Recently, glibc made a change to suppress sign-conversion warnings in
FD_SET (glibc commit ceb9e56b3d1). This uncovered an issue with the
kernel's definition of __NFDBITS if applications #include
<linux/types.h> after including <sys/select.h>. A build failure would
be seen when passing the -Werror=sign-compare and -D_FORTIFY_SOURCE=2
flags to gcc.
It was suggested that the kernel should either match the glibc
definition of __NFDBITS or remove that entirely. The current in-kernel
uses of __NFDBITS can be replaced with BITS_PER_LONG, and there are no
uses of the related __FDELT and __FDMASK defines. Given that, we'll
continue the cleanup that was started with commit 8b3d1cda4f
("posix_types: Remove fd_set macros") and drop the remaining unused
macros.
Additionally, linux/time.h has similar macros defined that expand to
nothing so we'll remove those at the same time.
Reported-by: Jeff Law <law@redhat.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
CC: <stable@vger.kernel.org>
Signed-off-by: Josh Boyer <jwboyer@redhat.com>
[ .. and fix up whitespace as per akpm ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull networking changes from David S Miller:
1) Remove the ipv4 routing cache. Now lookups go directly into the FIB
trie and use prebuilt routes cached there.
No more garbage collection, no more rDOS attacks on the routing
cache. Instead we now get predictable and consistent performance,
no matter what the pattern of traffic we service.
This has been almost 2 years in the making. Special thanks to
Julian Anastasov, Eric Dumazet, Steffen Klassert, and others who
have helped along the way.
I'm sure that with a change of this magnitude there will be some
kind of fallout, but such things ought the be simple to fix at this
point. Luckily I'm not European so I'll be around all of August to
fix things :-)
The major stages of this work here are each fronted by a forced
merge commit whose commit message contains a top-level description
of the motivations and implementation issues.
2) Pre-demux of established ipv4 TCP sockets, saves a route demux on
input.
3) TCP SYN/ACK performance tweaks from Eric Dumazet.
4) Add namespace support for netfilter L4 conntrack helpers, from Gao
Feng.
5) Add config mechanism for Energy Efficient Ethernet to ethtool, from
Yuval Mintz.
6) Remove quadratic behavior from /proc/net/unix, from Eric Dumazet.
7) Support for connection tracker helpers in userspace, from Pablo
Neira Ayuso.
8) Allow userspace driven TX load balancing functions in TEAM driver,
from Jiri Pirko.
9) Kill off NLMSG_PUT and RTA_PUT macros, more gross stuff with
embedded gotos.
10) TCP Small Queues, essentially minimize the amount of TCP data queued
up in the packet scheduler layer. Whereas the existing BQL (Byte
Queue Limits) limits the pkt_sched --> netdevice queuing levels,
this controls the TCP --> pkt_sched queueing levels.
From Eric Dumazet.
11) Reduce the number of get_page/put_page ops done on SKB fragments,
from Alexander Duyck.
12) Implement protection against blind resets in TCP (RFC 5961), from
Eric Dumazet.
13) Support the client side of TCP Fast Open, basically the ability to
send data in the SYN exchange, from Yuchung Cheng.
Basically, the sender queues up data with a sendmsg() call using
MSG_FASTOPEN, then they do the connect() which emits the queued up
fastopen data.
14) Avoid all the problems we get into in TCP when timers or PMTU events
hit a locked socket. The TCP Small Queues changes added a
tcp_release_cb() that allows us to queue work up to the
release_sock() caller, and that's what we use here too. From Eric
Dumazet.
15) Zero copy on TX support for TUN driver, from Michael S. Tsirkin.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1870 commits)
genetlink: define lockdep_genl_is_held() when CONFIG_LOCKDEP
r8169: revert "add byte queue limit support".
ipv4: Change rt->rt_iif encoding.
net: Make skb->skb_iif always track skb->dev
ipv4: Prepare for change of rt->rt_iif encoding.
ipv4: Remove all RTCF_DIRECTSRC handliing.
ipv4: Really ignore ICMP address requests/replies.
decnet: Don't set RTCF_DIRECTSRC.
net/ipv4/ip_vti.c: Fix __rcu warnings detected by sparse.
ipv4: Remove redundant assignment
rds: set correct msg_namelen
openvswitch: potential NULL deref in sample()
tcp: dont drop MTU reduction indications
bnx2x: Add new 57840 device IDs
tcp: avoid oops in tcp_metrics and reset tcpm_stamp
niu: Change niu_rbr_fill() to use unlikely() to check niu_rbr_add_page() return value
niu: Fix to check for dma mapping errors.
net: Fix references to out-of-scope variables in put_cmsg_compat()
net: ethernet: davinci_emac: add pm_runtime support
net: ethernet: davinci_emac: Remove unnecessary #include
...
