linux-toradex.git, branch v4.9.132

Linux 4.9.132

2018-10-10T06:53:23+00:00

dm thin metadata: fix __udivdi3 undefined on 32-bit

2018-10-10T06:53:23+00:00

commit 013ad043906b2befd4a9bfb06219ed9fedd92716 upstream.

sector_div() is only viable for use with sector_t.
dm_block_t is typedef'd to uint64_t -- so use div_u64() instead.

Fixes: 3ab918281 ("dm thin metadata: try to avoid ever aborting transactions")
Signed-off-by: Mike Snitzer 
Cc: Sudip Mukherjee 
Signed-off-by: Greg Kroah-Hartman

ocfs2: fix locking for res->tracking and dlm->tracking_list

2018-10-10T06:53:23+00:00

commit cbe355f57c8074bc4f452e5b6e35509044c6fa23 upstream.

In dlm_init_lockres() we access and modify res->tracking and
dlm->tracking_list without holding dlm->track_lock.  This can cause list
corruptions and can end up in kernel panic.

Fix this by locking res->tracking and dlm->tracking_list with
dlm->track_lock instead of dlm->spinlock.

Link: http://lkml.kernel.org/r/1529951192-4686-1-git-send-email-ashish.samant@oracle.com
Signed-off-by: Ashish Samant 
Reviewed-by: Changwei Ge 
Acked-by: Joseph Qi 
Acked-by: Jun Piao 
Cc: Mark Fasheh 
Cc: Joel Becker 
Cc: Junxiao Bi 
Cc: Changwei Ge 
Cc: 
Signed-off-by: Andrew Morton 
Signed-off-by: Greg Kroah-Hartman

proc: restrict kernel stack dumps to root

2018-10-10T06:53:23+00:00

commit f8a00cef17206ecd1b30d3d9f99e10d9fa707aa7 upstream.

Currently, you can use /proc/self/task/*/stack to cause a stack walk on
a task you control while it is running on another CPU.  That means that
the stack can change under the stack walker.  The stack walker does
have guards against going completely off the rails and into random
kernel memory, but it can interpret random data from your kernel stack
as instruction pointers and stack pointers.  This can cause exposure of
kernel stack contents to userspace.

Restrict the ability to inspect kernel stacks of arbitrary tasks to root
in order to prevent a local attacker from exploiting racy stack unwinding
to leak kernel task stack contents.  See the added comment for a longer
rationale.

There don't seem to be any users of this userspace API that can't
gracefully bail out if reading from the file fails.  Therefore, I believe
that this change is unlikely to break things.  In the case that this patch
does end up needing a revert, the next-best solution might be to fake a
single-entry stack based on wchan.

Link: http://lkml.kernel.org/r/20180927153316.200286-1-jannh@google.com
Fixes: 2ec220e27f50 ("proc: add /proc/*/stack")
Signed-off-by: Jann Horn 
Acked-by: Kees Cook 
Cc: Alexey Dobriyan 
Cc: Ken Chen 
Cc: Will Deacon 
Cc: Laura Abbott 
Cc: Andy Lutomirski 
Cc: Catalin Marinas 
Cc: Josh Poimboeuf 
Cc: Thomas Gleixner 
Cc: Ingo Molnar 
Cc: "H . Peter Anvin" 
Cc: 
Signed-off-by: Andrew Morton 
Signed-off-by: Greg Kroah-Hartman

gpiolib: Free the last requested descriptor

2018-10-10T06:53:23+00:00

commit 19a4fbffc94e41abaa2a623a25ce2641d69eccf0 upstream.

The current code only frees N-1 gpios if an error occurs during
gpiod_set_transitory, gpiod_direction_output or gpiod_direction_input.
Leading to gpios that cannot be used by userspace nor other drivers.

Cc: Timur Tabi 
Cc: stable@vger.kernel.org
Fixes: ab3dbcf78f60f46d ("gpioib: do not free unrequested descriptors)
Reported-by: Jan Lorenzen 
Reported-by: Jim Paris 
Signed-off-by: Ricardo Ribalda Delgado 
Signed-off-by: Linus Walleij 
Signed-off-by: Greg Kroah-Hartman

crypto: mxs-dcp - Fix wait logic on chan threads

2018-10-10T06:53:23+00:00

commit d80771c08363ad7fbf0f56f5301e7ca65065c582 upstream.

