This caused a significant performance degredation when using generic XDP
with multiple queues.
Fixes: f5cedc84a3 ("gve: Add transmit and receive support")
Signed-off-by: Tao Liu <xliutaox@google.com>
Link: https://lore.kernel.org/r/20220207175901.2486596-1-jeroendb@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The SMC calling convention designates R0-R7 as input registers in
AArch32 mode, and this conflicts with the compiler's use of R7 as a
frame pointer when building in Thumb2 mode. Generally, we don't enable
the frame pointer, and GCC happily enables the -pg profiling hooks
without them. However, Clang refuses, and errors out with the message
below:
drivers/firmware/arm_scmi/smc.c:152:2: error: write to reserved register 'R7'
arm_smccc_1_1_invoke(scmi_info->func_id, 0, 0, 0, 0, 0, 0, 0, &res);
^
include/linux/arm-smccc.h:550:4: note: expanded from macro 'arm_smccc_1_1_invoke'
arm_smccc_1_1_smc(__VA_ARGS__); \
^
Let's just disable ftrace for the compilation unit when building this
configuration.
Link: https://lore.kernel.org/r/20220203082204.1176734-11-ardb@kernel.org
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Replace header->pages * PAGE_SIZE with new header->size.
Fixes: ed2d9e1a26 ("bpf: Use size instead of pages in bpf_binary_header")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Song Liu <song@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220208220509.4180389-2-song@kernel.org
The btf__resolve_size() function returns negative error codes so
"elem_size" must be signed for the error handling to work.
Fixes: 920d16af9b ("libbpf: BTF dumper support for typed data")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/bpf/20220208071552.GB10495@kili
With the newly introduced aux-bus under the TI SN65DSI86 the panel
node should be described as a child instead of a standalone node, move
it there.
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Tested-by: Steev Klimaszewski <steev@kali.org>
Link: https://lore.kernel.org/r/20220208041606.144039-2-bjorn.andersson@linaro.org
The Lenovo Yoga C630 uses the PWM controller in the TI SN65DSI86 bridge
chip to provide a signal for the backlight control and has TLMM GPIO 11
attached to some regulator that drives the backlight.
Unfortunately the regulator attached to this gpio is also powering the
camera, so turning off backlight result in the detachment of the camera
as well.
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Tested-by: Steev Klimaszewski <steev@kali.org>
Link: https://lore.kernel.org/r/20220208041606.144039-1-bjorn.andersson@linaro.org
Buttonpads are expected to map the INPUT_PROP_BUTTONPAD property bit
and the BTN_LEFT key bit.
As explained in the specification, where a device has a button type
value of 0 (click-pad) or 1 (pressure-pad) there should not be
discrete buttons:
https://docs.microsoft.com/en-us/windows-hardware/design/component-guidelines/touchpad-windows-precision-touchpad-collection#device-capabilities-feature-report
However, some drivers map the BTN_RIGHT and/or BTN_MIDDLE key bits even
though the device is a buttonpad and therefore does not have those
buttons.
This behavior has forced userspace applications like libinput to
implement different workarounds and quirks to detect buttonpads and
offer to the user the right set of features and configuration options.
For more information:
https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/726
In order to avoid this issue clear the BTN_RIGHT and BTN_MIDDLE key
bits when the input device is register if the INPUT_PROP_BUTTONPAD
property bit is set.
Notice that this change will not affect udev because it does not check
for buttons. See systemd/src/udev/udev-builtin-input_id.c.
List of known affected hardware:
- Chuwi AeroBook Plus
- Chuwi Gemibook
- Framework Laptop
- GPD Win Max
- Huawei MateBook 2020
- Prestigio Smartbook 141 C2
- Purism Librem 14v1
- StarLite Mk II - AMI firmware
- StarLite Mk II - Coreboot firmware
- StarLite Mk III - AMI firmware
- StarLite Mk III - Coreboot firmware
- StarLabTop Mk IV - AMI firmware
- StarLabTop Mk IV - Coreboot firmware
- StarBook Mk V
Acked-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: José Expósito <jose.exposito89@gmail.com>
Link: https://lore.kernel.org/r/20220208174806.17183-1-jose.exposito89@gmail.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
The kernel parameter "tp_printk_stop_on_boot" starts with "tp_printk" which is
the same as another kernel parameter "tp_printk". If "tp_printk" setup is
called before the "tp_printk_stop_on_boot", it will override the latter
and keep it from being set.
This is similar to other kernel parameter issues, such as:
Commit 745a600cf1 ("um: console: Ignore console= option")
or init/do_mounts.c:45 (setup function of "ro" kernel param)
Fix it by checking for a "_" right after the "tp_printk" and if that
exists do not process the parameter.
Link: https://lkml.kernel.org/r/20220208195421.969326-1-jsyoo5b@gmail.com
Signed-off-by: JaeSang Yoo <jsyoo5b@gmail.com>
[ Fixed up change log and added space after if condition ]
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Currently we can get a warning on systems with eDP backlights like so:
nv_backlight: invalid backlight type
WARNING: CPU: 4 PID: 454 at drivers/video/backlight/backlight.c:420
backlight_device_register+0x226/0x250
This happens as a result of us not filling out props.type for the eDP
backlight, even though we do it for all other backlight types.
Since nothing in our driver uses anything but BACKLIGHT_RAW, let's take the
props\.type assignments out of the codepaths for individual backlight types
and just set it unconditionally to prevent this from happening again.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Fixes: 6eca310e89 ("drm/nouveau/kms/nv50-: Add basic DPCD backlight support for nouveau")
Cc: <stable@vger.kernel.org> # v5.15+
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220204193319.451119-1-lyude@redhat.com
It seems that some laptops will report having both an eDP and LVDS
connector, even though only the LVDS connector is actually hooked up. This
can lead to issues with backlight registration if the eDP connector ends up
getting registered before the LVDS connector, as the backlight device will
then be registered to the eDP connector instead of the LVDS connector.
So, fix this by only registering the backlight on connectors that are
reported as being connected.
Signed-off-by: Lyude Paul <lyude@redhat.com>
Fixes: 6eca310e89 ("drm/nouveau/kms/nv50-: Add basic DPCD backlight support for nouveau")
Bugzilla: https://gitlab.freedesktop.org/drm/nouveau/-/issues/137
Cc: <stable@vger.kernel.org> # v5.15+
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220204180504.328999-1-lyude@redhat.com
Two subtests in ksyms_module.c are not qualified as static, so these
subtests are exported as standalone tests in tests.h and lead to
confusion for the output of "./test_progs -t ksyms_module".
By using the following command ...
grep "^void \(serial_\)\?test_[a-zA-Z0-9_]\+(\(void\)\?)" \
tools/testing/selftests/bpf/prog_tests/*.c | \
awk -F : '{print $1}' | sort | uniq -c | awk '$1 != 1'
... one finds out that other tests also have a similar problem, so
fix these tests by marking subtests in these tests as static.
Signed-off-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/bpf/20220208065444.648778-1-houtao1@huawei.com
In cifsd thread, we should continue to call cifs_reconnect
whenever server->tcpStatus is marked as CifsNeedReconnect.
This was inexplicably removed by one of my recent commits.
Fixing that here.
Fixes: a05885ce13 ("cifs: fix the connection state transitions with multichannel")
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
This commit adjusts RUDE01 to 3 CPUs and TRACE01 to 5 CPUs in order to
test Tasks RCU's ability to handle non-power-of-two numbers of CPUs.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Thie commit adds kernel boot parameters to the SRCU-N and SRCU-P
rcutorture scenarios to cause SRCU-N to test contention-based resizing
and SRCU-P to test init_srcu_struct()-time resizing. Note that this
also tests never-resizing because the contention-based resizing normally
takes some minutes to make the shift.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
This commit fixes a couple of typos: s/--doall/--do-all/ and
s/--doallmodconfig/--do-allmodconfig/.
[ paulmck: Add Fixes: supplied by Paul Menzel. ]
Fixes: a115a775a8 ("torture: Add "make allmodconfig" to torture.sh")
Reported-by: Paul Menzel <pmenzel@molgen.mpg.de>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Encrypted files traditionally haven't supported DIO, due to the need to
encrypt/decrypt the data. However, when the encryption is implemented
using inline encryption (blk-crypto) instead of the traditional
filesystem-layer encryption, it is straightforward to support DIO.
Therefore, make f2fs support DIO on files that are using inline
encryption. Since f2fs uses iomap for DIO, and fscrypt support was
already added to iomap DIO, this just requires two small changes:
- Let DIO proceed when supported, by checking fscrypt_dio_supported()
instead of assuming that encrypted files never support DIO.
- In f2fs_iomap_begin(), use fscrypt_limit_io_blocks() to limit the
length of the mapping in the rare case where a DUN discontiguity
occurs in the middle of an extent. The iomap DIO implementation
requires this, since it assumes that it can submit a bio covering (up
to) the whole mapping, without checking fscrypt constraints itself.
