Marek and Guenter reported that commit 287905e68d ("driver core:
Expose device link details in sysfs") caused sleeping/scheduling while
atomic warnings.
BUG: sleeping function called from invalid context at kernel/locking/mutex.c:935
in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 12, name: kworker/0:1
2 locks held by kworker/0:1/12:
#0: ee8074a8 ((wq_completion)rcu_gp){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc
#1: ee921f20 ((work_completion)(&sdp->work)){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc
Preemption disabled at:
[<c01b10f0>] srcu_invoke_callbacks+0xc0/0x154
----- 8< ----- SNIP
[<c064590c>] (device_del) from [<c0645c9c>] (device_unregister+0x24/0x64)
[<c0645c9c>] (device_unregister) from [<c01b10fc>] (srcu_invoke_callbacks+0xcc/0x154)
[<c01b10fc>] (srcu_invoke_callbacks) from [<c01493c4>] (process_one_work+0x234/0x7dc)
[<c01493c4>] (process_one_work) from [<c01499b0>] (worker_thread+0x44/0x51c)
[<c01499b0>] (worker_thread) from [<c0150bf4>] (kthread+0x158/0x1a0)
[<c0150bf4>] (kthread) from [<c0100114>] (ret_from_fork+0x14/0x20)
Exception stack(0xee921fb0 to 0xee921ff8)
This was caused by the device link device being released in the context
of srcu_invoke_callbacks(). There is no need to wait till the RCU
callback to release the device link device. So release the device
earlier and move the call_srcu() into the device release code. That way,
the memory will get freed only after the device is released AND the RCU
callback is called.
Fixes: 287905e68d ("driver core: Expose device link details in sysfs")
Reported-by: Marek Szyprowski <m.szyprowski@samsung.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org>
Signed-off-by: Saravana Kannan <saravanak@google.com>
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20200716214523.2924704-1-saravanak@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The following commit:
14533a16c4 ("thermal/cpu-cooling, sched/core: Move the arch_set_thermal_pressure() API to generic scheduler code")
moved the definition of arch_set_thermal_pressure() to sched/core.c, but
kept its declaration in linux/arch_topology.h. When building e.g. an x86
kernel with CONFIG_SCHED_THERMAL_PRESSURE=y, cpufreq_cooling.c ends up
getting the declaration of arch_set_thermal_pressure() from
include/linux/arch_topology.h, which is somewhat awkward.
On top of this, sched/core.c unconditionally defines
o The thermal_pressure percpu variable
o arch_set_thermal_pressure()
while arch_scale_thermal_pressure() does nothing unless redefined by the
architecture.
arch_*() functions are meant to be defined by architectures, so revert the
aforementioned commit and re-implement it in a way that keeps
arch_set_thermal_pressure() architecture-definable, and doesn't define the
thermal pressure percpu variable for kernels that don't need
it (CONFIG_SCHED_THERMAL_PRESSURE=n).
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200712165917.9168-2-valentin.schneider@arm.com
Pull regmap fixes from Mark Brown:
"A couple of substantial fixes here, one from Doug which fixes the
debugfs code for MMIO regmaps (fortunately not the common case) and
one from Marc fixing lookups of multiple regmaps for the same device
(a very unusual case).
There's also a fix for Kconfig to ensure we enable SoundWire properly"
* tag 'regmap-fix-v5.8-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
regmap: debugfs: Don't sleep while atomic for fast_io regmaps
regmap: add missing dependency on SoundWire
regmap: dev_get_regmap_match(): fix string comparison
If a regmap has "fast_io" set then its lock function uses a spinlock.
That doesn't work so well with the functions:
* regmap_cache_only_write_file()
* regmap_cache_bypass_write_file()
Both of the above functions have the pattern:
1. Lock the regmap.
2. Call:
debugfs_write_file_bool()
copy_from_user()
__might_fault()
__might_sleep()
Let's reorder things a bit so that we do all of our sleepable
functions before we grab the lock.
Fixes: d3dc5430d6 ("regmap: debugfs: Allow writes to cache state settings")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20200715164611.1.I35b3533e8a80efde0cec1cc70f71e1e74b2fa0da@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
Udev rules that depend on the power/wakeup attribute don't get triggered
correctly if device_set_wakeup_capable is called after the device is
created. This can happen for several reasons (driver sets wakeup after
device is created, wakeup is changed on parent device, etc) and it seems
reasonable to emit a changed event when adding or removing attributes on
the device.
Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
This would be useful to check if a device is not probing because it's
waiting for a supplier to be added and then linked to before it can
probe.
To reduce sysfs clutter, this file is added only if it can ever be 1.
So, if fw_devlink is disabled or set to permissive, this file is not
added. Also, this file is removed once the device probes as it's no
longer relevant.
Signed-off-by: Saravana Kannan <saravanak@google.com>
Link: https://lore.kernel.org/r/20200521191800.136035-4-saravanak@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This can be used to check if a device supports sync_state() callbacks
and therefore keeps resources left on by the bootloader enabled till all
its consumers have probed.
