Some of the chips supported by the pca953x driver need the most
significant bit in the address word set to automatically increment the
address pointer on subsequent reads and writes (example: PCA9505). With
this bit unset the same register is read multiple times on a multi-byte
read sequence. Other chips must not have this bit set and autoincrement
always (example: PCA9555).
Up to now this AI bit was interpreted to be part of the address, which
resulted in inconsistent regmap caching when a register was written with
AI set and then read without it. This happened for the PCA9505 in
pca953x_gpio_set_multiple() where pca953x_read_regs() bulk read from the
cache for registers 0x8-0xc and then wrote to registers 0x88-0x8c. (Side
note: reading 5 values from offset 0x8 yiels OP0 5 times because AI must
be set to get OP0-OP4, which is another bug that is resolved here as a
by-product.) The same problem happens when calls to gpio_set_value() and
gpio_set_array_value() were mixed.
With this patch the AI bit is always set for chips that support it. This
works as there are no code locations that make use of the behaviour with
AI unset (for the chips that support it).
Note that the call to pca953x_setup_gpio() had to be done a bit earlier
to make the NBANK macro work.
The history of this bug is a bit complicated. Commit b32cecb46bdc
("gpio: pca953x: Extract the register address mangling to single
function") changed which chips and functions are affected. Commit 3b00691cc46a ("gpio: pca953x: hack to fix 24 bit gpio expanders") used
some duct tape to make the driver at least appear to work. Commit 49427232764d ("gpio: pca953x: Perform basic regmap conversion")
introduced the caching. Commit b4818afeacbd ("gpio: pca953x: Add
set_multiple to allow multiple bits to be set in one write.") introduced
the .set_multiple() callback which didn't work for chips that need the
AI bit which was fixed later for some chips in 8958262af3fb ("gpio:
pca953x: Repair multi-byte IO address increment on PCA9575"). So I'm
sorry, I don't know which commit I should pick for a Fixes: line.
Tested-by: Marcel Gudert <m.gudert@eckelmann.de> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Tested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
CLKC_RESET_VID_DIVIDER_CNTL_RESET_N_POST and
CLKC_RESET_VID_DIVIDER_CNTL_RESET_N_PRE are active low. This means:
- asserting them requires setting the register value to 0
- de-asserting them requires setting the register value to 1
Set the register value accordingly for these two reset lines by setting
the inverted the register value compared to all other reset lines.
Use hdmi_pll_lvds_out as parent of the vid_pll_in_sel clock. It's not
easy to see that the vendor kernel does the same, but it actually does.
meson_clk_pll_ops in mainline still cannot fully recalculate all rates
from the HDMI PLL registers because some register bits (at the time of
writing it's unknown which bits are used for this) double the HDMI PLL
output rate (compared to simply considering M, N and FRAC) for some (but
not all) PLL settings.
Update the vid_pll_in_sel parent so our clock calculation works for
simple clock settings like the CVBS output (where no rate doubling is
going on). The PLL ops need to be fixed later on for more complex clock
settings (all HDMI rates).
Fixes: 6cb57c678bb70 ("clk: meson: meson8b: add the read-only video clock trees") Suggested-by: Neil Armstrong <narmstrong@baylibre.com> Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com> Signed-off-by: Jerome Brunet <jbrunet@baylibre.com> Link: https://lore.kernel.org/r/20200417184127.1319871-2-martin.blumenstingl@googlemail.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Potentially, hvc_open() can be called in parallel when two tasks calls
open() on /dev/hvcX. In such a scenario, if the hp->ops->notifier_add()
callback in the function fails, where it sets the tty->driver_data to
NULL, the parallel hvc_open() can see this NULL and cause a memory abort.
Hence, serialize hvc_open and check if tty->private_data is NULL before
proceeding ahead.
The issue can be easily reproduced by launching two tasks simultaneously
that does nothing but open() and close() on /dev/hvcX.
For example:
$ ./simple_open_close /dev/hvc0 & ./simple_open_close /dev/hvc0 &
qdio_establish() calls qdio_setup_thinint() via qdio_setup_irq().
If the subsequent qdio_establish_thinint() fails, we miss to put the
DSCI again. Thus the DSCI isn't available for re-use. Given enough of
such errors, we could end up with having only the shared DSCI available.
Merge qdio_setup_thinint() into qdio_establish_thinint(), and deal with
such an error internally.
The linked list entry from FIFO is peeked at
queue_pending_output_urbs() but the actual element pop-out is
performed outside the spinlock, and it's potentially racy.
Do delete the link at the right place inside the spinlock.
For computation of the the next frame size current value of fs/fps and
accumulated fractional parts of fs/fps are used, where values are stored
in Q16.16 format. This is quite natural for computing frame size for
asynchronous endpoints driven by explicit feedback, since in this case
fs/fps is a value provided by the feedback endpoint and it's already in
the Q format. If an error is accumulated over time, the device can
adjust fs/fps value to prevent buffer overruns/underruns.
But for synchronous endpoints the accuracy provided by these computations
is not enough. Due to accumulated error the driver periodically produces
frames with incorrect size (+/- 1 audio sample).
This patch fixes this issue by implementing a different algorithm for
frame size computation. It is based on accumulating of the remainders
from division fs/fps and it doesn't accumulate errors over time. This
new method is enabled for synchronous and adaptive playback endpoints.
gasket_sysfs_register_store() invokes get_mapping(), which returns a
reference of the specified gasket_sysfs_mapping object to "mapping" with
increased refcnt.
When gasket_sysfs_register_store() returns, local variable "mapping"
becomes invalid, so the refcount should be decreased to keep refcount
balanced.
The reference counting issue happens in one exception handling path of
gasket_sysfs_register_store(). When gasket_dev is NULL, the function
forgets to decrease the refcnt increased by get_mapping(), causing a
refcnt leak.
Fix this issue by calling put_mapping() when gasket_dev is NULL.
gasket_sysfs_put_attr() invokes get_mapping(), which returns a reference
of the specified gasket_sysfs_mapping object to "mapping" with increased
refcnt.
When gasket_sysfs_put_attr() returns, local variable "mapping" becomes
invalid, so the refcount should be decreased to keep refcount balanced.
