linux-stable.git/fs/proc, branch v3.16.45

mm: larger stack guard gap, between vmas

2017-07-02T16:13:02+00:00

commit 1be7107fbe18eed3e319a6c3e83c78254b693acb upstream.

Stack guard page is a useful feature to reduce a risk of stack smashing
into a different mapping. We have been using a single page gap which
is sufficient to prevent having stack adjacent to a different mapping.
But this seems to be insufficient in the light of the stack usage in
userspace. E.g. glibc uses as large as 64kB alloca() in many commonly
used functions. Others use constructs liks gid_t buffer[NGROUPS_MAX]
which is 256kB or stack strings with MAX_ARG_STRLEN.

This will become especially dangerous for suid binaries and the default
no limit for the stack size limit because those applications can be
tricked to consume a large portion of the stack and a single glibc call
could jump over the guard page. These attacks are not theoretical,
unfortunatelly.

Make those attacks less probable by increasing the stack guard gap
to 1MB (on systems with 4k pages; but make it depend on the page size
because systems with larger base pages might cap stack allocations in
the PAGE_SIZE units) which should cover larger alloca() and VLA stack
allocations. It is obviously not a full fix because the problem is
somehow inherent, but it should reduce attack space a lot.

One could argue that the gap size should be configurable from userspace,
but that can be done later when somebody finds that the new 1MB is wrong
for some special case applications.  For now, add a kernel command line
option (stack_guard_gap) to specify the stack gap size (in page units).

Implementation wise, first delete all the old code for stack guard page:
because although we could get away with accounting one extra page in a
stack vma, accounting a larger gap can break userspace - case in point,
a program run with "ulimit -S -v 20000" failed when the 1MB gap was
counted for RLIMIT_AS; similar problems could come with RLIMIT_MLOCK
and strict non-overcommit mode.

Instead of keeping gap inside the stack vma, maintain the stack guard
gap as a gap between vmas: using vm_start_gap() in place of vm_start
(or vm_end_gap() in place of vm_end if VM_GROWSUP) in just those few
places which need to respect the gap - mainly arch_get_unmapped_area(),
and and the vma tree's subtree_gap support for that.

Original-patch-by: Oleg Nesterov 
Original-patch-by: Michal Hocko 
Signed-off-by: Hugh Dickins 
Acked-by: Michal Hocko 
Tested-by: Helge Deller  # parisc
Signed-off-by: Linus Torvalds 
[Hugh Dickins: Backported to 3.16]
Signed-off-by: Ben Hutchings

vfs: Commit to never having exectuables on proc and sysfs.

2017-04-04T21:21:56+00:00

commit 22f6b4d34fcf039c63a94e7670e0da24f8575a5a upstream.

Today proc and sysfs do not contain any executable files.  Several
applications today mount proc or sysfs without noexec and nosuid and
then depend on there being no exectuables files on proc or sysfs.
Having any executable files show on proc or sysfs would cause
a user space visible regression, and most likely security problems.

Therefore commit to never allowing executables on proc and sysfs by
adding a new flag to mark them as filesystems without executables and
enforce that flag.

Test the flag where MNT_NOEXEC is tested today, so that the only user
visible effect will be that exectuables will be treated as if the
execute bit is cleared.

The filesystems proc and sysfs do not currently incoporate any
executable files so this does not result in any user visible effects.

This makes it unnecessary to vet changes to proc and sysfs tightly for
adding exectuable files or changes to chattr that would modify
existing files, as no matter what the individual file say they will
not be treated as exectuable files by the vfs.

Not having to vet changes to closely is important as without this we
are only one proc_create call (or another goof up in the
implementation of notify_change) from having problematic executables
on proc.  Those mistakes are all too easy to make and would create
a situation where there are security issues or the assumptions of
some program having to be broken (and cause userspace regressions).