Co-developed-by: Satya Tangirala <satyat@google.com>
Signed-off-by: Satya Tangirala <satyat@google.com>
Acked-by: Jaegeuk Kim <jaegeuk@kernel.org>
Link: https://lore.kernel.org/r/20220128233940.79464-5-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Encrypted files traditionally haven't supported DIO, due to the need to
encrypt/decrypt the data. However, when the encryption is implemented
using inline encryption (blk-crypto) instead of the traditional
filesystem-layer encryption, it is straightforward to support DIO.
Therefore, make ext4 support DIO on files that are using inline
encryption. Since ext4 uses iomap for DIO, and fscrypt support was
already added to iomap DIO, this just requires two small changes:
- Let DIO proceed when supported, by checking fscrypt_dio_supported()
instead of assuming that encrypted files never support DIO.
- In ext4_iomap_begin(), use fscrypt_limit_io_blocks() to limit the
length of the mapping in the rare case where a DUN discontiguity
occurs in the middle of an extent. The iomap DIO implementation
requires this, since it assumes that it can submit a bio covering (up
to) the whole mapping, without checking fscrypt constraints itself.
Co-developed-by: Satya Tangirala <satyat@google.com>
Signed-off-by: Satya Tangirala <satyat@google.com>
Acked-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jaegeuk Kim <jaegeuk@kernel.org>
Link: https://lore.kernel.org/r/20220128233940.79464-4-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Encrypted files traditionally haven't supported DIO, due to the need to
encrypt/decrypt the data. However, when the encryption is implemented
using inline encryption (blk-crypto) instead of the traditional
filesystem-layer encryption, it is straightforward to support DIO.
Add support for this to the iomap DIO implementation by calling
fscrypt_set_bio_crypt_ctx() to set encryption contexts on the bios.
Don't check for the rare case where a DUN (crypto data unit number)
discontiguity creates a boundary that bios must not cross. Instead,
filesystems are expected to handle this in ->iomap_begin() by limiting
the length of the mapping so that iomap doesn't have to worry about it.
Co-developed-by: Satya Tangirala <satyat@google.com>
Signed-off-by: Satya Tangirala <satyat@google.com>
Acked-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220128233940.79464-3-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Encrypted files traditionally haven't supported DIO, due to the need to
encrypt/decrypt the data. However, when the encryption is implemented
using inline encryption (blk-crypto) instead of the traditional
filesystem-layer encryption, it is straightforward to support DIO.
In preparation for supporting this, add the following functions:
- fscrypt_dio_supported() checks whether a DIO request is supported as
far as encryption is concerned. Encrypted files will only support DIO
when inline encryption is used and the I/O request is properly
aligned; this function checks these preconditions.
- fscrypt_limit_io_blocks() limits the length of a bio to avoid crossing
a place in the file that a bio with an encryption context cannot
cross due to a DUN discontiguity. This function is needed by
filesystems that use the iomap DIO implementation (which operates
directly on logical ranges, so it won't use fscrypt_mergeable_bio())
and that support FSCRYPT_POLICY_FLAG_IV_INO_LBLK_32.
Co-developed-by: Satya Tangirala <satyat@google.com>
Signed-off-by: Satya Tangirala <satyat@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220128233940.79464-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
Hi,
the DMA trace implementation on AMD platform assumes that the stream_tag pointer
is pointing the stream_tag member of struct sof_ipc_dma_trace_params_ext, which
is true at the moment, but it can not be guarantied and a change in the dtrace
core can cause out of bound accesses for AMD.
For this reason, change the API to pass the struct itself which will remove the
assumption and makes it clear from both sides what is expected to be sent via the
parameter list.
This opens up a window to clean up the intel and AMD implementation at the same
time.
Regards,
Peter
---
Peter Ujfalusi (2):
ASoC: SOF: intel: hda-trace: Pass the dma buffer pointer to
hda_dsp_trace_prepare
ASoC: SOF: dma-trace: Pass pointer to params_ext struct in
trace_init()
sound/soc/sof/amd/acp-trace.c | 38 ++++++++-------------------------
sound/soc/sof/amd/acp.h | 3 ++-
sound/soc/sof/intel/hda-trace.c | 17 ++++++++-------
sound/soc/sof/intel/hda.h | 3 ++-
sound/soc/sof/ops.h | 4 ++--
sound/soc/sof/sof-priv.h | 2 +-
sound/soc/sof/trace.c | 2 +-
7 files changed, 26 insertions(+), 43 deletions(-)
--
2.35.0
Merge series from Mark Brown <broonie@kernel.org>:
This series removes a bunch of spurious selects of gpiolib that were
causing noise in randconfig build tests.
