| Message ID | 20201123173846.15045-1-lhenriques@suse.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] ceph: add ceph.caps vxattr | expand |
On Mon, 2020-11-23 at 17:38 +0000, Luis Henriques wrote: > Add a new vxattr that allows userspace to list the caps for a specific > directory or file. > > Signed-off-by: Luis Henriques <lhenriques@suse.de> > --- > Hi! > > Here's a version that also shows the caps in hexadecimal format, as > suggested by Jeff. IMO the parenthesis and the '0x' prefix help the > readability, but they may make it a bit harder for scripts to parsing the > output. I'm OK dropping those. > > Cheers, Looks good, merged into "testing". I did make a slight change to the format -- instead of putting the hex value in parenthesis, I separated the two fields with a /, which I think should make things easier for scripts to parse. You should be able to do something like this to get at the hex value for testing: $ getfattr -n ceph.caps foo | cut -d / -f2 Let me know if you see issues with that and we can revisit the format. Thanks!
Jeff Layton <jlayton@kernel.org> writes: > On Mon, 2020-11-23 at 17:38 +0000, Luis Henriques wrote: >> Add a new vxattr that allows userspace to list the caps for a specific >> directory or file. >> >> Signed-off-by: Luis Henriques <lhenriques@suse.de> >> --- >> Hi! >> >> Here's a version that also shows the caps in hexadecimal format, as >> suggested by Jeff. IMO the parenthesis and the '0x' prefix help the >> readability, but they may make it a bit harder for scripts to parsing the >> output. I'm OK dropping those. >> >> Cheers, > > Looks good, merged into "testing". Awesome, thanks! > I did make a slight change to the format -- instead of putting the hex > value in parenthesis, I separated the two fields with a /, which I think > should make things easier for scripts to parse. > > You should be able to do something like this to get at the hex value for > testing: > > $ getfattr -n ceph.caps foo | cut -d / -f2 > > Let me know if you see issues with that and we can revisit the format. Sure, I'm OK with that. Or even simply dropping any separator, having only a space/tab between the string and the hex value. Another option I saw was to have two vxattrs: ceph.caps.string and ceph.caps.int. But that's probably overkill. Cheers,
diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c index 197cb1234341..aec9bd5c8e77 100644 --- a/fs/ceph/xattr.c +++ b/fs/ceph/xattr.c @@ -303,6 +303,19 @@ static ssize_t ceph_vxattrcb_snap_btime(struct ceph_inode_info *ci, char *val, ci->i_snap_btime.tv_nsec); } +static ssize_t ceph_vxattrcb_caps(struct ceph_inode_info *ci, char *val, + size_t size) +{ + int issued; + + spin_lock(&ci->i_ceph_lock); + issued = __ceph_caps_issued(ci, NULL); + spin_unlock(&ci->i_ceph_lock); + + return ceph_fmt_xattr(val, size, "%s (0x%x)", + ceph_cap_string(issued), issued); +} + #define CEPH_XATTR_NAME(_type, _name) XATTR_CEPH_PREFIX #_type "." #_name #define CEPH_XATTR_NAME2(_type, _name, _name2) \ XATTR_CEPH_PREFIX #_type "." #_name "." #_name2 @@ -378,6 +391,13 @@ static struct ceph_vxattr ceph_dir_vxattrs[] = { .exists_cb = ceph_vxattrcb_snap_btime_exists, .flags = VXATTR_FLAG_READONLY, }, + { + .name = "ceph.caps", + .name_size = sizeof("ceph.caps"), + .getxattr_cb = ceph_vxattrcb_caps, + .exists_cb = NULL, + .flags = VXATTR_FLAG_HIDDEN, + }, { .name = NULL, 0 } /* Required table terminator */ }; @@ -403,6 +423,13 @@ static struct ceph_vxattr ceph_file_vxattrs[] = { .exists_cb = ceph_vxattrcb_snap_btime_exists, .flags = VXATTR_FLAG_READONLY, }, + { + .name = "ceph.caps", + .name_size = sizeof("ceph.caps"), + .getxattr_cb = ceph_vxattrcb_caps, + .exists_cb = NULL, + .flags = VXATTR_FLAG_HIDDEN, + }, { .name = NULL, 0 } /* Required table terminator */ };
Add a new vxattr that allows userspace to list the caps for a specific directory or file. Signed-off-by: Luis Henriques <lhenriques@suse.de> --- Hi! Here's a version that also shows the caps in hexadecimal format, as suggested by Jeff. IMO the parenthesis and the '0x' prefix help the readability, but they may make it a bit harder for scripts to parsing the output. I'm OK dropping those. Cheers,