Signed-off-by: "Eric W. Biederman" 
[bwh: Backported to 3.16: we don't have super_block::s_iflags; use
 file_system_type::fs_flags instead]
Signed-off-by: Ben Hutchings

sysctl: Drop reference added by grab_header in proc_sys_readdir

2017-02-23T03:54:47+00:00

commit 93362fa47fe98b62e4a34ab408c4a418432e7939 upstream.

Fixes CVE-2016-9191, proc_sys_readdir doesn't drop reference
added by grab_header when return from !dir_emit_dots path.
It can cause any path called unregister_sysctl_table will
wait forever.

The calltrace of CVE-2016-9191:

[ 5535.960522] Call Trace:
[ 5535.963265]  [] schedule+0x3f/0xa0
[ 5535.968817]  [] schedule_timeout+0x3db/0x6f0
[ 5535.975346]  [] ? wait_for_completion+0x45/0x130
[ 5535.982256]  [] wait_for_completion+0xc3/0x130
[ 5535.988972]  [] ? wake_up_q+0x80/0x80
[ 5535.994804]  [] drop_sysctl_table+0xc4/0xe0
[ 5536.001227]  [] drop_sysctl_table+0x77/0xe0
[ 5536.007648]  [] unregister_sysctl_table+0x4d/0xa0
[ 5536.014654]  [] unregister_sysctl_table+0x7f/0xa0
[ 5536.021657]  [] unregister_sched_domain_sysctl+0x15/0x40
[ 5536.029344]  [] partition_sched_domains+0x44/0x450
[ 5536.036447]  [] ? __mutex_unlock_slowpath+0x111/0x1f0
[ 5536.043844]  [] rebuild_sched_domains_locked+0x64/0xb0
[ 5536.051336]  [] update_flag+0x11d/0x210
[ 5536.057373]  [] ? mutex_lock_nested+0x2df/0x450
[ 5536.064186]  [] ? cpuset_css_offline+0x1b/0x60
[ 5536.070899]  [] ? trace_hardirqs_on+0xd/0x10
[ 5536.077420]  [] ? mutex_lock_nested+0x2df/0x450
[ 5536.084234]  [] ? css_killed_work_fn+0x25/0x220
[ 5536.091049]  [] cpuset_css_offline+0x35/0x60
[ 5536.097571]  [] css_killed_work_fn+0x5c/0x220
[ 5536.104207]  [] process_one_work+0x1df/0x710
[ 5536.110736]  [] ? process_one_work+0x160/0x710
[ 5536.117461]  [] worker_thread+0x12b/0x4a0
[ 5536.123697]  [] ? process_one_work+0x710/0x710
[ 5536.130426]  [] kthread+0xfe/0x120
[ 5536.135991]  [] ret_from_fork+0x1f/0x40
[ 5536.142041]  [] ? kthread_create_on_node+0x230/0x230

One cgroup maintainer mentioned that "cgroup is trying to offline
a cpuset css, which takes place under cgroup_mutex.  The offlining
ends up trying to drain active usages of a sysctl table which apprently
is not happening."
The real reason is that proc_sys_readdir doesn't drop reference added
by grab_header when return from !dir_emit_dots path. So this cpuset
offline path will wait here forever.

See here for details: http://www.openwall.com/lists/oss-security/2016/11/04/13

Fixes: f0c3b5093add ("[readdir] convert procfs")
Reported-by: CAI Qian 
Tested-by: Yang Shukui 
Signed-off-by: Zhou Chengming 
Acked-by: Al Viro 
Signed-off-by: Eric W. Biederman 
Signed-off-by: Ben Hutchings

fs: Give dentry to inode_change_ok() instead of inode

2017-02-23T03:53:52+00:00

commit 31051c85b5e2aaaf6315f74c72a732673632a905 upstream.

inode_change_ok() will be resposible for clearing capabilities and IMA
extended attributes and as such will need dentry. Give it as an argument
to inode_change_ok() instead of an inode. Also rename inode_change_ok()
to setattr_prepare() to better relect that it does also some
modifications in addition to checks.

