lock_page_memcg()/unlock_page_memcg() use spin_lock_irqsave/restore() if
the page's memcg is undergoing move accounting, which occurs when a
process leaves its memcg for a new one that has
memory.move_charge_at_immigrate set.
unlocked_inode_to_wb_begin,end() use spin_lock_irq/spin_unlock_irq() if
the given inode is switching writeback domains. Switches occur when
enough writes are issued from a new domain.
This existing pattern is thus suspicious:
lock_page_memcg(page);
unlocked_inode_to_wb_begin(inode, &locked);
...
unlocked_inode_to_wb_end(inode, locked);
unlock_page_memcg(page);
If both inode switch and process memcg migration are both in-flight then
unlocked_inode_to_wb_end() will unconditionally enable interrupts while
still holding the lock_page_memcg() irq spinlock. This suggests the
possibility of deadlock if an interrupt occurs before unlock_page_memcg().
Due to configuration limitations this deadlock is not currently possible
because we don't mix cgroup writeback (a cgroupv2 feature) and
memory.move_charge_at_immigrate (a cgroupv1 feature).
If the kernel is hacked to always claim inode switching and memcg
moving_account, then this script triggers lockup in less than a minute:
cd /mnt/cgroup/memory
mkdir a b
echo 1 > a/memory.move_charge_at_immigrate
echo 1 > b/memory.move_charge_at_immigrate
(
echo $BASHPID > a/cgroup.procs
while true; do
dd if=/dev/zero of=/mnt/big bs=1M count=256
done
) &
while true; do
sync
done &
sleep 1h &
SLEEP=$!
while true; do
echo $SLEEP > a/cgroup.procs
echo $SLEEP > b/cgroup.procs
done
The deadlock does not seem possible, so it's debatable if there's any
reason to modify the kernel. I suggest we should to prevent future
surprises. And Wang Long said "this deadlock occurs three times in our
environment", so there's more reason to apply this, even to stable.
Stable 4.4 has minor conflicts applying this patch. For a clean 4.4 patch
see "[PATCH for-4.4] writeback: safer lock nesting"
https://lkml.org/lkml/2018/4/11/146
Wang Long said "this deadlock occurs three times in our environment"
[gthelen@google.com: v4] Link: http://lkml.kernel.org/r/20180411084653.254724-1-gthelen@google.com
[akpm@linux-foundation.org: comment tweaks, struct initialization simplification]
Change-Id: Ibb773e8045852978f6207074491d262f1b3fb613 Link: http://lkml.kernel.org/r/20180410005908.167976-1-gthelen@google.com Fixes: 682aa8e1a6a1 ("writeback: implement unlocked_inode_to_wb transaction and use it for stat updates") Signed-off-by: Greg Thelen <gthelen@google.com> Reported-by: Wang Long <wanglong19@meituan.com> Acked-by: Wang Long <wanglong19@meituan.com> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Andrew Morton <akpm@linux-foundation.org> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Tejun Heo <tj@kernel.org> Cc: Nicholas Piggin <npiggin@gmail.com> Cc: <stable@vger.kernel.org> [v4.2+] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[natechancellor: Adjust context due to lack of b93b016313b3b] Signed-off-by: Nathan Chancellor <natechancellor@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
i2c_hid_command() returns non-zero in error cases (the actual
errno). Error handling in for I2C_HID_QUIRK_RESEND_REPORT_DESCR
case in i2c_hid_resume() had the check inverted; fix that.
Fixes: 3e83eda467 ("HID: i2c-hid: Fix resume issue on Raydium touchscreen device") Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Cc: Aaron Ma <aaron.ma@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is one issue relates to Coarse Power Gating(CPG) on KBL NUC in GVT-g,
vgpu can't get the correct default context by updating the registers before
inhibit context submission. It always get back the hardware default value
unless the inhibit context submission happened before the 1st time
forcewake put. With this wrong default context, vgpu will run with
incorrect state and meet unknown issues.
The solution is initialize these mmios by adding lri command in ring buffer
of the inhibit context, then gpu hardware has no chance to go down RC6 when
lri commands are right being executed, and then vgpu can get correct
default context for further use.
v3:
- fix code fault, use 'for' to loop through mmio render list(Zhenyu)
v4:
- save the count of engine mmio need to be restored for inhibit context and
refine some comments. (Kevin)
v5:
- code rebase
Cc: Kevin Tian <kevin.tian@intel.com> Cc: Zhenyu Wang <zhenyuw@linux.intel.com> Signed-off-by: Weinan Li <weinan.z.li@intel.com> Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com> Signed-off-by: Changbin Du <changbin.du@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
f2fs specifies the __GFP_ZERO flag for allocating some of its pages.
Unfortunately, the page cache also uses the mapping's GFP flags for
allocating radix tree nodes. It always masked off the __GFP_HIGHMEM
flag, and masks off __GFP_ZERO in some paths, but not all. That causes
radix tree nodes to be allocated with a NULL list_head, which causes
backtraces like:
The __GFP_DMA and __GFP_DMA32 flags would also be able to sneak through
if they are ever used. Fix them all by using GFP_RECLAIM_MASK at the
innermost location, and remove it from earlier in the callchain.
Link: http://lkml.kernel.org/r/20180411060320.14458-2-willy@infradead.org Fixes: 449dd6984d0e ("mm: keep page cache radix tree nodes in check") Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com> Reported-by: Chris Fries <cfries@google.com> Debugged-by: Minchan Kim <minchan@kernel.org> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Acked-by: Michal Hocko <mhocko@suse.com> Reviewed-by: Jan Kara <jack@suse.cz> Cc: <stable@vger.kernel.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>
The autofs file system mkdir inode operation blindly sets the created
directory mode to S_IFDIR | 0555, ingoring the passed in mode, which can
cause selinux dac_override denials.
But the function also checks if the caller is the daemon (as no-one else
should be able to do anything here) so there's no point in not honouring
the passed in mode, allowing the daemon to set appropriate mode when
required.
Link: http://lkml.kernel.org/r/152361593601.8051.14014139124905996173.stgit@pluto.themaw.net Signed-off-by: Ian Kent <raven@themaw.net> Cc: <stable@vger.kernel.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>
MAP_SYNC is a nop for device-dax. Allow MAP_SYNC to succeed on device-dax
to eliminate special casing between device-dax and fs-dax as to when the
flag can be specified. Device-dax users already implicitly assume that they do
not need to call fsync(), and this enables them to explicitly check for this
capability.
Cc: <stable@vger.kernel.org> Fixes: b6fb293f2497 ("mm: Define MAP_SYNC and VM_SYNC flags") Signed-off-by: Dave Jiang <dave.jiang@intel.com> Reviewed-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The new support for the standard _LSR and _LSW methods neglected to also
update the nvdimm_init_config_data() and nvdimm_set_config_data() to
return the translated error code from failed commands. This precision is
necessary because the locked status that was previously returned on
ND_CMD_GET_CONFIG_SIZE commands is now returned on
ND_CMD_{GET,SET}_CONFIG_DATA commands.
If the kernel misses this indication it can inadvertently fall back to
label-less mode when it should otherwise avoid all access to locked
regions.
Cc: <stable@vger.kernel.org> Fixes: 4b27db7e26cd ("acpi, nfit: add support for the _LSI, _LSR, and...") Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We want it only for the stuff created by SB_KERNMOUNT mounts, *not* for
their copies. As it is, creating a deep stack of bindings of /proc/*/ns/*
somewhere in a new namespace and exiting yields a stack overflow.
Cc: stable@kernel.org Reported-by: Alexander Aring <aring@mojatatu.com> Bisected-by: Kirill Tkhai <ktkhai@virtuozzo.com> Tested-by: Kirill Tkhai <ktkhai@virtuozzo.com> Tested-by: Alexander Aring <aring@mojatatu.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
if we ever hit rpc_gssd_dummy_depopulate() dentry passed to
it has refcount equal to 1. __rpc_rmpipe() drops it and
dput() done after that hits an already freed dentry.
Cc: stable@kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Turns out the VLV/CHV fixed function sprite CSC expects full range
data as input. We've been feeding it limited range data to it all
along. To expand the data out to full range we'll use the color
correction registers (brightness, contrast, and saturation).
On CHV pipe B we were actually doing the right thing already because we
progammed the custom CSC matrix to do expect limited range input. Now
that well pre-expand the data out with the color correction unit, we
need to change the CSC matrix to operate with full range input instead.
This should make the sprite output of the other pipes match the sprite
output of pipe B reasonably well. Looking at the resulting pipe CRCs,
there can be a slight difference in the output, but as I don't know
the formula used by the fixed function CSC of the other pipes, I don't
think it's worth the effort to try to match the output exactly. It
might not even be possible due to difference in internal precision etc.
One slight caveat here is that the color correction registers are single
bufferred, so we should really be updating them during vblank, but we
still don't have a mechanism for that, so just toss in another FIXME.
v2: Rebase
v3: s/bri/brightness/ s/con/contrast/ (Shashank)
v4: Clarify the constants and math (Shashank)
Cc: Harry Wentland <harry.wentland@amd.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Daniel Stone <daniel@fooishbar.org> Cc: Russell King - ARM Linux <linux@armlinux.org.uk> Cc: Ilia Mirkin <imirkin@alum.mit.edu> Cc: Hans Verkuil <hverkuil@xs4all.nl> Cc: Shashank Sharma <shashank.sharma@intel.com> Cc: Uma Shankar <uma.shankar@intel.com> Cc: Jyri Sarha <jsarha@ti.com> Cc: "Tang, Jun" <jun.tang@intel.com> Reported-by: "Tang, Jun" <jun.tang@intel.com> Cc: stable@vger.kernel.org Fixes: 7f1f3851feb0 ("drm/i915: sprite support for ValleyView v4") Reviewed-by: Shashank Sharma <shashank.sharma@intel.com> Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180214192327.3250-5-ville.syrjala@linux.intel.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drm/i915/gen9+: Set same power state before hibernation image
save/restore
during hibernation/suspend the power domain functionality got disabled,
after which resume could leave it incorrectly disabled if the ACPI
target state was S0 during suspend and i915 was not loaded by the loader
kernel.
This was caused by not considering if we resumed from hibernation as the
condition for power domains reiniting.
Fix this by simply tracking if we suspended power domains during system
suspend and reinit power domains accordingly during resume. This will
result in reiniting power domains always when resuming from hibernation,
regardless of the platform and whether or not i915 is loaded by the
loader kernel.
The reason we didn't catch this earlier is that the enabled/disabled
state of power domains during PMSG_FREEZE/PMSG_QUIESCE is platform
and kernel config dependent: on my SKL the target state is S4
during PMSG_FREEZE and (with the driver loaded in the loader kernel)
S0 during PMSG_QUIESCE. On the reporter's machine it's S0 during
PMSG_FREEZE but (contrary to this) power domains are not initialized
during PMSG_QUIESCE since i915 is not loaded in the loader kernel, or
it's loaded but without the DMC firmware being available.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105196 Reported-and-tested-by: amn-bas@hotmail.com Fixes: dd9f31c7a388 ("drm/i915/gen9+: Set same power state before hibernation image save/restore") Cc: amn-bas@hotmail.com Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: <stable@vger.kernel.org> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20180322143642.26883-1-imre.deak@intel.com
(cherry picked from commit 0f90603c33bdf6575cfdc81edd53f3f13ba166fb) Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit c31165d7400b ("mmc: sdhci-pci: Add support for HS200 tuning mode
on AMD, eMMC-4.5.1") added a HS200 tuning method for use with AMD SDHCI
controllers. As described in the commit subject, this tuning is specific
for HS200. However, as implemented, this method is used for all host
timings, because platform_execute_tuning, if it exists, is called
unconditionally by sdhci_execute_tuning(). This breaks tuning when using
the AMD controller with, for example, a DDR50 SD card.
Instead, we can implement an amd execute_tuning wrapper callback, and
then conditionally do the HS200 specific tuning for HS200, and otherwise
call back to the standard sdhci_execute_tuning().
Signed-off-by: Daniel Kurtz <djkurtz@chromium.org> Acked-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com> Acked-by: Adrian Hunter <adrian.hunter@intel.com> Fixes: c31165d7400b ("mmc: sdhci-pci: Add support for HS200 tuning mode on AMD, eMMC-4.5.1") Cc: stable@vger.kernel.org # v4.11+ Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When event on child inodes are sent to the parent inode mark and
parent inode mark was not marked with FAN_EVENT_ON_CHILD, the event
will not be delivered to the listener process. However, if the same
process also has a mount mark, the event to the parent inode will be
delivered regadless of the mount mark mask.
This behavior is incorrect in the case where the mount mark mask does
not contain the specific event type. For example, the process adds
a mark on a directory with mask FAN_MODIFY (without FAN_EVENT_ON_CHILD)
and a mount mark with mask FAN_CLOSE_NOWRITE (without FAN_ONDIR).
A modify event on a file inside that directory (and inside that mount)
should not create a FAN_MODIFY event, because neither of the marks
requested to get that event on the file.