The reference counting issue happens in one path of
gasket_sysfs_put_attr(). When mapping attribute is unknown, the function
forgets to decrease the refcnt increased by get_mapping(), causing a
refcnt leak.
Fix this issue by calling put_mapping() when put attribute fails due to
unknown attribute.
As far as the device is concerned the dma address is the physical
address. There is no need to convert it to a physical address,
especially not using dma-direct internals that are not available
to drivers and which will interact badly with IOMMUs. Last but not
least the commit introducing it claimed to just fix a type issue,
but actually changed behavior.
Fixes: 6e37ccf78a532 ("firmware: qcom_scm: Use proper types for dma mappings") Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org> Signed-off-by: Christoph Hellwig <hch@lst.de> Link: https://lore.kernel.org/r/20200414123136.441454-1-hch@lst.de Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
In virtiofs (unlike in regular fuse) processing of async replies is
serialized. This can result in a deadlock in rare corner cases when
there's a circular dependency between the completion of two or more async
replies.
Such a deadlock can be reproduced with xfstests:generic/503 if TEST_DIR ==
SCRATCH_MNT (which is a misconfiguration):
- Process A is waiting for page lock in worker thread context and blocked
(virtio_fs_requests_done_work()).
- Process B is holding page lock and waiting for pending writes to
finish (fuse_wait_on_page_writeback()).
- Write requests are waiting in virtqueue and can't complete because
worker thread is blocked on page lock (process A).
Fix this by creating a unique work_struct for each async reply that can
block (O_DIRECT read).
The rza1l_swio_entries referred to the wrong array rza1h_swio_pins,
which was intended to be rza1l_swio_pins. So let's fix it.
This is detected by the following gcc warning:
drivers/pinctrl/pinctrl-rza1.c:401:35: warning: ‘rza1l_swio_pins’
defined but not used [-Wunused-const-variable=]
static const struct rza1_swio_pin rza1l_swio_pins[] = {
^~~~~~~~~~~~~~~
Fixes: 039bc58e73b77723 ("pinctrl: rza1: Add support for RZ/A1L") Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Jason Yan <yanaijie@huawei.com> Link: https://lore.kernel.org/r/20200417111604.19143-1-yanaijie@huawei.com Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Sasha Levin <sashal@kernel.org>
The MFW may make a call to qed and then to qedf for protocol statistics
while the function is still probing. If this happens it's possible that
some members of the struct qedf_ctx may not be fully initialized which can
result in a NULL pointer dereference or general protection fault.
To prevent this, add a new flag call QEDF_PROBING and set it when the
__qedf_probe() function is active. Then in the qedf_get_protocol_tlv_data()
function we can check if the function is still probing and return
immediantely before any uninitialized structures can be touched.
Link: https://lore.kernel.org/r/20200416084314.18851-9-skashyap@marvell.com Signed-off-by: Chad Dupuis <cdupuis@marvell.com> Signed-off-by: Saurav Kashyap <skashyap@marvell.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The commit 3d2613c4289f
("GPIO: gpio-dwapb: Enable platform driver binding to MFD driver")
introduced a use of the platform driver but missed to add the following line
to it:
MODULE_ALIAS("platform:gpio-dwapb");
Add this to get driver loaded automatically if platform device is registered.
Fixes: 8b8061fcbfae ("ARM: dts: sun8i: h2+: add support for Banana Pi M2 Zero board") Signed-off-by: Vincent Stehlé <vincent.stehle@laposte.net> Cc: Icenowy Zheng <icenowy@aosc.io> Cc: Maxime Ripard <mripard@kernel.org> Signed-off-by: Maxime Ripard <maxime@cerno.tech> Signed-off-by: Sasha Levin <sashal@kernel.org>
The thermal trip points have unit name but no reg property, so we can
remove them. It also fixes the following warnings from 'make dtbs_check'
after adding the thermal yaml bindings.
arch/arm64/boot/dts/qcom/apq8016-sbc.dt.yaml: thermal-zones:
gpu-thermal:trips: 'trip-point@0' does not match any of the regexes:
'^[a-zA-Z][a-zA-Z0-9\\-_]{0,63}$', 'pinctrl-[0-9]+'
arch/arm64/boot/dts/qcom/apq8016-sbc.dt.yaml: thermal-zones:
camera-thermal:trips: 'trip-point@0' does not match any of the regexes:
'^[a-zA-Z][a-zA-Z0-9\\-_]{0,63}$', 'pinctrl-[0-9]+'
arch/arm64/boot/dts/qcom/apq8016-sbc.dt.yaml: thermal-zones:
modem-thermal:trips: 'trip-point@0' does not match any of the regexes:
'^[a-zA-Z][a-zA-Z0-9\\-_]{0,63}$', 'pinctrl-[0-9]+'
arch/arm64/boot/dts/qcom/msm8916-mtp.dt.yaml: thermal-zones:
gpu-thermal:trips: 'trip-point@0' does not match any of the regexes:
'^[a-zA-Z][a-zA-Z0-9\\-_]{0,63}$', 'pinctrl-[0-9]+'
arch/arm64/boot/dts/qcom/msm8916-mtp.dt.yaml: thermal-zones:
camera-thermal:trips: 'trip-point@0' does not match any of the regexes:
'^[a-zA-Z][a-zA-Z0-9\\-_]{0,63}$', 'pinctrl-[0-9]+'
arch/arm64/boot/dts/qcom/msm8916-mtp.dt.yaml: thermal-zones:
modem-thermal:trips: 'trip-point@0' does not match any of the regexes:
'^[a-zA-Z][a-zA-Z0-9\\-_]{0,63}$', 'pinctrl-[0-9]+'
For an unreachable target, offload_work is not initialized and the endpoint
state is set to OFLDCONN_NONE. This results in a WARN_ON due to the check
of the work function field being set to zero.
