Message ID | 20170411172235.9361-1-billodo@redhat.com (mailing list archive) |
---|---|
State | Superseded, archived |
Headers | show |
On 4/11/17 12:22 PM, Bill O'Donnell wrote: > xfs_growfs manpage clearly states that the filesystem must > be mounted to be grown. Current behavior allows xfs_growfs > to proceed if the filesystem /containing/ the path > of the desired target is mounted. This is not the specified > behavior. Instead, also check the targeted fs argument against > the entry found in the fstable lookup. Unless the targeted > fs is actually mounted, reject the command. > > In order to cover bind-mounts, create a new lookup function > based on the mountpoints instead of just the device name. > > Signed-off-by: Bill O'Donnell <billodo@redhat.com> > --- > > v2: in order to properly handle relative pathnames, symlinks, > and bind-mounts, use realpath to establish canonical path name. > This also requires the introduction of a lookup function based > on the target mountpoint. I still need to review this one carefully, but have you made any progress towards an xfstest for this behavior? Thanks, -Eric -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Apr 19, 2017 at 05:05:36PM -0500, Eric Sandeen wrote: > On 4/11/17 12:22 PM, Bill O'Donnell wrote: > > xfs_growfs manpage clearly states that the filesystem must > > be mounted to be grown. Current behavior allows xfs_growfs > > to proceed if the filesystem /containing/ the path > > of the desired target is mounted. This is not the specified > > behavior. Instead, also check the targeted fs argument against > > the entry found in the fstable lookup. Unless the targeted > > fs is actually mounted, reject the command. > > > > In order to cover bind-mounts, create a new lookup function > > based on the mountpoints instead of just the device name. > > > > Signed-off-by: Bill O'Donnell <billodo@redhat.com> > > --- > > > > v2: in order to properly handle relative pathnames, symlinks, > > and bind-mounts, use realpath to establish canonical path name. > > This also requires the introduction of a lookup function based > > on the target mountpoint. > > I still need to review this one carefully, but have you made any > progress towards an xfstest for this behavior? xfstest is still a work in progress. Hopefully will have one by end of this week. Thanks- Bill > > Thanks, > -Eric > -- > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 4/11/17 12:22 PM, Bill O'Donnell wrote: > xfs_growfs manpage clearly states that the filesystem must > be mounted to be grown. Current behavior allows xfs_growfs > to proceed if the filesystem /containing/ the path > of the desired target is mounted. This is not the specified > behavior. Instead, also check the targeted fs argument against > the entry found in the fstable lookup. Unless the targeted > fs is actually mounted, reject the command. > > In order to cover bind-mounts, create a new lookup function > based on the mountpoints instead of just the device name. > > Signed-off-by: Bill O'Donnell <billodo@redhat.com> > --- > > v2: in order to properly handle relative pathnames, symlinks, > and bind-mounts, use realpath to establish canonical path name. > This also requires the introduction of a lookup function based > on the target mountpoint. > > growfs/xfs_growfs.c | 10 +++++++++- > include/path.h | 1 + > libxcmd/paths.c | 21 +++++++++++++++++++++ > 3 files changed, 31 insertions(+), 1 deletion(-) > > diff --git a/growfs/xfs_growfs.c b/growfs/xfs_growfs.c > index a294e14..a61e2f1 100644 > --- a/growfs/xfs_growfs.c > +++ b/growfs/xfs_growfs.c > @@ -133,6 +133,7 @@ main(int argc, char **argv) > int spinodes; > int rmapbt_enabled; > int reflink_enabled; > + char rpath[PATH_MAX]; > > progname = basename(argv[0]); > setlocale(LC_ALL, ""); > @@ -202,7 +203,14 @@ main(int argc, char **argv) > aflag = 1; > > fs_table_initialise(0, NULL, 0, NULL); > - fs = fs_table_lookup(argv[optind], FS_MOUNT_POINT); > + > + if (!realpath(argv[optind], rpath)) { > + fprintf(stderr, _("%s: %s is not a mounted XFS filesystem\n"), > + progname, argv[optind]); > + return 1; Probably not the clearest error message; if realpath failed, we really don't know much at all about the path yet. (_("Path resolution failed for %s: %s\n"), path, strerror(errno)) or something like that would be more informative if it fails.. Otherwise, I think this looks ok. I was a little unsure about keeping the old fs_table_lookup behavior around, but given that xfs_growfs is the only utility (I think) which /explicitly/ says it requires a mount point (and can have irreversible consequences if it's pointed at the wrong thing) that's probably ok. One other suggestion would be to document the difference between the 2 lookup functions a bit more, making it more clear that one will find the fs table entry for the fs hosting the file at *path, and the other will find it only if it /is/ the mount point. I had a concern that while you're passing a realpath'd path to the new function, the mount points in the table aren't realpath'd. But I /think/ that still works; if I mount onto a symlinked dir, it's the realpath (symlink target) that