locks: require that flock->l_pid be set to 0 for file-private locks
Subsystems (2)
PATCH 1/1 locks: require that flock->l_pid be set to 0 for file-private locks
Neil Brown suggested potentially overloading the l_pid value as a "lock context" field for file-private locks. While I don't think we will probably want to do that here, it's probably a good idea to ensure that in the future we could extend this API without breaking existing callers.
Typically the l_pid value is ignored for incoming struct flock arguments, serving mainly as a place to return the pid of the owner if there is a conflicting lock. For file-private locks, require that it currently be set to 0 and return EINVAL if it isn't. If we eventually want to make a non-zero l_pid mean something, then this will help ensure that we don't break legacy programs that are using file-private locks.
--- fs/locks.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+)
fs/locks.c +24 -0
@@ -1931,6 +1931,10 @@ int fcntl_getlk(struct file *filp, unsigned int cmd, struct flock __user *l) goto out;if (cmd ==F_GETLKP ) { + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_GETLK ; file_lock.fl_flags |=FL_FILE_PVT ; file_lock.fl_owner = (fl_owner_t )filp;
@@ -2062,11 +2066,19 @@ again:*/ switch (cmd) {case F_SETLKP : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLK ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;break ;case F_SETLKPW : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLKW ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;
@@ -2121,6 +2133,10 @@ int fcntl_getlk64(struct file *filp, unsigned int cmd, struct flock64 __user *l) goto out;if (cmd ==F_GETLKP ) { + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_GETLK64 ; file_lock.fl_flags |=FL_FILE_PVT ; file_lock.fl_owner = (fl_owner_t )filp;
@@ -2185,11 +2201,19 @@ again:*/ switch (cmd) {case F_SETLKP : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLK64 ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;break ;case F_SETLKPW : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLKW64 ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;
Signature · 1 line
1.8.5.3
Quoted 6 lines · Jeff Layton
Neil Brown suggested potentially overloading the
l_pidvalue as a "lock context" field for file-private locks. While I don't think we will probably want to do that here, it's probably a good idea to ensure that in the future we could extend this API without breaking existing callers.
Quoted 6 lines
Typically the
l_pidvalue is ignored for incoming struct flock arguments, serving mainly as a place to return the pid of the owner if there is a conflicting lock. For file-private locks, require that it currently be set to 0 and return EINVAL if it isn't. If we eventually want to make a non-zerol_pidmean something, then this will help ensure that we don't break legacy programs that are using file-private locks.
Makes sense to me.
But, could you add move most of this initialization to a helper function? Or, better, just add it to flock64_to_posix_lock? That would
a) remove some code duplication, and b) give a single place to add a comment documenting the above rationale for the new check.
--b.
Quoted 76 lines
Cc Neil Brown <[email protected]>Signed-off-by Jeff Layton <[email protected]>--- fs/locks.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+)fs/locks.c +24
-0@@ -1931,6 +1931,10 @@ int fcntl_getlk(struct file *filp, unsigned int cmd, struct flock __user *l) goto out;if (cmd ==F_GETLKP ) { + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_GETLK ; file_lock.fl_flags |=FL_FILE_PVT ; file_lock.fl_owner = (fl_owner_t )filp;@@ -2062,11 +2066,19 @@ again:*/ switch (cmd) {case F_SETLKP : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLK ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;break ;case F_SETLKPW : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLKW ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;@@ -2121,6 +2133,10 @@ int fcntl_getlk64(struct file *filp, unsigned int cmd, struct flock64 __user *l) goto out;if (cmd ==F_GETLKP ) { + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_GETLK64 ; file_lock.fl_flags |=FL_FILE_PVT ; file_lock.fl_owner = (fl_owner_t )filp;@@ -2185,11 +2201,19 @@ again:*/ switch (cmd) {case F_SETLKP : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLK64 ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;break ;case F_SETLKPW : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLKW64 ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;Signature · 1 line
1.8.5.3
Quoted 14 lines · On Tue, 4 Mar 2014 14:22:12 -0500 "J. Bruce Fields" <[email protected]>…
Quoted 6 lines · Jeff Layton
Neil Brown suggested potentially overloading the
l_pidvalue as a "lock context" field for file-private locks. While I don't think we will probably want to do that here, it's probably a good idea to ensure that in the future we could extend this API without breaking existing callers.Quoted 6 lines
Typically the
l_pidvalue is ignored for incoming struct flock arguments, serving mainly as a place to return the pid of the owner if there is a conflicting lock. For file-private locks, require that it currently be set to 0 and return EINVAL if it isn't. If we eventually want to make a non-zerol_pidmean something, then this will help ensure that we don't break legacy programs that are using file-private locks.
Quoted 9 lines
Makes sense to me.
But, could you add move most of this initialization to a helper function? Or, better, just add it to
flock64_to_posix_lock? That would
a) remove some code duplication, and b) give a single place to add a comment documenting the above rationale for the new check.--b.
Well we could consolidate it, but I don't think it'll make the patch any smaller. We could add this to flock_to_posix_lock but you'd need to know whether this is for a classic or file-private lock. We can't reasonably enforce the l_pid == 0 requirement in the classic case.
So, we could do that but we'd have to start passing flock_to_posix_lock the cmd value or a flag or something...
Quoted 76 lines
Quoted 76 lines
Cc Neil Brown <[email protected]>Signed-off-by Jeff Layton <[email protected]>--- fs/locks.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+)fs/locks.c +24
-0@@ -1931,6 +1931,10 @@ int fcntl_getlk(struct file *filp, unsigned int cmd, struct flock __user *l) goto out;if (cmd ==F_GETLKP ) { + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_GETLK ; file_lock.fl_flags |=FL_FILE_PVT ; file_lock.fl_owner = (fl_owner_t )filp;@@ -2062,11 +2066,19 @@ again:*/ switch (cmd) {case F_SETLKP : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLK ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;break ;case F_SETLKPW : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLKW ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;@@ -2121,6 +2133,10 @@ int fcntl_getlk64(struct file *filp, unsigned int cmd, struct flock64 __user *l) goto out;if (cmd ==F_GETLKP ) { + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_GETLK64 ; file_lock.fl_flags |=FL_FILE_PVT ; file_lock.fl_owner = (fl_owner_t )filp;@@ -2185,11 +2201,19 @@ again:*/ switch (cmd) {case F_SETLKP : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLK64 ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;break ;case F_SETLKPW : + error = -EINVAL ; +if (flock.l_pid !=0 ) + goto out; + cmd =F_SETLKW64 ; file_lock->fl_flags |=FL_FILE_PVT ; file_lock->fl_owner = (fl_owner_t )filp;Signature · 1 line
1.8.5.3
Signature · 1 line
Jeff Layton <[email protected]>