Fixes: 1968f5eed54c ("fanotify: use both marks when possible") Cc: stable <stable@vger.kernel.org> Signed-off-by: Amir Goldstein <amir73il@gmail.com> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
OSTA UDF specification does not mention whether the CS0 charset in case
of two bytes per character encoding should be treated in UTF-16 or
UCS-2. The sample code in the standard does not treat UTF-16 surrogates
in any special way but on systems such as Windows which work in UTF-16
internally, filenames would be treated as being in UTF-16 effectively.
In Linux it is more difficult to handle characters outside of Base
Multilingual plane (beyond 0xffff) as NLS framework works with 2-byte
characters only. Just make sure we don't leak UTF-16 surrogates into the
resulting string when loading names from the filesystem for now.
CC: stable@vger.kernel.org # >= v4.6 Reported-by: Mingye Wang <arthur200126@gmail.com> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we patch an alternate feature section, we have to adjust any
relative branches that branch out of the alternate section.
But currently we have a bug if we have a branch that points to past
the last instruction of the alternate section, eg:
FTR_SECTION_ELSE
1: b 2f
or 6,6,6
2:
ALT_FTR_SECTION_END(...)
nop
This will result in a relative branch at 1 with a target that equals
the end of the alternate section.
That branch does not need adjusting when it's moved to the non-else
location. Currently we do adjust it, resulting in a branch that goes
off into the link-time location of the else section, which is junk.
The fix is to not patch branches that have a target == end of the
alternate section.
When setting up a CPU, we "push" (activate) a pool VP for it.
However it's an error to do so if it already has an active
pool VP.
This happens when doing soft CPU hotplug on powernv since we
don't tear down the CPU on unplug. The HW flags the error which
gets captured by the diagnostics.
Fix this by making sure to "pull" out any already active pool
first.
Fixes: 243e25112d06 ("powerpc/xive: Native exploitation of the XIVE interrupt controller") Cc: stable@vger.kernel.org # v4.12+ Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On boot we save the configuration space of PCIe bridges. We do this so
when we get an EEH event and everything gets reset that we can restore
them.
Unfortunately we save this state before we've enabled the MMIO space
on the bridges. Hence if we have to reset the bridge when we come back
MMIO is not enabled and we end up taking an PE freeze when the driver
starts accessing again.
This patch forces the memory/MMIO and bus mastering on when restoring
bridges on EEH. Ideally we'd do this correctly by saving the
configuration space writes later, but that will have to come later in
a larger EEH rewrite. For now we have this simple fix.
The original bug can be triggered on a boston machine by doing:
echo 0x8000000000000000 > /sys/kernel/debug/powerpc/PCI0001/err_injct_outbound
On boston, this PHB has a PCIe switch on it. Without this patch,
you'll see two EEH events, 1 expected and 1 the failure we are fixing
here. The second EEH event causes the anything under the PHB to
disappear (i.e. the i40e eth).
With this patch, only 1 EEH event occurs and devices properly recover.
Fixes: 652defed4875 ("powerpc/eeh: Check PCIe link after reset") Cc: stable@vger.kernel.org # v3.11+ Reported-by: Pridhiviraj Paidipeddi <ppaidipe@linux.vnet.ibm.com> Signed-off-by: Michael Neuling <mikey@neuling.org> Acked-by: Russell Currey <ruscur@russell.cc> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The label .Llast_fixup\@ is jumped to on page fault within the final
byte set loop of memset (on < MIPSR6 architectures). For some reason, in
this fault handler, the v1 register is randomly set to a2 & STORMASK.
This clobbers v1 for the calling function. This can be observed with the
following test code:
static int __init __attribute__((optimize("O0"))) test_clear_user(void)
{
register int t asm("v1");
char *test;
int j, k;
pr_info("\n\n\nTesting clear_user\n");
test = vmalloc(PAGE_SIZE);
for (j = 256; j < 512; j++) {
t = 0xa5a5a5a5;
if ((k = clear_user(test + PAGE_SIZE - 256, j)) != j - 256) {
pr_err("clear_user (%px %d) returned %d\n", test + PAGE_SIZE - 256, j, k);
}
if (t != 0xa5a5a5a5) {
pr_err("v1 was clobbered to 0x%x!\n", t);
}
}
return 0;
}
late_initcall(test_clear_user);
Which demonstrates that v1 is indeed clobbered (MIPS64):
Testing clear_user
v1 was clobbered to 0x1!
v1 was clobbered to 0x2!
v1 was clobbered to 0x3!
v1 was clobbered to 0x4!
v1 was clobbered to 0x5!
v1 was clobbered to 0x6!
v1 was clobbered to 0x7!
Since the number of bytes that could not be set is already contained in
a2, the andi placing a value in v1 is not necessary and actively
harmful in clobbering v1.
Reported-by: James Hogan <jhogan@kernel.org> Signed-off-by: Matt Redfearn <matt.redfearn@mips.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org Cc: stable@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/19109/ Signed-off-by: James Hogan <jhogan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The __clear_user function is defined to return the number of bytes that
could not be cleared. From the underlying memset / bzero implementation
this means setting register a2 to that number on return. Currently if a
page fault is triggered within the memset_partial block, the value
loaded into a2 on return is meaningless.
The label .Lpartial_fixup\@ is jumped to on page fault. In order to work
out how many bytes failed to copy, the exception handler should find how
many bytes left in the partial block (andi a2, STORMASK), add that to
the partial block end address (a2), and subtract the faulting address to
get the remainder. Currently it incorrectly subtracts the partial block
start address (t1), which has additionally been clobbered to generate a
jump target in memset_partial. Fix this by adding the block end address
instead.
This issue was found with the following test code:
int j, k;
for (j = 0; j < 512; j++) {
if ((k = clear_user(NULL, j)) != j) {
pr_err("clear_user (NULL %d) returned %d\n", j, k);
}
}
Which now passes on Creator Ci40 (MIPS32) and Cavium Octeon II (MIPS64).
Suggested-by: James Hogan <jhogan@kernel.org> Signed-off-by: Matt Redfearn <matt.redfearn@mips.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org Cc: stable@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/19108/ Signed-off-by: James Hogan <jhogan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The MIPS kernel memset / bzero implementation includes a small_memset
branch which is used when the region to be set is smaller than a long (4
bytes on 32bit, 8 bytes on 64bit). The current small_memset
implementation uses a simple store byte loop to write the destination.
There are 2 issues with this implementation:
1. When EVA mode is active, user and kernel address spaces may overlap.
Currently the use of the sb instruction means kernel mode addressing is
always used and an intended write to userspace may actually overwrite
some critical kernel data.
2. If the write triggers a page fault, for example by calling
__clear_user(NULL, 2), instead of gracefully handling the fault, an OOPS
is triggered.
Fix these issues by replacing the sb instruction with the EX() macro,
which will emit EVA compatible instuctions as required. Additionally
implement a fault fixup for small_memset which sets a2 to the number of
bytes that could not be cleared (as defined by __clear_user).
Reported-by: Chuanhua Lei <chuanhua.lei@intel.com> Signed-off-by: Matt Redfearn <matt.redfearn@mips.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: linux-mips@linux-mips.org Cc: stable@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/18975/ Signed-off-by: James Hogan <jhogan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Doing `ioctl(HIDIOCGFEATURE)` in a tight loop on a hidraw device
and then disconnecting the device, or unloading the driver, can
cause a NULL pointer dereference.
When a hidraw device is destroyed it sets 0 to `dev->exist`.
Most functions check 'dev->exist' before doing its work, but
`hidraw_get_report()` was missing that check.
The commit 581c4484769e ("HID: input: map digitizer battery usage")
assumed that devices having input (qas opposed to feature) report for
battery strength would report the data on their own, without the need to
be polled by the kernel; unfortunately it is not so. Many wireless mice
do not send unsolicited reports with battery strength data and have to
be polled explicitly. As a complication, stylus devices on digitizers
are not normally connected to the base and thus can not be polled - the
base can only determine battery strength in the stylus when it is in
proximity.
To solve this issue, we add a special flag that tells the kernel
to avoid polling the device (and expect unsolicited reports) and set it
when report field with physical usage of digitizer stylus (HID_DG_STYLUS).
Unless this flag is set, and we have not seen the unsolicited reports,
the kernel will attempt to poll the device when userspace attempts to
read "capacity" and "state" attributes of power_supply object
corresponding to the devices battery.
Fixes: 581c4484769e ("HID: input: map digitizer battery usage")
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=198095 Cc: stable@vger.kernel.org Reported-and-tested-by: Martin van Es <martin@mrvanes.com> Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by: Jiri Kosina <jkosina@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
add_device_randomness() use of crng_fast_load() was highly
problematic. Some callers of add_device_randomness() can pass in a
large amount of static information. This would immediately promote
the crng_init state from 0 to 1, without really doing much to
initialize the primary_crng's internal state with something even
vaguely unpredictable.
Since we don't have the speed constraints of add_interrupt_randomness(),
we can do a better job mixing in the what unpredictability a device
driver or architecture maintainer might see fit to give us, and do it
in a way which does not bump the crng_init_cnt variable.
Also, since add_device_randomness() doesn't bump any entropy
accounting in crng_init state 0, mix the device randomness into the
input_pool entropy pool as well. This is related to CVE-2018-1108.
Reported-by: Jann Horn <jannh@google.com> Fixes: ee7998c50c26 ("random: do not ignore early device randomness") Cc: stable@kernel.org # 4.13+ Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
0: The CRNG is not initialized at all
1: The CRNG has a small amount of entropy, hopefully good enough for
early-boot, non-cryptographical use cases
2: The CRNG is fully initialized and we are sure it is safe for
cryptographic use cases.
The crng_ready() function should only return true once we are in the
last state. This addresses CVE-2018-1108.
There are two front mics on this machine, if we don't adjust the
location for one of them, they will have the same mixer name,
pulseaudio can't handle this situation.
After applying this FIXUP, they will have different mixer name,
then pulseaudio can handle them correctly.
Otherwise, the pin will be regarded as microphone, and the jack name
is "Mic Phantom", it is always on in the pulseaudio even nothing is
plugged into the jack. So the UI is confusing to users since the
microphone always shows up in the UI even there is no microphone
plugged.
After adding this flag, the jack name is "Headset Mic Phantom", then
the pulseaudio can handle its detection correctly.
Some rawmidi compat ioctls lack of the input substream checks
(although they do check only for rfile->output). This many eventually
lead to an Oops as NULL substream is passed to the rawmidi core
functions.
Fix it by adding the proper checks before each function call.
Sending MIDI messages to a PODxt through the USB connection shows
"usb_submit_urb failed" in dmesg and the message is not received by
the POD.
The error is caused because in the funcion send_midi_async() in midi.c
there is a call to usb_sndbulkpipe() for endpoint 3 OUT, but the PODxt
USB descriptor shows that this endpoint it's an interrupt endpoint.
Patch tested with PODxt only.
[ The bug has been present from the very beginning in the staging
driver time, but Fixes below points to the commit moving to sound/
directory so that the fix can be cleanly applied -- tiwai ]
Two years ago I tried an AMD Radeon E8860 embedded GPU with the drm driver.
The dmesg output included driver warnings about an invalid PCIe lane width.
Tracking the problem back led to si_set_pcie_lane_width_in_smc().
The calculation of the lane widths via ATOM_PPLIB_PCIE_LINK_WIDTH_MASK and
ATOM_PPLIB_PCIE_LINK_WIDTH_SHIFT macros did not increment the resulting
value, per the comment in pptable.h ("lanes - 1"), and per usage elsewhere.
Applying the increment silenced the warnings.
The code has not changed since, so either my analysis was incorrect or the
bug has gone unnoticed. Hence submitting this as an RFC.
Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Chunming Zhou <david1.zhou@amd.com> Signed-off-by: Paul Parsons <lost.distance@yahoo.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Calling request_irq() followed by disable_irq() is usually a bad idea,
specially if the interrupt can be pending, and you're not yet in a
position to handle it.
This is exactly what happens on my kevin system when rebooting in a
second kernel using kexec: Some interrupt is left pending from
the previous kernel, and we take it too early, before disable_irq()
could do anything.
Let's clear the pending interrupts as we initialize the HW, and move
the interrupt request after that point. This ensures that we're in
a sane state when the interrupt is requested.
The calculation of the lane widths via ATOM_PPLIB_PCIE_LINK_WIDTH_MASK and
ATOM_PPLIB_PCIE_LINK_WIDTH_SHIFT macros did not increment the resulting
value, per the comment in pptable.h ("lanes - 1"), and per usage elsewhere.
Port of the radeon fix to amdgpu.
Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Chunming Zhou <david1.zhou@amd.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=102553 Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If these bos are evicted and are in the validated list
things blow up, so do not put them in there. Notably,
that tries to add the bo to the LRU twice, which results
in a BUG_ON in ttm_bo.c.
