Wrong d_type presented by xfs_io
Hi guys,
I've been looking through an issue raised by Brian, regarding my RENAME_EXCHANGE patchset, where, apparently, the d_type of some files are reseted during the exchange. Brian presented an example here:
http://oss.sgi.com/archives/xfs/2014-11/msg00170.html
When trying to debug it, I've been trying to create and rename some FS objects to try to reproduce the issue, but, all the files I've been creating are shown as DT_UNKNOWN type:
ex:
/dev/vdb2 on /xfs/scratch type xfs (rw,relatime,seclabel,attr2,inode64,noquota)
[root@localhost scratch]# mkdir /xfs/scratch/dir1
[root@localhost scratch]# xfs_io -c "readdir -v" /xfs/scratch/
00000006: d_ino: 0x00000083 d_off: 0x00000200 d_reclen: 0x18 d_type: DT_UNKNOWNd_name: dir1
[root@localhost scratch]# mknod /xfs/scratch/mybdev b 253 19
00000008: d_ino: 0x00000084 d_off: 0x00000200 d_reclen: 0x20 d_type: DT_UNKNOWNd_name: mybdev
So, anyone knows if is there any problem with xfs_io, or, if I'm doing something wrong here?
[root@localhost scratch]# xfs_io -Vxfs_io version 3.2.1
or maybe, we have a bug somewhere else.
Cheers
Signature · 6 lines
Carlos _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs
Quoted 6 lines · Carlos Maiolino
Hi guys,
I've been looking through an issue raised by Brian, regarding my
RENAME_EXCHANGEpatchset, where, apparently, thed_typeof some files are reseted during the exchange. Brian presented an example here:
Quoted 6 lines
http://oss.sgi.com/archives/xfs/2014-11/msg00170.html
When trying to debug it, I've been trying to create and rename some FS objects to try to reproduce the issue, but, all the files I've been creating are shown as
DT_UNKNOWNtype:
Hi Carlos,
I suspect you aren't using a v5 superblock. You can do that by passing '-m crc=1' to mkfs. Does d_type work once you're on v5?
Brian
Quoted 23 lines
ex:
/dev/vdb2on /xfs/scratch type xfs (rw,relatime,seclabel,attr2,inode64,noquota)[root@localhost scratch]# mkdir /xfs/scratch/dir1
[root@localhost scratch]#
xfs_io-c "readdir -v" /xfs/scratch/00000006:
d_ino:0x00000083d_off:0x00000200d_reclen:0x18d_type:DT_UNKNOWNd_name: dir1[root@localhost scratch]# mknod /xfs/scratch/mybdev b 253 19
00000008:
d_ino:0x00000084d_off:0x00000200d_reclen:0x20d_type:DT_UNKNOWNd_name: mybdevSo, anyone knows if is there any problem with
xfs_io, or, if I'm doing something wrong here?
Quoted 16 lines
[root@localhost scratch]#
xfs_io-Vxfs_ioversion 3.2.1or maybe, we have a bug somewhere else.
Cheers
Signature · 6 lines
Carlos _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 14 lines · Brian Foster
Quoted 6 lines · Carlos Maiolino
Hi guys,
I've been looking through an issue raised by Brian, regarding my
RENAME_EXCHANGEpatchset, where, apparently, thed_typeof some files are reseted during the exchange. Brian presented an example here:Quoted 6 lines
http://oss.sgi.com/archives/xfs/2014-11/msg00170.html
When trying to debug it, I've been trying to create and rename some FS objects to try to reproduce the issue, but, all the files I've been creating are shown as
DT_UNKNOWNtype:
Quoted 5 lines
Hi Carlos,
I suspect you aren't using a v5 superblock. You can do that by passing '-m crc=1' to mkfs. Does
d_typework once you're on v5?
Yes, it works, thanks man :)
Quoted 46 lines
Brian
Quoted 23 lines
ex:
/dev/vdb2on /xfs/scratch type xfs (rw,relatime,seclabel,attr2,inode64,noquota)[root@localhost scratch]# mkdir /xfs/scratch/dir1
[root@localhost scratch]#
xfs_io-c "readdir -v" /xfs/scratch/00000006:
d_ino:0x00000083d_off:0x00000200d_reclen:0x18d_type:DT_UNKNOWNd_name: dir1[root@localhost scratch]# mknod /xfs/scratch/mybdev b 253 19
00000008:
d_ino:0x00000084d_off:0x00000200d_reclen:0x20d_type:DT_UNKNOWNd_name: mybdevSo, anyone knows if is there any problem with
xfs_io, or, if I'm doing something wrong here?Quoted 16 lines
[root@localhost scratch]#
xfs_io-Vxfs_ioversion 3.2.1or maybe, we have a bug somewhere else.
Cheers
Signature · 6 lines
Carlos _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Signature · 6 lines
Carlos _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs