The problem that this patch aims to fix is vfsmount refcounting scalability.
We need to take a reference on the vfsmount for every successful path lookup,
which often go to the same mount point.
The fundamental difficulty is that a "simple" reference count can never be made
scalable, because any time a reference is dropped, we must check whether that
was the last reference. To do that requires communication with all other CPUs
that may have taken a reference count.
We can make refcounts more scalable in a couple of ways, involving keeping
distributed counters, and checking for the global-zero condition less
frequently.
- check the global sum once every interval (this will delay zero detection
for some interval, so it's probably a showstopper for vfsmounts).
- keep a local count and only taking the global sum when local reaches 0 (this
is difficult for vfsmounts, because we can't hold preempt off for the life of
a reference, so a counter would need to be per-thread or tied strongly to a
particular CPU which requires more locking).
- keep a local difference of increments and decrements, which allows us to sum
the total difference and hence find the refcount when summing all CPUs. Then,
keep a single integer "long" refcount for slow and long lasting references,
and only take the global sum of local counters when the long refcount is 0.
This last scheme is what I implemented here. Attached mounts and process root
and working directory references are "long" references, and everything else is
a short reference.
This allows scalable vfsmount references during path walking over mounted
subtrees and unattached (lazy umounted) mounts with processes still running
in them.
This results in one fewer atomic op in the fastpath: mntget is now just a
per-CPU inc, rather than an atomic inc; and mntput just requires a spinlock
and non-atomic decrement in the common case. However code is otherwise bigger
and heavier, so single threaded performance is basically a wash.
Signed-off-by: Nick Piggin <npiggin@kernel.dk>
Pull removal of fsnotify marks into generic_shutdown_super().
Split umount-time work into a new function - evict_inodes().
Make sure that invalidate_inodes() will be able to cope with
I_FREEING once we change locking in iput().
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
The number of inodes allocated does not need to be tied to the
addition or removal of an inode to/from a list. If we are not tied
to a list lock, we could update the counters when inodes are
initialised or destroyed, but to do that we need to convert the
counters to be per-cpu (i.e. independent of a lock). This means that
we have the freedom to change the list/locking implementation
without needing to care about the counters.
Based on a patch originally from Eric Dumazet.
[AV: cleaned up a bit, fixed build breakage on weird configs
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
fs: brlock vfsmount_lock
Use a brlock for the vfsmount lock. It must be taken for write whenever
modifying the mount hash or associated fields, and may be taken for read when
performing mount hash lookups.
A new lock is added for the mnt-id allocator, so it doesn't need to take
the heavy vfsmount write-lock.
The number of atomics should remain the same for fastpath rlock cases, though
code would be slightly slower due to per-cpu access. Scalability is not not be
much improved in common cases yet, due to other locks (ie. dcache_lock) getting
in the way. However path lookups crossing mountpoints should be one case where
scalability is improved (currently requiring the global lock).
The slowpath is slower due to use of brlock. On a 64 core, 64 socket, 32 node
Altix system (high latency to remote nodes), a simple umount microbenchmark
(mount --bind mnt mnt2 ; umount mnt2 loop 1000 times), before this patch it
took 6.8s, afterwards took 7.1s, about 5% slower.
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Nick Piggin <npiggin@kernel.dk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
tty: fix fu_list abuse
tty code abuses fu_list, which causes a bug in remount,ro handling.
If a tty device node is opened on a filesystem, then the last link to the inode
removed, the filesystem will be allowed to be remounted readonly. This is
because fs_may_remount_ro does not find the 0 link tty inode on the file sb
list (because the tty code incorrectly removed it to use for its own purpose).
This can result in a filesystem with errors after it is marked "clean".
Taking idea from Christoph's initial patch, allocate a tty private struct
at file->private_data and put our required list fields in there, linking
file and tty. This makes tty nodes behave the same way as other device nodes
and avoid meddling with the vfs, and avoids this bug.
The error handling is not trivial in the tty code, so for this bugfix, I take
the simple approach of using __GFP_NOFAIL and don't worry about memory errors.
This is not a problem because our allocator doesn't fail small allocs as a rule
anyway. So proper error handling is left as an exercise for tty hackers.
[ Arguably filesystem's device inode would ideally be divorced from the
driver's pseudo inode when it is opened, but in practice it's not clear whether
that will ever be worth implementing. ]
Cc: linux-kernel@vger.kernel.org
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Nick Piggin <npiggin@kernel.dk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Just set f_flags when shoving struct file into nameidata; don't
postpone that until __dentry_open(). do_filp_open() has correct
value; lookup_instantiate_filp() doesn't - we lose the difference
between O_RDWR and 3 by that point.
We still set .intent.open.flags, so no fs code needs to be changed.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
All users outside of fs/ of get_empty_filp() have been removed. This patch
moves the definition from the include/ directory to internal.h so no new
users crop up and removes the EXPORT_SYMBOL. I'd love to see open intents
stop using it too, but that's a problem for another day and a smarter
developer!
Signed-off-by: Eric Paris <eparis@redhat.com>
Acked-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
sys_mount() reads/copies a whole page for its "type" parameter. When
do_mount_root() passes a kernel address that points to an object which is
smaller than a whole page, copy_mount_options() will happily go past this
memory object, possibly dereferencing "wild" pointers that could be in any
state (hence the kmemcheck warning, which shows that parts of the next
page are not even allocated).
(The likelihood of something going wrong here is pretty low -- first of
all this only applies to kernel calls to sys_mount(), which are mostly
found in the boot code. Secondly, I guess if the page was not mapped,
exact_copy_from_user() _would_ in fact handle it correctly because of its
access_ok(), etc. checks.)
But it is much nicer to avoid the dubious reads altogether, by stopping as
soon as we find a NUL byte. Is there a good reason why we can't do
something like this, using the already existing strndup_from_user()?
[akpm@linux-foundation.org: make copy_mount_string() static]
[AV: fix compat mount breakage, which involves undoing akpm's change above]
Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Vegard Nossum <vegard.nossum@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: al <al@dizzy.pdmi.ras.ru>
It is unnecessarily fragile to have two places (fsync_super() and do_sync())
doing data integrity sync of the filesystem. Alter __fsync_super() to
accommodate needs of both callers and use it. So after this patch
__fsync_super() is the only place where we gather all the calls needed to
properly send all data on a filesystem to disk.
Nice bonus is that we get a complete livelock avoidance and write_supers()
is now only used for periodic writeback of superblocks.
sync_blockdevs() introduced a couple of patches ago is gone now.
[build fixes folded]
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
So far, do_sync() called:
sync_inodes(0);
sync_supers();
sync_filesystems(0);
sync_filesystems(1);
sync_inodes(1);
This ordering makes it kind of hard for filesystems as sync_inodes(0) need not
submit all the IO (for example it skips inodes with I_SYNC set) so e.g. forcing
transaction to disk in ->sync_fs() is not really enough. Therefore sys_sync has
not been completely reliable on some filesystems (ext3, ext4, reiserfs, ocfs2
and others are hit by this) when racing e.g. with background writeback. A
similar problem hits also other filesystems (e.g. ext2) because of
write_supers() being called before the sync_inodes(1).
Change the ordering of calls in do_sync() - this requires a new function
sync_blockdevs() to preserve the property that block devices are always synced
after write_super() / sync_fs() call.
The same issue is fixed in __fsync_super() function used on umount /
remount read-only.
[AV: build fixes]
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
This function walks the s_files lock, and operates primarily on the
files in a superblock, so it better belongs here (eg. see also
fs_may_remount_ro).
[AV: ... and it shouldn't be static after that move]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* all changes of current->fs are done under task_lock and write_lock of
old fs->lock
* refcount is not atomic anymore (same protection)
* its decrements are done when removing reference from current; at the
same time we decide whether to free it.
* put_fs_struct() is gone
* new field - ->in_exec. Set by check_unsafe_exec() if we are trying to do
execve() and only subthreads share fs_struct. Cleared when finishing exec
(success and failure alike). Makes CLONE_FS fail with -EAGAIN if set.
* check_unsafe_exec() may fail with -EAGAIN if another execve() from subthread
is in progress.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Pure code move; two new helper functions for nfsd and daemonize
(unshare_fs_struct() and daemonize_fs_struct() resp.; for now -
the same code as used to be in callers). unshare_fs_struct()
exported (for nfsd, as copy_fs_struct()/exit_fs() used to be),
copy_fs_struct() and exit_fs() don't need exports anymore.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Joe Malicki reports that setuid sometimes doesn't: very rarely,
a setuid root program does not get root euid; and, by the way,
they have a health check running lsof every few minutes.
Right, check_unsafe_exec() notes whether the files_struct is being
shared by more threads than will get killed by the exec, and if so
sets LSM_UNSAFE_SHARE to make bprm_set_creds() careful about euid.
But /proc/<pid>/fd and /proc/<pid>/fdinfo lookups make transient
use of get_files_struct(), which also raises that sharing count.
There's a rather simple fix for this: exec's check on files->count
has been redundant ever since 2.6.1 made it unshare_files() (except
while compat_do_execve() omitted to do so) - just remove that check.
[Note to -stable: this patch will not apply before 2.6.29: earlier
releases should just remove the files->count line from unsafe_exec().]
Reported-by: Joe Malicki <jmalicki@metacarta.com>
Narrowed-down-by: Michael Itz <mitz@metacarta.com>
Tested-by: Joe Malicki <jmalicki@metacarta.com>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The patch:
commit a6f76f23d2
CRED: Make execve() take advantage of copy-on-write credentials
moved the place in which the 'safeness' of a SUID/SGID exec was performed to
before de_thread() was called. This means that LSM_UNSAFE_SHARE is now
calculated incorrectly. This flag is set if any of the usage counts for
fs_struct, files_struct and sighand_struct are greater than 1 at the time the
determination is made. All of which are true for threads created by the
pthread library.
However, since we wish to make the security calculation before irrevocably
damaging the process so that we can return it an error code in the case where
we decide we want to reject the exec request on this basis, we have to make the
determination before calling de_thread().
So, instead, we count up the number of threads (CLONE_THREAD) that are sharing
our fs_struct (CLONE_FS), files_struct (CLONE_FILES) and sighand_structs
(CLONE_SIGHAND/CLONE_THREAD) with us. These will be killed by de_thread() and
so can be discounted by check_unsafe_exec().
We do have to be careful because CLONE_THREAD does not imply FS or FILES.
We _assume_ that there will be no extra references to these structs held by the
threads we're going to kill.
This can be tested with the attached pair of programs. Build the two programs
using the Makefile supplied, and run ./test1 as a non-root user. If
successful, you should see something like:
[dhowells@andromeda tmp]$ ./test1
--TEST1--
uid=4043, euid=4043 suid=4043
exec ./test2
--TEST2--
uid=4043, euid=0 suid=0
SUCCESS - Correct effective user ID
and if unsuccessful, something like:
[dhowells@andromeda tmp]$ ./test1
--TEST1--
uid=4043, euid=4043 suid=4043
exec ./test2
--TEST2--
uid=4043, euid=4043 suid=4043
ERROR - Incorrect effective user ID!
The non-root user ID you see will depend on the user you run as.
[test1.c]
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <pthread.h>
static void *thread_func(void *arg)
{
while (1) {}
}
int main(int argc, char **argv)
{
pthread_t tid;
uid_t uid, euid, suid;
printf("--TEST1--\n");
getresuid(&uid, &euid, &suid);
printf("uid=%d, euid=%d suid=%d\n", uid, euid, suid);
if (pthread_create(&tid, NULL, thread_func, NULL) < 0) {
perror("pthread_create");
exit(1);
}
printf("exec ./test2\n");
execlp("./test2", "test2", NULL);
perror("./test2");
_exit(1);
}
[test2.c]
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
int main(int argc, char **argv)
{
uid_t uid, euid, suid;
getresuid(&uid, &euid, &suid);
printf("--TEST2--\n");
printf("uid=%d, euid=%d suid=%d\n", uid, euid, suid);
if (euid != 0) {
fprintf(stderr, "ERROR - Incorrect effective user ID!\n");
exit(1);
}
printf("SUCCESS - Correct effective user ID\n");
exit(0);
}
[Makefile]
CFLAGS = -D_GNU_SOURCE -Wall -Werror -Wunused
all: test1 test2
test1: test1.c
gcc $(CFLAGS) -o test1 test1.c -lpthread
test2: test2.c
gcc $(CFLAGS) -o test2 test2.c
sudo chown root.root test2
sudo chmod +s test2
Reported-by: David Smith <dsmith@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: David Smith <dsmith@redhat.com>
Signed-off-by: James Morris <jmorris@namei.org>
Make execve() take advantage of copy-on-write credentials, allowing it to set
up the credentials in advance, and then commit the whole lot after the point
of no return.
This patch and the preceding patches have been tested with the LTP SELinux
testsuite.
This patch makes several logical sets of alteration:
(1) execve().
The credential bits from struct linux_binprm are, for the most part,
replaced with a single credentials pointer (bprm->cred). This means that
all the creds can be calculated in advance and then applied at the point
of no return with no possibility of failure.
I would like to replace bprm->cap_effective with:
cap_isclear(bprm->cap_effective)
but this seems impossible due to special behaviour for processes of pid 1
(they always retain their parent's capability masks where normally they'd
be changed - see cap_bprm_set_creds()).
The following sequence of events now happens:
(a) At the start of do_execve, the current task's cred_exec_mutex is
locked to prevent PTRACE_ATTACH from obsoleting the calculation of
creds that we make.
(a) prepare_exec_creds() is then called to make a copy of the current
task's credentials and prepare it. This copy is then assigned to
bprm->cred.
This renders security_bprm_alloc() and security_bprm_free()
unnecessary, and so they've been removed.
(b) The determination of unsafe execution is now performed immediately
after (a) rather than later on in the code. The result is stored in
bprm->unsafe for future reference.
(c) prepare_binprm() is called, possibly multiple times.
(i) This applies the result of set[ug]id binaries to the new creds
attached to bprm->cred. Personality bit clearance is recorded,
but now deferred on the basis that the exec procedure may yet
fail.
(ii) This then calls the new security_bprm_set_creds(). This should
calculate the new LSM and capability credentials into *bprm->cred.
This folds together security_bprm_set() and parts of
security_bprm_apply_creds() (these two have been removed).
Anything that might fail must be done at this point.
(iii) bprm->cred_prepared is set to 1.
bprm->cred_prepared is 0 on the first pass of the security
calculations, and 1 on all subsequent passes. This allows SELinux
in (ii) to base its calculations only on the initial script and
not on the interpreter.
(d) flush_old_exec() is called to commit the task to execution. This
performs the following steps with regard to credentials:
(i) Clear pdeath_signal and set dumpable on certain circumstances that
may not be covered by commit_creds().
(ii) Clear any bits in current->personality that were deferred from
(c.i).
(e) install_exec_creds() [compute_creds() as was] is called to install the
new credentials. This performs the following steps with regard to
credentials:
(i) Calls security_bprm_committing_creds() to apply any security
requirements, such as flushing unauthorised files in SELinux, that
must be done before the credentials are changed.
This is made up of bits of security_bprm_apply_creds() and
security_bprm_post_apply_creds(), both of which have been removed.
This function is not allowed to fail; anything that might fail
must have been done in (c.ii).
(ii) Calls commit_creds() to apply the new credentials in a single
assignment (more or less). Possibly pdeath_signal and dumpable
should be part of struct creds.
(iii) Unlocks the task's cred_replace_mutex, thus allowing
PTRACE_ATTACH to take place.
(iv) Clears The bprm->cred pointer as the credentials it was holding
are now immutable.
(v) Calls security_bprm_committed_creds() to apply any security
alterations that must be done after the creds have been changed.
SELinux uses this to flush signals and signal handlers.
(f) If an error occurs before (d.i), bprm_free() will call abort_creds()
to destroy the proposed new credentials and will then unlock
cred_replace_mutex. No changes to the credentials will have been
made.
(2) LSM interface.
A number of functions have been changed, added or removed:
(*) security_bprm_alloc(), ->bprm_alloc_security()
(*) security_bprm_free(), ->bprm_free_security()
Removed in favour of preparing new credentials and modifying those.
(*) security_bprm_apply_creds(), ->bprm_apply_creds()
(*) security_bprm_post_apply_creds(), ->bprm_post_apply_creds()
Removed; split between security_bprm_set_creds(),
security_bprm_committing_creds() and security_bprm_committed_creds().
(*) security_bprm_set(), ->bprm_set_security()
Removed; folded into security_bprm_set_creds().
(*) security_bprm_set_creds(), ->bprm_set_creds()
New. The new credentials in bprm->creds should be checked and set up
as appropriate. bprm->cred_prepared is 0 on the first call, 1 on the
second and subsequent calls.