While for the bo_list an alternative would be to not allow
always valid bos in there, that does not work for the user
fence.
v2: Fixed whitespace issue pointed out by checkpatch.pl
Signed-off-by: Bas Nieuwenhuizen <basni@chromium.org> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When device boots with T > T_trip_1 and requests interrupt,
the race condition takes place. The interrupt comes before
THERMAL_DEVICE_ENABLED is set. This leads to an attempt to
reading sensor value from irq and disabling the sensor, based on
the data->mode field, which expected to be THERMAL_DEVICE_ENABLED,
but still stays as THERMAL_DEVICE_DISABLED. Afher this issue
sensor is never re-enabled, as the driver state is wrong.
Fix this problem by setting the 'data' members prior to
requesting the interrupts.
Since the offset for both registers, PWMDWIDTH and PWMTHRES, used to
control PWM4 or PWM5 are distinct from the other PWMs, whose wrong
programming on PWM hardware causes waveform cannot be output as expected.
Thus, the patch adds the extra condition for fixing up the weird case to
let PWM4 or PWM5 able to work on MT7623.
v1 -> v2: use pwm45_fixup naming instead of pwm45_quirk
v2 -> v3: add more tags for Reviewed-by, Fixes, and Cc stable
Cc: stable@vger.kernel.org Fixes: caf065f8fd58 ("pwm: Add MediaTek PWM support") Signed-off-by: Sean Wang <sean.wang@mediatek.com> Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com> Cc: Zhi Mao <zhi.mao@mediatek.com> Cc: John Crispin <john@phrozen.org> Cc: Matthias Brugger <matthias.bgg@gmail.com> Signed-off-by: Thierry Reding <thierry.reding@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In order to enable a PLL, not only the PLL has to be powered up and
locked, but you also have to de-assert the reset signal. The last part
was missing. Add it so PLLs that were not enabled by the FW/bootloader
can be enabled from Linux.
Fixes: 41691b8862e2 ("clk: bcm2835: Add support for programming the audio domain clocks") Cc: <stable@vger.kernel.org> Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com> Reviewed-by: Eric Anholt <eric@anholt.net> Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The clock for which all PWM devices on MT7623 or MT2701 actually depending
on has to be divided by four from its parent clock axi_sel in the clock
path prior to PWM devices.
Consequently, adding a fixed-factor clock axisel_d4 as one-fourth of
clock axi_sel allows that PWM devices can have the correct resolution
calculation.
Cc: stable@vger.kernel.org Fixes: e9862118272a ("clk: mediatek: Add MT2701 clock support") Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we build this driver with on x86-32, gcc produces a false-positive warning:
drivers/clk/renesas/clk-sh73a0.c: In function 'sh73a0_cpg_clocks_init':
drivers/clk/renesas/clk-sh73a0.c:155:10: error: 'parent_name' may be used uninitialized in this function [-Werror=maybe-uninitialized]
return clk_register_fixed_factor(NULL, name, parent_name, 0,
We can work around that warning by adding a fake initialization, I tried
and failed to come up with any better workaround. This is currently one
of few remaining warnings for a 4.14.y randconfig build, so it would be
good to also have it backported at least to that version. Older versions
have more randconfig warnings, so we might not care.
I had not noticed this earlier, because one patch in my randconfig test
tree removes the '-ffreestanding' option on x86-32, and that avoids
the warning. The -ffreestanding flag was originally global but moved
into arch/i386 by Andi Kleen in commit 6edfba1b33c7 ("[PATCH] x86_64:
Don't define string functions to builtin") as a 'temporary workaround'.
Like many temporary hacks, this turned out to be rather long-lived, from
all I can tell we still need a simple fix to asm/string_32.h before it
can be removed, but I'm not sure about how to best do that.
Clearfog boards can come with a CPU clocked at 1600MHz (commercial)
or 1333MHz (industrial).
They have also some dip-switches to select a different clock (666, 800,
1066, 1200).
The funny thing is that the recovery button is on the MPP34 fq selector.
So, when booting an industrial board with this button down, the frequency
666MHz is selected (and the kernel didn't boot).
This patch add all the missing clocks.
The only mode I didn't test is 2GHz (uboot found 4294MHz instead :/ ).
Fixes: 0e85aeced4d6 ("clk: mvebu: add clock support for Armada 380/385") Cc: <stable@vger.kernel.org> # 3.16.x: 9593f4f56cf5: clk: mvebu: armada-38x: add support for 1866MHz variants Cc: <stable@vger.kernel.org> # 3.16.x Signed-off-by: Richard Genoud <richard.genoud@gmail.com> Acked-by: Gregory CLEMENT <gregory.clement@bootlin.com> Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Per PCIe r3.1, sec 2.2.6.2 and 7.8.4, a Requester may not use 8-bit Tags
unless its Extended Tag Field Enable is set, but all Receivers/Completers
must handle 8-bit Tags correctly regardless of their Extended Tag Field
Enable.
Some devices do not handle 8-bit Tags as Completers, so add a quirk for
them. If we find such a device, we disable Extended Tags for the entire
hierarchy to make peer-to-peer DMA possible.
The Broadcom HT1100/HT2000/HT2100 seems to have issues with handling 8-bit
tags. Mark it as broken.
This fixes Xorg hangs and unresponsive keyboards with errors like this:
radeon 0000:06:00.0: GPU lockup (current fence id 0x000000000000000e last fence id 0x0000000000000
[drm:r600_ring_test [radeon]] *ERROR* radeon: ring 0 test failed (scratch(0x8504)=0xCAFEDEAD)
[drm:r600_resume [radeon]] *ERROR* r600 startup failed on resume
Fixes: 60db3a4d8cc9 ("PCI: Enable PCIe Extended Tags if supported") Link: https://bugzilla.kernel.org/show_bug.cgi?id=196197 Signed-off-by: Sinan Kaya <okaya@codeaurora.org> Signed-off-by: Bjorn Helgaas <helgaas@kernel.org> CC: stable@vger.kernel.org # v4.11: 62ce94a7a5a5 PCI: Mark Broadcom HT2100 Root Port Extended Tags as broken CC: stable@vger.kernel.org # v4.11 Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A spinlock is held while updating the internal copy of the IRQ mask,
but not while writing it to the actual IMASK register. After the lock
is released, an IRQ can occur before the IMASK register is written.
If handling this IRQ causes the mask to be changed, when the handler
returns back to the middle of the first mask update, a stale value
will be written to the mask register.
If this causes an IRQ to become unmasked that cannot have its status
cleared by writing a 1 to it in the IREG register, e.g. the SDIO IRQ,
then we can end up stuck with the same IRQ repeatedly being fired but
not handled. Normally the MMC IRQ handler attempts to clear any
unexpected IRQs by writing IREG, but for those that cannot be cleared
in this way then the IRQ will just repeatedly fire.