This can also be used to check if sync_state() has been called for a
device or whether it is still trying to keep resources enabled because
they were left enabled by the bootloader and all its consumers haven't
probed yet.
Signed-off-by: Saravana Kannan <saravanak@google.com>
Link: https://lore.kernel.org/r/20200521191800.136035-3-saravanak@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It's helpful to be able to look at device link details from sysfs. So,
expose it in sysfs.
Say device-A is supplier of device-B. These are the additional files
this patch would create:
/sys/class/devlink/device-A:device-B/
auto_remove_on
consumer/ -> .../device-B/
runtime_pm
status
supplier/ -> .../device-A/
sync_state_only
/sys/devices/.../device-A/
consumer:device-B/ -> /sys/class/devlink/device-A:device-B/
/sys/devices/.../device-B/
supplier:device-A/ -> /sys/class/devlink/device-A:device-B/
That way:
To get a list of all the device link in the system:
ls /sys/class/devlink/
To get the consumer names and links of a device:
ls -d /sys/devices/.../device-X/consumer:*
To get the supplier names and links of a device:
ls -d /sys/devices/.../device-X/supplier:*
Signed-off-by: Saravana Kannan <saravanak@google.com>
Link: https://lore.kernel.org/r/20200521191800.136035-2-saravanak@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With the earlier patch in this series, all devices that deferred probe
due to fw_devlink_pause() will have their probes delayed till the
deferred probe thread is kicked off during late_initcall. This will also
affect all their consumers.
This delayed probing in unnecessary. So this patch just keeps track of
the devices that had their probe deferred due to fw_devlink_pause() and
attempts to probe them once during fw_devlink_resume().
Fixes: 716a7a2596 ("driver core: fw_devlink: Add support for batching fwnode parsing")
Signed-off-by: Saravana Kannan <saravanak@google.com>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200701194259.3337652-4-saravanak@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current deferred probe implementation can mess up suspend/resume
ordering if deferred probe thread is kicked off in parallel with the
main initcall thread (kernel_init thread) [1].
For example:
Say device-B is a consumer of device-A.
Initcall thread Deferred probe thread
=============== =====================
1. device-A is added.
2. device-B is added.
3. dpm_list is now [device-A, device-B].
4. driver-A defers probe of device-A.
5. device-A is moved to
end of dpm_list
6. dpm_list is now
[device-B, device-A]
7. driver-B is registereed and probes device-B.
8. dpm_list stays as [device-B, device-A].
The reverse order of dpm_list is used for suspend. So in this case
device-A would incorrectly get suspended before device-B.
Commit 716a7a2596 ("driver core: fw_devlink: Add support for batching
fwnode parsing") kicked off the deferred probe thread early during boot
to run in parallel with the initcall thread and caused suspend/resume
regressions. This patch removes the parallel run of the deferred probe
thread to avoid the suspend/resume regressions.
[1] - https://lore.kernel.org/lkml/CAGETcx8W96KAw-d_siTX4qHB_-7ddk0miYRDQeHE6E0_8qx-6Q@mail.gmail.com/
Fixes: 716a7a2596 ("driver core: fw_devlink: Add support for batching fwnode parsing")
Signed-off-by: Saravana Kannan <saravanak@google.com>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200701194259.3337652-2-saravanak@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The genpd infrastructure uses the terms master/slave, but such uses have
no external exposures (not even in Documentation/driver-api/pm/*) and are
not mandated by nor associated with any external specifications. Change
the language used through-out to parent/child.
There was one possible exception in the debugfs node
"pm_genpd/pm_genpd_summary" but its path has no hits outside of the
kernel itself when performing a code search[1], and it seems even this
single usage has been non-functional since it was introduced due to a
typo in the Python ("apend" instead of correct "append"). Fix the typo
while we're at it.
Link: https://codesearch.debian.net/ # [1]
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
This function is used by dev_get_regmap() to retrieve a regmap for the
specified device. If the device has more than one regmap, the name parameter
can be used to specify one.
The code here uses a pointer comparison to check for equal strings. This
however will probably always fail, as the regmap->name is allocated via
kstrdup_const() from the regmap's config->name.
Fix this by using strcmp() instead.
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://lore.kernel.org/r/20200703103315.267996-1-mkl@pengutronix.de
Signed-off-by: Mark Brown <broonie@kernel.org>
Pull power management fixes from Rafael Wysocki:
"These fix a recent regression that broke suspend-to-idle on some x86
systems, fix the intel_pstate driver to correctly let the platform
firmware control CPU performance in some cases and add __init
annotations to a couple of functions.
Specifics:
- Make sure that the _TIF_POLLING_NRFLAG is clear before entering the
last phase of suspend-to-idle to avoid wakeup issues on some x86
systems (Chen Yu, Rafael Wysocki).
- Cover one more case in which the intel_pstate driver should let the
platform firmware control the CPU frequency and refuse to load
(Srinivas Pandruvada).
- Add __init annotations to 2 functions in the power management core
(Christophe JAILLET)"
* tag 'pm-5.8-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
cpuidle: Rearrange s2idle-specific idle state entry code
PM: sleep: core: mark 2 functions as __init to save some memory
cpufreq: intel_pstate: Add one more OOB control bit
PM: s2idle: Clear _TIF_POLLING_NRFLAG before suspend to idle
'early_resume_init()' and 'late_resume_init() 'are only called respectively
via 'early_resume_init' and 'late_resume_init'.
They can be marked as __init to save a few bytes of memory.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
[ rjw: Subject edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Pull regmap fixes from Mark Brown:
"A few small fixes, none of which are likely to have any substantial
impact here - the most substantial one is a fix for a long standing
memory leak on devices that use register patching which will only have
an impact if the device is removed and re-added"
* tag 'regmap-fix-v5.8-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
regmap: Fix memory leak from regmap_register_patch
regmap: fix the kerneldoc for regmap_test_bits()
regmap: fix alignment issue
DT implementation of fw_devlink needs this function to detect cycles. So
make it available.
Signed-off-by: Saravana Kannan <saravanak@google.com>
Tested-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Rob Herring <robh@kernel.org>
Pull Kunit updates from Shuah Khan:
"This consists of:
- Several config fragment fixes from Anders Roxell to improve test
coverage.
- Improvements to kunit run script to use defconfig as default and
restructure the code for config/build/exec/parse from Vitor Massaru
Iha and David Gow.
- Miscellaneous documentation warn fix"
* tag 'linux-kselftest-kunit-5.8-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
security: apparmor: default KUNIT_* fragments to KUNIT_ALL_TESTS
fs: ext4: default KUNIT_* fragments to KUNIT_ALL_TESTS
drivers: base: default KUNIT_* fragments to KUNIT_ALL_TESTS
lib: Kconfig.debug: default KUNIT_* fragments to KUNIT_ALL_TESTS
kunit: default KUNIT_* fragments to KUNIT_ALL_TESTS
kunit: Kconfig: enable a KUNIT_ALL_TESTS fragment
kunit: Fix TabError, remove defconfig code and handle when there is no kunitconfig
kunit: use KUnit defconfig by default
kunit: use --build_dir=.kunit as default
Documentation: test.h - fix warnings
kunit: kunit_tool: Separate out config/build/exec/parse
Merge even more updates from Andrew Morton:
- a kernel-wide sweep of show_stack()
- pagetable cleanups
- abstract out accesses to mmap_sem - prep for mmap_sem scalability work
- hch's user acess work
Subsystems affected by this patch series: debug, mm/pagemap, mm/maccess,
mm/documentation.
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (93 commits)
include/linux/cache.h: expand documentation over __read_mostly
maccess: return -ERANGE when probe_kernel_read() fails
x86: use non-set_fs based maccess routines
maccess: allow architectures to provide kernel probing directly
maccess: move user access routines together
maccess: always use strict semantics for probe_kernel_read
maccess: remove strncpy_from_unsafe
tracing/kprobes: handle mixed kernel/userspace probes better
bpf: rework the compat kernel probe handling
bpf:bpf_seq_printf(): handle potentially unsafe format string better
bpf: handle the compat string in bpf_trace_copy_string better
bpf: factor out a bpf_trace_copy_string helper
maccess: unify the probe kernel arch hooks
maccess: remove probe_read_common and probe_write_common
maccess: rename strnlen_unsafe_user to strnlen_user_nofault
maccess: rename strncpy_from_unsafe_strict to strncpy_from_kernel_nofault
maccess: rename strncpy_from_unsafe_user to strncpy_from_user_nofault
maccess: update the top of file comment
maccess: clarify kerneldoc comments
maccess: remove duplicate kerneldoc comments
...
Pull x86 srbds fixes from Thomas Gleixner:
"The 9th episode of the dime novel "The performance killer" with the
subtitle "Slow Randomizing Boosts Denial of Service".
SRBDS is an MDS-like speculative side channel that can leak bits from
the random number generator (RNG) across cores and threads. New
microcode serializes the processor access during the execution of
RDRAND and RDSEED. This ensures that the shared buffer is overwritten
before it is released for reuse. This is equivalent to a full bus
lock, which means that many threads running the RNG instructions in
parallel have the same effect as the same amount of threads issuing a
locked instruction targeting an address which requires locking of two
cachelines at once.
The mitigation support comes with the usual pile of unpleasant
ingredients:
- command line options
- sysfs file
- microcode checks
- a list of vulnerable CPUs identified by model and stepping this
time which requires stepping match support for the cpu match logic.
- the inevitable slowdown of affected CPUs"
* branch 'x86/srbds' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/speculation: Add Ivy Bridge to affected list
x86/speculation: Add SRBDS vulnerability and mitigation documentation
x86/speculation: Add Special Register Buffer Data Sampling (SRBDS) mitigation
x86/cpu: Add 'table' argument to cpu_matches()