Mark Brown (6):
ASoC: dmic: Remove spurious gpiolib select
ASoC: rt9120: Remove spurious gpiolib select
ASoC: simple-amplifier: Remove spurious gpiolib select
ASoC: max9759: Remove spurious gpiolib select
ASoC: zl38060: Remove spurious gpiolib select
ASoC: simple-mux: Depend on gpiolib rather than selecting it
sound/soc/codecs/Kconfig | 8 ++------
1 file changed, 2 insertions(+), 6 deletions(-)
base-commit: e783362eb5
--
2.30.2
asm/svm.h is the correct place for all values that are defined in
the SVM spec, and that includes AVIC.
Also add some values from the spec that were not defined before
and will be soon useful.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20220207155447.840194-10-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
kvm_apic_update_apicv is called when AVIC is still active, thus IRR bits
can be set by the CPU after it is called, and don't cause the irr_pending
to be set to true.
Also logic in avic_kick_target_vcpu doesn't expect a race with this
function so to make it simple, just keep irr_pending set to true and
let the next interrupt injection to the guest clear it.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20220207155447.840194-9-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fix a corner case in which the L1 hypervisor intercepts
interrupts (INTERCEPT_INTR) and either doesn't set
virtual interrupt masking (V_INTR_MASKING) or enters a
nested guest with EFLAGS.IF disabled prior to the entry.
In this case, despite the fact that L1 intercepts the interrupts,
KVM still needs to set up an interrupt window to wait before
injecting the INTR vmexit.
Currently the KVM instead enters an endless loop of 'req_immediate_exit'.
Exactly the same issue also happens for SMIs and NMI.
Fix this as well.
Note that on VMX, this case is impossible as there is only
'vmexit on external interrupts' execution control which either set,
in which case both host and guest's EFLAGS.IF
are ignored, or not set, in which case no VMexits are delivered.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20220207155447.840194-8-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM already honours few clean bits thus it makes sense
to let the nested guest know about it.
Note that KVM also doesn't check if the hardware supports
clean bits, and therefore nested KVM was
already setting clean bits and L0 KVM
was already honouring them.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20220207155447.840194-6-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
While RSM induced VM entries are not full VM entries,
they still need to be followed by actual VM entry to complete it,
unlike setting the nested state.
This patch fixes boot of hyperv and SMM enabled
windows VM running nested on KVM, which fail due
to this issue combined with lack of dirty bit setting.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Cc: stable@vger.kernel.org
Message-Id: <20220207155447.840194-5-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
While usually, restoring the smm state makes the KVM enter
the nested guest thus a different vmcb (vmcb02 vs vmcb01),
KVM should still mark it as dirty, since hardware
can in theory cache multiple vmcbs.
Failure to do so, combined with lack of setting the
nested_run_pending (which is fixed in the next patch),
might make KVM re-enter vmcb01, which was just exited from,
with completely different set of guest state registers
(SMM vs non SMM) and without proper dirty bits set,
which results in the CPU reusing stale IDTR pointer
which leads to a guest shutdown on any interrupt.
On the real hardware this usually doesn't happen,
but when running nested, L0's KVM does check and
honour few dirty bits, causing this issue to happen.
This patch fixes boot of hyperv and SMM enabled
windows VM running nested on KVM.
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Cc: stable@vger.kernel.org
Message-Id: <20220207155447.840194-4-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Turns out that due to review feedback and/or rebases
I accidentally moved the call to nested_svm_load_cr3 to be too early,
before the NPT is enabled, which is very wrong to do.
KVM can't even access guest memory at that point as nested NPT
is needed for that, and of course it won't initialize the walk_mmu,
which is main issue the patch was addressing.
Fix this for real.
Fixes: 232f75d3b4 ("KVM: nSVM: call nested_svm_load_cr3 on nested state load")
Cc: stable@vger.kernel.org
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20220207155447.840194-3-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When the guest doesn't enable paging, and NPT/EPT is disabled, we
use guest't paging CR3's as KVM's shadow paging pointer and
we are technically in direct mode as if we were to use NPT/EPT.
In direct mode we create SPTEs with user mode permissions
because usually in the direct mode the NPT/EPT doesn't
need to restrict access based on guest CPL
(there are MBE/GMET extenstions for that but KVM doesn't use them).