Pull the big VFS changes from Al Viro:
"This one is *big* and changes quite a few things around VFS. What's in there:
- the first of two really major architecture changes - death to open
intents.
The former is finally there; it was very long in making, but with
Miklos getting through really hard and messy final push in
fs/namei.c, we finally have it. Unlike his variant, this one
doesn't introduce struct opendata; what we have instead is
->atomic_open() taking preallocated struct file * and passing
everything via its fields.
Instead of returning struct file *, it returns -E... on error, 0
on success and 1 in "deal with it yourself" case (e.g. symlink
found on server, etc.).
See comments before fs/namei.c:atomic_open(). That made a lot of
goodies finally possible and quite a few are in that pile:
->lookup(), ->d_revalidate() and ->create() do not get struct
nameidata * anymore; ->lookup() and ->d_revalidate() get lookup
flags instead, ->create() gets "do we want it exclusive" flag.
With the introduction of new helper (kern_path_locked()) we are rid
of all struct nameidata instances outside of fs/namei.c; it's still
visible in namei.h, but not for long. Come the next cycle,
declaration will move either to fs/internal.h or to fs/namei.c
itself. [me, miklos, hch]
- The second major change: behaviour of final fput(). Now we have
__fput() done without any locks held by caller *and* not from deep
in call stack.
That obviously lifts a lot of constraints on the locking in there.
Moreover, it's legal now to call fput() from atomic contexts (which
has immediately simplified life for aio.c). We also don't need
anti-recursion logics in __scm_destroy() anymore.
There is a price, though - the damn thing has become partially
asynchronous. For fput() from normal process we are guaranteed
that pending __fput() will be done before the caller returns to
userland, exits or gets stopped for ptrace.
For kernel threads and atomic contexts it's done via
schedule_work(), so theoretically we might need a way to make sure
it's finished; so far only one such place had been found, but there
might be more.
There's flush_delayed_fput() (do all pending __fput()) and there's
__fput_sync() (fput() analog doing __fput() immediately). I hope
we won't need them often; see warnings in fs/file_table.c for
details. [me, based on task_work series from Oleg merged last
cycle]
- sync series from Jan
- large part of "death to sync_supers()" work from Artem; the only
bits missing here are exofs and ext4 ones. As far as I understand,
those are going via the exofs and ext4 trees resp.; once they are
in, we can put ->write_super() to the rest, along with the thread
calling it.
- preparatory bits from unionmount series (from dhowells).
- assorted cleanups and fixes all over the place, as usual.
This is not the last pile for this cycle; there's at least jlayton's
ESTALE work and fsfreeze series (the latter - in dire need of fixes,
so I'm not sure it'll make the cut this cycle). I'll probably throw
symlink/hardlink restrictions stuff from Kees into the next pile, too.
Plus there's a lot of misc patches I hadn't thrown into that one -
it's large enough as it is..."
* 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (127 commits)
ext4: switch EXT4_IOC_RESIZE_FS to mnt_want_write_file()
btrfs: switch btrfs_ioctl_balance() to mnt_want_write_file()
switch dentry_open() to struct path, make it grab references itself
spufs: shift dget/mntget towards dentry_open()
zoran: don't bother with struct file * in zoran_map
ecryptfs: don't reinvent the wheels, please - use struct completion
don't expose I_NEW inodes via dentry->d_inode
tidy up namei.c a bit
unobfuscate follow_up() a bit
ext3: pass custom EOF to generic_file_llseek_size()
ext4: use core vfs llseek code for dir seeks
vfs: allow custom EOF in generic_file_llseek code
vfs: Avoid unnecessary WB_SYNC_NONE writeback during sys_sync and reorder sync passes
vfs: Remove unnecessary flushing of block devices
vfs: Make sys_sync writeout also block device inodes
vfs: Create function for iterating over block devices
vfs: Reorder operations during sys_sync
quota: Move quota syncing to ->sync_fs method
quota: Split dquot_quota_sync() to writeback and cache flushing part
vfs: Move noop_backing_dev_info check from sync into writeback
...
