The recent change for USB-audio disconnection race fixes introduced a
mutex deadlock again. There is a circular dependency between
chip->shutdown_rwsem and pcm->open_mutex, depicted like below, when a
device is opened during the disconnection operation:
B. snd_pcm_open() ->
pcm->open_mutex ->
snd_usb_pcm_open() ->
chip->shutdown_rwsem (read)
Since the chip->shutdown_rwsem protection in the case A is required
only for turning on the chip->shutdown flag and it doesn't have to be
taken for the whole operation, we can reduce its window in
snd_usb_audio_disconnect().
Signed-off-by: Mirko Lindner <mlindner@marvell.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The r8169 driver currently limits the DMA burst for TX to 1024 bytes. I have
a box where this prevents the interface from using the gigabit line to its full
potential. This patch solves the problem by setting TX_DMA_BURST to unlimited.
The box has an ASRock B75M motherboard with on-board RTL8168evl/8111evl
(XID 0c900880). TSO is enabled.
I used netperf (TCP_STREAM test) to measure the dependency of TX throughput
on MTU. I did it for three different values of TX_DMA_BURST ('5'=512, '6'=1024,
'7'=unlimited). This chart shows the results:
http://michich.fedorapeople.org/r8169/r8169-effects-of-TX_DMA_BURST.png
Interesting points:
- With the current DMA burst limit (1024):
- at the default MTU=1500 I get only 842 Mbit/s.
- when going from small MTU, the performance rises monotonically with
increasing MTU only up to a peak at MTU=1076 (908 MBit/s). Then there's
a sudden drop to 762 MBit/s from which the throughput rises monotonically
again with further MTU increases.
- With a smaller DMA burst limit (512):
- there's a similar peak at MTU=1076 and another one at MTU=564.
- With unlimited DMA burst:
- at the default MTU=1500 I get nice 940 Mbit/s.
- the throughput rises monotonically with increasing MTU with no strange
peaks.
Notice that the peaks occur at MTU sizes that are multiples of the DMA burst
limit plus 52. Why 52? Because:
20 (IP header) + 20 (TCP header) + 12 (TCP options) = 52
The Realtek-provided r8168 driver (v8.032.00) uses unlimited TX DMA burst too,
except for CFG_METHOD_1 where the TX DMA burst is set to 512 bytes.
CFG_METHOD_1 appears to be the oldest MAC version of "RTL8168B/8111B",
i.e. RTL_GIGA_MAC_VER_11 in r8169. Not sure if this MAC version really needs
the smaller burst limit, or if any other versions have similar requirements.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com> Acked-by: Francois Romieu <romieu@fr.zoreil.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
This regression was spotted between Debian squeeze and Debian wheezy
kernels (respectively based on 2.6.32 and 3.2). More info about
Wake-on-LAN issues with Realtek's 816x chipsets can be found in the
following thread: http://marc.info/?t=132079219400004
RTL_GIGA_MAC_VER_35 includes no multicast hardware filter.
Signed-off-by: Nathan Walp <faceprint@faceprint.com> Suggested-by: Hayes Wang <hayeswang@realtek.com> Acked-by: Francois Romieu <romieu@fr.zoreil.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The issue occurs when eCryptfs is mounted with a cipher supported by
the crypto subsystem but not by eCryptfs. The mount succeeds and an
error does not occur until a write. This change checks for eCryptfs
cipher support at mount time.
Resolves Launchpad issue #338914, reported by Tyler Hicks in 03/2009.
https://bugs.launchpad.net/ecryptfs/+bug/338914
Signed-off-by: Tim Sally <tsally@atomicpeace.com> Signed-off-by: Tyler Hicks <tyhicks@canonical.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
When the eCryptfs mount options do not include '-o acl', but the lower
filesystem's mount options do include 'acl', the MS_POSIXACL flag is not
flipped on in the eCryptfs super block flags. This flag is what the VFS
checks in do_last() when deciding if the current umask should be applied
to a newly created inode's mode or not. When a default POSIX ACL mask is
set on a directory, the current umask is incorrectly applied to new
inodes created in the directory. This patch ignores the MS_POSIXACL flag
passed into ecryptfs_mount() and sets the flag on the eCryptfs super
block depending on the flag's presence on the lower super block.
Additionally, it is incorrect to allow a writeable eCryptfs mount on top
of a read-only lower mount. This missing check did not allow writes to
the read-only lower mount because permissions checks are still performed
on the lower filesystem's objects but it is best to simply not allow a
rw mount on top of ro mount. However, a ro eCryptfs mount on top of a rw
mount is valid and still allowed.
As documented in RFC4861 (Neighbor Discovery for IP version 6) 7.2.6.,
unsolicited neighbour advertisements should be sent to the all-nodes
multicast address.
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
This buggy patch was a feature fix and has reached most stable
branches.
When skb->sk is NULL and when packet fanout is used, there is a
crash in match_fanout_group where skb->sk is accessed.
This patch fixes the issue by returning false as soon as the
socket is NULL: this correspond to the wanted behavior because
the kernel as to resend the skb to all the listening socket in
this case.
Signed-off-by: Eric Leblond <eric@regit.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
When creating an L2TPv3 Ethernet session, if register_netdev() should fail for
any reason (for example, automatic naming for "l2tpeth%d" interfaces hits the
32k-interface limit), the netdev is freed in the error path. However, the
l2tp_eth_sess structure's dev pointer is left uncleared, and this results in
l2tp_eth_delete() then attempting to unregister the same netdev later in the
session teardown. This results in an oops.
To avoid this, clear the session dev pointer in the error path.
Signed-off-by: Tom Parkin <tparkin@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Reading TCP stats when using TCP Illinois congestion control algorithm
can cause a divide by zero kernel oops.
The division by zero occur in tcp_illinois_info() at:
do_div(t, ca->cnt_rtt);
where ca->cnt_rtt can become zero (when rtt_reset is called)
Steps to Reproduce:
1. Register tcp_illinois:
# sysctl -w net.ipv4.tcp_congestion_control=illinois
2. Monitor internal TCP information via command "ss -i"
# watch -d ss -i
3. Establish new TCP conn to machine
Either it fails at the initial conn, or else it needs to wait
for a loss or a reset.
This is only related to reading stats. The function avg_delay() also
performs the same divide, but is guarded with a (ca->cnt_rtt > 0) at its
calling point in update_params(). Thus, simply fix tcp_illinois_info().
Function tcp_illinois_info() / get_info() is called without
socket lock. Thus, eliminate any race condition on ca->cnt_rtt
by using a local stack variable. Simply reuse info.tcpv_rttcnt,
as its already set to ca->cnt_rtt.
Function avg_delay() is not affected by this race condition, as
its called with the socket lock.
Cc: Petr Matousek <pmatouse@redhat.com> Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com> Acked-by: Eric Dumazet <edumazet@google.com> Acked-by: Stephen Hemminger <shemminger@vyatta.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Driver anchors the tx urbs and defers the urb submission if
a transmit request comes when the interface is suspended.
Anchoring urb increments the urb reference count. These
deferred urbs are later accessed by calling usb_get_from_anchor()
for submission during interface resume. usb_get_from_anchor()
unanchors the urb but urb reference count remains same.
This causes the urb reference count to remain non-zero
after usb_free_urb() gets called and urb never gets freed.
Hence call usb_put_urb() after anchoring the urb to properly
balance the reference count for these deferred urbs. Also,
unanchor these deferred urbs during disconnect, to free them
up.
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org> Acked-by: Oliver Neukum <oneukum@suse.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Commit a02e4b7dae4551(Demark default hoplimit as zero) only changes the
hoplimit checking condition and default value in ip6_dst_hoplimit, not
zeros all hoplimit default value.
Keep the zeroing ip6_template_metrics[RTAX_HOPLIMIT - 1] to force it as
const, cause as a37e6e344910(net: force dst_default_metrics to const
section)
Signed-off-by: Li RongQing <roy.qing.li@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
tcp_ioctl() tries to take into account if tcp socket received a FIN
to report correct number bytes in receive queue.
But its flaky because if the application ate the last skb,
we return 1 instead of 0.
Correct way to detect that FIN was received is to test SOCK_DONE.
Reported-by: Elliot Hughes <enh@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Neal Cardwell <ncardwell@google.com> Cc: Tom Herbert <therbert@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
On some suspend/resume operations involving wimax device, we have
noticed some intermittent memory corruptions in netlink code.
Stéphane Marchesin tracked this corruption in netlink_update_listeners()
and suggested a patch.
It appears netlink_release() should use kfree_rcu() instead of kfree()
for the listeners structure as it may be used by other cpus using RCU
protection.
netlink_release() must set to NULL the listeners pointer when
it is about to be freed.
Also have to protect netlink_update_listeners() and
netlink_has_listeners() if listeners is NULL.
Add a nl_deref_protected() lockdep helper to properly document which
locks protects us.
Reported-by: Jonathan Kliegman <kliegs@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Stéphane Marchesin <marcheu@google.com> Cc: Sam Leffler <sleffler@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Zijie Pan <zijie.pan@6wind.com> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com> Acked-by: Vlad Yasevich <vyasevich@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Reviewed-by: Brian Paul <brianp@vmware.com> Reviewed-by: Dmitry Torokhov <dtor@vmware.com> Cc: linux-graphics-maintainer@vmware.com Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The device would not reset properly when resuming from hibernation.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Reviewed-by: Brian Paul <brianp@vmware.com> Reviewed-by: Dmitry Torokhov <dtor@vmware.com> Cc: linux-graphics-maintainer@vmware.com Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
In kswapd(), set current->reclaim_state to NULL before returning, as
current->reclaim_state holds reference to variable on kswapd()'s stack.
In rare cases, while returning from kswapd() during memory offlining,
__free_slab() and freepages() can access the dangling pointer of
current->reclaim_state.
Signed-off-by: Takamori Yamaguchi <takamori.yamaguchi@jp.sony.com> Signed-off-by: Aaditya Kumar <aaditya.kumar@ap.sony.com> Acked-by: David Rientjes <rientjes@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
but never applied it. Repeated attempts over time to actually get him
to apply it have never had a reply from anyone who has raised it
So apply it anyway
Signed-off-by: Alan Cox <alan@linux.intel.com> Reported-by: Anders Blomdell <anders.blomdell@control.lth.se> Cc: Eric Paris <eparis@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Commit 4439647 ("xfs: reset buffer pointers before freeing them") in
3.0-rc1 introduced a regression when recovering log buffers that
wrapped around the end of log. The second part of the log buffer at
the start of the physical log was being read into the header buffer
rather than the data buffer, and hence recovery was seeing garbage
in the data buffer when it got to the region of the log buffer that
was incorrectly read.
Reported-by: Torsten Kaiser <just.for.lkml@googlemail.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
There are uncovered cases whether the card refcount introduced by the
commit a0830dbd isn't properly increased or decreased:
- OSS PCM and mixer success paths
- When lookup function gets NULL
VT1802 codec provides the invalid connection lists of NID 0x24 and
0x33 containing the routes to a non-exist widget 0x3e. This confuses
the auto-parser. Fix it up in the driver by overriding these
connections.
Reported-by: Massimo Del Fedele <max@veneto.com> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
In via_auto_fill_adc_nids(), the parser tries to fill dac_nids[] at
the point of the current line-out (i). When no valid path is found
for this output, this results in dac = 0, thus it creates a hole in
dac_nids[]. This confuses is_empty_dac() and trims the detected DAC
in later reference.
This patch fixes the bug by appending DAC properly to dac_nids[] in
via_auto_fill_adc_nids().
Reported-by: Massimo Del Fedele <max@veneto.com> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Don't assume bank 0 is selected at device probe time. This may not be
the case. Force bank selection at first register access to guarantee
that we read the right registers upon driver loading.
Signed-off-by: Jean Delvare <khali@linux-fr.org> Reviewed-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Several bug reports suggest that the forcibly resetting IEC958 status
bits is required for AD codecs to get the SPDIF output working
properly after changing streams.
Correctly enable the digital microphones with the right bits in the
right coeffecient registers on Cirrus CS4206/7 codecs. It also
prevents misconfiguring ADC1/2.
This fixes the digital mic on the Macbook Pro 10,1/Retina.
Based-on-patch-by: Alexander Stein <alexander.stein@systec-electronic.com> Signed-off-by: Daniel J Blueman <daniel@quora.org> Signed-off-by: Takashi Iwai <tiwai@suse.de>
[bwh: Backported to 3.2: adjust context] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The userspace cifs.idmap program generally works with the wbclient libs
to generate binary SIDs in userspace. That program defines the struct
that holds these values as having a max of 15 subauthorities. The kernel
idmapping code however limits that value to 5.
When the kernel copies those values around though, it doesn't sanity
check the num_subauths value handed back from userspace or from the
server. It's possible therefore for userspace to hand us back a bogus
num_subauths value (or one that's valid, but greater than 5) that could
cause the kernel to walk off the end of the cifs_sid->sub_auths array.
Fix this by defining a new routine for copying sids and using that in
all of the places that copy it. If we end up with a sid that's longer
than expected then this approach will just lop off the "extra" subauths,
but that's basically what the code does today already. Better approaches
might be to fix this code to reject SIDs with >5 subauths, or fix it
to handle the subauths array dynamically.
At the same time, change the kernel to check the length of the data
returned by userspace. If it's shorter than struct cifs_sid, reject it
and return -EIO. If that happens we'll end up with fields that are
basically uninitialized.
Long term, it might make sense to redefine cifs_sid using a flexarray at
the end, to allow for variable-length subauth lists, and teach the code
to handle the case where the subauths array being passed in from
userspace is shorter than 5 elements.
Note too, that I don't consider this a security issue since you'd need
a compromised cifs.idmap program. If you have that, you can do all sorts
of nefarious stuff. Still, this is probably reasonable for stable.
Reviewed-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com> Signed-off-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The sleeping code in iscsi_target_tx_thread() is susceptible to the classic
missed wakeup race:
- TX thread finishes handle_immediate_queue() and handle_response_queue(),
thinks both queues are empty.
- Another thread adds a queue entry and does wake_up_process(), which does
nothing because the TX thread is still awake.
- TX thread does schedule_timeout() and sleeps forever.
In practice this can kill an iSCSI connection if for example an initiator
does single-threaded writes and the target misses the wakeup window when
queueing an R2T; in this case the connection will be stuck until the
initiator loses patience and does some task management operation (or kills
the connection entirely).
Fix this by converting to wait_event_interruptible(), which does not
suffer from this sort of race.
Signed-off-by: Roland Dreier <roland@purestorage.com> Cc: Andy Grover <agrover@redhat.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The expression (max_sectors * block_size) might overflow a u32
(indeed, since iblock sets max_hw_sectors to UINT_MAX, it is
guaranteed to overflow and end up with a much-too-small result in many
common cases). Fix this by doing an equivalent calculation that
doesn't require multiplication.
While we're touching this code, avoid splitting a printk format across
two lines and use pr_info(...) instead of printk(KERN_INFO ...).
Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
If the call to core_dev_release_virtual_lun0() fails, then nothing
sets ret to anything other than 0, so even though everything is
torn down and freed, target_core_init_configfs() will seem to succeed
and the module will be loaded. Fix this by passing the return value
on up the chain.
Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Since commit c7f404b ('vfs: new superblock methods to override
/proc/*/mount{s,info}'), nfs_path() is used to generate the mounted
device name reported back to userland.
nfs_path() always generates a trailing slash when the given dentry is
the root of an NFS mount, but userland may expect the original device
name to be returned verbatim (as it used to be). Make this
canonicalisation optional and change the callers accordingly.
[jrnieder@gmail.com: use flag instead of bool argument] Reported-and-tested-by: Chris Hiestand <chiestand@salk.edu>
Reference: http://bugs.debian.org/669314 Signed-off-by: Ben Hutchings <ben@decadent.org.uk> Signed-off-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
[bwh: Backported to 3.2:
- Adjust context
- nfs_show_devname() still takes a pointer to struct vfsmount]
In very busy v3 environment, rpc.mountd can respond to the NULL
procedure but not the MNT procedure in a timely manner causing
the MNT procedure to time out. The problem is the mount system
call returns EIO which causes the mount to fail, instead of
ETIMEDOUT, which would cause the mount to be retried.
This patch sets the RPC_TASK_SOFT|RPC_TASK_TIMEOUT flags to
the rpc_call_sync() call in nfs_mount() which causes
ETIMEDOUT to be returned on timed out connections.
Signed-off-by: Steve Dickson <steved@redhat.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The DNS resolver's use of the sunrpc cache involves a 'ttl' number
(relative) rather that a timeout (absolute). This confused me when
I wrote
commit c5b29f885afe890f953f7f23424045cdad31d3e4
"sunrpc: use seconds since boot in expiry cache"
and I managed to break it. The effect is that any TTL is interpreted
as 0, and nothing useful gets into the cache.
This patch removes the use of get_expiry() - which really expects an
expiry time - and uses get_uint() instead, treating the int correctly
as a ttl.
This fixes a regression that has been present since 2.6.37, causing
certain NFS accesses in certain environments to incorrectly fail.
Reported-by: Chuck Lever <chuck.lever@oracle.com> Tested-by: Chuck Lever <chuck.lever@oracle.com> Signed-off-by: NeilBrown <neilb@suse.de> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Currently, we will schedule session recovery and then return to the
caller of nfs4_handle_exception. This works for most cases, but causes
a hang on the following test case:
Client Server
------ ------
Open file over NFS v4.1
Write to file
Expire client
Try to lock file
The server will return NFS4ERR_BADSESSION, prompting the client to
schedule recovery. However, the client will continue placing lock
attempts and the open recovery never seems to be scheduled. The
simplest solution is to wait for session recovery to run before retrying
the lock.
Signed-off-by: Bryan Schumaker <bjschuma@netapp.com> Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
[bwh: Backported to 3.2: adjust context] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
As Mukesh explained it, the MMUEXT_TLB_FLUSH_ALL allows the
hypervisor to do a TLB flush on all active vCPUs. If instead
we were using the generic one (which ends up being xen_flush_tlb)
we end up making the MMUEXT_TLB_FLUSH_LOCAL hypercall. But
before we make that hypercall the kernel will IPI all of the
vCPUs (even those that were asleep from the hypervisor
perspective). The end result is that we needlessly wake them
up and do a TLB flush when we can just let the hypervisor
do it correctly.
This patch gives around 50% speed improvement when migrating
idle guest's from one host to another.
Microsoft Digital Media Keyboard 3000 has two interfaces, and the
second one has a report descriptor with a bug. The second collection
says:
05 01 -- global; usage page -- 01 -- Generic Desktop Controls
09 80 -- local; usage -- 80 -- System Control
a1 01 -- main; collection -- 01 -- application
I.e. it makes us think that there are all kinds of usages of system
control. That the keyboard is a not only a keyboard, but also a
joystick, mouse, gamepad, keypad, etc. The same as for the Wireless
Desktop Receiver, this should be Physical Min/Max. So fix that
appropriately.
The tsc40 driver announces it supports the pressure event, but will never
send one. The announcement will cause tslib to wait for such events and
sending all touch events with a pressure of 0. Removing the announcement
will make tslib fall back to emulating the pressure on touch events so
everything works as expected.
Masaki found and patched a kallsyms issue: the last symbol in a
module's symtab wasn't transferred. This is because we manually copy
the zero'th entry (which is always empty) then copy the rest in a loop
starting at 1, though from src[0]. His fix was minimal, I prefer to
rewrite the loops in more standard form.
There are two loops: one to get the size, and one to copy. Make these
identical: always count entry 0 and any defined symbol in an allocated
non-init section.
LKML: http://lkml.org/lkml/2012/10/24/27 Reported-by: Masaki Kimura <masaki.kimura.kz@hitachi.com> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
[bwh: Backported to 3.2: we're still using a bitmap to compress the string
table] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The ath9k xmit functions for AMPDUs can send frames as non-aggregate in case
only one frame is currently available. The client will then answer using a
normal Ack instead of a BlockAck. This acknowledgement has no TID stored and
therefore the hardware is not able to provide us the corresponding TID.
The TID set by the hardware in the tx status descriptor has to be seen as
undefined and not as a valid TID value for normal acknowledgements. Doing
otherwise results in a massive amount of retransmissions and stalls of
connections.
Users may experience low bandwidth and complete connection stalls in
environments with transfers using multiple TIDs.
Signed-off-by: Sven Eckelmann <sven@narfation.org> Signed-off-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de> Acked-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
[bwh: Backported to 3.2: adjust context] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Some hardware has correct (!= 0xff) value of tssi_bounds[4] in the
EEPROM, but step is equal to 0xff. This results on ridiculous delta
calculations and completely broke TX power settings.
Reported-and-tested-by: Pavel Lucik <pavel.lucik@gmail.com> Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-by: Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
map->kmap_ops allocated in gntdev_alloc_map() wasn't freed by
gntdev_put_map().
Add a gntdev_free_map() helper function to free everything allocated
by gntdev_alloc_map().
Signed-off-by: David Vrabel <david.vrabel@citrix.com> Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
bf->bf_next is only while buffers are chained as part of an A-MPDU
in the tx queue. When a tid queue is flushed (e.g. on tearing down
an aggregation session), frames can be enqueued again as normal
transmission, without bf_next being cleared. This can lead to the
old pointer being dereferenced again later.
This patch might fix crashes and "Failed to stop TX DMA!" messages.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
[bwh: Backported to 3.2: adjust context] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The code to allow EAPOL frames even when the station
isn't yet marked associated needs to check that the
incoming frame is long enough and due to paged RX it
also can't assume skb->data contains the right data,
it must use skb_copy_bits(). Fix this to avoid using
data that doesn't really exist.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
A number of places in the mesh code don't check that
the frame data is present and in the skb header when
trying to access. Add those checks and the necessary
pskb_may_pull() calls. This prevents accessing data
that doesn't actually exist.
To do this, export ieee80211_get_mesh_hdrlen() to be
able to use it in mac80211.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Due to pskb_may_pull() checking the skb length, all
non-management frames are checked on input whether
their 802.11 header is fully present. Also add that
check for management frames and remove a check that
is now duplicate. This prevents accessing skb data
beyond the frame end.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The mesh header can have address extension by a 4th
or a 5th and 6th address, but never both. Drop such
frames in 802.11 -> 802.3 conversion along with any
frames that have the wrong extension.
Reviewed-by: Javier Cardona <javier@cozybit.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The 'ssid' field of the cfg80211_ibss_params is a u8 pointer and
its length is likely to be less than IEEE80211_MAX_SSID_LEN most
of the time.
This patch fixes the ssid copy in ieee80211_ibss_join() by using
the SSID length to prevent it from reading beyond the string.
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
[rewrapped commit message, small rewording] Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Per IEEE Std. 802.11-2012, Sec 8.2.4.4.1, the sequence Control field is
not present in control frames. We noticed this problem when processing
Block Ack Requests.
Signed-off-by: Javier Cardona <javier@cozybit.com> Signed-off-by: Javier Lopez <jlopex@cozybit.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Doing otherwise is wrong, and may wreak havoc on the mpp tables,
specially if the frame is encrypted.
Reported-by: Chaoxing Lin <Chaoxing.Lin@ultra-3eti.com> Signed-off-by: Javier Cardona <javier@cozybit.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
[bwh: Backported to 3.2: we have a large block conditional on
IEEE80211_RX_RA_MATCH rather than a goto conditional on the opposite,
so delete the condition] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The i2c core driver will turn the platform device ID to busnum
When using platfrom device ID as -1, it means dynamically assigned
the busnum. When writing code, we need to make sure the busnum,
and call i2c_register_board_info(int busnum, ...) to register device
if using -1, we do not know the value of busnum
In order to solve this issue, set the platform device ID as a fix number
Here using 0 to match the busnum used in i2c_regsiter_board_info()
Signed-off-by: Bo Shen <voice.shen@atmel.com> Acked-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com> Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com> Acked-by: Ludovic Desroches <ludovic.desroches@atmel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Newer at91sam9g10 SoC revision can't be detected, so the kernel can't boot with
this kind of kernel panic:
"AT91: Impossible to detect the SOC type"
CPU: ARM926EJ-S [41069265] revision 5 (ARMv5TEJ), cr=00053177
CPU: VIVT data cache, VIVT instruction cache
Machine: Atmel AT91SAM9G10-EK
Ignoring tag cmdline (using the default kernel command line)
bootconsole [earlycon0] enabled
Memory policy: ECC disabled, Data cache writeback
Kernel panic - not syncing: AT91: Impossible to detect the SOC type
[<c00133d4>] (unwind_backtrace+0x0/0xe0) from [<c02366dc>] (panic+0x78/0x1cc)
[<c02366dc>] (panic+0x78/0x1cc) from [<c02fa35c>] (at91_map_io+0x90/0xc8)
[<c02fa35c>] (at91_map_io+0x90/0xc8) from [<c02f9860>] (paging_init+0x564/0x6d0)
[<c02f9860>] (paging_init+0x564/0x6d0) from [<c02f7914>] (setup_arch+0x464/0x704)
[<c02f7914>] (setup_arch+0x464/0x704) from [<c02f44f8>] (start_kernel+0x6c/0x2d4)
[<c02f44f8>] (start_kernel+0x6c/0x2d4) from [<20008040>] (0x20008040)
The reason for this is that the Debug Unit Chip ID Register has changed between
Engineering Sample and definitive revision of the SoC. Changing the check of
cidr to socid will address the problem. We do not integrate this check to the
list just above because we also have to make sure that the extended id is
disregarded.
Signed-off-by: Ivan Shugov <ivan.shugov@gmail.com>
[nicolas.ferre@atmel.com: change commit message] Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com> Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
cryptd_queue_worker attempts to prevent simultaneous accesses to crypto
workqueue by cryptd_enqueue_request using preempt_disable/preempt_enable.
However cryptd_enqueue_request might be called from softirq context,
so add local_bh_disable/local_bh_enable to prevent data corruption and
panics.
Bug report at http://marc.info/?l=linux-crypto-vger&m=134858649616319&w=2
v2:
- Disable software interrupts instead of hardware interrupts
Reported-by: Gurucharan Shetty <gurucharan.shetty@gmail.com> Signed-off-by: Jussi Kivilinna <jussi.kivilinna@mbnet.fi> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
drm/i915: make sure we write all the DIP data bytes
we need to clear the entire sdvo buffer to avoid upsetting the
display.
Since infoframe buffer writing is now a bit more elaborate, extract it
into it's own function. This will be useful if we ever get around to
properly update the ELD for sdvo. Also #define proper names for the
two buffer indexes with fixed usage.
v2: Cite the right commit above, spotted by Paulo Zanoni.
v3: I'm too stupid to paste the right commit.
v4: Ben Hutchings noticed that I've failed to handle an underflow in
my loop logic, breaking it for i >= length + 8. Since I've just lost C
programmer license, use his solution. Also, make the frustrated 0-base
buffer size a notch more clear.
Reported-and-tested-by: Jürg Billeter <j@bitron.ch>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=25732 Cc: Paulo Zanoni <przanoni@gmail.com> Cc: Ben Hutchings <ben@decadent.org.uk> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
At least the worst offenders:
- SDVO specifies that the encoder should compute the ecc. Testing also
shows that we must not send the ecc field, so copy the dip_infoframe
struct to a temporay place and avoid the ecc field. This way the avi
infoframe is exactly 17 bytes long, which agrees with what the spec
mandates as a minimal storage capacity (with the ecc field it would
be 18 bytes).
- Only 17 when sending the avi infoframe. The SDVO spec explicitly
says that sending more data than what the device announces results
in undefined behaviour.
- Add __attribute__((packed)) to the avi and spd infoframes, for
otherwise they're wrongly aligned. Noticed because the avi infoframe
ended up being 18 bytes large instead of 17. We haven't noticed this
yet because we don't use the uint16_t fields yet (which are the only
ones that would be wrongly aligned).
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=25732 Tested-by: Peter Ross <pross@xvid.org> (G35 SDVO-HDMI) Reviewed-by: Eugeni Dodonov <eugeni.dodonov@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The overlay on the i830M has a peculiar failure mode: It works the
first time around after boot-up, but consistenly hangs the second time
it's used.
Chris Wilson has dug out a nice errata:
"1.5.12 Clock Gating Disable for Display Register
Address Offset: 06200h–06203h
"Bit 3
Ovrunit Clock Gating Disable.
0 = Clock gating controlled by unit enabling logic
1 = Disable clock gating function
DevALM Errata ALM049: Overlay Clock Gating Must be Disabled: Overlay
& L2 Cache clock gating must be disabled in order to prevent device
hangs when turning off overlay.SW must turn off Ovrunit clock gating
(6200h) and L2 Cache clock gating (C8h)."
Now I've nowhere found that 0xc8 register and hence couldn't apply the
l2 cache workaround. But I've remembered that part of the magic that
the OVERLAY_ON/OFF commands are supposed to do is to rearrange cache
allocations so that the overlay scaler has some scratch space.
And while pondering how that could explain the hang the 2nd time we
enable the overlay, I've remembered that the old ums overlay code did
_not_ issue the OVERLAY_OFF cmd.
And indeed, disabling the OFF cmd results in the overlay working
flawlessly, so I guess we can workaround the lack of the above
workaround by simply never disabling the overlay engine once it's
enabled.
Note that we have the first part of the above w/a already implemented
in i830_init_clock_gating - leave that as-is to avoid surprises.
v2: Add a comment in the code.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47827 Tested-by: Rhys <rhyspuk@gmail.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[bwh: Backported to 3.2:
- Adjust context
- s/intel_ring_emit(ring, /OUT_RING(/] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
No driver initializes chan->max_antenna_gain to something sensible, and
the only place where it is being used right now is inside ath9k. This
leads to ath9k potentially using less tx power than it can use, which can
decrease performance/range in some rare cases.
Rather than going through every single driver, this patch initializes
chan->orig_mag in wiphy_register(), ignoring whatever value the driver
left in there. If a driver for some reason wishes to limit it independent
from regulatory rulesets, it can do so internally.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
nlmsg_parse() might return an error, so test its return value before
potential random memory accesses.
Errors introduced in commit 115c9b81928 (rtnetlink: Fix problem with
buffer allocation)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Cc: Greg Rose <gregory.v.rose@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Implement a new netlink attribute type IFLA_EXT_MASK. The mask
is a 32 bit value that can be used to indicate to the kernel that
certain extended ifinfo values are requested by the user application.
At this time the only mask value defined is RTEXT_FILTER_VF to
indicate that the user wants the ifinfo dump to send information
about the VFs belonging to the interface.
This patch fixes a bug in which certain applications do not have
large enough buffers to accommodate the extra information returned
by the kernel with large numbers of SR-IOV virtual functions.
Those applications will not send the new netlink attribute with
the interface info dump request netlink messages so they will
not get unexpectedly large request buffers returned by the kernel.
Modifies the rtnl_calcit function to traverse the list of net
devices and compute the minimum buffer size that can hold the
info dumps of all matching devices based upon the filter passed
in via the new netlink attribute filter mask. If no filter
mask is sent then the buffer allocation defaults to NLMSG_GOODSIZE.
With this change it is possible to add yet to be defined netlink
attributes to the dump request which should make it fairly extensible
in the future.
Signed-off-by: Greg Rose <gregory.v.rose@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 3.2: drop the change in do_setlink() that reverts
commit f18da14565819ba43b8321237e2426a2914cc2ef, which we never applied] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Add missing index that may have led us to enabling
more crtcs than necessary.
May also fix:
https://bugs.freedesktop.org/show_bug.cgi?id=56139
Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Michel Dänzer <michel.daenzer@amd.com> Signed-off-by: Dave Airlie <airlied@redhat.com>
[bwh: Backported to 3.2: adjust context] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
When disconnect callback is called, each component should wake up
sleepers and check card->shutdown flag for avoiding the endless sleep
blocking the proper resource release.
Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
For more strict protection for wild disconnections, a refcount is
introduced to the card instance, and let it up/down when an object is
referred via snd_lookup_*() in the open ops.
The free-after-last-close check is also changed to check this refcount
instead of the empty list, too.
Close some races at disconnection of a USB audio device by adding the
chip->shutdown_mutex and chip->shutdown check at appropriate places.
The spots to put bandaids are:
- PCM prepare, hw_params and hw_free
- where the usb device is accessed for communication or get speed, in
mixer.c and others; the device speed is now cached in subs->speed
instead of accessing to chip->dev
The accesses in PCM open and close don't need the mutex protection
because these are already handled in the core PCM disconnection code.
The autosuspend/autoresume codes are still uncovered by this patch
because of possible mutex deadlocks. They'll be covered by the
upcoming change to rwsem.
Also the mixer codes are untouched, too. These will be fixed in
another patch, too.
Fix races at PCM disconnection:
- while a PCM device is being opened or closed
- while the PCM state is being changed without lock in prepare,
hw_params, hw_free ops
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
[jrnieder@gmail.com: backport to 3.2: make fbcon suspend/resume
handling conditional in a vague hope that this will approximate what
the original does for 3.3+] Signed-off-by: Jonathan Nieder <jrnieder@gmail.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The X86_32-only disable_hlt/enable_hlt mechanism was used by the
32-bit floppy driver. Its effect was to replace the use of the
HLT instruction inside default_idle() with cpu_relax() - essentially
it turned off the use of HLT.
This workaround was commented in the code as:
"disable hlt during certain critical i/o operations"
"This halt magic was a workaround for ancient floppy DMA
wreckage. It should be safe to remove."
H. Peter Anvin additionally adds:
"To the best of my knowledge, no-hlt only existed because of
flaky power distributions on 386/486 systems which were sold to
run DOS. Since DOS did no power management of any kind,
including HLT, the power draw was fairly uniform; when exposed
to the much hhigher noise levels you got when Linux used HLT
caused some of these systems to fail.
They were by far in the minority even back then."
Alan Cox further says:
"Also for the Cyrix 5510 which tended to go castors up if a HLT
occurred during a DMA cycle and on a few other boxes HLT during
DMA tended to go astray.
Do we care ? I doubt it. The 5510 was pretty obscure, the 5520
fixed it, the 5530 is probably the oldest still in any kind of
use."
So, let's finally drop this.
Signed-off-by: Len Brown <len.brown@intel.com> Signed-off-by: Josh Boyer <jwboyer@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Acked-by: "H. Peter Anvin" <hpa@zytor.com> Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk> Cc: Stephen Hemminger <shemminger@vyatta.com Cc: Linus Torvalds <torvalds@linux-foundation.org> Link: http://lkml.kernel.org/n/tip-3rhk9bzf0x9rljkv488tloib@git.kernel.org
[ If anyone cares then alternative instruction patching could be
used to replace HLT with a one-byte NOP instruction. Much simpler. ] Signed-off-by: Ingo Molnar <mingo@kernel.org>
[bwh: Backported to 3.2: adjust filename, context] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
This is not required in mainline since commit f1e91e1640d808d332498a6b09b2bcd01462eff9 ('Bluetooth: Always compile
SCO and L2CAP in Bluetooth Core') removed that option.
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
This USB V.92/V.32bis Controllered Modem have the USB vendor ID 0x0572
and device ID 0x1340. It need the NO_UNION_NORMAL quirk to be recognized.
Reference:
http://www.conexant.com/servlets/DownloadServlet/DSH-201723-005.pdf?docid=1725&revid=5
See idVendor and idProduct in table 6-1. Device Descriptors
Signed-off-by: Jean-Christian de Rivaz <jc@eclis.ch> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
tpm_write calls tpm_transmit without checking the return value and
assigns the return value unconditionally to chip->pending_data, even if
it's an error value.
This causes three bugs.
So if we write to /dev/tpm0 with a tpm_param_size bigger than
TPM_BUFSIZE=0x1000 (e.g. 0x100a)
and a bufsize also bigger than TPM_BUFSIZE (e.g. 0x100a)
tpm_transmit returns -E2BIG which is assigned to chip->pending_data as
-7, but tpm_write returns that TPM_BUFSIZE bytes have been successfully
been written to the TPM, altough this is not true (bug #1).
As we did write more than than TPM_BUFSIZE bytes but tpm_write reports
that only TPM_BUFSIZE bytes have been written the vfs tries to write
the remaining bytes (in this case 10 bytes) to the tpm device driver via
tpm_write which then blocks at
/* cannot perform a write until the read has cleared
either via tpm_read or a user_read_timer timeout */
while (atomic_read(&chip->data_pending) != 0)
msleep(TPM_TIMEOUT);
for 60 seconds, since data_pending is -7 and nobody is able to
read it (since tpm_read luckily checks if data_pending is greater than
0) (#bug 2).
After that the remaining bytes are written to the TPM which are
interpreted by the tpm as a normal command. (bug #3)
So if the last bytes of the command stream happen to be a e.g.
tpm_force_clear this gets accidentally sent to the TPM.
This patch fixes all three bugs, by propagating the error code of
tpm_write and returning -E2BIG if the input buffer is too big,
since the response from the tpm for a truncated value is bogus anyway.
Moreover it returns -EBUSY to userspace if there is a response ready to be
read.
Signed-off-by: Peter Huewe <peter.huewe@infineon.com> Signed-off-by: Kent Yoder <key@linux.vnet.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Existing code assumes that del_timer returns true for alive conntrack
entries. However, this is not true if reliable events are enabled.
In that case, del_timer may return true for entries that were
just inserted in the dying list. Note that packets / ctnetlink may
hold references to conntrack entries that were just inserted to such
list.
This patch fixes the issue by adding an independent timer for
event delivery. This increases the size of the ecache extension.
Still we can revisit this later and use variable size extensions
to allocate this area on demand.
Tested-by: Oliver Smith <olipro@8.c.9.b.0.7.4.0.1.0.0.2.ip6.arpa> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Acked-by: David Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The compat ioctl for VIDEO_SET_SPU_PALETTE was missing an error check
while converting ioctl arguments. This could lead to leaking kernel
stack contents into userspace.
Patch extracted from existing fix in grsecurity.
Signed-off-by: Kees Cook <keescook@chromium.org> Cc: David Miller <davem@davemloft.net> Cc: Brad Spengler <spender@grsecurity.net> Cc: PaX Team <pageexec@freemail.hu> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Fix possible overflow of the buffer used for expanding environment
variables when building file list.
In the extremely unlikely case of an attacker having control over the
environment variables visible to gen_init_cpio, control over the
contents of the file gen_init_cpio parses, and gen_init_cpio was built
without compiler hardening, the attacker can gain arbitrary execution
control via a stack buffer overflow.
Signed-off-by: Jan Luebbe <jlu@pengutronix.de> Cc: Alessandro Zummo <a.zummo@towertech.it> Cc: Roland Stigge <stigge@antcom.de> Cc: Grant Likely <grant.likely@secretlab.ca> Tested-by: Roland Stigge <stigge@antcom.de> Cc: Sascha Hauer <kernel@pengutronix.de> Cc: Russell King <linux@arm.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The genalloc code uses the bitmap API from include/linux/bitmap.h and
lib/bitmap.c, which is based on long values. Both bitmap_set from
lib/bitmap.c and bitmap_set_ll, which is the lockless version from
genalloc.c, use BITMAP_LAST_WORD_MASK to set the first bits in a long in
the bitmap.