Reviewed-by: Christoph Hellwig 
Signed-off-by: Jan Kara 
[bwh: Backported to 3.16:
 - Drop changes to orangefs, overlayfs
 - Adjust filenames, context
 - In nfsd, pass dentry to nfsd_sanitize_attrs()
 - Update ext3 as well]
Signed-off-by: Ben Hutchings

Revert "fs: Give dentry to inode_change_ok() instead of inode"

2017-02-23T03:53:52+00:00

This reverts commit be9df699432235753c3824b0f5a27d46de7fdc9e, which was
commit 31051c85b5e2aaaf6315f74c72a732673632a905 upstream.  The backport
breaks fuse and makes a mess of xfs, which can be improved by picking
further upstream commits as I should have done in the first place.

Signed-off-by: Ben Hutchings

fs: Give dentry to inode_change_ok() instead of inode

2016-11-20T01:17:38+00:00

commit 31051c85b5e2aaaf6315f74c72a732673632a905 upstream.

inode_change_ok() will be resposible for clearing capabilities and IMA
extended attributes and as such will need dentry. Give it as an argument
to inode_change_ok() instead of an inode. Also rename inode_change_ok()
to setattr_prepare() to better relect that it does also some
modifications in addition to checks.

Reviewed-by: Christoph Hellwig 
Signed-off-by: Jan Kara 
[bwh: Backported to 3.16:
 - Drop changes to orangefs, overlayfs
 - Adjust filenames, context
 - In fuse, pass dentry to fuse_do_setattr()
 - In nfsd, pass dentry to nfsd_sanitize_attrs()
 - In xfs, pass dentry to xfs_setattr_nonsize() and xfs_setattr_size()
 - Update ext3 as well]
Signed-off-by: Ben Hutchings

proc: prevent stacking filesystems on top

2016-08-22T21:38:27+00:00

commit e54ad7f1ee263ffa5a2de9c609d58dfa27b21cd9 upstream.

This prevents stacking filesystems (ecryptfs and overlayfs) from using
procfs as lower filesystem.  There is too much magic going on inside
procfs, and there is no good reason to stack stuff on top of procfs.

(For example, procfs does access checks in VFS open handlers, and
ecryptfs by design calls open handlers from a kernel thread that doesn't
drop privileges or so.)

Signed-off-by: Jann Horn 
Signed-off-by: Linus Torvalds 
Signed-off-by: Ben Hutchings

proc: prevent accessing /proc//environ until it's ready

2016-06-15T20:29:31+00:00

commit 8148a73c9901a8794a50f950083c00ccf97d43b3 upstream.

If /proc//environ gets read before the envp[] array is fully set up
in create_{aout,elf,elf_fdpic,flat}_tables(), we might end up trying to
read more bytes than are actually written, as env_start will already be
set but env_end will still be zero, making the range calculation
underflow, allowing to read beyond the end of what has been written.

Fix this as it is done for /proc//cmdline by testing env_end for
zero.  It is, apparently, intentionally set last in create_*_tables().

This bug was found by the PaX size_overflow plugin that detected the
arithmetic underflow of 'this_len = env_end - (env_start + src)' when
env_end is still zero.

The expected consequence is that userland trying to access
/proc//environ of a not yet fully set up process may get
inconsistent data as we're in the middle of copying in the environment
variables.

Fixes: https://forums.grsecurity.net/viewtopic.php?f=3&t=4363
Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=116461
Signed-off-by: Mathias Krause 
Cc: Emese Revfy 
Cc: Pax Team 
Cc: Al Viro 
Cc: Mateusz Guzik 
Cc: Alexey Dobriyan 
Cc: Cyrill Gorcunov 
Cc: Jarod Wilson 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Ben Hutchings

fs/proc, core/debug: Don't expose absolute kernel addresses via wchan

2015-12-13T17:49:51+00:00

commit b2f73922d119686323f14fbbe46587f863852328 upstream.