Pull SELinux regression fixes from James Morris.
Andrew Morton has a box that hit that open perms problem.
I also renamed the "epollwakeup" selinux name for the new capability to
be "block_suspend", to match the rename done by commit d9914cf661
("PM: Rename CAP_EPOLLWAKEUP to CAP_BLOCK_SUSPEND").
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
SELinux: do not check open perms if they are not known to policy
SELinux: include definition of new capabilities
When I introduced open perms policy didn't understand them and I
implemented them as a policycap. When I added the checking of open perm
to truncate I forgot to conditionalize it on the userspace defined
policy capability. Running an old policy with a new kernel will not
check open on open(2) but will check it on truncate. Conditionalize the
truncate check the same as the open check.
Signed-off-by: Eric Paris <eparis@redhat.com>
Cc: stable@vger.kernel.org # 3.4.x
Signed-off-by: James Morris <james.l.morris@oracle.com>
The kernel has added CAP_WAKE_ALARM and CAP_EPOLLWAKEUP. We need to
define these in SELinux so they can be mediated by policy.
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: James Morris <james.l.morris@oracle.com>
This patch adds the following structure:
struct netlink_kernel_cfg {
unsigned int groups;
void (*input)(struct sk_buff *skb);
struct mutex *cb_mutex;
};
That can be passed to netlink_kernel_create to set optional configurations
for netlink kernel sockets.
I've populated this structure by looking for NULL and zero parameters at the
existing code. The remaining parameters that always need to be set are still
left in the original interface.
That includes optional parameters for the netlink socket creation. This allows
easy extensibility of this interface in the future.
This patch also adapts all callers to use this new interface.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is a cleanup. Use NFPROTO_* for consistency with other
netfilter code.
Signed-off-by: Alban Crequy <alban.crequy@collabora.co.uk>
Reviewed-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Reviewed-by: Vincent Sanders <vincent.sanders@collabora.co.uk>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
a) %d does _not_ produce a page worth of output
b) snprintf() doesn't return negatives - it used to in old glibc, but
that's the kernel...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Pull security subsystem updates from James Morris:
"New notable features:
- The seccomp work from Will Drewry
- PR_{GET,SET}_NO_NEW_PRIVS from Andy Lutomirski
- Longer security labels for Smack from Casey Schaufler
- Additional ptrace restriction modes for Yama by Kees Cook"
Fix up trivial context conflicts in arch/x86/Kconfig and include/linux/filter.h
* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security: (65 commits)
apparmor: fix long path failure due to disconnected path
apparmor: fix profile lookup for unconfined
ima: fix filename hint to reflect script interpreter name
KEYS: Don't check for NULL key pointer in key_validate()
Smack: allow for significantly longer Smack labels v4
gfp flags for security_inode_alloc()?
Smack: recursive tramsmute
Yama: replace capable() with ns_capable()
TOMOYO: Accept manager programs which do not start with / .
KEYS: Add invalidation support
KEYS: Do LRU discard in full keyrings
KEYS: Permit in-place link replacement in keyring list
KEYS: Perform RCU synchronisation on keys prior to key destruction
KEYS: Announce key type (un)registration
KEYS: Reorganise keys Makefile
KEYS: Move the key config into security/keys/Kconfig
KEYS: Use the compat keyctl() syscall wrapper on Sparc64 for Sparc32 compat
Yama: remove an unused variable
samples/seccomp: fix dependencies on arch macros
Yama: add additional ptrace scopes
...
This patch removes ip_queue support which was marked as obsolete
years ago. The nfnetlink_queue modules provides more advanced
user-space packet queueing mechanism.
This patch also removes capability code included in SELinux that
refers to ip_queue. Otherwise, we break compilation.
Several warning has been sent regarding this to the mailing list
in the past month without anyone rising the hand to stop this
with some strong argument.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
With this change, calling
prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)
disables privilege granting operations at execve-time. For example, a
process will not be able to execute a setuid binary to change their uid
or gid if this bit is set. The same is true for file capabilities.