Warning (unit_address_vs_reg): /oscillator@0: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /oscillator@1: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /oscillator@2: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /thermal-zones/cpu_thermal/trips/trip-point@0: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /thermal-zones/cpu_thermal/trips/trip-point@1: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /thermal-zones/cpu_thermal/trips/cpu_crit@0: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /thermal-zones/cpu_thermal/cooling-maps/map@0: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /thermal-zones/cpu_thermal/cooling-maps/map@1: node has a unit name, but no reg property
Warning (unit_address_vs_reg): /reserved-memory/vpu_dma_mem_region: node has a reg or ranges property, but no unit name
Warning (simple_bus_reg): /soc/pinctrl@10005000: simple-bus unit address format error, expected "1000b000"
Warning (simple_bus_reg): /soc/interrupt-controller@10220000: simple-bus unit address format error, expected "10221000"
Since the purgatory is a special stand-alone binary, various profiling
and sanitizing options must be disabled. Having these options enabled
typically will cause dependencies on various special symbols exported by
special libs / stubs used by these frameworks. Since the purgatory is
special, it is not linked against these stubs causing missing symbols in
the purgatory if these options are not disabled.
Sync the set of disabled profiling and sanitizing options with that from
drivers/firmware/efi/libstub/Makefile, adding
-DDISABLE_BRANCH_PROFILING to the CFLAGS and setting:
GCOV_PROFILE := n
UBSAN_SANITIZE := n
This fixes broken references to ftrace_likely_update() when
CONFIG_TRACE_BRANCH_PROFILING is enabled and to __gcov_init() and
__gcov_exit() when CONFIG_GCOV_KERNEL is enabled.
The subset test is not taking into account the unconfined exception
which will cause profile transitions in the stacked confinement
case to fail when no_new_privs is applied.
This fixes a regression introduced in the fix for
https://bugs.launchpad.net/bugs/1839037
The adapter info MAD is used to send the client info and receive the host
info as a response. A persistent buffer is used and as such the client info
is overwritten after the response. During the course of a normal adapter
reset the client info is refreshed in the buffer in preparation for sending
the adapter info MAD.
However, in the special case of LPM where we reenable the CRQ instead of a
full CRQ teardown and reset we fail to refresh the client info in the
adapter info buffer. As a result, after Live Partition Migration (LPM) we
erroneously report the host's info as our own.
When sockhash gets destroyed while sockets are still linked to it, we will
walk the bucket lists and delete the links. However, we are not freeing the
list elements after processing them, leaking the memory.
The leak can be triggered by close()'ing a sockhash map when it still
contains sockets, and observed with kmemleak:
Currently apparmor_sk_clone_security() does not check for existing
label/peer in the 'new' struct sock; it just overwrites it, if any
(with another reference to the label of the source sock.)
The label reference count leak is observed if apparmor_sock_graft()
is called previously: this sets the 'ctx->label' field by getting
a reference to the current label (later overwritten, without put.)
Apparently both calls are done on their own right, especially for
other LSMs, being introduced in 2010/2014, before apparmor socket
mediation in 2017 (see commits [1,2,3,4]).
So, it looks OK there! Let's fix the reference leak in apparmor.
Test-case:
---------
Exercise that code path enough to overflow label reference count.
While trying to use the lantiq_gswip driver on one of my boards I made
a mistake when specifying the phy-mode (because the out-of-tree driver
wants phy-mode "gmii" or "mii" for the internal PHYs). In this case the
following error is printed multiple times:
Unsupported interface: 3
While it gives at least a hint at what may be wrong it is not very user
friendly. Print the human readable phy-mode and also which port is
configured incorrectly (this hardware supports ports 0..6) to improve
the cases where someone made a mistake.
Fixes: 14fceff4771e51 ("net: dsa: Add Lantiq / Intel DSA driver for vrx200") Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com> Acked-by: Hauke Mehrtens <hauke@hauke-m.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
1. If a task is attached to a unconfined profile that is not the
ns->unconfined profile then. Mode the mode is always reported
as -
$ ps -Z
LABEL PID TTY TIME CMD
unconfined 1287 pts/0 00:00:01 bash
test (-) 1892 pts/0 00:00:00 ps
instead of the correct value of (unconfined) as shown below
$ ps -Z
LABEL PID TTY TIME CMD
unconfined 2483 pts/0 00:00:01 bash
test (unconfined) 3591 pts/0 00:00:00 ps
2. if a task is confined by a stack of profiles that are unconfined
the output of label mode is again the incorrect value of (-) like
above, instead of (unconfined). This is because the visibile
profile count increment is skipped by the special casing of
unconfined.
Fixes: f1bd904175e8 ("apparmor: add the base fns() for domain labels") Signed-off-by: John Johansen <john.johansen@canonical.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When System.map was generated, the kernel used mksysmap to
filter the kernel symbols, but all the symbols with the
second letter 'L' in the kernel were filtered out, not just
the symbols starting with 'dot + L'.
For example:
ashimida@ubuntu:~/linux$ cat System.map |grep ' .L'
ashimida@ubuntu:~/linux$ nm -n vmlinux |grep ' .L' ffff0000088028e0 t bLength_show
...... ffff0000092e0408 b PLLP_OUTC_lock ffff0000092e0410 b PLLP_OUTA_lock
The original intent should be to filter out all local symbols
starting with '.L', so the dot should be escaped.
Fixes: 00902e984732 ("mksysmap: Add h8300 local symbol pattern") Signed-off-by: ashimida <ashimida@linux.alibaba.com> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When the commit adding ntb_default_port_number() and
ntb_default_peer_port_number() entered the kernel there was no
users of it so it was impossible to tell what the API needed.
When a user finally landed a year later (ntb_pingpong) there were
more NTB topologies were created and no consideration was considered
to how other drivers had changed.
Now that there is a user it can be fixed to provide a sensible default
for the legacy drivers that do not implement ntb_{peer_}port_number().
Seeing ntb_pingpong doesn't check error codes returning EINVAL was also
not sensible.
Patches for ntb_pingpong and ntb_perf follow (which are broken
otherwise) to support hardware that doesn't have port numbers. This is
important not only to not break support with existing drivers but for
the cross link topology which, due to its perfect symmetry, cannot
assign unique port numbers to each side.
Fixes: 1e5301196a88 ("NTB: Add indexed ports NTB API") Signed-off-by: Logan Gunthorpe <logang@deltatee.com> Acked-by: Allen Hubbe <allenbh@gmail.com> Tested-by: Alexander Fomichev <fomichev.ru@gmail.com> Signed-off-by: Jon Mason <jdmason@kudzu.us> Signed-off-by: Sasha Levin <sashal@kernel.org>
This commit fixes pingpong support for existing drivers that do not
implement ntb_default_port_number() and ntb_default_peer_port_number().
This is required for hardware (like the crosslink topology of
switchtec) which cannot assign reasonable port numbers to each port due
to its perfect symmetry.
Instead of picking the doorbell to use based on the the index of the
peer, we use the peer's port number. This is a bit clearer and easier
to understand.
Fixes: c7aeb0afdcc2 ("NTB: ntb_pp: Add full multi-port NTB API support") Signed-off-by: Logan Gunthorpe <logang@deltatee.com> Acked-by: Allen Hubbe <allenbh@gmail.com> Tested-by: Alexander Fomichev <fomichev.ru@gmail.com> Signed-off-by: Jon Mason <jdmason@kudzu.us> Signed-off-by: Sasha Levin <sashal@kernel.org>
If register_netdev(dev) fails, free_netdev(dev) needs
to be called, otherwise a memory leak will occur.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Wang Hai <wanghai38@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The pages backing page-table allocations for SRMMU are allocated via
memblock as part of the "nocache" region initialisation during
srmmu_paging_init() and should not be freed even if a later call to
pgtable_pte_page_ctor() fails.
Remove the broken call to __free_page().
Cc: David S. Miller <davem@davemloft.net> Cc: Kirill A. Shutemov <kirill@shutemov.name> Fixes: 1ae9ae5f7df7 ("sparc: handle pgtable_page_ctor() fail") Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Even in failed case of pm_runtime_get_sync(), the usage_count is
incremented. In order to keep the usage_count with correct value call
appropriate pm_runtime_put().
'mem=" option is an easy way to put high pressure on memory during
some test. Hence after applying the memory limit, instead of total
mem, the actual usable memory should be considered when reserving mem
for crashkernel. Otherwise the boot up may experience OOM issue.
E.g. it would reserve 4G prior to the change and 512M afterward, if
passing
crashkernel="2G-4G:384M,4G-16G:512M,16G-64G:1G,64G-128G:2G,128G-:4G",
and mem=5G on a 256G machine.
This issue is powerpc specific because it puts higher priority on
fadump and kdump reservation than on "mem=". Referring the following
code:
if (fadump_reserve_mem() == 0)
reserve_crashkernel();
...
/* Ensure that total memory size is page-aligned. */
limit = ALIGN(memory_limit ?: memblock_phys_mem_size(), PAGE_SIZE);
memblock_enforce_memory_limit(limit);
While on other arches, the effect of "mem=" takes a higher priority
and pass through memblock_phys_mem_size() before calling
reserve_crashkernel().
where its target list contains NB names ("FS0" & "FS1") rather than
FQDN ones ("FS0.FOO.COM" & "FS1.FOO.COM"), we end up connecting to
\FOO\share1 but server->hostname will have "FOO.COM". The reason is
because both "FS0" and "FS0.FOO.COM" resolve to same IP address and
they share same TCP server connection, but "FS0.FOO.COM" was the first
hostname set -- which is OK.
However, if the echo thread timeouts and we still have a good
connection to "FS0", in cifs_reconnect()
it successfully reconnects to "FS0" server but does not set up next
DFS target - which should be the same target server "\FS0\share1" -
and server->hostname remains set to "FS0.FOO.COM" rather than "FS0",
as reconn_inval_dfs_target() would have it set to "FS0" if called
earlier.
Finally, in __smb2_reconnect(), the reconnect of tcons would fail
because tcon->ses->server->hostname (FS0.FOO.COM) does not match DFS
target's hostname (FS0).
Fix that by calling reconn_inval_dfs_target() before
generic_ip_connect() so server->hostname will get updated correctly
prior to reconnecting its tcons in __smb2_reconnect().
With "cifs: handle hostnames that resolve to same ip in failover"
patch
- The above problem would not occur.
- We could save an DNS query to find out that they both resolve to
the same ip address.
Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz> Reviewed-by: Aurelien Aptel <aaptel@suse.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
kobject_init_and_add() takes reference even when it fails. If this
function returns an error, kobject_put() must be called to properly clean
up the memory associated with the object. Previous
commit b8eb718348b8 ("net-sysfs: Fix reference count leak in
rx|netdev_queue_add_kobject") fixed a similar problem.
Link: https://lore.kernel.org/r/20200528030231.9082-1-wu000273@umn.edu Signed-off-by: Qiushi Wu <wu000273@umn.edu> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Versions of VMD with the Host Physical Address shadow register use this
register to calculate the bus address offset needed to do guest
passthrough of the domain. This register shadows the Host Physical
Address registers including the resource type bits. After calculating
the offset, the extra resource type bits lead to the VMD resources being
over-provisioned at the front and under-provisioned at the back.
nfsd4_process_cb_update() invokes svc_xprt_get(), which increases the
refcount of the "c->cn_xprt".
The reference counting issue happens in one exception handling path of
nfsd4_process_cb_update(). When setup callback client failed, the
function forgets to decrease the refcnt increased by svc_xprt_get(),
causing a refcnt leak.
Fix this issue by calling svc_xprt_put() when setup callback client
failed.
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn> Signed-off-by: Xin Tan <tanxin.ctf@gmail.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 2b206ee6b0df ("powerpc/perf/hv-24x7: Display change in counter
values")' added to print _change_ in the counter value rather then raw
value for 24x7 counters. Incase of transactions, the event count
is set to 0 at the beginning of the transaction. It also sets
the event's prev_count to the raw value at the time of initialization.
Because of setting event count to 0, we are seeing some weird behaviour,
whenever we run multiple 24x7 events at a time.
As we are setting event_count to 0, for interval case, overall event_count is not
coming in incremental order. As we may can get new delta lesser then previous count.
Because of which when we print intervals, we are getting negative value which create
these large values.