So the /proc/PID/stat 'wchan' field (the 30th field, which contains
the absolute kernel address of the kernel function a task is blocked in)
leaks absolute kernel addresses to unprivileged user-space:

        seq_put_decimal_ull(m, ' ', wchan);

The absolute address might also leak via /proc/PID/wchan as well, if
KALLSYMS is turned off or if the symbol lookup fails for some reason:

static int proc_pid_wchan(struct seq_file *m, struct pid_namespace *ns,
                          struct pid *pid, struct task_struct *task)
{
        unsigned long wchan;
        char symname[KSYM_NAME_LEN];

        wchan = get_wchan(task);

        if (lookup_symbol_name(wchan, symname) < 0) {
                if (!ptrace_may_access(task, PTRACE_MODE_READ))
                        return 0;
                seq_printf(m, "%lu", wchan);
        } else {
                seq_printf(m, "%s", symname);
        }

        return 0;
}

This isn't ideal, because for example it trivially leaks the KASLR offset
to any local attacker:

  fomalhaut:~> printf "%016lx\n" $(cat /proc/$$/stat | cut -d' ' -f35)
  ffffffff8123b380

Most real-life uses of wchan are symbolic:

  ps -eo pid:10,tid:10,wchan:30,comm

and procps uses /proc/PID/wchan, not the absolute address in /proc/PID/stat:

  triton:~/tip> strace -f ps -eo pid:10,tid:10,wchan:30,comm 2>&1 | grep wchan | tail -1
  open("/proc/30833/wchan", O_RDONLY)     = 6

There's one compatibility quirk here: procps relies on whether the
absolute value is non-zero - and we can provide that functionality
by outputing "0" or "1" depending on whether the task is blocked
(whether there's a wchan address).

These days there appears to be very little legitimate reason
user-space would be interested in  the absolute address. The
absolute address is mostly historic: from the days when we
didn't have kallsyms and user-space procps had to do the
decoding itself via the System.map.

So this patch sets all numeric output to "0" or "1" and keeps only
symbolic output, in /proc/PID/wchan.

( The absolute sleep address can generally still be profiled via
  perf, by tasks with sufficient privileges. )

Reviewed-by: Thomas Gleixner 
Acked-by: Kees Cook 
Acked-by: Linus Torvalds 
Cc: Al Viro 
Cc: Alexander Potapenko 
Cc: Andrey Konovalov 
Cc: Andrey Ryabinin 
Cc: Andy Lutomirski 
Cc: Andy Lutomirski 
Cc: Borislav Petkov 
Cc: Denys Vlasenko 
Cc: Dmitry Vyukov 
Cc: Kostya Serebryany 
Cc: Mike Galbraith 
Cc: Peter Zijlstra 
Cc: Peter Zijlstra 
Cc: Sasha Levin 
Cc: kasan-dev 
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/20150930135917.GA3285@gmail.com
Signed-off-by: Ingo Molnar 
[ kamal: backport to 3.16-stable: proc_pid_wchan context ]
Signed-off-by: Kamal Mostafa 
Signed-off-by: Luis Henriques

proc: actually make proc_fd_permission() thread-friendly

2015-12-13T17:49:12+00:00

commit 54708d2858e79a2bdda10bf8a20c80eb96c20613 upstream.

The commit 96d0df79f264 ("proc: make proc_fd_permission() thread-friendly")
fixed the access to /proc/self/fd from sub-threads, but introduced another
problem: a sub-thread can't access /proc//fd/ or /proc/thread-self/fd
if generic_permission() fails.

Change proc_fd_permission() to check same_thread_group(pid_task(), current).

Fixes: 96d0df79f264 ("proc: make proc_fd_permission() thread-friendly")
Reported-by: "Jin, Yihua" 
Signed-off-by: Oleg Nesterov 
Cc: "Eric W. Biederman" 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Luis Henriques