Additionally, LSM_UNSAFE_NO_NEW_PRIVS is defined to ensure that
LSMs respect the requested behavior.
To determine if the NO_NEW_PRIVS bit is set, a task may call
prctl(PR_GET_NO_NEW_PRIVS, 0, 0, 0, 0);
It returns 1 if set and 0 if it is not set. If any of the arguments are
non-zero, it will return -1 and set errno to -EINVAL.
(PR_SET_NO_NEW_PRIVS behaves similarly.)
This functionality is desired for the proposed seccomp filter patch
series. By using PR_SET_NO_NEW_PRIVS, it allows a task to modify the
system call behavior for itself and its child tasks without being
able to impact the behavior of a more privileged task.
Another potential use is making certain privileged operations
unprivileged. For example, chroot may be considered "safe" if it cannot
affect privileged tasks.
Note, this patch causes execve to fail when PR_SET_NO_NEW_PRIVS is
set and AppArmor is in use. It is fixed in a subsequent patch.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Will Drewry <wad@chromium.org>
Acked-by: Eric Paris <eparis@redhat.com>
Acked-by: Kees Cook <keescook@chromium.org>
v18: updated change desc
v17: using new define values as per 3.4
Signed-off-by: James Morris <james.l.morris@oracle.com>
avc_add_callback now just used for registering reset functions
in initcalls, and the callback functions just did reset operations.
So, reducing the arguments to only one event is enough now.
Signed-off-by: Wanlong Gao <gaowanlong@cn.fujitsu.com>
Signed-off-by: Eric Paris <eparis@redhat.com>
avc_add_callback now only called from initcalls, so replace the
weak GFP_ATOMIC to GFP_KERNEL, and mark this function __init
to make a warning when not been called from initcalls.
Signed-off-by: Wanlong Gao <gaowanlong@cn.fujitsu.com>
Signed-off-by: Eric Paris <eparis@redhat.com>
We no longer need the distinction. We only need data after we decide to do an
audit. So turn the "late" audit data into just "data" and remove what we
currently have as "data".
Signed-off-by: Eric Paris <eparis@redhat.com>
It isn't needed. If you don't set the type of the data associated with
that type it is a pretty obvious programming bug. So why waste the cycles?
Signed-off-by: Eric Paris <eparis@redhat.com>
There are no legitimate users. Always use current and get back some stack
space for the common_audit_data.
Signed-off-by: Eric Paris <eparis@redhat.com>
selinux_inode_has_perm is a hot path. Instead of declaring the
common_audit_data on the stack move it to a noinline function only used in
the rare case we need to send an audit message.
Signed-off-by: Eric Paris <eparis@redhat.com>
We pay a rather large overhead initializing the common_audit_data.
Since we only need this information if we actually emit an audit
message there is little need to set it up in the hot path. This patch
splits the functionality of avc_has_perm() into avc_has_perm_noaudit(),
avc_audit_required() and slow_avc_audit(). But we take care of setting
up to audit between required() and the actual audit call. Thus saving
measurable time in a hot path.
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: Eric Paris <eparis@redhat.com>
We reset the bool names and values array to NULL, but do not reset the
number of entries in these arrays to 0. If we error out and then get back
into this function we will walk these NULL pointers based on the belief
that they are non-zero length.
Signed-off-by: Eric Paris <eparis@redhat.com>
cc: stable@kernel.org
I'm not really sure what the idea behind the sel_div function is, but it's
useless. Since a and b are both unsigned, it's impossible for a % b < 0.
That means that part of the function never does anything. Thus it's just a
normal /. Just do that instead. I don't even understand what that operation
was supposed to mean in the signed case however....
If it was signed:
sel_div(-2, 4) == ((-2 / 4) - ((-2 % 4) < 0))
((0) - ((-2) < 0))
((0) - (1))
(-1)
What actually happens:
sel_div(-2, 4) == ((18446744073709551614 / 4) - ((18446744073709551614 % 4) < 0))
((4611686018427387903) - ((2 < 0))
(4611686018427387903 - 0)
((unsigned int)4611686018427387903)
(4294967295)
Neither makes a whole ton of sense to me. So I'm getting rid of the
function entirely.
