               isize=512    agcount=4, agsize=327680 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1
data     =                       bsize=4096   blocks=1310720, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0 ftype=1
log      =internal log           bsize=4096   blocks=12800, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0
$ sudo ~/packages/mkfs.xfs -f -m crc=1 -n ftype=0 /dev/vda
Usage: mkfs.xfs
/* blocksize */         [-b log=n|size=num]
/* metadata */          [-m crc=[0|1]
/* data subvol */       [-d agcount=n,agsize=n,file,name=xxx,size=num,
                            (sunit=value,swidth=value|su=num,sw=num),
                            sectlog=n|sectsize=num
/* force overwrite */   [-f]
/* inode size */        [-i log=n|perblock=n|size=num,maxpct=n,attr=0|1|2,
                            projid32bit=0|1]
/* no discard */        [-K]
/* log subvol */        [-l agnum=n,internal,size=num,logdev=xxx,version=n
                            sunit=value|su=num,sectlog=n|sectsize=num,
                            lazy-count=0|1]
/* label */             [-L label (maximum 12 characters)]
/* naming */            [-n log=n|size=num,version=2|ci,ftype=0|1]
/* no-op info only */   [-N]
/* prototype file */    [-p fname]
/* quiet */             [-q]
/* realtime subvol */   [-r extsize=num,size=num,rtdev=xxx]
/* sectorsize */        [-s log=n|size=num]
/* version */           [-V]
                        devicename
<devicename> is required unless -d name=xxx is given.
<num> is xxx (bytes), xxxs (sectors), xxxb (fs blocks), xxxk (xxx KiB),
      xxxm (xxx MiB), xxxg (xxx GiB), xxxt (xxx TiB) or xxxp (xxx PiB).
<value> is xxx (512 byte blocks).
$

Further, just calling usage here doesn't tell the user what they did wrong that caused mkfs to abort. It should output what the error in the parameters is, especially as there is no indication in the usage message that ftype is not valid with crcs. Same for the usage call in the -m crc parsing.

Quoted 33 lines
man/man8/mkfs.xfs.8 +10 -2
 	}
 	validate_log_size(logblocks, blocklog, min_logblocks);
 
+	/*
+	 * dirent filetype field always enabled on v5 superblocks
+	 */
+	if (crcs_enabled) {
+		sbp->sb_features_incompat = XFS_SB_FEAT_INCOMPAT_FTYPE;
+		dirftype = 1;
+	}
+
 	if (!qflag || Nflag) {
 		printf(_(
 		   "meta-data=%-22s isize=%-6d agcount=%lld, agsize=%lld blks\n"
@@ -2441,7 +2466,7 @@ _("size %s specified for log subvolume i
 		   "         =%-22s crc=%u\n"
 		   "data     =%-22s bsize=%-6u blocks=%llu, imaxpct=%u\n"
 		   "         =%-22s sunit=%-6u swidth=%u blks\n"
-		   "naming   =version %-14u bsize=%-6u ascii-ci=%d\n"
+		   "naming   =version %-14u bsize=%-6u ascii-ci=%d ftype=%d\n"
 		   "log      =%-22s bsize=%-6d blocks=%lld, version=%d\n"
 		   "         =%-22s sectsz=%-5u sunit=%d blks, lazy-count=%d\n"
 		   "realtime =%-22s extsz=%-6d blocks=%lld, rtextents=%lld\n"),
@@ -2450,7 +2475,7 @@ _("size %s specified for log subvolume i
 			"", crcs_enabled,
 			"", blocksize, (long long)dblocks, imaxpct,
 			"", dsunit, dswidth,
-			dirversion, dirblocksize, nci,
+			dirversion, dirblocksize, nci, dirftype,
 			logfile, 1 << blocklog, (long long)logblocks,
 			logversion, "", lsectorsize, lsunit, lazy_sb_counters,
 			rtfile, rtextblocks << blocklog,