This patch removes part where we set event_count to 0 in function
'h_24x7_event_read'. There won't be much impact as we do set event->hw.prev_count
to the raw value at the time of initialization to print change value.
The commit citied in the Fixes line wasn't complete and solved
only part of the problems. Update the mlx5_ib to properly support
MLX5_CMD_OP_INIT2INIT_QP command in the DEVX, that is required when
modify the QP tx_port_affinity.
Fixes: 819f7427bafd ("RDMA/mlx5: Add init2init as a modify command") Link: https://lore.kernel.org/r/20200527135703.482501-1-leon@kernel.org Signed-off-by: Mark Zhang <markz@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
vhost-scsi pre-allocates the maximum sg entries per command and if a
command requires more than VHOST_SCSI_PREALLOC_SGLS entries, then that
command is failed by it. This patch lets vhost communicate the max sg limit
when it registers vhost_scsi_ops with TCM. With this change, TCM would
report the max sg entries through "Block Limits" VPD page which will be
typically queried by the SCSI initiator during device discovery. By knowing
this limit, the initiator could ensure the maximum transfer length is less
than or equal to what is reported by vhost-scsi.
Link: https://lore.kernel.org/r/1590166317-953-1-git-send-email-sudhakar.panneerselvam@oracle.com Cc: Michael S. Tsirkin <mst@redhat.com> Cc: Jason Wang <jasowang@redhat.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Stefan Hajnoczi <stefanha@redhat.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Signed-off-by: Sudhakar Panneerselvam <sudhakar.panneerselvam@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In order to create or activate a new node, lpfc_els_unsol_buffer() invokes
lpfc_nlp_init() or lpfc_enable_node() or lpfc_nlp_get(), all of them will
return a reference of the specified lpfc_nodelist object to "ndlp" with
increased refcnt.
When lpfc_els_unsol_buffer() returns, local variable "ndlp" becomes
invalid, so the refcount should be decreased to keep refcount balanced.
The reference counting issue happens in one exception handling path of
lpfc_els_unsol_buffer(). When "ndlp" in DEV_LOSS, the function forgets to
decrease the refcnt increased by lpfc_nlp_init() or lpfc_enable_node() or
lpfc_nlp_get(), causing a refcnt leak.
Fix this issue by calling lpfc_nlp_put() when "ndlp" in DEV_LOSS.
Link: https://lore.kernel.org/r/1590416184-52592-1-git-send-email-xiyuyang19@fudan.edu.cn Reviewed-by: Daniel Wagner <dwagner@suse.de> Reviewed-by: James Smart <james.smart@broadcom.com> Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn> Signed-off-by: Xin Tan <tanxin.ctf@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
There wasn't any clean up done if cxgb3_alloc_atid() failed and also the
original code didn't release "csk->l2t".
Link: https://lore.kernel.org/r/20200521121221.GA247492@mwanda Fixes: 6f7efaabefeb ("[SCSI] cxgb3i: change cxgb3i to use libcxgbi") Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
WM8994 chip has built-in regulators, which might be used for chip
operation. They are controlled by a separate wm8994-regulator driver,
which should be loaded before this driver calls regulator_get(), because
that driver also provides consumer-supply mapping for the them. If that
driver is not yet loaded, regulator core substitute them with dummy
regulator, what breaks chip operation, because the built-in regulators are
never enabled. Fix this by annotating this driver with MODULE_SOFTDEP()
"pre" dependency to "wm8994_regulator" module.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Charles Keepax <ckeepax@opensource.cirrus.com> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The threaded interrupt handler may still be called after the
usb_gadget_disconnect is called, it causes the structures used
at interrupt handler was freed before it uses, eg the
usb_request. This issue usually occurs we remove the udc function
during the transfer. Below is the example when doing stress
test for android switch function, the EP0's request is freed
by .unbind (configfs_composite_unbind -> composite_dev_cleanup),
but the threaded handler accesses this request during handling
setup packet request.
In fact, there is no protection between unbind the udc
and udc interrupt handling, so we have to avoid the interrupt
handler is occurred or scheduled during the .unbind flow.
init: Sending signal 9 to service 'adbd' (pid 18077) process group...
android_work: did not send uevent (0 0 000000007bec2039)
libprocessgroup: Successfully killed process cgroup uid 0 pid 18077 in 6ms
init: Service 'adbd' (pid 18077) received signal 9
init: Sending signal 9 to service 'adbd' (pid 18077) process group...
libprocessgroup: Successfully killed process cgroup uid 0 pid 18077 in 0ms
init: processing action (init.svc.adbd=stopped) from (/init.usb.configfs.rc:14)
init: Received control message 'start' for 'adbd' from pid: 399 (/vendor/bin/hw/android.hardware.usb@1.
Reviewed-by: Jun Li <jun.li@nxp.com> Signed-off-by: Peter Chen <peter.chen@nxp.com> Signed-off-by: Felipe Balbi <balbi@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Add missed acpi_gpiochip_free_interrupts() call when unregistering ports.
While at it, drop extra check to call acpi_gpiochip_request_interrupts().
There is no need to have an additional check to call
acpi_gpiochip_request_interrupts(). Even without any interrupts available
the registered ACPI Event handlers can be useful for debugging purposes.
This patch fixes two bit conflicts in the pci-bridge-emul driver:
1. Bit 3 of Device Status (19 of Device Control) is marked as both
Write-1-to-Clear and Read-Only. It should be Write-1-to-Clear.
The Read-Only and Reserved bitmasks are shifted by 1 bit due to this
error.
2. Bit 12 of Slot Control is marked as both Read-Write and Reserved.
It should be Read-Write.
We found out that after phy up, the hardware reports another oob interrupt
but did not follow a phy up interrupt:
oob ready -> phy up -> DEV found -> oob read -> wait phy up -> timeout
We run link reset when wait phy up timeout, and it send a normal disk into
reset processing. So we made some circumvention action in the code, so that
this abnormal oob interrupt will not start the timer to wait for phy up.
Missing INIT2INIT entry in the list of modify commands caused DEVX
applications to be unable to modify_qp for this transition state. Add the
MLX5_CMD_OP_INIT2INIT_QP opcode to the list of allowed DEVX opcodes.
Fixes: e662e14d801b ("IB/mlx5: Add DEVX support for modify and query commands") Link: https://lore.kernel.org/r/20200513095550.211345-1-leon@kernel.org Signed-off-by: Aharon Landau <aharonl@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
On some QCOM platforms like SC7180, SDM845 and SM8150,
reading TMC mode register without proper coresight power
management can lead to async exceptions like the one in
the call trace below in tmc_read_prepare_etb(). This can
happen if the user tries to read the TMC etf data via
device node without setting up source and the sink first.
Fix this by having a check for coresight sysfs mode
before reading TMC mode management register.
The Arm Ltd. boards were using an outdated address convention in the DT
node names, by separating the high from the low 32-bits of an address by
a comma.
Remove the comma from the node name suffix to be DT spec compliant.
Since commit dcebd755926b ("block: use bio_for_each_bvec() to compute
multi-page bvec count"), the kernel will bug_on on the PS3 because
bio_split() is called with sectors == 0:
The problem originates from setting the segment boundary of the
request queue to -1UL. This makes get_max_segment_size() return zero
when offset is zero, whatever the max segment size. The test with
BLK_SEG_BOUNDARY_MASK fails and 'mask - (mask & offset) + 1' overflows
to zero in the return statement.
Not setting the segment boundary and using the default
value (BLK_SEG_BOUNDARY_MASK) fixes the problem.
Trying to change Link Status register does not have any effect as this
is a read-only register. Trying to overwrite bits for Negotiated Link
Width does not make sense.
In future proper change of link width can be done via Lane Count Select
bits in PCIe Control 0 register.
Trying to unconditionally enable ASPM L0s via ASPM Control bits in Link
Control register is wrong. There should be at least some detection if
endpoint supports L0s as isn't mandatory.
Moreover ASPM Control bits in Link Control register are controlled by
pcie/aspm.c code which sets it according to system ASPM settings,
immediately after aardvark driver probes. So setting these bits by
aardvark driver has no long running effect.
Remove code which touches ASPM L0s bits from this driver and let
kernel's ASPM implementation to set ASPM state properly.
Some users are reporting issues that this code is problematic for some
Intel wifi cards and removing it fixes them, see e.g.:
https://bugzilla.kernel.org/show_bug.cgi?id=196339
If problems with Intel wifi cards occur even after this commit, then
pcie/aspm.c code could be modified / hooked to not enable ASPM L0s state
for affected problematic cards.
Link: https://lore.kernel.org/r/20200430080625.26070-3-pali@kernel.org Tested-by: Tomasz Maciej Nowak <tmn505@gmail.com> Signed-off-by: Pali Rohár <pali@kernel.org> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Acked-by: Rob Herring <robh@kernel.org> Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The GIC DT nodes for the fastmodels were not fully compliant with the
DT binding, which has certain expectations about child nodes and their
size and address cells values.
Use smaller #address-cells and #size-cells values, as the binding
requests, and adjust the reg properties accordingly.
This requires adjusting the interrupt nexus nodes as well, as one
field of the interrupt-map property depends on the GIC's address-size.
Since the .dts files share interrupt nexus nodes across different
interrupt controllers (GICv2 vs. GICv3), we need to use the only
commonly allowed #address-size value of <1> for both.
This provides a better separation between runtime and PM sleep
callbacks.
Only do nothing if given runtime flag is set and calback is not set.
With the current implementation, if PM sleep callback is set but runtime
callback is not set then at runtime resume we reload the firmware even
if we do not support runtime resume callback.
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com> Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Link: https://lore.kernel.org/r/20200515135958.17511-2-kai.vehmanen@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
On SoCs with Standby Control Registers (STBCRs) instead of Module Stop
Control Registers (MSTPCRs), the suspend handler saves the wrong
registers, and the resume handler prints the wrong register in an error
message.
Fortunately this cannot happen yet, as the suspend/resume code is used
on PSCI systems only, and systems with STBCRs (RZ/A1 and RZ/A2) do not
use PSCI. Still, it is better to fix this, to avoid this becoming a
problem in the future.
Distinguish between STBCRs and MSTPCRs where needed. Replace the
useless printing of the virtual register address in the resume error
message by printing the register index.
This fixes a problem with using the GPIO as an interrupt on Jaguar2
(and similar), as the register layout of the platforms with 64 GPIO's
are pairwise, such that the original offset must be multiplied with
the platform stride.
The GIC DT nodes for the Juno boards were not fully compliant with
the DT binding, which has certain expectations about child nodes and
their size and address cells values.
Use smaller #address-cells and #size-cells values, as the binding
requests, and adjust the reg properties accordingly.
This requires adjusting the interrupt nexus nodes as well, as one
field of the interrupt-map property depends on the GIC's address-size.
Use sdhci-caps-mask to forbid SDR104 mode on the SDIO capable SDHCI
controller. Without this the device cannot pass electromagnetic
interference certifications.
Fixes: 7109d817db2e ("arm64: dts: marvell: add DTS for Turris Mox") Signed-off-by: Marek Behún <marek.behun@nic.cz> Cc: Gregory CLEMENT <gregory.clement@bootlin.com> Signed-off-by: Gregory CLEMENT <gregory.clement@bootlin.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
SCSI LUN passthrough code such as qemu's "scsi-block" device model
pass every IO to the host via SG_IO ioctls. Currently, dm-multipath
calls choose_pgpath() only in the block IO code path, not in the ioctl
code path (unless current_pgpath is NULL). This has the effect that no
path switching and thus no load balancing is done for SCSI-passthrough
IO, unless the active path fails.
Fix this by using the same logic in multipath_prepare_ioctl() as in
multipath_clone_and_map().
Note: The allegedly best path selection algorithm, service-time,
still wouldn't work perfectly, because the io size of the current
request is always set to 0. Changing that for the IO passthrough
case would require the ioctl cmd and arg to be passed to dm's
prepare_ioctl() method.
Signed-off-by: Martin Wilck <mwilck@suse.com> Reviewed-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
fastrpc_invoke_ctx can have refcount of 2 in error path where
rpmsg_send() fails to send invoke message. decrement the refcount
properly in the error path to fix this leak.
This also fixes below static checker warning:
drivers/misc/fastrpc.c:990 fastrpc_internal_invoke()
warn: 'ctx->refcount.refcount.ref.counter' not decremented on lines: 990.
Fixes: c68cfb718c8f ("misc: fastrpc: Add support for context") Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org> Link: https://lore.kernel.org/r/20200512110930.2550-1-srinivas.kandagatla@linaro.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
fastrpc_channel_ctx is not freed if misc_register() fails, this would
lead to a memory leak. Fix this leak by adding kfree in misc_register()
error path.
Standard 8250 UART ports are designed in a way so they can communicate
with baud rates up to 1/16 of a reference frequency. It's expected from
most of the currently supported UART controllers. That's why the former
version of serial8250_get_baud_rate() method called uart_get_baud_rate()
with min and max baud rates passed as (port->uartclk / 16 / UART_DIV_MAX)
and ((port->uartclk + tolerance) / 16) respectively. Doing otherwise, like
it was suggested in commit ("serial: 8250_mtk: support big baud rate."),
caused acceptance of bauds, which was higher than the normal UART
controllers actually supported. As a result if some user-space program
requested to set a baud greater than (uartclk / 16) it would have been
permitted without truncation, but then serial8250_get_divisor(baud)
(which calls uart_get_divisor() to get the reference clock divisor) would
have returned a zero divisor. Setting zero divisor will cause an
unpredictable effect varying from chip to chip. In case of DW APB UART the
communications just stop.
Lets fix this problem by getting back the limitation of (uartclk +
tolerance) / 16 maximum baud supported by the generic 8250 port. Mediatek
8250 UART ports driver developer shouldn't have touched it in the first
place notably seeing he already provided a custom version of set_termios()
callback in that glue-driver which took into account the extended baud
rate values and accordingly updated the standard and vendor-specific
divisor latch registers anyway.
Fixes: 81bb549fdf14 ("serial: 8250_mtk: support big baud rate.") Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru> Cc: Alexey Malahov <Alexey.Malahov@baikalelectronics.ru> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Paul Burton <paulburton@kernel.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Long Cheng <long.cheng@mediatek.com> Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Cc: Maxime Ripard <mripard@kernel.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Will Deacon <will@kernel.org> Cc: Russell King <linux@armlinux.org.uk> Cc: linux-mips@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org Cc: linux-mediatek@lists.infradead.org Link: https://lore.kernel.org/r/20200506233136.11842-2-Sergey.Semin@baikalelectronics.ru Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The TOP 'aclk*_isp', 'aclk550_cam', 'gscl_wa' and 'gscl_wb' clocks must
be kept enabled all the time to allow proper access to power management
control for the ISP and CAM power domains. The last two clocks, although
related to GScaler device and GSCL power domain, provides also the
I_WRAP_CLK signal to MIPI CSIS0/1 devices, which are a part of CAM power
domain and are needed for proper power on/off sequence.
Currently there are no drivers for the devices, which are part of CAM and
ISP power domains yet. This patch only fixes the race between disabling
the unused power domains and disabling unused clocks, which randomly
resulted in the following error during boot:
Power domain CAM disable failed
Power domain ISP disable failed
Fixes: 318fa46cc60d ("clk/samsung: exynos542x: mark some clocks as critical") Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Chanwoo Choi <cw00.choi@samsung.com> Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The application processor accessing the mpss region when the Q6 modem is
running will lead to an XPU violation. Fix this by un-mapping the mpss
segments post copy during mpss authentication and coredumps.
Tested-by: Evan Green <evgreen@chromium.org> Signed-off-by: Sibi Sankar <sibis@codeaurora.org> Link: https://lore.kernel.org/r/20200415071619.6052-1-sibis@codeaurora.org Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
If we timeout during a message transfer, the control register may
contain bits that cause an action to be set. Read-modify-writing the
register leaving these bits set may trigger the hardware to attempt
one of these actions unintentionally.
Always clear these bits when cleaning up after a message or after
a timeout.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk> Signed-off-by: Wolfram Sang <wsa@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Limit the output of humidity compensation to the range between 0 and 100
percent.
Depending on the calibration parameters of the individual sensor it
happens, that a humidity above 100 percent or below 0 percent is
calculated, which don't make sense in terms of relative humidity.
Add a clamp to the compensation formula as described in the datasheet of
the sensor in chapter 4.2.3.
Although this clamp is documented, it was never in the driver of the
kernel.
It depends on the circumstances (calibration parameters, temperature,
humidity) if one can see a value above 100 percent without the clamp.
The writer of this patch was working with this type of sensor without
noting this error. So it seems to be a rare event when this bug occures.
Signed-off-by: Andreas Klinger <ak@it-klinger.de> Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In function mc13xxx_rtc_probe, the mc13xxx_unlock() is called
before rtc_register_device(). But in the error path of
rtc_register_device(), the mc13xxx_unlock() is called again,
which causes a double-unlock problem. Thus add a call of the
function “mc13xxx_lock” in an if branch for the completion
of the exception handling.
Since commit b6eba86030bf ("Input: edt-ft5x06 - add offset support for
ev-ft5726") offset-x and offset-y is supported. Devices using those
offset parameters don't support the offset parameter so we need to add
the NO_REGISTER check for edt_ft5x06_ts_get_defaults().
Fixes: b6eba86030bf ("Input: edt-ft5x06 - add offset support for ev-ft5726") Signed-off-by: Marco Felsch <m.felsch@pengutronix.de> Link: https://lore.kernel.org/r/20200227112819.16754-2-m.felsch@pengutronix.de Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The goal of the following command sequence is to restart the adapter.
However, the tgt_stop flag remains set, indicating that the adapter is
still in stopping state even after re-enabling it.
qlt_handle_cmd_for_atio() rejects the request to send commands because the
adapter is in the stopping state:
kernel: PID 0:qla_target.c:4442 qlt_handle_cmd_for_atio(): tgt_stop 0x1, tgt_stopped 0x0
kernel: qla2xxx [0001:00:02.0]-3861:1: PID 0:qla_target.c:4447: New command while device c000000005314600 is shutting down
kernel: qla2xxx [0001:00:02.0]-e85f:1: PID 0:qla_target.c:5728: qla_target: Unable to send command to target
This patch calls qla_stop_phase2() in addition to qlt_stop_phase1() in
tcm_qla2xxx_tpg_enable_store() and tcm_qla2xxx_npiv_tpg_enable_store(). The
qlt_stop_phase1() marks adapter as stopping (tgt_stop == 0x1, tgt_stopped
== 0x0) but qlt_stop_phase2() marks adapter as stopped (tgt_stop == 0x0,
tgt_stopped == 0x1).
Link: https://lore.kernel.org/r/52be1e8a3537f6c5407eae3edd4c8e08a9545ea5.camel@yadro.com Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com> Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com> Signed-off-by: Viacheslav Dubeyko <v.dubeiko@yadro.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When resizing a BAR, pci_reassign_bridge_resources() is invoked to bring
the bridge windows of parent bridges in line with the new BAR assignment.
This assumes the device whose BAR is being resized lives on a subordinate
bus, but this is not necessarily the case. A device may live on the root
bus, in which case dev->bus->self is NULL, and passing a NULL pci_dev
pointer to pci_reassign_bridge_resources() will cause it to crash.
So let's make the call to pci_reassign_bridge_resources() conditional on
whether dev->bus->self is non-NULL in the first place.
Fixes: 8bb705e3e79d84e7 ("PCI: Add pci_resize_resource() for resizing BARs") Link: https://lore.kernel.org/r/20200421162256.26887-1-ardb@kernel.org Signed-off-by: Ard Biesheuvel <ardb@kernel.org> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The GX and AXG SCP sram nodes were using invalid compatible and
node names for the sram entries.
Fixup the sram entries node names, and use proper compatible for them.
It notably fixes:
sram@c8000000: 'scp-shmem@0', 'scp-shmem@200' do not match any of the regexes: '^([a-z]*-)?sram(-section)?@[a-f0-9]+$', 'pinctrl-[0-9]+'
Smatch complains that the "path_data->handle" variable is user controlled.
It comes from iscsi_set_path() so that seems possible. It's harmless to
add a limit check.
The qedi->ep_tbl[] array has qedi->max_active_conns elements (which is
always ISCSI_MAX_SESS_PER_HBA (4096) elements). The array is allocated in
the qedi_cm_alloc_mem() function.
Link: https://lore.kernel.org/r/20200428131939.GA696531@mwanda Fixes: ace7f46ba5fd ("scsi: qedi: Add QLogic FastLinQ offload iSCSI driver framework.") Acked-by: Manish Rangankar <mrangankar@marvell.com> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
In case scsi_setup_fs_cmnd() fails we're not freeing the sgtables allocated
by scsi_init_io(), thus we leak the allocated memory.
Free the sgtables allocated by scsi_init_io() in case scsi_setup_fs_cmnd()
fails.
Technically scsi_setup_scsi_cmnd() does not suffer from this problem as it
can only fail if scsi_init_io() fails, so it does not have sgtables
allocated. But to maintain symmetry and as a measure of defensive
programming, free the sgtables on scsi_setup_scsi_cmnd() failure as well.
scsi_mq_free_sgtables() has safeguards against double-freeing of memory so
this is safe to do.
While we're at it, rename scsi_mq_free_sgtables() to scsi_free_sgtables().
The CMA and DMA_CMA Kconfig options need to be selected
by the Integrator in order to produce boot console on some
Integrator systems.
The REGULATOR and REGULATOR_FIXED_VOLTAGE need to be
selected in order to boot the system from an external
MMC card when using MMCI/PL181 from the device tree
probe path.
Select these things directly from the Kconfig so we are
sure to be able to bring the systems up with console
from any device tree.
davinci_mcasp_get_dma_type() invokes dma_request_chan(), which returns a
reference of the specified dma_chan object to "chan" with increased
refcnt.
When davinci_mcasp_get_dma_type() returns, local variable "chan" becomes
invalid, so the refcount should be decreased to keep refcount balanced.
The reference counting issue happens in one exception handling path of
davinci_mcasp_get_dma_type(). When chan device is NULL, the function
forgets to decrease the refcnt increased by dma_request_chan(), causing
a refcnt leak.
Fix this issue by calling dma_release_channel() when chan device is
NULL.
Fix this by ensuring that the regulators are disabled, if enabled, on
probe failure.
Finally, ensure that the vddio regulator is disabled in the driver
remove handler.
Signed-off-by: Jon Hunter <jonathanh@nvidia.com> Reviewed-by: Daniel Thompson <daniel.thompson@linaro.org> Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
During the process of debugging a processor derived from the msm8916 which
we found the new processor was not starting one of its PLLs.
After tracing the addresses and writes that downstream was doing and
comparing to upstream it became obvious that we were writing to a different
register location than downstream when trying to configure the PLL.
This error is also present in upstream msm8916.
As an example clk-pll.c::clk_pll_recalc_rate wants to write to
pll->config_reg updating the bit-field POST_DIV_RATIO. That bit-field is
defined in PLL_USER_CTL not in PLL_CONFIG_CTL. Taking the BIMC PLL as an
example
If ida_simple_get() returns an error when called in rproc_alloc(),
put_device() is called to clean things up. By this time the rproc
device type has been assigned, with rproc_type_release() as the
release function.
The first thing rproc_type_release() does is call:
idr_destroy(&rproc->notifyids);
But at the time the ida_simple_get() call is made, the notifyids
field in the remoteproc structure has not been initialized.
I'm not actually sure this case causes an observable problem, but
it's incorrect. Fix this by initializing the notifyids field before
calling ida_simple_get() in rproc_alloc().
Fixes: b5ab5e24e960 ("remoteproc: maintain a generic child device for each rproc") Signed-off-by: Alex Elder <elder@linaro.org> Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org> Reviewed-by: Suman Anna <s-anna@ti.com> Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org> Link: https://lore.kernel.org/r/20200415204858.2448-2-mathieu.poirier@linaro.org Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / users / dwmw2 / linux.git / log