Signed-off-by: Eric Paris <eparis@redhat.com>
It's possible that the caller passed a NULL for scontext. However if this
is a defered mapping we might still attempt to call *scontext=kstrdup().
This is bad. Instead just return the len.
Signed-off-by: Eric Paris <eparis@redhat.com>
We know that some yum operation is causing CAP_MAC_ADMIN failures. This
implies that an RPM is laying down (or attempting to lay down) a file with
an invalid label. The problem is that we don't have any information to
track down the cause. This patch will cause such a failure to report the
failed label in an SELINUX_ERR audit message. This is similar to the
SELINUX_ERR reports on invalid transitions and things like that. It should
help run down problems on what is trying to set invalid labels in the
future.
Resulting records look something like:
type=AVC msg=audit(1319659241.138:71): avc: denied { mac_admin } for pid=2594 comm="chcon" capability=33 scontext=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 tcontext=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 tclass=capability2
type=SELINUX_ERR msg=audit(1319659241.138:71): op=setxattr invalid_context=unconfined_u:object_r:hello:s0
type=SYSCALL msg=audit(1319659241.138:71): arch=c000003e syscall=188 success=no exit=-22 a0=a2c0e0 a1=390341b79b a2=a2d620 a3=1f items=1 ppid=2519 pid=2594 auid=0 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=pts0 ses=1 comm="chcon" exe="/usr/bin/chcon" subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 key=(null)
type=CWD msg=audit(1319659241.138:71): cwd="/root" type=PATH msg=audit(1319659241.138:71): item=0 name="test" inode=785879 dev=fc:03 mode=0100644 ouid=0 ogid=0 rdev=00:00 obj=unconfined_u:object_r:admin_home_t:s0
Signed-off-by: Eric Paris <eparis@redhat.com>
In RH BZ 578841 we realized that the SELinux sandbox program was allowed to
truncate files outside of the sandbox. The reason is because sandbox
confinement is determined almost entirely by the 'open' permission. The idea
was that if the sandbox was unable to open() files it would be unable to do
harm to those files. This turns out to be false in light of syscalls like
truncate() and chmod() which don't require a previous open() call. I looked
at the syscalls that did not have an associated 'open' check and found that
truncate(), did not have a seperate permission and even if it did have a
separate permission such a permission owuld be inadequate for use by
sandbox (since it owuld have to be granted so liberally as to be useless).
This patch checks the OPEN permission on truncate. I think a better solution
for sandbox is a whole new permission, but at least this fixes what we have
today.
Signed-off-by: Eric Paris <eparis@redhat.com>
Because Fedora shipped userspace based on my development tree we now
have policy version 27 in the wild defining only default user, role, and
range. Thus to add default_type we need a policy.28.
Signed-off-by: Eric Paris <eparis@redhat.com>
When new objects are created we have great and flexible rules to
determine the type of the new object. We aren't quite as flexible or
mature when it comes to determining the user, role, and range. This
patch adds a new ability to specify the place a new objects user, role,
and range should come from. For users and roles it can come from either
the source or the target of the operation. aka for files the user can
either come from the source (the running process and todays default) or
it can come from the target (aka the parent directory of the new file)
examples always are done with
directory context: system_u:object_r:mnt_t:s0-s0:c0.c512
process context: unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023
[no rule]
unconfined_u:object_r:mnt_t:s0 test_none
[default user source]
unconfined_u:object_r:mnt_t:s0 test_user_source
[default user target]
system_u:object_r:mnt_t:s0 test_user_target
[default role source]
unconfined_u:unconfined_r:mnt_t:s0 test_role_source
[default role target]
unconfined_u:object_r:mnt_t:s0 test_role_target
[default range source low]
unconfined_u:object_r:mnt_t:s0 test_range_source_low
[default range source high]
unconfined_u:object_r:mnt_t:s0:c0.c1023 test_range_source_high
[default range source low-high]
unconfined_u:object_r:mnt_t:s0-s0:c0.c1023 test_range_source_low-high
[default range target low]
unconfined_u:object_r:mnt_t:s0 test_range_target_low
[default range target high]
unconfined_u:object_r:mnt_t:s0:c0.c512 test_range_target_high
[default range target low-high]
unconfined_u:object_r:mnt_t:s0-s0:c0.c512 test_range_target_low-high
Signed-off-by: Eric Paris <eparis@redhat.com>
There is no reason the DAC perms on reading the policy file need to be root
only. There are selinux checks which should control this access.
Signed-off-by: Eric Paris <eparis@redhat.com>
It just bloats the audit data structure for no good reason, since the
only time those fields are filled are just before calling the
common_lsm_audit() function, which is also the only user of those
fields.
So just make them be the arguments to common_lsm_audit(), rather than
bloating that structure that is passed around everywhere, and is
initialized in hot paths.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Instead of declaring the entire selinux_audit_data on the stack when we
start an operation on declare it on the stack if we are going to use it.
We know it's usefulness at the end of the security decision and can declare
it there.
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
After shrinking the common_audit_data stack usage for private LSM data I'm
not going to shrink the data union. To do this I'm going to move anything
larger than 2 void * ptrs to it's own structure and require it to be declared
separately on the calling stack. Thus hot paths which don't need more than
a couple pointer don't have to declare space to hold large unneeded
structures. I could get this down to one void * by dealing with the key
struct and the struct path. We'll see if that is helpful after taking care of
networking.
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus found that the gigantic size of the common audit data caused a big
perf hit on something as simple as running stat() in a loop. This patch
requires LSMs to declare the LSM specific portion separately rather than
doing it in a union. Thus each LSM can be responsible for shrinking their
portion and don't have to pay a penalty just because other LSMs have a
bigger space requirement.
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull second try at vfs part d#2 from Al Viro:
"Miklos' first series (with do_lookup() rewrite split into edible
chunks) + assorted bits and pieces.
The 'untangling of do_lookup()' series is is a splitup of what used to
be a monolithic patch from Miklos, so this series is basically "how do
I convince myself that his patch is correct (or find a hole in it)".
No holes found and I like the resulting cleanup, so in it went..."
Changes from try 1: Fix a boot problem with selinux, and commit messages
prettied up a bit.
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (24 commits)
vfs: fix out-of-date dentry_unhash() comment
vfs: split __lookup_hash
untangling do_lookup() - take __lookup_hash()-calling case out of line.
untangling do_lookup() - switch to calling __lookup_hash()
untangling do_lookup() - merge d_alloc_and_lookup() callers
untangling do_lookup() - merge failure exits in !dentry case
untangling do_lookup() - massage !dentry case towards __lookup_hash()
untangling do_lookup() - get rid of need_reval in !dentry case
untangling do_lookup() - eliminate a loop.
untangling do_lookup() - expand the area under ->i_mutex
untangling do_lookup() - isolate !dentry stuff from the rest of it.
vfs: move MAY_EXEC check from __lookup_hash()
vfs: don't revalidate just looked up dentry
vfs: fix d_need_lookup/d_revalidate order in do_lookup
ext3: move headers to fs/ext3/
migrate ext2_fs.h guts to fs/ext2/ext2.h
new helper: ext2_image_size()
get rid of pointless includes of ext2_fs.h
ext2: No longer export ext2_fs.h to user space
mtdchar: kill persistently held vfsmount
...
Now that all the slow-path code is gone from these functions, we can
inline them into the main caller - avc_has_perm_flags().
Now the compiler can see that 'avc' is allocated on the stack for this
case, which helps register pressure a bit. It also actually shrinks the
total stack frame, because the stack frame that avc_has_perm_flags()
always needed (for that 'avc' allocation) is now sufficient for the
inlined functions too.
Inlining isn't bad - but mindless inlining of cold code (see the
previous commit) is.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The selinux AVC paths remain some of the hottest (and deepest) codepaths
at filename lookup time, and we make it worse by having the slow path
cases take up I$ and stack space even when they don't trigger. Gcc
tends to always want to inline functions that are just called once -
never mind that this might make for slower and worse code in the caller.
So this tries to improve on it a bit by making the slow-path cases
explicitly separate functions that are marked noinline, causing gcc to
at least no longer allocate stack space for them unless they are
actually called. It also seems to help register allocation a tiny bit,
since gcc now doesn't take the slow case code into account.
Uninlining the slow path may also allow us to inline the remaining hot
path into the one caller that actually matters: avc_has_perm_flags().
I'll have to look at that separately, but both avc_audit() and
avc_has_perm_noaudit() are now small and lean enough that inlining them
may make sense.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
