Update the size used in 'dma_free_coherent()' in order to match the one
used in the corresponding 'dma_alloc_coherent()', in
'spider_net_init_chain()'.
Fixes: d4ed8f8d1fb7 ("Spidernet DMA coalescing") Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Update the size used in 'dma_free_coherent()' in order to match the one
used in the corresponding 'dma_alloc_coherent()'.
Fixes: 369a782af0f1 ("net: sgi: ioc3-eth: ensure tx ring is 16k aligned.") Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
On an error exit path, a negative error code should be returned
instead of a positive return value.
Fixes: 0c45d7fe12c7e ("liquidio: fix use of pf in pass-through mode in a virtual machine") Cc: Rick Farrington <ricardo.farrington@cavium.com> Signed-off-by: Tianjia Zhang <tianjia.zhang@linux.alibaba.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
In function hw_atl_a0_hw_multicast_list_set(), when an invalid
request is encountered, a negative error code should be returned.
Fixes: bab6de8fd180b ("net: ethernet: aquantia: Atlantic A0 and B0 specific functions") Cc: David VomLehn <vomlehn@texas.net> Signed-off-by: Tianjia Zhang <tianjia.zhang@linux.alibaba.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
I was attempting to use pid filtering with function_graph, but it wasn't
allowing anything to make it through. Turns out ftrace_trace_task
returns false if ftrace_ignore_pid is not-empty, which isn't correct
anymore. We're now setting it to FTRACE_PID_IGNORE if we need to ignore
that pid, otherwise it's set to the pid (which is weird considering the
name) or to FTRACE_PID_TRACE. Fix the check to check for !=
FTRACE_PID_IGNORE. With this we can now use function_graph with pid
filtering.
Link: https://lkml.kernel.org/r/20200725005048.1790-1-josef@toxicpanda.com Fixes: 717e3f5ebc82 ("ftrace: Make function trace pid filtering a bit more exact") Signed-off-by: Josef Bacik <josef@toxicpanda.com> Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The WARN_*() macros are intended to catch impossible situations
from the SW point of view. They gave a little in case HW<->SW interface
is out-of-sync.
Such out-of-sync scenario can be due to SW errors that are not part
of this flow or because some HW errors, where dump stack won't help
either.
This specific WARN_ON() is useless because mlx5_core code is prepared
to handle such situations and will unfold everything correctly while
providing enough information to the users to understand why FS is not
working.
The DR TX state machine supports the following order:
modify header, push vlan and encapsulation.
Instead fs_dr would pass:
push vlan, modify header and encapsulation.
The above caused the rule creation to fail on invalid action
sequence provided error.
Fixes: 6a48faeeca10 ("net/mlx5: Add direct rule fs_cmd implementation") Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed-by: Maor Gottlieb <maorg@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The '&&' command seems to have a bad effect when $(cmd_$(1)) exits with
non-zero effect: the command failure is masked (despite `set -e`) and all but
the first command of $(dep-cmd) is executed (successfully, as they are mostly
printfs), thus overall returning 0 in the end.
This means in practice that despite compilation errors, tools's build Makefile
will return success. We see this very reliably with libbpf's Makefile, which
doesn't get compilation error propagated properly. This in turns causes issues
with selftests build, as well as bpftool and other projects that rely on
building libbpf.
The fix is simple: don't use &&. Given `set -e`, we don't need to chain
commands with &&. The shell will exit on first failure, giving desired
behavior and propagating error properly.
Fixes: 275e2d95591e ("tools build: Move dependency copy into function") Signed-off-by: Andrii Nakryiko <andriin@fb.com> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Acked-by: Jiri Olsa <jolsa@redhat.com> Link: https://lore.kernel.org/bpf/20200731024244.872574-1-andriin@fb.com Signed-off-by: Sasha Levin <sashal@kernel.org>
wl1251_event_ps_report() should not always return 0 because
wl1251_ps_set_mode() may fail. Change it to return 'ret'.
Fixes: f7ad1eed4d4b ("wl1251: retry power save entry") Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Wang Hai <wanghai38@huawei.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/20200730073939.33704-1-wanghai38@huawei.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Add the missing platform_device_unregister() before return from
qtnf_core_mac_alloc() in the error handling case.
Fixes: 616f5701f4ab ("qtnfmac: assign each wiphy to its own virtual platform device") Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Wang Hai <wanghai38@huawei.com> Reviewed-by: Sergey Matyukevich <geomatsi@gmail.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/20200730064910.37589-1-wanghai38@huawei.com Signed-off-by: Sasha Levin <sashal@kernel.org>
The coex mechanism used to skip upon the freeze flag is raised.
That will cause the coex mechanism being skipped unexpectedly.
Coex only wanted to keep the TDMA table from being changed by
BT side.
So, check the freeze and reason, if the coex reason is coming
from BT info, skip it, to make sure the coex triggered by Wifi
itself can work.
This is required for the AP mode, while the control flow is
different with STA mode. When starting an AP mode, the AP mode
needs to start working immedaitely after leaving IPS, and the
freeze flag could be raised. If the coex info is skipped, then
the AP mode will not set the antenna owner, leads to TX stuck.
Fix the transmission is not sent with short GI under
some conditions even if the receiver supports short GI.
If VHT capability IE exists in the beacon, the original
code uses the short GI for 80M field as driver's short GI
setting for transmission, even the current bandwidth is
not 80MHz.
Short GI supported fields for 20M/40M are informed in HT
capability information element, and short GI supported
field for 80M is informed in VHT capability information
element.
These three fields may be set to different values.
Driver needs to record each short GI support field for
each bandwidth, and send correct info depends on current
bandwidth to the WiFi firmware.
Fixes: e3037485c68e ("rtw88: new Realtek 802.11ac driver") Signed-off-by: Tsang-Shian Lin <thlin@realtek.com> Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/20200717064937.27966-3-yhchuang@realtek.com Signed-off-by: Sasha Levin <sashal@kernel.org>
In the ice_init_hw_tbls, if the devm_kcalloc for es->written fails, catch
that error and bail out gracefully, instead of continuing with a NULL
pointer.
This fix has been added to address memory leak issues resulting from
triggering a sudden driver reset which does not allow us to follow our
normal removal flows for SW XLT entries for advanced features.
- Adding call to destroy flow profile locks when clearing SW XLT tables.
- Extraction sequence entries were not correctly cleared previously
which could cause ownership conflicts for repeated reset-replay calls.
Discard events that don't contain any entries. This shouldn't happen,
but subsequent code relies on being able to use entry 0. So better
be safe than accessing garbage.
Fixes: b4d72c08b358 ("qeth: bridgeport support - basic control") Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com> Reviewed-by: Alexandra Winter <wintera@linux.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
When preparing a buffer for RX refill, tolerate that it already has a
pool_entry attached. Otherwise we could easily leak such a pool_entry
when re-driving the RX refill after an error (from eg. do_qdio()).
This needs some minor adjustment in the code that drains RX buffer(s)
prior to RX refill and during teardown, so that ->pool_entry is NULLed
accordingly.
Fixes: 4a71df50047f ("qeth: new qeth device driver") Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The allocation order of things in soc_new_pcm_runtime was changed to
move the device_register before the allocation of the rtd structure.
This was to allow the rtd allocation to be managed by devm. However
currently the sysfs entries are added by device_register and their
visibility depends on variables within the rtd structure, this causes
the pmdown_time and dapm_widgets sysfs entries to be missing for all
rtds.
Correct this issue by manually calling device_add_groups after the
appropriate information is available.
Fixes: d918a37610b1 ("ASoC: soc-core: tidyup soc_new_pcm_runtime() alloc order") Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com> Link: https://lore.kernel.org/r/20200730120715.637-1-ckeepax@opensource.cirrus.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
After carefully checking, it appears that both tdmout and tdmin require the
rising edge of the sclk they get to be synchronized with the frame sync
event (which should be a rising edge of lrclk).
TDMIN was improperly set before this patch. Remove the sclk_invert quirk
which is no longer needed and fix the sclk phase.
Fixes: 1a11d88f499c ("ASoC: meson: add tdm formatter base driver") Signed-off-by: Jerome Brunet <jbrunet@baylibre.com> Link: https://lore.kernel.org/r/20200729154456.1983396-4-jbrunet@baylibre.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
After carefully checking the result provided by the TDMIN on the g12a and
sm1 SoC families, the TDMIN skew offset appears to be 3 instead of 2 on the
axg.
The size of the CPU affinity mask must be large enough for
systems with a very large number of CPUs. Otherwise, tests
which try to determine the first online CPU by calling
sched_getaffinity() will fail. This makes sure that the size
of the allocated affinity mask is dependent on the number of
CPUs as reported by get_nprocs_conf().
Fixes: 3752e453f6ba ("selftests/powerpc: Add tests of PMU EBBs") Reported-by: Shirisha Ganta <shiganta@in.ibm.com> Signed-off-by: Sandipan Das <sandipan@linux.ibm.com> Reviewed-by: Kamalesh Babulal <kamalesh@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/a408c4b8e9a23bb39b539417a21eb0ff47bb5127.1596084858.git.sandipan@linux.ibm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
In the unlikely event that the device tree lacks a /cpus node,
find_dlpar_cpus_to_add() oddly frees the cpu_drcs buffer it has been
passed before returning an error. Its only caller also frees the
buffer on error.
Remove the less conventional kfree() of a caller-supplied buffer from
find_dlpar_cpus_to_add().
The Armada 8K cpufreq driver needs the Armada AP CPU CLK
to work. This dependency is currently not satisfied and
the ARMADA_AP_CPU_CLK can not be selected independently.
Add it to the cpufreq Armada8k driver.
Fixes: f525a670533d ("cpufreq: ap806: add cpufreq driver for Armada 8K") Signed-off-by: Sven Auhagen <sven.auhagen@voleatech.de> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When dumping QPs bound to a counter, raw QPs should be allowed to dump
without the CAP_NET_RAW privilege. This is consistent with what "rdma res
show qp" does.
Fixes: c4ffee7c9bdb ("RDMA/netlink: Implement counter dumpit calback") Link: https://lore.kernel.org/r/20200727095828.496195-1-leon@kernel.org Signed-off-by: Mark Zhang <markz@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
On systems with large number of cpus, test fails trying to set
affinity by calling sched_setaffinity() with smaller size for affinity
mask. This patch fixes it by making sure that the size of allocated
affinity mask is dependent on the number of CPUs as reported by
get_nprocs().
Commit 866bfc75f40e ("powerpc: conditionally compile platform-specific
serial drivers") made some code depend on CONFIG_PPC_MPC52XX, which
doesn't exist.
For drivers that don't have the error handling callbacks we implement
recovery by removing the device and re-probing it. This causes the sysfs
directory for the PCI device to be removed which causes the following
spurious error to be printed when checking the PE state:
Breaking 0005:03:00.0...
./eeh-basic.sh: line 13: can't open /sys/bus/pci/devices/0005:03:00.0/eeh_pe_state: no such file
0005:03:00.0, waited 0/60
0005:03:00.0, waited 1/60
0005:03:00.0, waited 2/60
0005:03:00.0, waited 3/60
0005:03:00.0, waited 4/60
0005:03:00.0, waited 5/60
0005:03:00.0, waited 6/60
0005:03:00.0, waited 7/60
0005:03:00.0, Recovered after 8 seconds
We currently try to avoid this by checking if the PE state file exists
before reading from it. This is however inherently racy so re-work the
state checking so that we only read from the file once, and we squash any
errors that occur while reading.
In commit 8d3d7e2b35ea, we changed xfs_qm_dqpurge to bail out if we
can't lock the dquot buf to flush the dquot. This prevents the AIL from
blocking on the dquot, but it also forgets to clear the FREEING flag on
its way out. A subsequent purge attempt will see the FREEING flag is
set and bail out, which leads to dqpurge_all failing to purge all the
dquots.
(copy-pasting from Dave Chinner's identical patch)
This was found by inspection after having xfs/305 hang 1 in ~50
iterations in a quotaoff operation:
Returning -EAGAIN from xfs_qm_dqpurge() without clearing the
XFS_DQ_FREEING flag means the xfs_qm_dqpurge_all() code can never
free the dquot, and we loop forever waiting for the XFS_DQ_FREEING
flag to go away on the dquot that leaked it via -EAGAIN.
Fixes: 8d3d7e2b35ea ("xfs: trylock underlying buffer on dquot flush") Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Allison Collins <allison.henderson@oracle.com> Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The block reservation calculation for inode allocation is supposed
to consist of the blocks required for the inode chunk plus
(maxlevels-1) of the inode btree multiplied by the number of inode
btrees in the fs (2 when finobt is enabled, 1 otherwise).
Instead, the macro returns (ialloc_blocks + 2) due to a precedence
error in the calculation logic. This leads to block reservation
overruns via generic/531 on small block filesystems with finobt
enabled. Add braces to fix the calculation and reserve the
appropriate number of blocks.
Fixes: 9d43b180af67 ("xfs: update inode allocation/free transaction reservations for finobt") Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Alter the rtl8366_vlan_add() to call rtl8366_set_vlan()
inside the loop that goes over all VIDs since we now
properly support calling that function more than once.
Augment the loop to postincrement as this is more
intuitive.
The loop moved past the last VID but called
rtl8366_set_vlan() with the port number instead of
the VID, assuming a 1-to-1 correspondence between
ports and VIDs. This was also a bug.
Cc: DENG Qingfang <dqfext@gmail.com> Cc: Mauri Sandberg <sandberg@mailfence.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Fixes: d8652956cf37 ("net: dsa: realtek-smi: Add Realtek SMI driver") Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
The RTL8366 would not handle adding new members (ports) to
a VLAN: the code assumed that ->port_vlan_add() was only
called once for a single port. When intializing the
switch with .configure_vlan_while_not_filtering set to
true, the function is called numerous times for adding
all ports to VLAN1, which was something the code could
not handle.
Alter rtl8366_set_vlan() to just |= new members and
untagged flags to 4k and MC VLAN table entries alike.
This makes it possible to just add new ports to a
VLAN.
Put in some helpful debug code that can be used to find
any further bugs here.
Cc: DENG Qingfang <dqfext@gmail.com> Cc: Mauri Sandberg <sandberg@mailfence.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Fixes: d8652956cf37 ("net: dsa: realtek-smi: Add Realtek SMI driver") Signed-off-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Setting memdump state to idle prior to setting of callback function
pointer for command timeout to NULL,causing the issue.Now moved the
initialisation of memdump state to qca_setup().
Unregister from suspend notifications and cancel suspend preparations
before running hci_dev_do_close. Otherwise, the suspend notifier may
race with unregister and cause cmd_timeout even after hdev has been
freed.
For some reason they tend to squat on the very first CSR/
Cambridge Silicon Radio VID/PID instead of paying fees.
This is an extremely common problem; the issue goes as back as 2013
and these devices are only getting more popular, even rebranded by
reputable vendors and sold by retailers everywhere.
So, at this point in time there are hundreds of modern dongles reusing
the ID of what originally was an early Bluetooth 1.1 controller.
Linux is the only place where they don't work due to spotty checks
in our detection code. It only covered a minimum subset.
So what's the big idea? Take advantage of the fact that all CSR
chips report the same internal version as both the LMP sub-version and
HCI revision number. It always matches, couple that with the manufacturer
code, that rarely lies, and we now have a good idea of who is who.
Additionally, by compiling a list of user-reported HCI/lsusb dumps, and
searching around for legit CSR dongles in similar product ranges we can
find what CSR BlueCore firmware supported which Bluetooth versions.
That way we can narrow down ranges of fakes for each of them.
e.g. Real CSR dongles with LMP subversion 0x73 are old enough that
support BT 1.1 only; so it's a dead giveaway when some
third-party BT 4.0 dongle reuses it.
So, to sum things up; there are multiple classes of fake controllers
reusing the same 0A12:0001 VID/PID. This has been broken for a while.
Known 'fake' bcdDevices: 0x0100, 0x0134, 0x1915, 0x2520, 0x7558, 0x8891
IC markings on 0x7558: FR3191AHAL 749H15143 (???)
https://bugzilla.kernel.org/show_bug.cgi?id=60824
Fixes: 81cac64ba258ae (Deal with USB devices that are faking CSR vendor) Reported-by: Michał Wiśniewski <brylozketrzyn@gmail.com> Tested-by: Mike Johnson <yuyuyak@gmail.com> Tested-by: Ricardo Rodrigues <ekatonb@gmail.com> Tested-by: M.Hanny Sabbagh <mhsabbagh@outlook.com> Tested-by: Oussama BEN BRAHIM <b.brahim.oussama@gmail.com> Tested-by: Ismael Ferreras Morezuelas <swyterzone@gmail.com> Signed-off-by: Ismael Ferreras Morezuelas <swyterzone@gmail.com> Signed-off-by: Marcel Holtmann <marcel@holtmann.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
88pm860x_battery.c:522:19: warning: Assigned value is
garbage or undefined [core.uninitialized.Assign]
info->start_soc = soc;
^ ~~~
soc is set by calling calc_soc.
But calc_soc can return without setting soc.
So check the return status and bail similarly to other
checks in pm860x_init_battery and initialize soc to
silence the warning.
Fixes: a830d28b48bf ("power_supply: Enable battery-charger for 88pm860x") Signed-off-by: Tom Rix <trix@redhat.com> Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 1b79c5284439 ("PCI: cadence: Add host driver for Cadence PCIe
controller") in order to update Vendor ID, directly wrote to
PCI_VENDOR_ID register. However PCI_VENDOR_ID in root port configuration
space is read-only register and writing to it will have no effect.
Use local management register to configure Vendor ID and Subsystem Vendor
ID.
Link: https://lore.kernel.org/r/20200722110317.4744-10-kishon@ti.com Fixes: 1b79c5284439 ("PCI: cadence: Add host driver for Cadence PCIe controller") Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Reviewed-by: Rob Herring <robh@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
commit bd22885aa188 ("PCI: cadence: Refactor driver to use as a core
library") while refactoring the Cadence PCIe driver to be used as
library, removed pm_runtime_get_sync() from cdns_pcie_ep_setup()
and cdns_pcie_host_setup() but missed to remove the corresponding
pm_runtime_put_sync() in the error path. Fix it here.
Link: https://lore.kernel.org/r/20200722110317.4744-3-kishon@ti.com Fixes: bd22885aa188 ("PCI: cadence: Refactor driver to use as a core library") Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Reviewed-by: Rob Herring <robh@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The MV88E6097 chip does not support configuring jumbo frames. Prior to
commit 5f4366660d65 only the 6352, 6351, 6165 and 6320 chips configured
jumbo mode. The refactor accidentally added the function for the 6097.
Remove the erroneous function pointer assignment.
Fixes: 5f4366660d65 ("net: dsa: mv88e6xxx: Refactor setting of jumbo frames") Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
In theory, a panic can happen after a bus or host reset with dma_started
flag set. Fix this by halting the DMA before reinitializing the host.
Don't assume that ms->current_req is set when halt_dma() is invoked as it
may not hold for bus or host reset.
BTW, this particular Conner drive can be made to work by inhibiting
disconnect/reselect with 'mesh.resel_targets=0'.
To avoid a warning in free_irq, clear the affinity hint.
Link: https://lore.kernel.org/r/20200709133144.8363-1-thenzl@redhat.com Fixes: f0b9e7bdc309 ("scsi: megaraid_sas: Set affinity for high IOPS reply queues") Acked-by: Sumit Saxena <sumit.saxena@broadcom.com> Signed-off-by: Tomas Henzl <thenzl@redhat.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
As per UAC2 spec (ch. 4.7.2), wTotalLength of AC Interface
Header Descriptor "includes the combined length of this
descriptor header and all Clock Source, Unit and Terminal
descriptors."
Thus add its size to its wTotalLength.
Also after recent changes wTotalLength is calculated
dynamically, update static definition of uac2_ac_header_descriptor
accordingly
Fixes: 132fcb460839 ("usb: gadget: Add Audio Class 2.0 Driver") Signed-off-by: Ruslan Bilovol <ruslan.bilovol@gmail.com> Signed-off-by: Felipe Balbi <balbi@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When gadget registration fails, one should not call usb_del_gadget_udc().
Ensure this by setting gadget->udc to NULL. Also in case of a failure
there is no need to disable low-level hardware, so return immiedetly
instead of jumping to error_init label.
This fixes the following kernel NULL ptr dereference on gadget failure
(can be easily triggered with g_mass_storage without any module
parameters):
dwc2 12480000.hsotg: dwc2_check_params: Invalid parameter besl=1
dwc2 12480000.hsotg: dwc2_check_params: Invalid parameter g_np_tx_fifo_size=1024
dwc2 12480000.hsotg: EPs: 16, dedicated fifos, 7808 entries in SPRAM
Mass Storage Function, version: 2009/09/11
LUN: removable file: (no medium)
no file given for LUN0
g_mass_storage 12480000.hsotg: failed to start g_mass_storage: -22
8<--- cut here ---
Unable to handle kernel NULL pointer dereference at virtual address 00000104
pgd = (ptrval)
[00000104] *pgd=00000000
Internal error: Oops: 805 [#1] PREEMPT SMP ARM
Modules linked in:
CPU: 0 PID: 12 Comm: kworker/0:1 Not tainted 5.8.0-rc5 #3133
Hardware name: Samsung Exynos (Flattened Device Tree)
Workqueue: events deferred_probe_work_func
PC is at usb_del_gadget_udc+0x38/0xc4
LR is at __mutex_lock+0x31c/0xb18
...
Process kworker/0:1 (pid: 12, stack limit = 0x(ptrval))
Stack: (0xef121db0 to 0xef122000)
...
[<c076bf3c>] (usb_del_gadget_udc) from [<c0726bec>] (dwc2_hsotg_remove+0x10/0x20)
[<c0726bec>] (dwc2_hsotg_remove) from [<c0711208>] (dwc2_driver_probe+0x57c/0x69c)
[<c0711208>] (dwc2_driver_probe) from [<c06247c0>] (platform_drv_probe+0x6c/0xa4)
[<c06247c0>] (platform_drv_probe) from [<c0621df4>] (really_probe+0x200/0x48c)
[<c0621df4>] (really_probe) from [<c06221e8>] (driver_probe_device+0x78/0x1fc)
[<c06221e8>] (driver_probe_device) from [<c061fcd4>] (bus_for_each_drv+0x74/0xb8)
[<c061fcd4>] (bus_for_each_drv) from [<c0621b54>] (__device_attach+0xd4/0x16c)
[<c0621b54>] (__device_attach) from [<c0620c98>] (bus_probe_device+0x88/0x90)
[<c0620c98>] (bus_probe_device) from [<c06211b0>] (deferred_probe_work_func+0x3c/0xd0)
[<c06211b0>] (deferred_probe_work_func) from [<c0149280>] (process_one_work+0x234/0x7dc)
[<c0149280>] (process_one_work) from [<c014986c>] (worker_thread+0x44/0x51c)
[<c014986c>] (worker_thread) from [<c0150b1c>] (kthread+0x158/0x1a0)
[<c0150b1c>] (kthread) from [<c0100114>] (ret_from_fork+0x14/0x20)
Exception stack(0xef121fb0 to 0xef121ff8)
...
---[ end trace 9724c2fc7cc9c982 ]---
While fixing this also fix the double call to dwc2_lowlevel_hw_disable()
if dr_mode is set to USB_DR_MODE_PERIPHERAL. In such case low-level
hardware is already disabled before calling usb_add_gadget_udc(). That
function correctly preserves low-level hardware state, there is no need
for the second unconditional dwc2_lowlevel_hw_disable() call.
Fixes: 207324a321a8 ("usb: dwc2: Postponed gadget registration to the udc class driver") Acked-by: Minas Harutyunyan <hminas@synopsys.com> Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by: Felipe Balbi <balbi@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Update PCI device id for Jasper Lake processor thermal device.
With this proc_thermal driver is getting loaded and processor
thermal functionality works on Jasper Lake system.
if of_find_device_by_node() succeed, dwc3_octeon_device_init() doesn't have
a corresponding put_device(). Thus add put_device() to fix the exception
handling for this function implementation.
Fixes: 93e502b3c2d4 ("MIPS: OCTEON: Platform support for OCTEON III USB controller") Signed-off-by: Yu Kuai <yukuai3@huawei.com> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Commit 5bd70440cb0a ("ASoC: soc-dai: revert all changes to DAI
startup/shutdown sequence"), introduced a slight change of semantics
to DAI startup/shutdown. If startup() returns an error, shutdown()
is now called for the DAI.
This causes a deadlock in hdac_hda which issues a call to
snd_hda_codec_pcm_put() in case open fails. Upon error, soc_pcm_open()
will call shutdown(), and pcm_put() ends up getting called twice. Result
is a deadlock on pcm->open_mutex, as snd_device_free() gets called from
within snd_pcm_open(). Typical task backtrace looks like this:
Fixes: 5bd70440cb0a ("ASoC: soc-dai: revert all changes to DAI startup/shutdown sequence") Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com> Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by: Rander Wang <rander.wang@linux.intel.com> BugLink: https://github.com/thesofproject/linux/issues/2159 Link: https://lore.kernel.org/r/20200717101950.3885187-3-kai.vehmanen@linux.intel.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
"BPF_MAXINSNS: Maximum possible literals" unnecessarily falls back to
the interpreter because of failing sanity check in bpf_set_addr. The
problem is that there are a lot of branches that can be shrunk, and
doing so opens up the possibility to shrink even more. This process
does not converge after 3 passes, causing code offsets to change during
the codegen pass, which must never happen.
Fix by inserting nops during codegen pass in order to preserve code
offets.
"BPF_MAXINSNS: Maximum possible literals" test causes panic with
bpf_jit_harden = 2. The reason is that BPF_JMP | BPF_EXIT is always
emitted as brc, however, after removal of JITed image size
limitations, brcl might be required.
Both signed and unsigned variants of BPF_JMP | BPF_K require
sign-extending the immediate. JIT emits cgfi for the signed case,
which is correct, and clgfi for the unsigned case, which is not
correct: clgfi zero-extends the immediate.
s390 does not provide an instruction that does sign-extension and
unsigned comparison at the same time. Therefore, fix by first loading
the sign-extended immediate into work register REG_1 and proceeding
as if it's BPF_X.
The mvneta hardware appears to lock up in various random ways when
repeatedly switching speeds between 1G and 2.5G, which involves
reprogramming the COMPHY. It is not entirely clear why this happens,
but best guess is that reprogramming the COMPHY glitches mvneta clocks
causing the hardware to fail. It seems that rebooting resolves the
failure, but not down/up cycling the interface alone.
Various other approaches have been tried, such as trying to cleanly
power down the COMPHY and then take it back through the power up
initialisation, but this does not seem to help.
It was finally noticed that u-boot's last step when configuring a
COMPHY for "SGMII" mode was to poke at a register described as
"GBE_CONFIGURATION_REG", which is undocumented in any external
documentation. All that we have is the fact that u-boot sets a bit
corresponding to the "SGMII" lane at the end of COMPHY initialisation.
Experimentation shows that if we clear this bit prior to changing the
speed, and then set it afterwards, mvneta does not suffer this problem
on the SolidRun Clearfog when switching speeds between 1G and 2.5G.
This problem was found while script-testing phylink.
This fix also requires the corresponding change to DT to be effective.
See "ARM: dts: armada-38x: fix NETA lockup when repeatedly switching
speeds".
Fixes: 14dc100b4411 ("phy: armada38x: add common phy support") Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Link: https://lore.kernel.org/r/E1jxtRj-0003Tz-CG@rmk-PC.armlinux.org.uk Signed-off-by: Vinod Koul <vkoul@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fix potential memory leak in mcu message handler on error condition.
Fixes: 0e6a29e477f3 ("mt76: mt7615: add support to read temperature from mcu") Acked-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Sean Wang <sean.wang@mediatek.com> Signed-off-by: Felix Fietkau <nbd@nbd.name> Signed-off-by: Sasha Levin <sashal@kernel.org>
Compilation error:
arch/powerpc/perf/perf_regs.c:80:undefined reference to `.is_sier_available'
Currently is_sier_available() is part of core-book3s.c, which is added
to build based on CONFIG_PPC_PERF_CTRS.
A config with CONFIG_PERF_EVENTS and without CONFIG_PPC_PERF_CTRS will
have a build break because of missing is_sier_available().
In practice it only breaks when CONFIG_FSL_EMB_PERF_EVENT=n because
that also guards the usage of is_sier_available(). That only happens
with CONFIG_PPC_BOOK3E_64=y and CONFIG_FSL_SOC_BOOKE=n.
Patch adds is_sier_available() in asm/perf_event.h to fix the build
break for configs missing CONFIG_PPC_PERF_CTRS.
Fixes: 333804dc3b7a ("powerpc/perf: Update perf_regs structure to include SIER") Reported-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
[mpe: Add detail about CONFIG_FSL_SOC_BOOKE] Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20200614083604.302611-1-maddy@linux.ibm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
The ETM state save/restore incorrectly reads/writes some of the 64bit
registers (e.g, address comparators, vmid/cid comparators etc.) using
32bit accesses. Ensure we use the appropriate width accessors for
the registers.
Fixes: f188b5e76aae ("coresight: etm4x: Save/restore state across CPU low power states") Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Mike Leach <mike.leach@linaro.org> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org> Link: https://lore.kernel.org/r/20200716175746.3338735-18-mathieu.poirier@linaro.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Reading TMC mode register without proper coresight power
management can lead to exceptions like the one in the call
trace below in tmc_read_unprepare_etb() when the trace data
is read after the sink is disabled. So fix this by having
a check for coresight sysfs mode before reading TMC mode
management register in tmc_read_unprepare_etb() similar to
tmc_read_prepare_etb().
The function quirks_param_set() takes as argument a const char* pointer
to the new value of the usbcore.quirks parameter. It then casts this
pointer to a non-const char* pointer and passes it to the strsep()
function, which overwrites the value.
Fix this by creating a copy of the value using kstrdup() and letting
that copy be written to by strsep().
The display gpll0 branch clock inside GCC needs to always be enabled.
Otherwise the AHB clk (disp_cc_mdss_ahb_clk_src) for the display clk
controller (dispcc) will stop clocking while sourcing from gpll0 when
this branch inside GCC is turned off during unused clk disabling. We can
never turn this branch off because the AHB clk for the display subsystem
is needed to read/write any registers inside the display subsystem
including clk related ones. This makes this branch a really easy way to
turn off AHB access to the display subsystem and cause all sorts of
mayhem. Let's just make the clk ops keep the clk enabled forever and
ignore any attempts to disable this clk so that dispcc accesses keep
working.
Signed-off-by: Taniya Das <tdas@codeaurora.org> Reported-by: Evan Green <evgreen@chromium.org> Link: https://lore.kernel.org/r/1594796050-14511-1-git-send-email-tdas@codeaurora.org Fixes: 17269568f726 ("clk: qcom: Add Global Clock controller (GCC) driver for SC7180") Fixes: 06391eddb60a ("clk: qcom: Add Global Clock controller (GCC) driver for SDM845")
[sboyd@kernel.org: Fill out commit text more] Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The set-led command is eight bytes long and starts with a command byte
followed by six bytes of RGB data and ends with a byte encoding a
frequency (see iuu_led() and iuu_rgbf_fill_buffer()).
The led activity helpers had a few long-standing bugs which corrupted
the command packets by inserting a second command byte and thereby
offsetting the RGB data and dropping the frequency in non-xmas mode.
In xmas mode, a related off-by-one error left the frequency field
uninitialised.
The lantiq-ssc driver uses internally an own workqueue to wait till the
data is not only written out of the FIFO but really written to the wire.
This workqueue is flushed while the SPI subsystem is working in some
other system workqueue.
The system workqueue is marked as WQ_MEM_RECLAIM, but the workqueue in
the lantiq-ssc driver does not use WQ_MEM_RECLAIM for now. Add this flag
too to prevent this warning.
This fixes the following warning:
[ 2.975956] WARNING: CPU: 1 PID: 17 at kernel/workqueue.c:2614 check_flush_dependency+0x168/0x184
[ 2.984752] workqueue: WQ_MEM_RECLAIM kblockd:blk_mq_run_work_fn is flushing !WQ_MEM_RECLAIM 1e100800.spi:0x0
Fixes: 891b7c5fbf61 ("mtd_blkdevs: convert to blk-mq") Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de> Link: https://lore.kernel.org/r/20200717215648.20522-1-hauke@hauke-m.de Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Add missing regulator_disable() as devm_action to avoid dedicated
unbind() callback and fix the missing error handling.
Fixes: fcbc51e54d2a ("staging: drm/imx: Add support for Television Encoder (TVEv2)") Signed-off-by: Marco Felsch <m.felsch@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Component driver structures allocated with devm_kmalloc() in bind() are
freed automatically after unbind(). Since the contained drm structures
are accessed afterwards in drm_mode_config_cleanup(), move the
allocation into probe() to extend the driver structure's lifetime to the
lifetime of the device. This should eventually be changed to use drm
resource managed allocations with lifetime of the drm device.
We also need to ensure that all componets are available during the
unbind() so we need to call component_unbind_all() before we free
non-devres resources like planes.
Note this patch fixes the the use after free bug but introduces a
possible boot loop issue. The issue is triggered if the HDMI support is
enabled and a component driver always return -EPROBE_DEFER, see
discussion [1] for more details.
[1] https://lkml.org/lkml/2020/3/24/1467
Fixes: 17b5001b5143 ("imx-drm: convert to componentised device support") Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
[m.felsch@pengutronix: fix imx_tve_probe()]
[m.felsch@pengutronix: resort component_unbind_all())
[m.felsch@pengutronix: adapt commit message] Signed-off-by: Marco Felsch <m.felsch@pengutronix.de> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
We are wrongly using CPU_FTRS_POWER8 to check for P8 support. Instead, we should
use PVR value. Now considering we are using CPU_FTRS_POWER8, that
implies we returned true for P9 with older firmware. Keep the same behavior
by checking for P9 PVR value.
If CONFIG_DEBUG_SHIRQ was enabled, r8a77951-salvator-xs could boot
correctly. If we appended "earlycon keep_bootcon" to the kernel
command like, we could get kernel log like below.
SError Interrupt on CPU0, code 0xbf000002 -- SError
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.8.0-rc3-salvator-x-00505-g6c843129e6faaf01 #785
Hardware name: Renesas Salvator-X 2nd version board based on r8a77951 (DT)
pstate: 60400085 (nZCv daIf +PAN -UAO BTYPE=--)
pc : rcar_gen3_phy_usb2_irq+0x14/0x54
lr : free_irq+0xf4/0x27c
This means free_irq() calls the interrupt handler while PM runtime
is not getting if DEBUG_SHIRQ is enabled and rcar_gen3_phy_usb2_probe()
failed. To fix the issue, move the irq registration place to
rcar_gen3_phy_usb2_init() which is ready to handle the interrupts.
Note that after the commit 549b6b55b005 ("phy: renesas: rcar-gen3-usb2:
enable/disable independent irqs") which is merged into v5.2, since this
driver creates multiple phy instances, needs to check whether one of
phy instances is initialized. However, if we backport this patch to v5.1
or less, we don't need to check it because such kernel have single
phy instance.
A malicious USB device may lack endpoints the driver assumes to exist
Accessing them leads to NULL pointer accesses. This patch introduces
sanity checking.
Reported-and-tested-by: syzbot+cabfa4b5b05ff6be4ef0@syzkaller.appspotmail.com Signed-off-by: Oliver Neukum <oneukum@suse.com> Fixes: 866b8695d67e8 ("Staging: add the go7007 video driver") Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl> Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
axg_card_add_tdm_loopback() misses to call kfree() in an error path. We
can use devm_kasprintf() to fix the issue, also improve maintainability.
So use it instead.
Fixes: c84836d7f650 ("ASoC: meson: axg-card: use modern dai_link style") Signed-off-by: Jing Xiangfeng <jingxiangfeng@huawei.com> Reviewed-by: Jerome Brunet <jbrunet@baylibre.com> Link: https://lore.kernel.org/r/20200717082242.130627-1-jingxiangfeng@huawei.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently the masking of ret with 0xff and followed by a right shift
of 8 bits always leaves a zero result. It appears the mask of 0xff
is incorrect and should be 0xff00, but I don't have the hardware to
test this. Fix this to mask the upper 8 bits before shifting.
[ Not tested ]
Addresses-Coverity: ("Operands don't affect result") Fixes: 8fc8598e61f6 ("Staging: Added Realtek rtl8192u driver to staging") Signed-off-by: Colin Ian King <colin.king@canonical.com> Link: https://lore.kernel.org/r/20200716154720.1710252-1-colin.king@canonical.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The KEY_CHECK function only supports the uid, pcr, and keyrings
conditionals. Make this clear at policy load so that IMA policy authors
don't assume that other conditionals are supported.
The KEXEC_CMDLINE hook function only supports the pcr conditional. Make
this clear at policy load so that IMA policy authors don't assume that
other conditionals are supported.
Since KEXEC_CMDLINE's inception, ima_match_rules() has always returned
true on any loaded KEXEC_CMDLINE rule without any consideration for
other conditionals present in the rule. Make it clear that pcr is the
only supported KEXEC_CMDLINE conditional by returning an error during
policy load.
An example of why this is a problem can be explained with the following
rule:
dont_measure func=KEXEC_CMDLINE obj_type=foo_t
An IMA policy author would have assumed that rule is valid because the
parser accepted it but the result was that measurements for all
KEXEC_CMDLINE operations would be disabled.
Fixes: b0935123a183 ("IMA: Define a new hook to measure the kexec boot command line arguments") Signed-off-by: Tyler Hicks <tyhicks@linux.microsoft.com> Reviewed-by: Mimi Zohar <zohar@linux.ibm.com> Reviewed-by: Lakshmi Ramasubramanian <nramas@linux.microsoft.com> Signed-off-by: Mimi Zohar <zohar@linux.ibm.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Buffer based hook functions, such as KEXEC_CMDLINE and KEY_CHECK, can
only measure. The process_buffer_measurement() function quietly ignores
all actions except measure so make this behavior clear at the time of
policy load.
The parsing of the keyrings conditional had a check to ensure that it
was only specified with measure actions but the check should be on the
hook function and not the keyrings conditional since
"appraise func=KEY_CHECK" is not a valid rule.
Fixes: b0935123a183 ("IMA: Define a new hook to measure the kexec boot command line arguments") Fixes: 5808611cccb2 ("IMA: Add KEY_CHECK func to measure keys") Signed-off-by: Tyler Hicks <tyhicks@linux.microsoft.com> Signed-off-by: Mimi Zohar <zohar@linux.ibm.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Use ima_free_rule() to fix memory leaks of allocated ima_rule_entry
members, such as .fsname and .keyrings, when an error is encountered
during rule parsing.
Set the args_p pointer to NULL after freeing it in the error path of
ima_lsm_rule_init() so that it isn't freed twice.
This fixes a memory leak seen when loading an rule that contains an
additional piece of allocated memory, such as an fsname, followed by an
invalid conditional:
Fixes: f1b08bbcbdaf ("ima: define a new policy condition based on the filesystem name") Fixes: 2b60c0ecedf8 ("IMA: Read keyrings= option from the IMA policy") Signed-off-by: Tyler Hicks <tyhicks@linux.microsoft.com> Signed-off-by: Mimi Zohar <zohar@linux.ibm.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Create a function, ima_free_rule(), to free all memory associated with
an ima_rule_entry. Use the new function to fix memory leaks of allocated
ima_rule_entry members, such as .fsname and .keyrings, when deleting a
list of rules.
Make the existing ima_lsm_free_rule() function specific to the LSM
audit rule array of an ima_rule_entry and require that callers make an
additional call to kfree to free the ima_rule_entry itself.
This fixes a memory leak seen when loading by a valid rule that contains
an additional piece of allocated memory, such as an fsname, followed by
an invalid rule that triggers a policy load failure:
Fixes: f1b08bbcbdaf ("ima: define a new policy condition based on the filesystem name") Fixes: 2b60c0ecedf8 ("IMA: Read keyrings= option from the IMA policy") Signed-off-by: Tyler Hicks <tyhicks@linux.microsoft.com> Signed-off-by: Mimi Zohar <zohar@linux.ibm.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Ask the LSM to free its audit rule rather than directly calling kfree().
Both AppArmor and SELinux do additional work in their audit_rule_free()
hooks. Fix memory leaks by allowing the LSMs to perform necessary work.
rxe_post_send_kernel() iterates over linked list of wr's, until the
wr->next ptr is NULL. However if we've got an interrupt after last wr is
posted, control may be returned to the code after send completion callback
is executed and wr memory is freed.
As a result, wr->next pointer may contain incorrect value leading to
panic. Store the wr->next on the stack before posting it.
QP's with the same SRQ, working on different CQs and running in parallel
on different CPUs could lead to a race when maintaining the SRQ consumer
count, and leads to FW running out of SRQs. Update the consumer
atomically. Make sure the wqe_prod is updated after the sge_prod due to
FW requirements.
Fixes: 3491c9e799fb ("qedr: Add support for kernel mode SRQ's") Link: https://lore.kernel.org/r/20200708195526.31040-1-ybason@marvell.com Signed-off-by: Michal Kalderon <mkalderon@marvell.com> Signed-off-by: Yuval Basson <ybason@marvell.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The code in vdso_cpu_init that exposes the cpu and numa node to
userspace via SPRG_VDSO incorrctly masks the cpu to 12 bits. This means
that any kernel running on a box with more than 4096 threads (NR_CPUS
advertises a limit of of 8192 cpus) would expose userspace to two cpu
contexts running at the same time with the same cpu number.
Note: I'm not aware of any distro shipping a kernel with support for more
than 4096 threads today, nor of any system image that currently exceeds
4096 threads. Found via code browsing.
Fixes: 18ad51dd342a7eb09dbcd059d0b451b616d4dafc ("powerpc: Add VDSO version of getcpu") Signed-off-by: Milton Miller <miltonm@us.ibm.com> Signed-off-by: Anton Blanchard <anton@linux.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20200715233704.1352257-1-anton@ozlabs.org Signed-off-by: Sasha Levin <sashal@kernel.org>
Partition suspension, used for hibernation and migration, requires
that the OS place all but one of the LPAR's processor threads into one
of two states prior to calling the ibm,suspend-me RTAS function:
* the architected offline state (via RTAS stop-self); or
* the H_JOIN hcall, which does not return until the partition
resumes execution
Using H_CEDE as the offline mode, introduced by
commit 3aa565f53c39 ("powerpc/pseries: Add hooks to put the CPU into
an appropriate offline state"), means that any threads which are
offline from Linux's point of view must be moved to one of those two
states before a partition suspension can proceed.
This was eventually addressed in commit 120496ac2d2d ("powerpc: Bring
all threads online prior to migration/hibernation"), which added code
to temporarily bring up any offline processor threads so they can call
H_JOIN. Conceptually this is fine, but the implementation has had
multiple races with cpu hotplug operations initiated from user
space[1][2][3], the error handling is fragile, and it generates
user-visible cpu hotplug events which is a lot of noise for a platform
feature that's supposed to minimize disruption to workloads.
With commit 3aa565f53c39 ("powerpc/pseries: Add hooks to put the CPU
into an appropriate offline state") reverted, this code becomes
unnecessary, so remove it. Since any offline CPUs now are truly
offline from the platform's point of view, it is no longer necessary
to bring up CPUs only to have them call H_JOIN and then go offline
again upon resuming. Only active threads are required to call H_JOIN;
stopped threads can be left alone.
[1] commit a6717c01ddc2 ("powerpc/rtas: use device model APIs and
serialization during LPM")
[2] commit 9fb603050ffd ("powerpc/rtas: retry when cpu offline races
with suspend/migration")
[3] commit dfd718a2ed1f ("powerpc/rtas: Fix a potential race between
CPU-Offline & Migration")
Fixes: 120496ac2d2d ("powerpc: Bring all threads online prior to migration/hibernation") Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20200612051238.1007764-3-nathanl@linux.ibm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
This effectively reverts commit 3aa565f53c39 ("powerpc/pseries: Add
hooks to put the CPU into an appropriate offline state"), which added
an offline mode for CPUs which uses the H_CEDE hcall instead of the
architected stop-self RTAS function in order to facilitate "folding"
of dedicated mode processors on PowerVM platforms to achieve energy
savings. This has been the default offline mode since its
introduction.
There's nothing about stop-self that would prevent the hypervisor from
achieving the energy savings available via H_CEDE, so the original
premise of this change appears to be flawed.
I also have encountered the claim that the transition to and from
ceded state is much faster than stop-self/start-cpu. Certainly we
would not want to use stop-self as an *idle* mode. That is what H_CEDE
is for. However, this difference is insignificant in the context of
Linux CPU hotplug, where the latency of an offline or online operation
on current systems is on the order of 100ms, mainly attributable to
all the various subsystems' cpuhp callbacks.
The cede offline mode also prevents accurate accounting, as discussed
before:
https://lore.kernel.org/linuxppc-dev/1571740391-3251-1-git-send-email-ego@linux.vnet.ibm.com/
Unconditionally use stop-self to offline processor threads. This is
the architected method for offlining CPUs on PAPR systems.
The "cede_offline" boot parameter is rendered obsolete.
Removing this code enables the removal of the partition suspend code
which temporarily onlines all present CPUs.
Fixes: 3aa565f53c39 ("powerpc/pseries: Add hooks to put the CPU into an appropriate offline state") Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com> Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20200612051238.1007764-2-nathanl@linux.ibm.com Signed-off-by: Sasha Levin <sashal@kernel.org>
When creating an FS_MODIFY event on inode itself (not on parent)
the file_name argument should be NULL.
The change to send a non NULL name to inode itself was done on purpuse
as part of another commit, as Tejun writes: "...While at it, supply the
target file name to fsnotify() from kernfs_node->name.".
But this is wrong practice and inconsistent with inotify behavior when
watching a single file. When a child is being watched (as opposed to the
parent directory) the inotify event should contain the watch descriptor,
but not the file name.
Fixes: df6a58c5c5aa ("kernfs: don't depend on d_find_any_alias()...") Link: https://lore.kernel.org/r/20200708111156.24659-5-amir73il@gmail.com Acked-by: Tejun Heo <tj@kernel.org> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Amir Goldstein <amir73il@gmail.com> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Sasha Levin <sashal@kernel.org>
projects / thirdparty / kernel / stable.git / log