(*) security_bprm_committing_creds(), ->bprm_committing_creds()
(*) security_bprm_committed_creds(), ->bprm_committed_creds()
New. Apply the security effects of the new credentials. This
includes closing unauthorised files in SELinux. This function may not
fail. When the former is called, the creds haven't yet been applied
to the process; when the latter is called, they have.
The former may access bprm->cred, the latter may not.
(3) SELinux.
SELinux has a number of changes, in addition to those to support the LSM
interface changes mentioned above:
(a) The bprm_security_struct struct has been removed in favour of using
the credentials-under-construction approach.
(c) flush_unauthorized_files() now takes a cred pointer and passes it on
to inode_has_perm(), file_has_perm() and dentry_open().
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: James Morris <jmorris@namei.org>
Acked-by: Serge Hallyn <serue@us.ibm.com>
Signed-off-by: James Morris <jmorris@namei.org>
Merge all compat ioctl handling into compat_ioctl.c instead of splitting it
over compat.c and compat_ioctl.c. This also allows to get rid of ioctl32.h
Signed-off-by: Christoph Hellwig <hch@lst.de>
Looks-good-to: Andi Kleen <ak@suse.de>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Make it possible to disable the block layer. Not all embedded devices require
it, some can make do with just JFFS2, NFS, ramfs, etc - none of which require
the block layer to be present.
This patch does the following:
(*) Introduces CONFIG_BLOCK to disable the block layer, buffering and blockdev
support.
(*) Adds dependencies on CONFIG_BLOCK to any configuration item that controls
an item that uses the block layer. This includes:
(*) Block I/O tracing.
(*) Disk partition code.
(*) All filesystems that are block based, eg: Ext3, ReiserFS, ISOFS.
(*) The SCSI layer. As far as I can tell, even SCSI chardevs use the
block layer to do scheduling. Some drivers that use SCSI facilities -
such as USB storage - end up disabled indirectly from this.
(*) Various block-based device drivers, such as IDE and the old CDROM
drivers.
(*) MTD blockdev handling and FTL.
(*) JFFS - which uses set_bdev_super(), something it could avoid doing by
taking a leaf out of JFFS2's book.
(*) Makes most of the contents of linux/blkdev.h, linux/buffer_head.h and
linux/elevator.h contingent on CONFIG_BLOCK being set. sector_div() is,
however, still used in places, and so is still available.
(*) Also made contingent are the contents of linux/mpage.h, linux/genhd.h and
parts of linux/fs.h.
(*) Makes a number of files in fs/ contingent on CONFIG_BLOCK.
(*) Makes mm/bounce.c (bounce buffering) contingent on CONFIG_BLOCK.
(*) set_page_dirty() doesn't call __set_page_dirty_buffers() if CONFIG_BLOCK
is not enabled.
(*) fs/no-block.c is created to hold out-of-line stubs and things that are
required when CONFIG_BLOCK is not set:
(*) Default blockdev file operations (to give error ENODEV on opening).
(*) Makes some /proc changes:
(*) /proc/devices does not list any blockdevs.
(*) /proc/diskstats and /proc/partitions are contingent on CONFIG_BLOCK.
(*) Makes some compat ioctl handling contingent on CONFIG_BLOCK.
(*) If CONFIG_BLOCK is not defined, makes sys_quotactl() return -ENODEV if
given command other than Q_SYNC or if a special device is specified.
(*) In init/do_mounts.c, no reference is made to the blockdev routines if
CONFIG_BLOCK is not defined. This does not prohibit NFS roots or JFFS2.
(*) The bdflush, ioprio_set and ioprio_get syscalls can now be absent (return
error ENOSYS by way of cond_syscall if so).
(*) The seclvl_bd_claim() and seclvl_bd_release() security calls do nothing if
CONFIG_BLOCK is not set, since they can't then happen.
Signed-Off-By: David Howells <dhowells@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Move blockdev_superblock extern declaration from fs/fs-writeback.c to a
headerfile and remove the dependence on it by wrapping it in a macro.
Signed-Off-By: David Howells <dhowells@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Create a new header file, fs/internal.h, for common definitions local to the
sources in the fs/ directory.
Move extern definitions that should be in header files from fs/*.c to
fs/internal.h or other main header files where they span directories.
Signed-Off-By: David Howells <dhowells@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