ends up in /proc/mounts. Still, it might be safest to realpath the path found in the fs table as well prior to the comparison - if it's iterating over /etc/mtab instead of /proc/mounts on an older machine, I'm not sure what the paths end up looking like in that file. -Eric > + } > + > + fs = fs_table_lookup_mount(rpath); > if (!fs) { > fprintf(stderr, _("%s: %s is not a mounted XFS filesystem\n"), > progname, argv[optind]); > diff --git a/include/path.h b/include/path.h > index d077bac..1d3a902 100644 > --- a/include/path.h > +++ b/include/path.h > @@ -56,6 +56,7 @@ extern void fs_table_insert_project_path(char *__dir, uint __projid); > > > extern fs_path_t *fs_table_lookup(const char *__dir, uint __flags); > +extern fs_path_t *fs_table_lookup_mount(const char *__dir); > > typedef struct fs_cursor { > uint count; /* total count of mount entries */ > diff --git a/libxcmd/paths.c b/libxcmd/paths.c > index 816acc2..10927c6 100644 > --- a/libxcmd/paths.c > +++ b/libxcmd/paths.c > @@ -86,6 +86,27 @@ fs_table_lookup( > return NULL; > } > > +/* > + * Find the FS table entry for the given path. Compare > + * the path to mountpoint entries in the table. > + */ > +struct fs_path * > +fs_table_lookup_mount( > + const char *dir) > +{ > + uint i; > + dev_t dev = 0; > + > + if (fs_device_number(dir, &dev)) > + return NULL; > + > + for (i = 0; i < fs_count; i++) { > + if (strcmp(fs_table[i].fs_dir, dir) == 0) > + return &fs_table[i]; > + } > + return NULL; > +} > + > static int > fs_table_insert( > char *dir, > -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 4/20/17 5:22 PM, Eric Sandeen wrote: > On 4/11/17 12:22 PM, Bill O'Donnell wrote: >> +/* >> + * Find the FS table entry for the given path. Compare >> + * the path to mountpoint entries in the table. >> + */ >> +struct fs_path * >> +fs_table_lookup_mount( >> + const char *dir) >> +{ >> + uint i; >> + dev_t dev = 0; >> + >> + if (fs_device_number(dir, &dev)) >> + return NULL; >> + >> + for (i = 0; i < fs_count; i++) { >> + if (strcmp(fs_table[i].fs_dir, dir) == 0) >> + return &fs_table[i]; Oh, and I think we need to test the flags on the entry to be sure that it is a FS_MOUNT_POINT as well, and not a project path. -Eric >> + } >> + return NULL; >> +} >> + >> static int >> fs_table_insert( >> char *dir, >> > -- > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- To unsubscribe from this list: send the line "unsubscribe linux-xfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/growfs/xfs_growfs.c b/growfs/xfs_growfs.c index a294e14..a61e2f1 100644 --- a/growfs/xfs_growfs.c +++ b/growfs/xfs_growfs.c @@ -133,6 +133,7 @@ main(int argc, char **argv) int spinodes; int rmapbt_enabled; int reflink_enabled; + char rpath[PATH_MAX]; progname = basename(argv[0]); setlocale(LC_ALL, ""); @@ -202,7 +203,14 @@ main(int argc, char **argv) aflag = 1; fs_table_initialise(0, NULL, 0, NULL); - fs = fs_table_lookup(argv[optind], FS_MOUNT_POINT); + + if (!realpath(argv[optind], rpath)) { + fprintf(stderr, _("%s: %s is not a mounted XFS filesystem\n"), + progname, argv[optind]); + return 1; + } + + fs = fs_table_lookup_mount(rpath); if (!fs) { fprintf(stderr, _("%s: %s is not a mounted XFS filesystem\n"), progname, argv[optind]); diff --git a/include/path.h b/include/path.h index d077bac..1d3a902 100644 --- a/include/path.h +++ b/include/path.h @@ -56,6 +56,7 @@ extern void fs_table_insert_project_path(char *__dir, uint __projid); extern fs_path_t *fs_table_lookup(const char *__dir, uint __flags); +extern fs_path_t *fs_table_lookup_mount(const char *__dir); typedef struct fs_cursor { uint count; /* total count of mount entries */ diff --git a/libxcmd/paths.c b/libxcmd/paths.c index 816acc2..10927c6 100644 --- a/libxcmd/paths.c +++ b/libxcmd/paths.c @@ -86,6 +86,27 @@ fs_table_lookup( return NULL; } +/* + * Find the FS table entry for the given path. Compare + * the path to mountpoint entries in the table. + */ +struct fs_path * +fs_table_lookup_mount( + const char *dir) +{ + uint i; + dev_t dev = 0; + + if (fs_device_number(dir, &dev)) + return NULL; + + for (i = 0; i < fs_count; i++) { + if (strcmp(fs_table[i].fs_dir, dir) == 0) + return &fs_table[i]; + } + return NULL; +} + static int fs_table_insert( char *dir,
xfs_growfs manpage clearly states that the filesystem must be mounted to be grown. Current behavior allows xfs_growfs to proceed if the filesystem /containing/ the path of the desired target is mounted. This is not the specified behavior. Instead, also check the targeted fs argument against the entry found in the fstable lookup. Unless the targeted fs is actually mounted, reject the command. In order to cover bind-mounts, create a new lookup function based on the mountpoints instead of just the device name. Signed-off-by: Bill O'Donnell <billodo@redhat.com> --- v2: in order to properly handle relative pathnames, symlinks, and bind-mounts, use realpath to establish canonical path name. This also requires the introduction of a lookup function based on the target mountpoint. growfs/xfs_growfs.c | 10 +++++++++- include/path.h | 1 + libxcmd/paths.c | 21 +++++++++++++++++++++ 3 files changed, 31 insertions(+), 1 deletion(-)