man/man8/mkfs.xfs.8 _("size %s specified for log subvolume i · line 2543

Quoted 8 lines
man/man8/mkfs.xfs.8 +3 -1
@@ -2512,8 +2537,10 @@ _("size %s specified for log subvolume i
 		sbp->sb_logsectlog = 0;
 		sbp->sb_logsectsize = 0;
 	}
+
 	sbp->sb_features2 = XFS_SB_VERSION2_MKFS(crcs_enabled, lazy_sb_counters,
-					attrversion == 2, !projid16bit, 0);
+				   attrversion == 2, !projid16bit, 0,
+				   (crcs_enabled && dirftype));

That will never set the ftype bit on v4 superblocks - it will only ever get set on v5 superblocks:

$ sudo ~/packages/mkfs.xfs -f -n ftype=1 /dev/vda
meta-data=/dev/vda               isize=256    agcount=4, agsize=327680 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=0
data     =                       bsize=4096   blocks=1310720, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0 ftype=1
                                                         ^^^^^^^
log      =internal log           bsize=4096   blocks=12800, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0
$ sudo xfs_db -r -c version /dev/vda
versionnum [0xb4a4+0x8a] = V4,NLINK,ALIGN,DIRV2,LOGV2,EXTFLG,MOREBITS,ATTR2,LAZYSBCOUNT,PROJID32BIT
$

The sb_features2 field has the value of 0x8a, which is:

#define XFS_SB_VERSION2_LAZYSBCOUNTBIT  0x00000002      /* Superblk counters */  
#define XFS_SB_VERSION2_ATTR2BIT        0x00000008      /* Inline attr rework */ 
#define XFS_SB_VERSION2_PROJID32BIT     0x00000080      /* 32 bit project id */  

Clearly the ftype feature bit is not set:

#define XFS_SB_VERSION2_FTYPE 0x00000200 /* inode type in dir */

So any testing you've done with this mkfs patch hasn't tested dtype enabled dirents on v4 filesystems at all....

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#6 lore raw
In reply to Mark Tinguely
Quoted 9 lines · Mark Tinguely

Here are the patches that enable the inode in the directory feature in v4 superblocks.

patch 1: add the entries to xfs_sb.h (sync with kernel)
patch 2: add the XFS_FSOP_GEOM_FLAGS_FTYPE to xfs_fs.h (sync with kernel)
add the entry to repair so that xfs_info reports the feature
patch 3: add the feature to mkfs.xfs and manual page.
note: this new feature is ignored for superblock v5
automatically turns on this feature.

xfs_db needs to have the version string output updated, it also needs support for dirent dtype dumping for block, leaf and node form directories.

Cheers,

Dave.

Signature · 7 lines
Dave Chinner
[email protected]

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
#5 lore raw

Re: [PATCH 3/3] xfsprog: add mkfs.xfs sb v4 support for dirent filetype field

In reply to Dave Chinner
Quoted 7 lines · Dave Chinner
Quoted 6 lines · Mark Tinguely

Add directory inode type feature to mkfs.xfs and its manual page.

In sb v4, "mkfs.xfs -n ftype=1" turns on the feature. The feature is still automatically turned on for sb v5. Reject the "ftype=1" request if used with the "crc=1" setting.

Quoted 8 lines

mkfs/xfs_mkfs.c _("size %s specified for log subvolume i · line 2543

Quoted 8 lines

---

mkfs/xfs_mkfs.c +3 -1
+
  	sbp->sb_features2 = XFS_SB_VERSION2_MKFS(crcs_enabled, lazy_sb_counters,
-					attrversion == 2, !projid16bit, 0);
+				   attrversion == 2, !projid16bit, 0,
+				   (crcs_enabled && dirftype));

Eek that change was made yesterday and should be:
(!crcs_enabled && dirftype)
so the feature bit is only turned on for v4 sb.

--Mark.

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs