If unable to ensure_next_mapping() we must add the current bio, which
was removed from the @bios list via bio_list_pop, back to the
deferred_bios list before all the remaining @bios.
Signed-off-by: Mike Snitzer <snitzer@redhat.com> Acked-by: Joe Thornber <ejt@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The persistent-data library used by dm-thin, dm-cache, etc is
transactional. If anything goes wrong, such as an io error when writing
new metadata or a power failure, then we roll back to the last
transaction.
Atomicity when committing a transaction is achieved by:
a) Never overwriting data from the previous transaction.
b) Writing the superblock last, after all other metadata has hit the
disk.
This commit and the following commit ("dm: take care to copy the space
map roots before locking the superblock") fix a bug associated with (b).
When committing it was possible for the superblock to still be written
in spite of an io error occurring during the preceeding metadata flush.
With these commits we're careful not to take the write lock out on the
superblock until after the metadata flush has completed.
Change the transaction manager's semantics for dm_tm_commit() to assume
all data has been flushed _before_ the single superblock that is passed
in.
As a prerequisite, split the block manager's block unlocking and
flushing by simplifying dm_bm_flush_and_unlock() to dm_bm_flush(). Now
the unlocking must be done separately.
This issue was discovered by forcing io errors at the crucial time
using dm-flakey.
Signed-off-by: Joe Thornber <ejt@redhat.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the discard block size is larger than the cache block size we will
not properly quiesce IO to a region that is about to be discarded. This
results in a race between a cache migration where no copy is needed, and
a write to an adjacent cache block that's within the same large discard
block.
Workaround this by limiting the discard_block_size to cache_block_size.
Also limit the max_discard_sectors to cache_block_size.
A more comprehensive fix that introduces range locking support in the
bio_prison and proper quiescing of a discard range that spans multiple
cache blocks is already in development.
Reported-by: Morgan Mears <Morgan.Mears@netapp.com> Signed-off-by: Mike Snitzer <snitzer@redhat.com> Acked-by: Joe Thornber <ejt@redhat.com> Acked-by: Heinz Mauelshagen <heinzm@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Crash detected on sam5d35 and its pmecc nand ecc controller.
The problem was a call to chip->ecc.hwctl from nand_write_subpage_hwecc
(nand_base.c) when we write a sub page.
chip->ecc.hwctl function is not set when we are using PMECC controller.
As a workaround, set NAND_NO_SUBPAGE_WRITE for PMECC controller in
order to disable sub page access in nand_write_page.
The functions for data copying copyarea_foreward_8bpp and
copyarea_backward_8bpp are buggy, they produce screen corruption.
This patch fixes the functions and moves the logic to one function
"copyarea_8bpp". For simplicity, the function only handles copying that
is aligned on 8 pixes. If we copy an unaligned area, generic function
cfb_copyarea is used.
Mode setting in the TGA driver is broken for these reasons:
- info->fix.line_length is set just once in tgafb_init_fix function. If
we change videomode, info->fix.line_length is not recalculated - so
the video mode is changed but the screen is corrupted because of wrong
info->fix.line_length.
- info->fix.smem_len is set in tgafb_init_fix to the size of the default
video mode (640x480). If we set a higher resolution,
info->fix.smem_len is smaller than the current screen size, preventing
the userspace program from mapping the framebuffer.
This patch fixes it:
- info->fix.line_length initialization is moved to tgafb_set_par so that
it is recalculated with each mode change.
- info->fix.smem_len is set to a fixed value representing the real
amount of video ram (the values are taken from xfree86 driver).
- add a check to tgafb_check_var to prevent us from setting a videomode
that doesn't fit into videoram.
- in tgafb_register, tgafb_init_fix is moved upwards, to be called
before fb_find_mode (because fb_find_mode already needs the videoram
size set in tgafb_init_fix).
The AS3722_GPIO_INV bit will always be blindly overwritten by
as3722_pinctrl_gpio_set_direction() and will be ignored when
setting the value of the GPIO in as3722_gpio_set() since the
enable_gpio_invert flag is never set. This will cause an
initially inverted GPIO to toggle when requested as an output,
which could be problematic if, for example, the GPIO controls
a critical regulator.
Instead of setting up the enable_gpio_invert flag, just leave
the invert bit alone and check it before setting the GPIO value.
Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Reviewed-by: Stephen Warren <swarren@nvidia.com> Tested-by: Stephen Warren <swarren@nvidia.com> Acked-by: Laxman Dewangan <ldewangan@nvidia.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The scenario here is that someone calls enable_irq_wake() from somewhere
in the code. This will result in the lockdep producing a backtrace as can
be seen below. In my case, this problem is triggered when using the wl1271
(TI WlCore) driver found in drivers/net/wireless/ti/ .
The problem cause is rather obvious from the backtrace, but let's outline
the dependency. enable_irq_wake() grabs the IRQ buslock in irq_set_irq_wake(),
which in turns calls mxs_gpio_set_wake_irq() . But mxs_gpio_set_wake_irq()
calls enable_irq_wake() again on the one-level-higher IRQ , thus it tries to
grab the IRQ buslock again in irq_set_irq_wake() . Because the spinlock in
irq_set_irq_wake()->irq_get_desc_buslock()->__irq_get_desc_lock() is not
marked as recursive, lockdep will spew the stuff below.
We know we can safely re-enter the lock, so use IRQ_GC_INIT_NESTED_LOCK to
fix the spew.
=============================================
[ INFO: possible recursive locking detected ] 3.10.33-00012-gf06b763-dirty #61 Not tainted
---------------------------------------------
kworker/0:1/18 is trying to acquire lock:
(&irq_desc_lock_class){-.-...}, at: [<c00685f0>] __irq_get_desc_lock+0x48/0x88
but task is already holding lock:
(&irq_desc_lock_class){-.-...}, at: [<c00685f0>] __irq_get_desc_lock+0x48/0x88
other info that might help us debug this:
Possible unsafe locking scenario:
packet_beacon is not initialized and hence packet_beacon
contains garbage from the stack, so set it to false.
Signed-off-by: Colin Ian King <colin.king@canonical.com> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is noting wrong with the above commit. The regression occurs because
V0 queue on RTL8192SE cards uses priority 6, not the usual 7. The fix is to
modify the rtl8192se routine that sets the correct transmit queue.
Reported-by: Alex Miller <almiller_1@yahoo.co.uk> Tested-by: Alex Miller <almiller_1@yahoo.co.uk> Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In commit f78bccd79ba3cd9d9664981b501d57bdb81ab8a4 entitled "rtlwifi:
rtl8192ce: Fix too long disable of IRQs", Olivier Langlois
<olivier@trillion01.com> fixed a problem caused by an extra long disabling
of interrupts. This patch makes the same fix for rtl8192se.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In commit f78bccd79ba3cd9d9664981b501d57bdb81ab8a4 entitled "rtlwifi:
rtl8192ce: Fix too long disable of IRQs", Olivier Langlois
<olivier@trillion01.com> fixed a problem caused by an extra long disabling
of interrupts. This patch makes the same fix for rtl8192cu.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In commit f78bccd79ba3cd9d9664981b501d57bdb81ab8a4 entitled "rtlwifi:
rtl8192ce: Fix too long disable of IRQs", Olivier Langlois
<olivier@trillion01.com> fixed a problem caused by an extra long disabling
of interrupts. This patch makes the same fix for rtl8188ee.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In commit f78bccd79ba3cd9d9664981b501d57bdb81ab8a4 entitled "rtlwifi:
rtl8192ce: Fix too long disable of IRQs", Olivier Langlois
<olivier@trillion01.com> fixed a problem caused by an extra long disabling
of interrupts. This patch makes the same fix for rtl8723ae.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A fl->fl_break_time of 0 has a special meaning to the lease break code
that basically means "never break the lease". knfsd uses this to ensure
that leases don't disappear out from under it.
Unfortunately, the code in __break_lease can end up passing this value
to wait_event_interruptible as a timeout, which prevents it from going
to sleep at all. This causes __break_lease to spin in a tight loop and
causes soft lockups.
Fix this by ensuring that we pass a minimum value of 1 as a timeout
instead.
Cc: J. Bruce Fields <bfields@fieldses.org> Reported-by: Terry Barnaby <terry1@beam.ltd.uk> Signed-off-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
qi->tqi_readyTime is written directly to registers that expect
microseconds as unit instead of TU.
When setting the CABQ ready time, cur_conf->beacon_interval is in TU, so
convert it to microseconds before passing it to ath9k_hw.
This should hopefully fix some Tx DMA issues with buffered multicast
frames in AP mode.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Their power value is initialized to zero. This patch fixes an issue
where the configured power drops to the minimum value when AP_VLAN
interfaces are created/removed.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jouni reported that when doing off-channel transmissions mixed
with on-channel transmissions, the on-channel ones ended up on
the off-channel in some cases.
The reason for that is that during the refactoring of the off-
channel code, I lost the part that stopped all activity and as
a consequence the on-channel frames (including data frames)
were no longer queued but would be transmitted on the temporary
channel.
Fix this by simply restoring the lost activity stop call.
Since Stanislaw's patch removing the quiescing code, mac80211 had
a race regarding suspend vs. authentication: as cfg80211 doesn't
track authentication attempts, it can't abort them. Therefore the
attempts may be kept running while suspending, which can lead to
all kinds of issues, in at least some cases causing an error in
iwlmvm firmware.
Fix this by aborting the authentication attempt when suspending.
commit de74a1d9032f4d37ea453ad2a647e1aff4cd2591
"mac80211: fix WPA with VLAN on AP side with ps-sta"
fixed an issue where queued multicast packets would
be sent out encrypted with the key of an other bss.
commit "7cbf9d017dbb5e3276de7d527925d42d4c11e732"
"mac80211: fix oops on mesh PS broadcast forwarding"
essentially reverted it, because vif.type cannot be AP_VLAN
due to the check to vif.type in ieee80211_get_buffered_bc before.
As the later commit intended to fix the MESH case, fix it
by checking for IFTYPE_AP instead of IFTYPE_AP_VLAN.
Fixes: 7cbf9d017dbb ("mac80211: fix oops on mesh PS broadcast forwarding") Signed-off-by: Michael Braun <michael-dev@fami-braun.de> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The bss struct might be freed in ieee80211_rx_bss_put(),
so we shouldn't use it afterwards.
Fixes: 817cee7675237 ("mac80211: track AP's beacon rate and give it to the driver") Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Back in 27f4d1f6bc32c2ed7b2c5080cbd58b14df622607 we refactored the CRUSH
code to allow adjustment of the retry counts on a per-pool basis. That
commit had an off-by-one bug: the previous "tries" counter was a *retry*
count, not a *try* count, but the new code was passing in 1 meaning
there should be no retries.
Fix the ftotal vs tries comparison to use < instead of <= to fix the
problem. Note that the original code used <= here, which means the
global "choose_total_tries" tunable is actually counting retries.
Compensate for that by adding 1 in crush_do_rule when we pull the tunable
into the local variable.
This was noticed looking at output from a user provided osdmap.
Unfortunately the map doesn't illustrate the change in mapping behavior
and I haven't managed to construct one yet that does. Inspection of the
crush debug output now aligns with prior versions, though.
The current firmware advertises support for uAPSD, but
critical bugs force us to disable the feature.
When a fixed firmware will be available, we will be able to
re-enable uAPSD.
Crucial/Micron M500 drives properly support queued DSM TRIM starting
with firmware MU05. Update the blacklist so we only disable queued trim
for older firmware releases.
Early M550 series drives suffer from the same issue as M500. A bugfix
firmware is in the pipeline but not ready yet. Until then, blacklist
queued trim for M550.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Cc: Chris Samuel <chris@csamuel.org> Cc: Marc MERLIN <marc@merlins.org> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In multiple MSI mode all AHCI ports (including dummy) get assigned
separate MSI vectors and (as result of execution
pci_enable_msi_exact() function) separate IRQ numbers, (mapped to the
MSI vectors).
Therefore, although interrupts from dummy ports are not desired they
are still enabled. We do not request IRQs for dummy ports, but that
only means we do not assign AHCI-specific ISRs to corresponding IRQ
numbers.
As result, dummy port interrupts still could come and traverse all the
way from the PCI device to the kernel, causing unnecessary overhead.
This update disables IRQs for dummy ports and prevents the described
issue.
Signed-off-by: Alexander Gordeev <agordeev@redhat.com> Signed-off-by: Tejun Heo <tj@kernel.org> Tested-by: David Milburn <dmilburn@redhat.com> Cc: linux-ide@vger.kernel.org Fixes: 5ca72c4f7c41 ("AHCI: Support multiple MSIs") Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The AHCI specification allows hardware to choose to revert to
single MSI mode when fewer messages are allocated than requested.
Yet, at least ICH10 chipset reverts to single MSI mode even when
enough messages are allocated in some cases (see below).
This update forces the driver to not rely on initialization of
multiple MSIs mode alone and always check if "MSI Revert to
Single Message" (MRSM) mode was enforced by the controller and
fallback to the single MSI mode in case it did.
That prevents a situation when the driver configured multiple
per-port IRQ handlers, but the controller sends all port's
interrupts to a single IRQ, which could easily screw up the
interrupt handling and lead to delays and possibly crashes.
The fix was tested on a 6-port controller that successfully
reverted to the single MSI mode:
00:1f.2 SATA controller: Intel Corporation 82801JI (ICH10 Family) SATA
AHCI Controller (prog-if 01 [AHCI 1.0])
Subsystem: Super Micro Computer Inc Device 10a7
Flags: bus master, 66MHz, medium devsel, latency 0, IRQ 101
I/O ports at f110 [size=8]
I/O ports at f100 [size=4]
I/O ports at f0f0 [size=8]
I/O ports at f0e0 [size=4]
I/O ports at f020 [size=32]
Memory at fbf00000 (32-bit, non-prefetchable) [size=2K]
Capabilities: [80] MSI: Enable+ Count=1/16 Maskable- 64bit-
Capabilities: [70] Power Management version 3
Capabilities: [a8] SATA HBA v1.0
Capabilities: [b0] PCI Advanced Features
Kernel driver in use: ahci
With 6 ports just 8 MSI vectors should be enough, but the adapter
enforces the MRSM mode when less than 16 vectors are written to
the Multiple Messages Enable PCI register. I instigated MRSM mode
by forcing @nvec to 8 in ahci_init_interrupts().
The AHCI spec allows implementations to issue commands in tag order
rather than FIFO order:
5.3.2.12 P:SelectCmd
HBA sets pSlotLoc = (pSlotLoc + 1) mod (CAP.NCS + 1)
or HBA selects the command to issue that has had the
PxCI bit set to '1' longer than any other command
pending to be issued.
The result is that commands posted sequentially (time-wise) may play out
of sequence when issued by hardware.
This behavior has likely been hidden by drives that arrange for commands
to complete in issue order. However, it appears recent drives (two from
different vendors that we have found so far) inflict out-of-order
completions as a matter of course. So, we need to take care to maintain
ordered submission, otherwise we risk triggering a drive to fall out of
sequential-io automation and back to random-io processing, which incurs
large latency and degrades throughput.
This issue was found in simple benchmarks where QD=2 seq-write
performance was 30-50% *greater* than QD=32 seq-write performance.
Tagging for -stable and making the change globally since it has a low
risk-to-reward ratio. Also, word is that recent versions of an unnamed
OS also does it this way now. So, drives in the field are already
experienced with this tag ordering scheme.
Cc: Dave Jiang <dave.jiang@intel.com> Cc: Ed Ciechanowski <ed.ciechanowski@intel.com> Reviewed-by: Matthew Wilcox <matthew.r.wilcox@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
System may crash in ahci_hw_interrupt() or ahci_thread_fn() when
accessing the interrupt status in a port's private_data if the port is
actually a DUMMY port.
00:1f.2 SATA controller: Intel Corporation 82801JI (ICH10 Family) SATA AHCI Controller
This reverts commit e3a8786c10e75903f1269474e21fe8cb49c3a670. While
this commit allows to use the mvneta driver as a module on some
configurations, it breaks other configurations even if mvneta is used
built-in.
This breakage is due to the fact that on some RGMII platforms, the PCS
bit has to be set, and on some other platforms, it has to be
cleared. At the moment, we lack informations to know exactly the
significance of this bit (the datasheet only says "enables PCS"), and
so we can't produce a patch that will work on all platforms at this
point. And since this change is breaking the network completely for
many users, it's much better to revert it for now. We'll come back
later with a proper fix that takes into account all platforms.
Basically:
* Armada XP GP is configured as RGMII-ID, and needs the PCS bit to be
set.
* Armada 370 Mirabox is configured as RGMII-ID, and needs the PCS bit
to be cleared.
And at the moment, we don't know how to make the distinction between
those two cases. One hint is that the Armada XP GP appears in fact to
be using a QSGMII connection with the PHY (Quad-SGMII), but
configuring it as SGMII doesn't work, while RGMII-ID works. This needs
more investigation, but in the mean time, let's unbreak the network
for all those users.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com> Reported-by: Arnaud Ebalard <arno@natisbad.org> Reported-by: Alexander Reuter <Alexander.Reuter@gmx.net> Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=73401 Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch fixes the hardware cursor on mach64 when font width is not a
multiple of 8 pixels.
If you load such a font, the cursor is expanded to the next 8-byte
boundary and a part of the next character after the cursor is not
visible.
For example, when you load a font with 12-pixel width, the cursor width
is 16 pixels and when the cursor is displayed, 4 pixels of the next
character are not visible.
The reason is this: atyfb_cursor is called with proper parameters to
load an image that is 12-pixel wide. However, the number is aligned on
the next 8-pixel boundary on the line
"unsigned int width = (cursor->image.width + 7) >> 3;" and the whole
function acts as it is was loading a 16-pixel image.
This patch fixes it so that the value written to the framebuffer is
padded with 0xaaaa (the transparent pattern) when the image size it not
a multiple of 8 pixels. The transparent pattern causes that the cursor
will not interfere with the next character.
The function cfb_copyarea is buggy when the copy operation is not aligned on
long boundary (4 bytes on 32-bit machines, 8 bytes on 64-bit machines).
How to reproduce:
- use x86-64 machine
- use a framebuffer driver without acceleration (for example uvesafb)
- set the framebuffer to 8-bit depth
(for example fbset -a 1024x768-60 -depth 8)
- load a font with character width that is not a multiple of 8 pixels
note: the console-tools package cannot load a font that has
width different from 8 pixels. You need to install the packages
"kbd" and "console-terminus" and use the program "setfont" to
set font width (for example: setfont Uni2-Terminus20x10)
- move some text left and right on the bash command line and you get a
screen corruption
To expose more bugs, put this line to the end of uvesafb_init_info:
info->flags |= FBINFO_HWACCEL_COPYAREA | FBINFO_READS_FAST;
- Now framebuffer console will use cfb_copyarea for console scrolling.
You get a screen corruption when console is scrolled.
This patch is a rewrite of cfb_copyarea. It fixes the bugs, with this
patch, console scrolling in 8-bit depth with a font width that is not a
multiple of 8 pixels works fine.
The cfb_copyarea code was very buggy and it looks like it was written
and never tried with non-8-pixel font.
There was a very small race window where resume to kernel mode from a
Exception Path (or pure kernel mode which is true for most of ARC
exceptions anyways), was not disabling interrupts in restore_regs,
clobbering the exception regs
Anton found the culprit call flow (after many sleepless nights)
| 1. we got a Trap from user land
| 2. started to service it.
| 3. While doing some stuff on user-land memory (I think it is padzero()),
| we got a DataTlbMiss
| 4. On return from it we are taking "resume_kernel_mode" path
| 5. NEED_RESHED is not set, so we go to "return from exception" path in
| restore regs.
| 6. there seems to be IRQ happening
Don't schedule an rpc_delay before checking to see if the task
is a SOFTCONN because the tk_callback from the delay (__rpc_atrun)
clears the task status before the rpc_exit_task can be run.
Signed-off-by: Steve Dickson <steved@redhat.com> Fixes: 561ec1603171c (SUNRPC: call_connect_status should recheck...) Link: http://lkml.kernel.org/r/5329CF7C.7090308@RedHat.com Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 9e1fda4ae158 ("ASoC: dapm: Implement mixer input auto-disable")
is trying to free the widget it allocated by snd_soc_dapm_new_control()
call in dapm_kcontrol_data_alloc() by adding kfree(data->widget) to
dapm_kcontrol_free().
This is causing a widget double free with auto-disabled DAPM kcontrols
in sound card unregistration because widgets are already freed before
dapm_kcontrol_free() is called.
Reason for that is all widgets are added into dapm->card->widgets list
in snd_soc_dapm_new_control() and freed in dapm_free_widgets() during
execution of snd_soc_dapm_free().
Now snd_soc_dapm_free() calls for different DAPM contexts happens before
snd_card_free() call from where the call chain to dapm_kcontrol_free()
begins:
The A register needs to be initialized to zero in the prolog if the
first instruction of the BPF program is BPF_S_LDX_B_MSH to prevent
leaking the content of %r5 to user space.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Using a notification type mask for the store event information chsc
is unsupported on some firmware levels. Retry SEI with that mask set
to zero (which is the old way of requesting only channel subsystem
related events).
Reported-and-tested-by: Stefan Haberland <stefan.haberland@de.ibm.com> Reviewed-by: Peter Oberparleiter <oberpar@linux.vnet.ibm.com> Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 41dd03a9 may cause Oops in rtas_stop_self().
The reason is that the rtas_args was moved into stack space. For a box
with more that 4GB RAM, the stack could easily be outside 32bit range,
but RTAS is 32bit.
So the patch moves rtas_args away from stack by adding static before
it.
Signed-off-by: Li Zhong <zhong@linux.vnet.ibm.com> Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This moves tm_recheckpoint to a C function and moves the tm_restore_sprs into
that function. It then adds IRQ disabling over the trechkpt critical section.
It also sets the TEXASR FS in the signals code to ensure this is never set now
that we explictly write the TM sprs in tm_recheckpoint.
Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On suspend, _scsih_suspend calls mpt2sas_base_free_resources, which
in turn calls pci_disable_device if the device is enabled prior to
suspending. However, _scsih_suspend also calls pci_disable_device
itself.
Thus, in the event that the device is enabled prior to suspending,
pci_disable_device will be called twice. This patch removes the
duplicate call to pci_disable_device in _scsi_suspend as it is both
unnecessary and results in a kernel oops.
virtscsi_init calls virtscsi_remove_vqs on err, even before initializing
the vqs. The latter calls virtscsi_set_affinity, so let's check the
pointer there before setting affinity on it.
This fixes a panic when setting device's num_queues=2 on RHEL 6.5:
cc -g -O2 -Wall -I. -I ../../usr/include/ -Wno-pointer-sign
-fno-strict-overflow -fno-strict-aliasing -fno-common -MMD
-U_FORTIFY_SOURCE -c -o virtio_test.o virtio_test.c
virtio_test.c: In function ‘run_test’:
virtio_test.c:176:7: error: expected ‘)’ before ‘r’
r = -1;
^
Fixes: 53c18c9906441 (virtio_test: verify if virtqueue_kick() succeeded) Cc: Heinz Graalfs <graalfs@linux.vnet.ibm.com> Signed-off-by: Joel Stanley <joel@jms.id.au> Acked-by: Michael S. Tsirkin <mst@redhat.com> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
BUG: soft lockup - CPU#0 stuck for 22s! [vballoon:367]
We have a wait_event_interruptible(), but the condition is always true
(more ballooning to do) so we don't ever sleep. We also have a
wait_event() for the host to ack, but that is also always true as QEMU
is synchronous for balloon operations.
The original MIPS hibernate code flushes cache and TLB entries in
swsusp_arch_resume(). But they are removed in Commit 44eeab67416711
(MIPS: Hibernation: Remove SMP TLB and cacheflushing code.). A cross-
CPU flush is surely unnecessary because all but the local CPU have
already been disabled. But a local flush (at least the TLB flush) is
needed. When we do hibernation on Loongson-3 with an E1000E NIC, it is
very easy to produce a kernel panic (kernel page fault, or unaligned
access). The root cause is E1000E driver use vzalloc_node() to allocate
pages, the stale TLB entries of the booting kernel will be misused by
the resumed target kernel.
Signed-off-by: Huacai Chen <chenhc@lemote.com> Cc: John Crispin <john@phrozen.org> Cc: Steven J. Hill <Steven.Hill@imgtec.com> Cc: Aurelien Jarno <aurelien@aurel32.net> Cc: linux-mips@linux-mips.org Cc: Fuxin Zhang <zhangfx@lemote.com> Cc: Zhangjin Wu <wuzhangjin@gmail.com>
Patchwork: https://patchwork.linux-mips.org/patch/6643/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
I noticed KVM is broken when KVM in-kernel XICS emulation
(CONFIG_KVM_XICS) is disabled.
The problem was introduced in 48eaef05 (KVM: PPC: Book3S HV: use
xics_wake_cpu only when defined). It used CONFIG_KVM_XICS to wrap
xics_wake_cpu, where CONFIG_PPC_ICP_NATIVE should have been
used.
Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Paul Mackerras <paulus@samba.org> Acked-by: Scott Wood <scottwood@freescale.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Previously a reserved instruction exception while in guest code would
cause a KVM internal error if kvm_mips_handle_ri() didn't recognise the
instruction (including a RDHWR from an unrecognised hardware register).
However the guest OS should really have the opportunity to catch the
exception so that it can take the appropriate actions such as sending a
SIGILL to the guest user process or emulating the instruction itself.
Therefore in these cases emulate a guest RI exception and only return
EMULATE_FAIL if that fails, being careful to revert the PC first in case
the exception occurred in a branch delay slot in which case the PC will
already point to the branch target.
Also turn the printk messages relating to these cases into kvm_debug
messages so that they aren't usually visible.
This allows crashme to run in the guest without killing the entire VM.
QE reported that they got the BUG_ON in ioapic_service to trigger.
I cannot reproduce it, but there are two reasons why this could happen.
The less likely but also easiest one, is when kvm_irq_delivery_to_apic
does not deliver to any APIC and returns -1.
Because irqe.shorthand == 0, the kvm_for_each_vcpu loop in that
function is never reached. However, you can target the similar loop in
kvm_irq_delivery_to_apic_fast; just program a zero logical destination
address into the IOAPIC, or an out-of-range physical destination address.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
get_user_pages(mm) is simply wrong if mm->mm_users == 0 and exit_mmap/etc
was already called (or is in progress), mm->mm_count can only pin mm->pgd
and mm_struct itself.
Change kvm_setup_async_pf/async_pf_execute to inc/dec mm->mm_users.
kvm_create_vm/kvm_destroy_vm play with ->mm_count too but this case looks
fine at first glance, it seems that this ->mm is only used to verify that
current->mm == kvm->mm.
The kvm/mmu code shared by arm and arm64 uses kalloc() to allocate
a bounce page (if hypervisor init code crosses page boundary) and
hypervisor PGDs. The problem is that kalloc() does not guarantee
the proper alignment. In the case of the bounce page, the page sized
buffer allocated may also cross a page boundary negating the purpose
and leading to a hang during kvm initialization. Likewise the PGDs
allocated may not meet the minimum alignment requirements of the
underlying MMU. This patch uses __get_free_page() to guarantee the
worst case alignment needs of the bounce page and PGDs on both arm
and arm64.
Signed-off-by: Mark Salter <msalter@redhat.com> Acked-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When dispatch SGI(mode == 0), that is the vcpu of VM should send
sgi to the cpu which the target_cpus list.
So, there must add the "break" to branch of case 0.
Signed-off-by: Haibin Wang <wanghaibin.wang@huawei.com> Acked-by: Marc Zyngier <marc.zyngier@arm.com> Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Do not leak kernel-only floppy_raw_cmd structure members to userspace.
This includes the linked-list pointer and the pointer to the allocated
DMA space.
Always clear out these floppy_raw_cmd struct members after copying the
entire structure from userspace so that the in-kernel version is always
valid and never left in an interdeterminate state.
The tty atomic_write_lock does not provide an exclusion guarantee for
the tty driver if the termios settings are LECHO & !OPOST. And since
it is unexpected and not allowed to call TTY buffer helpers like
tty_insert_flip_string concurrently, this may lead to crashes when
concurrect writers call pty_write. In that case the following two
writers:
* the ECHOing from a workqueue and
* pty_write from the process
race and can overflow the corresponding TTY buffer like follows.
If we look into tty_insert_flip_string_fixed_flag, there is:
int space = __tty_buffer_request_room(port, goal, flags);
struct tty_buffer *tb = port->buf.tail;
...
memcpy(char_buf_ptr(tb, tb->used), chars, space);
...
tb->used += space;
so the race of the two can result in something like this:
A B
__tty_buffer_request_room
__tty_buffer_request_room
memcpy(buf(tb->used), ...)
tb->used += space;
memcpy(buf(tb->used), ...) ->BOOM
B's memcpy is past the tty_buffer due to the previous A's tb->used
increment.
Since the N_TTY line discipline input processing can output
concurrently with a tty write, obtain the N_TTY ldisc output_lock to
serialize echo output with normal tty writes. This ensures the tty
buffer helper tty_insert_flip_string is not called concurrently and
everything is fine.
Note that this is nicely reproducible by an ordinary user using
forkpty and some setup around that (raw termios + ECHO). And it is
present in kernels at least after commit d945cb9cce20ac7143c2de8d88b187f62db99bdc (pty: Rework the pty layer to
use the normal buffering logic) in 2.6.31-rc3.
js: add more info to the commit log
js: switch to bool
js: lock unconditionally
js: lock only the tty->ops->write call
Commit 6a20dbd6caa2358716136144bf524331d70b1e03,
"tty: Fix race condition between __tty_buffer_request_room and flush_to_ldisc"
correctly identifies an unsafe race condition between
__tty_buffer_request_room() and flush_to_ldisc(), where the consumer
flush_to_ldisc() prematurely advances the head before consuming the
last of the data committed. For example:
CPU 0 | CPU 1
__tty_buffer_request_room | flush_to_ldisc
... | ...
| count = head->commit - head->read
n = tty_buffer_alloc() |
b->commit = b->used |
b->next = n |
| if (!count) /* T */
| if (head->next == NULL) /* F */
| buf->head = head->next
In this case, buf->head has been advanced but head->commit may have
been updated with a new value.
Instead of reintroducing an unnecessary lock, fix the race locklessly.
Read the commit-next pair in the reverse order of writing, which guarantees
the commit value read is the latest value written if the head is
advancing.
The sleep function was updated to put the serial port to sleep only when necessary.
This appears to resolve the errant behavior of the driver as described in
Kernel Bug 61961 – "My Exar Corp. XR17C/D152 Dual PCI UART modem does not
work with 3.8.0".
Signed-off-by: Michael Welling <mwelling@ieee.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When 'console=hvc0' is specified to the kernel parameter in x86 KVM guest,
hvc console is setup within a kthread. However, that will cause SEGV
and the boot will fail when the driver is builtin to the kernel,
because currently hvc_console_setup() is annotated with '__init'. This
patch removes '__init' to boot the guest successfully with 'console=hvc0'.
In "ARM: dts: am33xx: correcting dt node unit address for usb", the
usb_ctrl_mod and cppi41dma nodes were updated with the correct register
addresses. However, the dts files that reference these nodes were not
updated, and those devices are no longer being enabled.
This patch corrects the references for the affected dts files.
Signed-off-by: Leigh Brown <leigh@solinno.co.uk> Signed-off-by: Tony Lindgren <tony@atomide.com> Cc: Johan Hovold <jhovold@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To avoid memory fetch underflows with larger USB transfers, Tegra SoCs
need txfill_tuning's txfifothresh register field set to a non-default
value. Add a custom reset override in order to set this up.
These values are recommended practice for all Tegra chips. However,
I've only noticed practical problems when not setting them this way on
systems using Tegra124. Hence, CC: stable only for recent kernels which
actually support Tegra124.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Acked-by: Alan Stern <stern@rowland.harvard.edu> Tested-by: Alexandre Courbot <acourbot@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Tegra124 clock DT binding currently provides 3 clocks that don't
actually exist; 2 for NAND and one for UART5/UARTE. Delete these. While
this is technically an incompatible DT ABI change, nothing could have
used these clock IDs for anything practical, since the HW doesn't exist.
Tegra124 only has 4 UARTs. Parts of the documentation hint at a fifth
UART, but this appears to be left-over from earlier SoC documentation.
Remove the non-existent DT node for UART5.
We have observed a rare cycle state desync bug after Set TR Dequeue
Pointer commands on Intel LynxPoint xHCs (resulting in an endpoint that
doesn't fetch new TRBs and thus an unresponsive USB device). It always
triggers when a previous Set TR Dequeue Pointer command has set the
pointer to the final Link TRB of a segment, and then another URB gets
enqueued and cancelled again before it can be completed. Further
investigation showed that the xHC had returned the Link TRB in the TRB
Pointer field of the Transfer Event (CC == Stopped -- Length Invalid),
but when xhci_find_new_dequeue_state() later accesses the Endpoint
Context's TR Dequeue Pointer field it is set to the first TRB of the
next segment.
The driver expects those two values to be the same in this situation,
and uses the cycle state of the latter together with the address of the
former. This should be fine according to the XHCI specification, since
the endpoint ring should be stopped when returning the Transfer Event
and thus should not advance over the Link TRB before it gets restarted.
However, real-world XHCI implementations apparently don't really care
that much about these details, so the driver should follow a more
defensive approach to try to work around HC spec violations.
This patch removes the stopped_trb variable that had been used to store
the TRB Pointer from the last Transfer Event of a stopped TRB. Instead,
xhci_find_new_dequeue_state() now relies only on the Endpoint Context,
requiring a small amount of additional processing to find the virtual
address corresponding to the TR Dequeue Pointer. Some other parts of the
function were slightly rearranged to better fit into this model.
This patch should be backported to kernels as old as 2.6.31 that contain
the commit ae636747146ea97efa18e04576acd3416e2514f5 "USB: xhci: URB
cancellation support."
Signed-off-by: Julius Werner <jwerner@chromium.org> Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The function ext4_update_i_disksize() is used in only one place, in
the function mpage_map_and_submit_extent(). Move its code to simplify
the code paths, and also move the call to ext4_mark_inode_dirty() into
the i_data_sem's critical region, to be consistent with all of the
other places where we update i_disksize. That way, we also keep the
raw_inode's i_disksize protected, to avoid the following race:
CPU #1 CPU #2
down_write(&i_data_sem)
Modify i_disk_size
up_write(&i_data_sem)
down_write(&i_data_sem)
Modify i_disk_size
Copy i_disk_size to on-disk inode
up_write(&i_data_sem)
Copy i_disk_size to on-disk inode
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The reason for the warning is that buffer_head passed into
jbd2_journal_dirty_metadata() didn't have journal_head attached. This is
caused by the following race of two ext4_xattr_release_block() calls:
We fix the problem by moving ext4_handle_dirty_xattr_block() under the
buffer lock. Sadly this cannot be done in nojournal mode as that
function can call sync_dirty_buffer() which would deadlock. Luckily in
nojournal mode the race is harmless (we only dirty already freed buffer)
and thus for nojournal mode we leave the dirtying outside of the buffer
lock.
Reported-by: Sage Weil <sage@inktank.com> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ext4_end_bio() currently throws away the error that it receives. Chances
are this is part of a spate of errors, one of which will end up getting
the error returned to userspace somehow, but we shouldn't take that risk.
Also print out the errno to aid in debug.
Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Reviewed-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we try to get 2^32-1 block of the file which has the extent
(ee_block=2^32-2, ee_len=1) with FIBMAP ioctl, it causes BUG_ON
in ext4_ext_put_gap_in_cache().
To avoid the problem, ext4_map_blocks() needs to check the file logical block
number. ext4_ext_put_gap_in_cache() called via ext4_map_blocks() cannot
handle 2^32-1 because the maximum file logical block number is 2^32-2.
Note that ext4_ind_map_blocks() returns -EIO when the block number is invalid.
So ext4_map_blocks() should also return the same errno.
Commit 4550dd6c6b062 introduced for_each_bvec() which iterates over each
bvec attached to a bio or bip. However, the macro fails to check bi_size
before dereferencing which can lead to crashes while counting/mapping
integrity scatterlist segments.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Cc: Kent Overstreet <kmo@daterainc.com> Cc: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Jens Axboe <axboe@fb.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current mainline has copies propagated to *all* nodes, then
tears down the copies we made for nodes that do not contain
counterparts of the desired mountpoint. That sets the right
propagation graph for the copies (at teardown time we move
the slaves of removed node to a surviving peer or directly
to master), but we end up paying a fairly steep price in
useless allocations. It's fairly easy to create a situation
where N calls of mount(2) create exactly N bindings, with
O(N^2) vfsmounts allocated and freed in process.
Fortunately, it is possible to avoid those allocations/freeings.
The trick is to create copies in the right order and find which
one would've eventually become a master with the current algorithm.
It turns out to be possible in O(nodes getting propagation) time
and with no extra allocations at all.
One part is that we need to make sure that eventual master will be
created before its slaves, so we need to walk the propagation
tree in a different order - by peer groups. And iterate through
the peers before dealing with the next group.
Another thing is finding the (earlier) copy that will be a master
of one we are about to create; to do that we are (temporary) marking
the masters of mountpoints we are attaching the copies to.
Either we are in a peer of the last mountpoint we'd dealt with,
or we have the following situation: we are attaching to mountpoint M,
the last copy S_0 had been attached to M_0 and there are sequences
S_0...S_n, M_0...M_n such that S_{i+1} is a master of S_{i},
S_{i} mounted on M{i} and we need to create a slave of the first S_{k}
such that M is getting propagation from M_{k}. It means that the master
of M_{k} will be among the sequence of masters of M. On the
other hand, the nearest marked node in that sequence will either
be the master of M_{k} or the master of M_{k-1} (the latter -
in the case if M_{k-1} is a slave of something M gets propagation
from, but in a wrong peer group).
So we go through the sequence of masters of M until we find
a marked one (P). Let N be the one before it. Then we go through
the sequence of masters of S_0 until we find one (say, S) mounted
on a node D that has P as master and check if D is a peer of N.
If it is, S will be the master of new copy, if not - the master of S
will be.
That's it for the hard part; the rest is fairly simple. Iterator
is in next_group(), handling of one prospective mountpoint is
propagate_one().
It seems to survive all tests and gives a noticably better performance
than the current mainline for setups that are seriously using shared
subtrees.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If parent device does not have of_node set the s2mps11_clk_parse_dt()
returned NULL. This NULL was later passed to of_clk_add_provider() which
dereferenced it in pr_debug() call.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com> Signed-off-by: Mike Turquette <mturquette@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 9548906b2bb7 ('xattr: Constify ->name member of "struct xattr"')
missed that ocfs2 is calling kfree(xattr->name). As a result, kernel
panic occurs upon calling kfree(xattr->name) because xattr->name refers
static constant names. This patch removes kfree(xattr->name) from
ocfs2_mknod() and ocfs2_symlink().
Do not put bh when buffer_uptodate failed in ocfs2_write_block and
ocfs2_write_super_or_backup, because it will put bh in b_end_io.
Otherwise it will hit a warning "VFS: brelse: Trying to free free
buffer".
Signed-off-by: Alex Chen <alex.chen@huawei.com> Reviewed-by: Joseph Qi <joseph.qi@huawei.com> Reviewed-by: Srinivas Eeda <srinivas.eeda@oracle.com> Cc: Mark Fasheh <mfasheh@suse.com> Acked-by: Joel Becker <jlbec@evilplan.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a race window in dlm_do_recovery() between dlm_remaster_locks()
and dlm_reset_recovery() when the recovery master nearly finish the
recovery process for a dead node. After the master sends FINALIZE_RECO
message in dlm_remaster_locks(), another node may become the recovery
master for another dead node, and then send the BEGIN_RECO message to
all the nodes included the old master, in the handler of this message
dlm_begin_reco_handler() of old master, dlm->reco.dead_node and
dlm->reco.new_master will be set to the second dead node and the new
master, then in dlm_reset_recovery(), these two variables will be reset
to default value. This will cause new recovery master can not finish
the recovery process and hung, at last the whole cluster will hung for
recovery.
old recovery master: new recovery master:
dlm_remaster_locks()
become recovery master for
another dead node.
dlm_send_begin_reco_message()
dlm_begin_reco_handler()
{
if (dlm->reco.state & DLM_RECO_STATE_FINALIZE) {
return -EAGAIN;
}
dlm_set_reco_master(dlm, br->node_idx);
dlm_set_reco_dead_node(dlm, br->dead_node);
}
dlm_reset_recovery()
{
dlm_set_reco_dead_node(dlm, O2NM_INVALID_NODE_NUM);
dlm_set_reco_master(dlm, O2NM_INVALID_NODE_NUM);
}
will hang in dlm_remaster_locks() for
request dlm locks info
Before send FINALIZE_RECO message, recovery master should set
DLM_RECO_STATE_FINALIZE for itself and clear it after the recovery done,
this can break the race windows as the BEGIN_RECO messages will not be
handled before DLM_RECO_STATE_FINALIZE flag is cleared.
A similar race may happen between new recovery master and normal node
which is in dlm_finalize_reco_handler(), also fix it.
Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com> Reviewed-by: Srinivas Eeda <srinivas.eeda@oracle.com> Reviewed-by: Wengang Wang <wen.gang.wang@oracle.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Mark Fasheh <mfasheh@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This issue was introduced by commit 800deef3f6f8 ("ocfs2: use
list_for_each_entry where benefical") in 2007 where it replaced
list_for_each with list_for_each_entry. The variable "lock" will point
to invalid data if "tmpq" list is empty and a panic will be triggered
due to this. Sunil advised reverting it back, but the old version was
also not right. At the end of the outer for loop, that
list_for_each_entry will also set "lock" to an invalid data, then in the
next loop, if the "tmpq" list is empty, "lock" will be an stale invalid
data and cause the panic. So reverting the list_for_each back and reset
"lock" to NULL to fix this issue.
Another concern is that this seemes can not happen because the "tmpq"
list should not be empty. Let me describe how.
old lock resource owner(node 1): migratation target(node 2):
image there's lockres with a EX lock from node 2 in
granted list, a NR lock from node x with convert_type
EX in converting list.
dlm_empty_lockres() {
dlm_pick_migration_target() {
pick node 2 as target as its lock is the first one
in granted list.
}
dlm_migrate_lockres() {
dlm_mark_lockres_migrating() {
res->state |= DLM_LOCK_RES_BLOCK_DIRTY;
wait_event(dlm->ast_wq, !dlm_lockres_is_dirty(dlm, res));
//after the above code, we can not dirty lockres any more,
// so dlm_thread shuffle list will not run
downconvert lock from EX to NR
upconvert lock from NR to EX
<<< migration may schedule out here, then
<<< node 2 send down convert request to convert type from EX to
<<< NR, then send up convert request to convert type from NR to
<<< EX, at this time, lockres granted list is empty, and two locks
<<< in the converting list, node x up convert lock followed by
<<< node 2 up convert lock.
// will set lockres RES_MIGRATING flag, the following
// lock/unlock can not run
dlm_lockres_release_ast(dlm, res);
}
dlm_send_one_lockres()
dlm_process_recovery_data()
for (i=0; i<mres->num_locks; i++)
if (ml->node == dlm->node_num)
for (j = DLM_GRANTED_LIST; j <= DLM_BLOCKED_LIST; j++) {
list_for_each_entry(lock, tmpq, list)
if (lock) break; <<< lock is invalid as grant list is empty.
}
if (lock->ml.node != ml->node)
BUG() >>> crash here
}
I see the above locks status from a vmcore of our internal bug.
Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com> Reviewed-by: Wengang Wang <wen.gang.wang@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Reviewed-by: Srinivas Eeda <srinivas.eeda@oracle.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Mark Fasheh <mfasheh@suse.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As sysctl_hung_task_timeout_sec is unsigned long, when this value is
larger then LONG_MAX/HZ, the function schedule_timeout_interruptible in
watchdog will return immediately without sleep and with print :
When I decrease the value of nr_hugepage in procfs a lot, softlockup
happens. It is because there is no chance of context switch during this
process.
On the other hand, when I allocate a large number of hugepages, there is
some chance of context switch. Hence softlockup doesn't happen during
this process. So it's necessary to add the context switch in the
freeing process as same as allocating process to avoid softlockup.
When I freed 12 TB hugapages with kernel-2.6.32-358.el6, the freeing
process occupied a CPU over 150 seconds and following softlockup message
appeared twice or more.
BUG: soft lockup - CPU#16 stuck for 67s! [sh:12883] ...
Pid: 12883, comm: sh Not tainted 2.6.32-358.el6.x86_64 #1
Call Trace:
free_pool_huge_page+0xb8/0xd0
set_max_huge_pages+0x128/0x190
hugetlb_sysctl_handler_common+0x113/0x140
hugetlb_sysctl_handler+0x1e/0x20
proc_sys_call_handler+0x97/0xd0
proc_sys_write+0x14/0x20
vfs_write+0xb8/0x1a0
sys_write+0x51/0x90
__audit_syscall_exit+0x265/0x290
system_call_fastpath+0x16/0x1b
I have not confirmed this problem with upstream kernels because I am not
able to prepare the machine equipped with 12TB memory now. However I
confirmed that the amount of decreasing hugepages was directly
proportional to the amount of required time.
I measured required times on a smaller machine. It showed 130-145
hugepages decreased in a millisecond.
A BUG_ON(!PageLocked) was triggered in mlock_vma_page() by Sasha Levin
fuzzing with trinity. The call site try_to_unmap_cluster() does not lock
the pages other than its check_page parameter (which is already locked).
The BUG_ON in mlock_vma_page() is not documented and its purpose is
somewhat unclear, but apparently it serializes against page migration,
which could otherwise fail to transfer the PG_mlocked flag. This would
not be fatal, as the page would be eventually encountered again, but
NR_MLOCK accounting would become distorted nevertheless. This patch adds
a comment to the BUG_ON in mlock_vma_page() and munlock_vma_page() to that
effect.
The call site try_to_unmap_cluster() is fixed so that for page !=
check_page, trylock_page() is attempted (to avoid possible deadlocks as we
already have check_page locked) and mlock_vma_page() is performed only
upon success. If the page lock cannot be obtained, the page is left
without PG_mlocked, which is again not a problem in the whole unevictable
memory design.
Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Signed-off-by: Bob Liu <bob.liu@oracle.com> Reported-by: Sasha Levin <sasha.levin@oracle.com> Cc: Wanpeng Li <liwanp@linux.vnet.ibm.com> Cc: Michel Lespinasse <walken@google.com> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: David Rientjes <rientjes@google.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Hugh Dickins <hughd@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On NUMA systems, a node may start thrashing cache or even swap anonymous
pages while there are still free pages on remote nodes.
This is a result of commits 81c0a2bb515f ("mm: page_alloc: fair zone
allocator policy") and fff4068cba48 ("mm: page_alloc: revert NUMA aspect
of fair allocation policy").
Before those changes, the allocator would first try all allowed zones,
including those on remote nodes, before waking any kswapds. But now,
the allocator fastpath doubles as the fairness pass, which in turn can
only consider the local node to prevent remote spilling based on
exhausted fairness batches alone. Remote nodes are only considered in
the slowpath, after the kswapds are woken up. But if remote nodes still
have free memory, kswapd should not be woken to rebalance the local node
or it may thrash cash or swap prematurely.
Fix this by adding one more unfair pass over the zonelist that is
allowed to spill to remote nodes after the local fairness pass fails but
before entering the slowpath and waking the kswapds.
This also gets rid of the GFP_THISNODE exemption from the fairness
protocol because the unfair pass is no longer tied to kswapd, which
GFP_THISNODE is not allowed to wake up.
However, because remote spills can be more frequent now - we prefer them
over local kswapd reclaim - the allocation batches on remote nodes could
underflow more heavily. When resetting the batches, use
atomic_long_read() directly instead of zone_page_state() to calculate the
delta as the latter filters negative counter values.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Rik van Riel <riel@redhat.com> Acked-by: Mel Gorman <mgorman@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/sh/kernel/dumpstack.c: In function 'print_trace_address':
arch/sh/kernel/dumpstack.c:118:2: error: format not a string literal and no format arguments [-Werror=format-security]
Use the "%s" format so that it's impossible to interpret 'data' as a
format string.
Signed-off-by: Matt Fleming <matt.fleming@intel.com> Reported-by: Kees Cook <keescook@chromium.org> Acked-by: Kees Cook <keescook@chromium.org> Cc: Paul Mundt <lethal@linux-sh.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch fixes a bug where outstanding RDMA_READs with WRITE_PENDING
status require an extra target_put_sess_cmd() in isert_put_cmd() code
when called from isert_cq_tx_comp_err() + isert_cq_drain_comp_llist()
context during session shutdown.
The extra kref PUT is required so that transport_generic_free_cmd()
invokes the last target_put_sess_cmd() -> target_release_cmd_kref(),
which will complete(&se_cmd->cmd_wait_comp) the outstanding se_cmd
descriptor with WRITE_PENDING status, and awake the completion in
target_wait_for_sess_cmds() to invoke TFO->release_cmd().
The bug was manifesting itself in target_wait_for_sess_cmds() where
a se_cmd descriptor with WRITE_PENDING status would end up sleeping
indefinately.
Acked-by: Sagi Grimberg <sagig@mellanox.com> Cc: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch changes isert_conn_create_fastreg_pool() to follow
logic in iscsi_target_locate_portal() for determining how many
FRMR descriptors to allocate based upon the number of possible
per-session command slots that are available.
This addresses an OOPs in isert_reg_rdma() where due to the
use of ISCSI_DEF_XMIT_CMDS_MAX could end up returning a bogus
fast_reg_descriptor when the number of active tags exceeded
the original hardcoded max.
Note this also includes moving isert_conn_create_fastreg_pool()
from isert_connect_request() to isert_put_login_tx() before
posting the final Login Response PDU in order to determine the
se_nacl->queue_depth (eg: number of tags) per session the target
will be enforcing.
v2 changes:
- Move isert_conn->conn_fr_pool list_head init into
isert_conn_request()
v3 changes:
- Drop unnecessary list_empty() check in isert_reg_rdma()
(Sagi)
Cc: Sagi Grimberg <sagig@mellanox.com> Cc: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch fixes 2 issues in the fast completion path:
1) Possible double completions / double dma_unmap_sg() calls due to lack
of atomicity in the check and subsequent dereference of the upper layer
callback function. Fixed with cmpxchg before unmap and callback.
2) Regression in unaligned IO constraining workaround for p420m devices.
Fixed by checking if IO is unaligned and using proper semaphore if so.
When a driver doesn't have pre_reset, post_reset, or reset_resume
methods, the USB core unbinds that driver when its device undergoes a
reset or a reset-resume, and then rebinds it afterward.
The existing straightforward implementation can lead to problems,
because each interface gets unbound and rebound before the next
interface is handled. If a driver claims additional interfaces, the
claim may fail because the old binding instance may still own the
additional interface when the new instance tries to claim it.
This patch fixes the problem by first unbinding all the interfaces
that are marked (i.e., their needs_binding flag is set) and then
rebinding all of them.
The patch also makes the helper functions in driver.c a little more
uniform and adjusts some out-of-date comments.
Tests have shown that when a power-up transition is followed by other
PHY operations too quickly, the USB port appears dead. Waiting 1ms fixes
this problem.
Signed-off-by: Daniel Mack <zonque@gmail.com> Signed-off-by: Felipe Balbi <balbi@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>