That one uses (1 << bits) - 1, 0b111, if you are setting the first three
bits. This means that the API counts from the least significant bits
(LSB from now on) to the MSB. The LSB in the first long is bit 0, then.
The same works for the lookup functions.
The genalloc code uses longs for the bitmap, as it should. In
include/linux/genalloc.h, struct gen_pool_chunk has unsigned long
bits[0] as its last member. When allocating the struct, genalloc should
reserve enough space for the bitmap. This should be a proper number of
longs that can fit the amount of bits in the bitmap.
However, genalloc allocates an integer number of bytes that fit the
amount of bits, but may not be an integer amount of longs. 9 bytes, for
example, could be allocated for 70 bits.
This is a problem in itself if the Least Significat Bit in a long is in
the byte with the largest address, which happens in Big Endian machines.
This means genalloc is not allocating the byte in which it will try to
set or check for a bit.
This may end up in memory corruption, where genalloc will try to set the
bits it has not allocated. In fact, genalloc may not set these bits
because it may find them already set, because they were not zeroed since
they were not allocated. And that's what causes a BUG when
gen_pool_destroy is called and check for any set bits.
What really happens is that genalloc uses kmalloc_node with __GFP_ZERO
on gen_pool_add_virt. With SLAB and SLUB, this means the whole slab
will be cleared, not only the requested bytes. Since struct
gen_pool_chunk has a size that is a multiple of 8, and slab sizes are
multiples of 8, we get lucky and allocate and clear the right amount of
bytes.
Hower, this is not the case with SLOB or with older code that did memset
after allocating instead of using __GFP_ZERO.
So, a simple module as this (running 3.6.0), will cause a crash when
rmmod'ed.
module_init(foo_init);
module_exit(foo_exit);
[root@phantom-lp2 foo]# zcat /proc/config.gz | grep SLOB
CONFIG_SLOB=y
[root@phantom-lp2 foo]# insmod ./foo.ko
[root@phantom-lp2 foo]# rmmod foo
------------[ cut here ]------------
kernel BUG at lib/genalloc.c:243!
cpu 0x4: Vector: 700 (Program Check) at [c0000000bb0e7960]
pc: c0000000003cb50c: .gen_pool_destroy+0xac/0x110
lr: c0000000003cb4fc: .gen_pool_destroy+0x9c/0x110
sp: c0000000bb0e7be0
msr: 8000000000029032
current = 0xc0000000bb0e0000
paca = 0xc000000006d30e00 softe: 0 irq_happened: 0x01
pid = 13044, comm = rmmod
kernel BUG at lib/genalloc.c:243!
[c0000000bb0e7ca0] d000000004b00020 .foo_exit+0x20/0x38 [foo]
[c0000000bb0e7d20] c0000000000dff98 .SyS_delete_module+0x1a8/0x290
[c0000000bb0e7e30] c0000000000097d4 syscall_exit+0x0/0x94
--- Exception: c00 (System Call) at 000000800753d1a0
SP (fffd0b0e640) is in userspace
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com> Cc: Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Benjamin Gaignard <benjamin.gaignard@stericsson.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
On s390 any write to a page (even from kernel itself) sets architecture
specific page dirty bit. Thus when a page is written to via buffered
write, HW dirty bit gets set and when we later map and unmap the page,
page_remove_rmap() finds the dirty bit and calls set_page_dirty().
Dirtying of a page which shouldn't be dirty can cause all sorts of
problems to filesystems. The bug we observed in practice is that
buffers from the page get freed, so when the page gets later marked as
dirty and writeback writes it, XFS crashes due to an assertion
BUG_ON(!PagePrivate(page)) in page_buffers() called from
xfs_count_page_state().
Similar problem can also happen when zero_user_segment() call from
xfs_vm_writepage() (or block_write_full_page() for that matter) set the
hardware dirty bit during writeback, later buffers get freed, and then
page unmapped.
Fix the issue by ignoring s390 HW dirty bit for page cache pages of
mappings with mapping_cap_account_dirty(). This is safe because for
such mappings when a page gets marked as writeable in PTE it is also
marked dirty in do_wp_page() or do_page_fault(). When the dirty bit is
cleared by clear_page_dirty_for_io(), the page gets writeprotected in
page_mkclean(). So pagecache page is writeable if and only if it is
dirty.
Thanks to Hugh Dickins for pointing out mapping has to have
mapping_cap_account_dirty() for things to work and proposing a cleaned
up variant of the patch.
The patch has survived about two hours of running fsx-linux on tmpfs
while heavily swapping and several days of running on out build machines
where the original problem was triggered.
Signed-off-by: Jan Kara <jack@suse.cz> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[bwh: Backported to 3.2: adjust context; in particular there is no local
'anon' in page_remove_rmap()] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
flush_old_exec() clears PF_KTHREAD but forgets about PF_NOFREEZE.
Signed-off-by: Oleg Nesterov <oleg@redhat.com> Acked-by: Tejun Heo <tj@kernel.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
[bwh: Backported to 3.2: PF_FORKNOEXEC is cleared elsewhere] Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
The command cancellation code doesn't check whether find_trb_seg()
couldn't find the segment that contains the TRB to be canceled. This
could cause a NULL pointer deference later in the function when next_trb
is called. It's unlikely to happen unless something is wrong with the
command ring pointers, so add some debugging in case it happens.
This patch should be backported to stable kernels as old as 3.0, that
contain the commit b63f4053cc8aa22a98e3f9a97845afe6c15d0a0d "xHCI:
handle command after aborting the command ring".
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com> Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
projects / karo-tx-linux.git / log