This was resulting in lockups after a while of using Wi-Fi on the
CI20 (GitHub issue #19).
Resolve by holding the spinlock until after the IMASK register has
been updated.
Cc: stable@vger.kernel.org Link: https://github.com/MIPS/CI20_linux/issues/19 Fixes: 61bfbdb85687 ("MMC: Add support for the controller on JZ4740 SoCs.") Tested-by: Mathieu Malaterre <malat@debian.org> Signed-off-by: Alex Smith <alex.smith@imgtec.com> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Upon module load, mmc_block allocates a bus with bus_registeri() in
mmc_blk_init(). This reference never gets freed during module unload, which
leads to subsequent re-insertions of the module fails and a WARN() splat is
triggered.
Fix the bug by dropping the reference for the bus in mmc_blk_exit().
Signed-off-by: Alexander Kappner <agk@godking.net> Fixes: 97548575bef3 ("mmc: block: Convert RPMB to a character device") Cc: <stable@vger.kernel.org> Reviewed-by: Shawn Lin <shawn.lin@rock-chips.com> Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
glibc 2.26 removed the 'struct ucontext' to "improve" POSIX compliance
and break programs, including User Mode Linux. Fix User Mode Linux
by using POSIX ucontext_t.
This fixes:
arch/um/os-Linux/signal.c: In function 'hard_handler':
arch/um/os-Linux/signal.c:163:22: error: dereferencing pointer to incomplete type 'struct ucontext'
mcontext_t *mc = &uc->uc_mcontext;
arch/x86/um/stub_segv.c: In function 'stub_segv_handler':
arch/x86/um/stub_segv.c:16:13: error: dereferencing pointer to incomplete type 'struct ucontext'
&uc->uc_mcontext);
Cc: stable@vger.kernel.org Signed-off-by: Krzysztof Mazur <krzysiek@podlesie.net> Signed-off-by: Richard Weinberger <richard@nod.at> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Recent libcs have gotten a bit more strict, so we actually need to
include the right headers and use the right types. This enables UML to
compile again.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Cc: stable@vger.kernel.org Signed-off-by: Richard Weinberger <richard@nod.at> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The ring buffer is made up of a link list of pages. When making the ring
buffer bigger, it will allocate all the pages it needs before adding to the
ring buffer, and if it fails, it frees them and returns an error. This makes
increasing the ring buffer size an all or nothing action. When this was
first created, the pages were allocated with "NORETRY". This was to not
cause any Out-Of-Memory (OOM) actions from allocating the ring buffer. But
NORETRY was too strict, as the ring buffer would fail to expand even when
there's memory available, but was taken up in the page cache.
Commit 848618857d253 ("tracing/ring_buffer: Try harder to allocate") changed
the allocating from NORETRY to RETRY_MAYFAIL. The RETRY_MAYFAIL would
allocate from the page cache, but if there was no memory available, it would
simple fail the allocation and not trigger an OOM.
This worked fine, but had one problem. As the ring buffer would allocate one
page at a time, it could take up all memory in the system before it failed
to allocate and free that memory. If the allocation is happening and the
ring buffer allocates all memory and then tries to take more than available,
its allocation will not trigger an OOM, but if there's any allocation that
happens someplace else, that could trigger an OOM, even though once the ring
buffer's allocation fails, it would free up all the previous memory it tried
to allocate, and allow other memory allocations to succeed.
Commit d02bd27bd33dd ("mm/page_alloc.c: calculate 'available' memory in a
separate function") separated out si_mem_availble() as a separate function
that could be used to see how much memory is available in the system. Using
this function to make sure that the ring buffer could be allocated before it
tries to allocate pages we can avoid allocating all memory in the system and
making it vulnerable to OOMs if other allocations are taking place.
Link: http://lkml.kernel.org/r/1522320104-6573-1-git-send-email-zhaoyang.huang@spreadtrum.com CC: stable@vger.kernel.org Cc: linux-mm@kvack.org Fixes: 848618857d253 ("tracing/ring_buffer: Try harder to allocate")
Requires: d02bd27bd33dd ("mm/page_alloc.c: calculate 'available' memory in a separate function") Reported-by: Zhaoyang Huang <huangzhaoyang@gmail.com> Tested-by: Joel Fernandes <joelaf@google.com> Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Per the ACPI specification the only functional purpose for a DIMM
Control Region to be mapped into the system physical address space, from
an OSPM perspective, is to support block-apertures. However, there are
some BIOSen that publish DIMM Control Region SPA entries for pre-boot
environment consumption. Undo the kernel policy of generating disabled
'ndblk' regions when this configuration is detected.
Cc: <stable@vger.kernel.org> Fixes: 1f7df6f88b92 ("libnvdimm, nfit: regions (block-data-window...)") Reviewed-by: Toshi Kani <toshi.kani@hpe.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a small window whereby ARS scan requests can schedule work that
userspace will miss when polling scrub_show. Hold the init_mutex lock
over calls to report the status to close this potential escape. Also,
make sure that requests to cancel the ARS workqueue are treated as an
idle event.
Cc: <stable@vger.kernel.org> Cc: Vishal Verma <vishal.l.verma@intel.com> Fixes: 37b137ff8c83 ("nfit, libnvdimm: allow an ARS scrub...") Reviewed-by: Dave Jiang <dave.jiang@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit 841a915d20c7b2 ("printf: Do not have bprintf dereference pointers")
would preprocess various pointers that are dereferenced in the bprintf()
because the recording and printing are done at two different times. Some
pointers stayed dereferenced in the ring buffer because user space could
handle them (namely "%pS" and friends). Pointers that are not dereferenced
should not be processed immediately but instead just saved directly.
Cc: stable@vger.kernel.org Fixes: 841a915d20c7b2 ("printf: Do not have bprintf dereference pointers") Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
/*
* Give up if we don't find an instance of a uuid at each
* position (from 0 to nd_region->ndr_mappings - 1), or if we
* find a dimm with two instances of the same uuid.
*/
dev_err(&nd_region->dev, "%s missing label for %pUb\n",
dev_name(ndd->dev), nd_label->uuid);
At initialization time the 'dimm' driver caches a copy of the memory
device's label area and reserves address space for each of the
namespaces defined.
However, as can be seen below, the reservation occurs even when the
index blocks are invalid:
The Acer Acer Veriton X4110G has a TPM device detected as:
tpm_tis 00:0b: 1.2 TPM (device-id 0xFE, rev-id 71)
After the first S3 suspend, the following error appears during resume:
tpm tpm0: A TPM error(38) occurred continue selftest
Any following S3 suspend attempts will now fail with this error:
tpm tpm0: Error (38) sending savestate before suspend
PM: Device 00:0b failed to suspend: error 38
Error 38 is TPM_ERR_INVALID_POSTINIT which means the TPM is
not in the correct state. This indicates that the platform BIOS
is not sending the usual TPM_Startup command during S3 resume.
>From this point onwards, all TPM commands will fail.
The same issue was previously reported on Foxconn 6150BK8MC and
Sony Vaio TX3.
The platform behaviour seems broken here, but we should not break
suspend/resume because of this.
When the unexpected TPM state is encountered, set a flag to skip the
affected TPM_SaveState command on later suspends.
cxllib_handle_fault() is called by an external driver when it needs to
have the host resolve page faults for a buffer. The buffer can cover
several pages and VMAs. The function iterates over all the pages used
by the buffer, based on the page size of the VMA.
To ensure some stability while processing the faults, the thread T1
grabs the mm->mmap_sem semaphore with read access (R1). However, when
processing a page fault for a single page, one of the underlying
functions, copro_handle_mm_fault(), also grabs the same semaphore with
read access (R2). So the thread T1 takes the semaphore twice.
If another thread T2 tries to access the semaphore in write mode W1
(say, because it wants to allocate memory and calls 'brk'), then that
thread T2 will have to wait because there's a reader (R1). If the
thread T1 is processing a new page at that time, it won't get an
automatic grant at R2, because there's now a writer thread
waiting (T2). And we have a deadlock.
The timeline is:
1. thread T1 owns the semaphore with read access R1
2. thread T2 requests write access W1 and waits
3. thread T1 requests read access R2 and waits
The fix is for the thread T1 to release the semaphore R1 once it got
the information it needs from the current VMA. The address space/VMAs
could evolve while T1 iterates over the full buffer, but in the
unlikely case where T1 misses a page, the external driver will raise a
new page fault when retrying the memory access.
Fixes: 3ced8d730063 ("cxl: Export library to support IBM XSL") Cc: stable@vger.kernel.org # 4.13+ Signed-off-by: Frederic Barrat <fbarrat@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Despite the efforts made to correctly read the NDA and CUBC registers,
the order in which the registers are read could sometimes lead to an
inconsistent state.
Re-using the timeline from the comments, this following timing of
registers reads could lead to reading NDA with value "@desc2" and
CUBC with value "MAX desc1":
This is allowed by the condition ((check_nda == cur_nda) && initd),
despite cur_ubc and cur_nda being in the precise state we don't want.
This error leads to incorrect residue computation.
Fix it by inversing the order in which CUBC and INITD are read. This
makes sure that NDA and CUBC are always read together either _before_
INITD goes to 0 or _after_ it is back at 1.
The case where NDA is read before INITD is at 0 and CUBC is read after
INITD is back at 1 will be rejected by check_nda and cur_nda being
different.
Fix a race for "nosync" activations providing "aa.." device health
characters and "0/N" sync ratio rather than "AA..." and "N/N". Occurs
when status for the raid set is retrieved during resume before the MD
sync thread starts and clears the MD_RECOVERY_NEEDED flag.
gcc-4.4.4 has issues with initialization of anonymous unions:
drivers/infiniband/ulp/srpt/ib_srpt.c: In function 'srpt_zerolength_write':
drivers/infiniband/ulp/srpt/ib_srpt.c:854: error: unknown field 'wr_cqe' specified in initializer
drivers/infiniband/ulp/srpt/ib_srpt.c:854: warning: initialization makes integer from pointer without a cast
Work aound this.
Fixes: 2a78cb4db487 ("IB/srpt: Fix an out-of-bounds stack access in srpt_zerolength_write()") Cc: Bart Van Assche <bart.vanassche@wdc.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Jason Gunthorpe <jgg@mellanox.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
gcc-4.4.4 has issues with initialization of anonymous unions.
drivers/infiniband/core/verbs.c: In function '__ib_drain_sq':
drivers/infiniband/core/verbs.c:2204: error: unknown field 'wr_cqe' specified in initializer
drivers/infiniband/core/verbs.c:2204: warning: initialization makes integer from pointer without a cast
Work around this.
Fixes: a1ae7d0345edd5 ("RDMA/core: Avoid that ib_drain_qp() triggers an out-of-bounds stack access") Cc: Bart Van Assche <bart.vanassche@wdc.com> Cc: Steve Wise <swise@opengridcomputing.com> Cc: Sagi Grimberg <sagi@grimberg.me> Cc: Jason Gunthorpe <jgg@mellanox.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Ensure that cv_end is equal to ibdev->num_comp_vectors for the
NUMA node with the highest index. This patch improves spreading
of RDMA channels over completion vectors and thereby improves
performance, especially on systems with only a single NUMA node.
This patch drops support for the comp_vector login parameter by
ignoring the value of that parameter since I have not found a
good way to combine support for that parameter and automatic
spreading of RDMA channels over completion vectors.
Fixes: d92c0da71a35 ("IB/srp: Add multichannel support") Reported-by: Alexander Schmid <alex@modula-shop-systems.de> Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Cc: Alexander Schmid <alex@modula-shop-systems.de> Cc: stable@vger.kernel.org Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Before commit e494f6a72839 ("[SCSI] improved eh timeout handler") it
did not really matter whether or not abort handlers like srp_abort()
called .scsi_done() when returning another value than SUCCESS. Since
that commit however this matters. Hence only call .scsi_done() when
returning SUCCESS.
Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Cc: stable@vger.kernel.org Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The commit 02a5d6925cd3 ("ALSA: pcm: Avoid potential races between OSS
ioctls and read/write") split the PCM preparation code to a locked
version, and it added a sanity check of runtime->oss.prepare flag
along with the change. This leaded to an endless loop when the stream
gets XRUN: namely, snd_pcm_oss_write3() and co call
snd_pcm_oss_prepare() without setting runtime->oss.prepare flag and
the loop continues until the PCM state reaches to another one.
As the function is supposed to execute the preparation
unconditionally, drop the invalid state check there.
The PCM runtime object is created and freed dynamically at PCM stream
open / close time. This is tracked via substream->runtime, and it's
cleared at snd_pcm_detach_substream().
The runtime object assignment is protected by PCM open_mutex, so for
all PCM operations, it's safely handled. However, each PCM substream
provides also an ALSA timer interface, and user-space can access to
this while closing a PCM substream. This may eventually lead to a
UAF, as snd_pcm_timer_resolution() tries to access the runtime while
clearing it in other side.
Fortunately, it's the only concurrent access from the PCM timer, and
it merely reads runtime->timer_resolution field. So, we can avoid the
race by reordering kfree() and wrapping the substream->runtime
clearance with the corresponding timer lock.
The previous fix 40cab6e88cb0 ("ALSA: pcm: Return -EBUSY for OSS
ioctls changing busy streams") introduced some mutex unbalance; the
check of runtime->oss.rw_ref was inserted in a wrong place after the
mutex lock.
This patch fixes the inconsistency by rewriting with the helper
functions to lock/unlock parameters with the stream check.
OSS PCM stream management isn't modal but it allows ioctls issued at
any time for changing the parameters. In the previous hardening
patch ("ALSA: pcm: Avoid potential races between OSS ioctls and
read/write"), we covered these races and prevent the corruption by
protecting the concurrent accesses via params_lock mutex. However,
this means that some ioctls that try to change the stream parameter
(e.g. channels or format) would be blocked until the read/write
finishes, and it may take really long.
Basically changing the parameter while reading/writing is an invalid
operation, hence it's even more user-friendly from the API POV if it
returns -EBUSY in such a situation.
This patch adds such checks in the relevant ioctls with the addition
of read/write access refcount.
Although we apply the params_lock mutex to the whole read and write
operations as well as snd_pcm_oss_change_params(), we may still face
some races.
First off, the params_lock is taken inside the read and write loop.
This is intentional for avoiding the too long locking, but it allows
the in-between parameter change, which might lead to invalid
pointers. We check the readiness of the stream and set up via
snd_pcm_oss_make_ready() at the beginning of read and write, but it's
called only once, by assuming that it remains ready in the rest.
Second, many ioctls that may change the actual parameters
(i.e. setting runtime->oss.params=1) aren't protected, hence they can
be processed in a half-baked state.
This patch is an attempt to plug these holes. The stream readiness
check is moved inside the read/write inner loop, so that the stream is
always set up in a proper state before further processing. Also, each
ioctl that may change the parameter is wrapped with the params_lock
for avoiding the races.
The issues were triggered by syzkaller in a few different scenarios,
particularly the one below appearing as GPF in loopback_pos_update.
Michal Kalderon has found some corner cases around device unload
with active NFS mounts that I didn't have the imagination to test
when xprtrdma device removal was added last year.
- The ULP device removal handler is responsible for deallocating
the PD. That wasn't clear to me initially, and my own testing
suggested it was not necessary, but that is incorrect.
- The transport destruction path can no longer assume that there
is a valid ID.
- When destroying a transport, ensure that ib_free_cq() is not
invoked on a CQ that was already released.
Reported-by: Michal Kalderon <Michal.Kalderon@cavium.com> Fixes: bebd031866ca ("xprtrdma: Support unplugging an HCA from ...") Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Cc: stable@vger.kernel.org # v4.12+ Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With v4.15, on one of my NFS/RDMA clients I measured a nearly
doubling in the latency of small read and write system calls. There
was no change in server round trip time. The extra latency appears
in the whole RPC execution path.
"git bisect" settled on commit ccede7598588 ("xprtrdma: Spread reply
processing over more CPUs") .
After some experimentation, I found that leaving the WQ bound and
allowing the scheduler to pick the dispatch CPU seems to eliminate
the long latencies, and it does not introduce any new regressions.
The fix is implemented by reverting only the part of
commit ccede7598588 ("xprtrdma: Spread reply processing over more
CPUs") that dispatches RPC replies specifically on the CPU where the
matching RPC call was made.
Interestingly, saving the CPU number and later queuing reply
processing there was effective _only_ for a NFS READ and WRITE
request. On my NUMA client, in-kernel RPC reply processing for
asynchronous RPCs was dispatched on the same CPU where the RPC call
was made, as expected. However synchronous RPCs seem to get their
reply dispatched on some other CPU than where the call was placed,
every time.
Fixes: ccede7598588 ("xprtrdma: Spread reply processing over ... ") Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Cc: stable@vger.kernel.org # v4.15+ Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
==================================================================
BUG: KASAN: stack-out-of-bounds in rxe_post_send+0x77d/0x9b0 [rdma_rxe]
Read of size 8 at addr ffff880061aef860 by task 01/1080
Fixes: 765d67748bcf ("IB: new common API for draining queues") Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com> Cc: Steve Wise <swise@opengridcomputing.com> Cc: Sagi Grimberg <sagi@grimberg.me> Cc: stable@vger.kernel.org Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch avoids that KASAN reports the following when the SRP initiator
calls srp_post_send():
==================================================================
BUG: KASAN: stack-out-of-bounds in rxe_post_send+0x5c4/0x980 [rdma_rxe]
Read of size 8 at addr ffff880066606e30 by task 02-mq/1074
Check to make sure that ctx->cm_id->device is set before we use it.
Otherwise userspace can trigger a NULL dereference by doing
RDMA_USER_CM_CMD_SET_OPTION on an ID that is not bound to a device.
Cc: <stable@vger.kernel.org> Reported-by: <syzbot+a67bc93e14682d92fc2f@syzkaller.appspotmail.com> Signed-off-by: Roland Dreier <roland@purestorage.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
dm-crypt consumes an excessive amount memory when the user attempts to
zero a dm-crypt device with "blkdiscard -z". The command "blkdiscard -z"
calls the BLKZEROOUT ioctl, it goes to the function __blkdev_issue_zeroout,
__blkdev_issue_zeroout sends a large amount of write bios that contain
the zero page as their payload.
For each incoming page, dm-crypt allocates another page that holds the
encrypted data, so when processing "blkdiscard -z", dm-crypt tries to
allocate the amount of memory that is equal to the size of the device.
This can trigger OOM killer or cause system crash.
Fix this by limiting the amount of memory that dm-crypt allocates to 2%
of total system memory. This limit is system-wide and is divided by the
number of active dm-crypt devices and each device receives an equal
share.
A malicious user could force the directory pointer to be in an invalid
spot by using seekdir(2). Use the mechanism we already have to notice
if the directory has changed since the last time we called
ext4_readdir() to force a revalidation of the pointer.
Add explicit checks in ext4_xattr_block_get() just in case the
e_value_offs and e_value_size fields in the the xattr block are
corrupted in memory after the buffer_verified bit is set on the xattr
block.
Add some paranoia checks to make sure we don't stray beyond the end of
the valid memory region containing ext4 xattr entries while we are
scanning for a match.
Also rename the function to xattr_find_entry() since it is static and
thus only used in fs/ext4/xattr.c
projects / linux-stable / log