When compiling with CONFIG_DEBUG_ATOMIC_SLEEP=y the mxs-dcp driver
prints warnings such as:

WARNING: CPU: 0 PID: 120 at kernel/sched/core.c:7736 __might_sleep+0x98/0x9c
do not call blocking ops when !TASK_RUNNING; state=1 set at [<8081978c>] dcp_chan_thread_sha+0x3c/0x2ec

The problem is that blocking ops will manipulate current->state
themselves so it is not allowed to call them between
set_current_state(TASK_INTERRUPTIBLE) and schedule().

Fix this by converting the per-chan mutex to a spinlock (it only
protects tiny list ops anyway) and rearranging the wait logic so that
callbacks are called current->state as TASK_RUNNING. Those callbacks
will indeed call blocking ops themselves so this is required.

Cc: 
Signed-off-by: Leonard Crestez 
Signed-off-by: Herbert Xu 
Signed-off-by: Greg Kroah-Hartman

crypto: qat - Fix KASAN stack-out-of-bounds bug in adf_probe()

2018-10-10T06:53:23+00:00

commit ba439a6cbfa2936a6713f64cb499de7943673fe3 upstream.

The following KASAN warning was printed when booting a 64-bit kernel
on some systems with Intel CPUs:

[   44.512826] ==================================================================
[   44.520165] BUG: KASAN: stack-out-of-bounds in find_first_bit+0xb0/0xc0
[   44.526786] Read of size 8 at addr ffff88041e02fc50 by task kworker/0:2/124

[   44.535253] CPU: 0 PID: 124 Comm: kworker/0:2 Tainted: G               X --------- ---  4.18.0-12.el8.x86_64+debug #1
[   44.545858] Hardware name: Intel Corporation PURLEY/PURLEY, BIOS BKVDTRL1.86B.0005.D08.1712070559 12/07/2017
[   44.555682] Workqueue: events work_for_cpu_fn
[   44.560043] Call Trace:
[   44.562502]  dump_stack+0x9a/0xe9
[   44.565832]  print_address_description+0x65/0x22e
[   44.570683]  ? find_first_bit+0xb0/0xc0
[   44.570689]  kasan_report.cold.6+0x92/0x19f
[   44.578726]  find_first_bit+0xb0/0xc0
[   44.578737]  adf_probe+0x9eb/0x19a0 [qat_c62x]
[   44.578751]  ? adf_remove+0x110/0x110 [qat_c62x]
[   44.591490]  ? mark_held_locks+0xc8/0x140
[   44.591498]  ? _raw_spin_unlock+0x30/0x30
[   44.591505]  ? trace_hardirqs_on_caller+0x381/0x570
[   44.604418]  ? adf_remove+0x110/0x110 [qat_c62x]
[   44.604427]  local_pci_probe+0xd4/0x180
[   44.604432]  ? pci_device_shutdown+0x110/0x110
[   44.617386]  work_for_cpu_fn+0x51/0xa0
[   44.621145]  process_one_work+0x8fe/0x16e0
[   44.625263]  ? pwq_dec_nr_in_flight+0x2d0/0x2d0
[   44.629799]  ? lock_acquire+0x14c/0x400
[   44.633645]  ? move_linked_works+0x12e/0x2a0
[   44.637928]  worker_thread+0x536/0xb50
[   44.641690]  ? __kthread_parkme+0xb6/0x180
[   44.645796]  ? process_one_work+0x16e0/0x16e0
[   44.650160]  kthread+0x30c/0x3d0
[   44.653400]  ? kthread_create_worker_on_cpu+0xc0/0xc0
[   44.658457]  ret_from_fork+0x3a/0x50

[   44.663557] The buggy address belongs to the page:
[   44.668350] page:ffffea0010780bc0 count:0 mapcount:0 mapping:0000000000000000 index:0x0
[   44.676356] flags: 0x17ffffc0000000()
[   44.680023] raw: 0017ffffc0000000 ffffea0010780bc8 ffffea0010780bc8 0000000000000000
[   44.687769] raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
[   44.695510] page dumped because: kasan: bad access detected

[   44.702578] Memory state around the buggy address:
[   44.707372]  ffff88041e02fb00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[   44.714593]  ffff88041e02fb80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[   44.721810] >ffff88041e02fc00: 00 00 00 00 00 00 f1 f1 f1 f1 04 f2 f2 f2 f2 f2
[   44.729028]                                                  ^
[   44.734864]  ffff88041e02fc80: f2 f2 00 00 00 00 f3 f3 f3 f3 00 00 00 00 00 00
[   44.742082]  ffff88041e02fd00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
[   44.749299] ==================================================================

Looking into the code:

  int ret, bar_mask;
    :
  for_each_set_bit(bar_nr, (const unsigned long *)&bar_mask,

It is casting a 32-bit integer pointer to a 64-bit unsigned long
pointer. There are two problems here. First, the 32-bit pointer address
may not be 64-bit aligned. Secondly, it is accessing an extra 4 bytes.

This is fixed by changing the bar_mask type to unsigned long.

Cc: 
Signed-off-by: Waiman Long 
Signed-off-by: Herbert Xu 
Signed-off-by: Greg Kroah-Hartman

ALSA: hda/realtek - Cannot adjust speaker's volume on Dell XPS 27 7760

2018-10-10T06:53:22+00:00

commit 709ae62e8e6d9ac4df7dadb3b8ae432675c45ef9 upstream.

The issue is the same as commit dd9aa335c880 ("ALSA: hda/realtek - Can't
adjust speaker's volume on a Dell AIO"), the output requires to connect
to a node with Amp-out capability.

Applying the same fixup ALC298_FIXUP_SPK_VOLUME can fix the issue.

BugLink: https://bugs.launchpad.net/bugs/1775068
Signed-off-by: Kai-Heng Feng 
Cc: 
Signed-off-by: Takashi Iwai 
Signed-off-by: Greg Kroah-Hartman

smb2: fix missing files in root share directory listing

2018-10-10T06:53:22+00:00

commit 0595751f267994c3c7027377058e4185b3a28e75 upstream.

When mounting a Windows share that is the root of a drive (eg. C$)
the server does not return . and .. directory entries. This results in
the smb2 code path erroneously skipping the 2 first entries.

Pseudo-code of the readdir() code path:

cifs_readdir(struct file, struct dir_context)
    initiate_cifs_search            <-- if no reponse cached yet
        server->ops->query_dir_first

    dir_emit_dots
        dir_emit                    <-- adds "." and ".." if we're at pos=0

    find_cifs_entry
        initiate_cifs_search        <-- if pos < start of current response
                                         (restart search)
        server->ops->query_dir_next <-- if pos > end of current response
                                         (fetch next search res)

    for(...)                        <-- loops over cur response entries
                                          starting at pos
        cifs_filldir                <-- skip . and .., emit entry
            cifs_fill_dirent
            dir_emit
	pos++

A) dir_emit_dots() always adds . & ..
   and sets the current dir pos to 2 (0 and 1 are done).

Therefore we always want the index_to_find to be 2 regardless of if
the response has . and ..

B) smb1 code initializes index_of_last_entry with a +2 offset

  in cifssmb.c CIFSFindFirst():
		psrch_inf->index_of_last_entry = 2 /* skip . and .. */ +
			psrch_inf->entries_in_buffer;

Later in find_cifs_entry() we want to find the next dir entry at pos=2
as a result of (A)

	first_entry_in_buffer = cfile->srch_inf.index_of_last_entry -
					cfile->srch_inf.entries_in_buffer;

This var is the dir pos that the first entry in the buffer will
have therefore it must be 2 in the first call.

If we don't offset index_of_last_entry by 2 (like in (B)),
first_entry_in_buffer=0 but we were instructed to get pos=2 so this
code in find_cifs_entry() skips the 2 first which is ok for non-root
shares, as it skips . and .. from the response but is not ok for root
shares where the 2 first are actual files

		pos_in_buf = index_to_find - first_entry_in_buffer;
                // pos_in_buf=2
		// we skip 2 first response entries :(
		for (i = 0; (i < (pos_in_buf)) && (cur_ent != NULL); i++) {
			/* go entry by entry figuring out which is first */
			cur_ent = nxt_dir_entry(cur_ent, end_of_smb,
						cfile->srch_inf.info_level);
		}

C) cifs_filldir() skips . and .. so we can safely ignore them for now.

Sample program:

int main(int argc, char **argv)
{
	const char *path = argc >= 2 ? argv[1] : ".";
	DIR *dh;
	struct dirent *de;

	printf("listing path <%s>\n", path);
	dh = opendir(path);
	if (!dh) {
		printf("opendir error %d\n", errno);
		return 1;
	}

	while (1) {
		de = readdir(dh);
		if (!de) {
			if (errno) {
				printf("readdir error %d\n", errno);
				return 1;
			}
			printf("end of listing\n");
			break;
		}
		printf("off=%lu <%s>\n", de->d_off, de->d_name);
	}

	return 0;
}

Before the fix with SMB1 on root shares:

<.>            off=1
<..>           off=2
<$Recycle.Bin> off=3
      off=4

and on non-root shares:

<.>    off=1
<..>   off=4  <-- after adding .., the offsets jumps to +2 because
<2536> off=5       we skipped . and .. from response buffer (C)
<411>  off=6       but still incremented pos
 off=7
  off=8

Therefore the fix for smb2 is to mimic smb1 behaviour and offset the
index_of_last_entry by 2.

Test results comparing smb1 and smb2 before/after the fix on root
share, non-root shares and on large directories (ie. multi-response
dir listing):

PRE FIX
=======
pre-1-root VS pre-2-root:
        ERR pre-2-root is missing [bootmgr, $Recycle.Bin]
pre-1-nonroot VS pre-2-nonroot:
        OK~ same files, same order, different offsets
pre-1-nonroot-large VS pre-2-nonroot-large:
        OK~ same files, same order, different offsets

POST FIX
========
post-1-root VS post-2-root:
        OK same files, same order, same offsets
post-1-nonroot VS post-2-nonroot:
        OK same files, same order, same offsets
post-1-nonroot-large VS post-2-nonroot-large:
        OK same files, same order, same offsets

REGRESSION?
===========
pre-1-root VS post-1-root:
        OK same files, same order, same offsets
pre-1-nonroot VS post-1-nonroot:
        OK same files, same order, same offsets

BugLink: https://bugzilla.samba.org/show_bug.cgi?id=13107
Signed-off-by: Aurelien Aptel 
Signed-off-by: Paulo Alcantara 
Reviewed-by: Ronnie Sahlberg 
Signed-off-by: Steve French 
CC: Stable 
Signed-off-by: Greg Kroah-Hartman

sysfs: Do not return POSIX ACL xattrs via listxattr

2018-10-10T06:53:22+00:00

commit ffc4c92227db5699493e43eb140b4cb5904c30ff upstream.

Commit 786534b92f3c introduced a regression that caused listxattr to
return the POSIX ACL attribute names even though sysfs doesn't support
POSIX ACLs.  This happens because simple_xattr_list checks for NULL
i_acl / i_default_acl, but inode_init_always initializes those fields
to ACL_NOT_CACHED ((void *)-1).  For example:
    $ getfattr -m- -d /sys
    /sys: system.posix_acl_access: Operation not supported
    /sys: system.posix_acl_default: Operation not supported
Fix this in simple_xattr_list by checking if the filesystem supports POSIX ACLs.

Fixes: 786534b92f3c ("tmpfs: listxattr should include POSIX ACL xattrs")
Reported-by:  Marc Aurèle La France 
Tested-by: Marc Aurèle La France 
Signed-off-by: Andreas Gruenbacher 
Cc: stable@vger.kernel.org # v4.5+
Signed-off-by: Al Viro 
Signed-off-by: Greg Kroah-Hartman