In this special "use guest paging as direct" mode however,
and if CR4.SMAP/CR4.SMEP are enabled, that will make the CPU
fault on each access and KVM will enter endless loop of page faults.
Since page protection doesn't have any meaning in !PG case,
just don't passthrough these bits.
The fix is the same as was done for VMX in commit:
commit 656ec4a492 ("KVM: VMX: fix SMEP and SMAP without EPT")
This fixes the boot of windows 10 without NPT for good.
(Without this patch, BSP boots, but APs were stuck in endless
loop of page faults, causing the VM boot with 1 CPU)
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
Cc: stable@vger.kernel.org
Message-Id: <20220207155447.840194-2-mlevitsk@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Remove a WARN on an "AVIC IPI invalid target" exit, the WARN is trivial
to trigger from guest as it will fail on any destination APIC ID that
doesn't exist from the guest's perspective.
Don't bother recording anything in the kernel log, the common tracepoint
for kvm_avic_incomplete_ipi() is sufficient for debugging.
This reverts commit 37ef0c4414.
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20220204214205.3306634-2-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Currently, call_rcu_tasks_generic() sets ->percpu_enqueue_shift to
order_base_2(nr_cpu_ids) upon encountering sufficient contention.
This does not shift to use of non-CPU-0 callback queues as intended, but
rather continues using only CPU 0's queue. Although this does provide
some decrease in contention due to spreading work over multiple locks,
it is not the dramatic decrease that was intended.
This commit therefore makes call_rcu_tasks_generic() set
->percpu_enqueue_shift to 0.
Reported-by: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
The ilog2() function can be used to generate a shift count, but it will
generate the same count for a power of two as for one greater than a power
of two. This results in shift counts that are larger than necessary for
systems with a power-of-two number of CPUs because the CPUs are numbered
from zero, so that the maximum CPU number is one less than that power
of two.
This commit therefore substitutes order_base_2(), which appears to have
been designed for exactly this use case.
Suggested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
The pattern "rdp->grpmask & rcu_rnp_online_cpus(rnp)" occurs frequently
in RCU code in order to determine whether rdp->cpu is online from an
RCU perspective. This commit therefore creates an rcu_rdp_cpu_online()
function to replace it.
[ paulmck: Apply kernel test robot unused-variable feedback. ]
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
This commit removes the cpus_read_lock() and cpus_read_unlock() calls
from rcu_barrier(), thus allowing CPUs to come and go during the course
of rcu_barrier() execution. Posting of the ->barrier_head callbacks does
synchronize with portions of RCU's CPU-hotplug notifiers, but these locks
are held for short time periods on both sides. Thus, full CPU-hotplug
operations could both start and finish during the execution of a given
rcu_barrier() invocation.
Additional synchronization is provided by a global ->barrier_lock.
Since the ->barrier_lock is only used during rcu_barrier() execution and
during onlining/offlining a CPU, the contention for this lock should
be low. It might be tempting to make use of a per-CPU lock just on
general principles, but straightforward attempts to do this have the
problems shown below.
Initial state: 3 CPUs present, CPU 0 and CPU1 do not have
any callback and CPU2 has callbacks.
1. CPU0 calls rcu_barrier().
2. CPU1 starts offlining for CPU2. CPU1 calls
rcutree_migrate_callbacks(). rcu_barrier_entrain() is called
from rcutree_migrate_callbacks(), with CPU2's rdp->barrier_lock.
It does not entrain ->barrier_head for CPU2, as rcu_barrier()
on CPU0 hasn't started the barrier sequence (by calling
rcu_seq_start(&rcu_state.barrier_sequence)) yet.
3. CPU0 starts new barrier sequence. It iterates over
CPU0 and CPU1, after acquiring their per-cpu ->barrier_lock
and finds 0 segcblist length. It updates ->barrier_seq_snap
for CPU0 and CPU1 and continues loop iteration to CPU2.
for_each_possible_cpu(cpu) {
raw_spin_lock_irqsave(&rdp->barrier_lock, flags);
if (!rcu_segcblist_n_cbs(&rdp->cblist)) {
WRITE_ONCE(rdp->barrier_seq_snap, gseq);
raw_spin_unlock_irqrestore(&rdp->barrier_lock, flags);
rcu_barrier_trace(TPS("NQ"), cpu, rcu_state.barrier_sequence);
continue;
}
4. rcutree_migrate_callbacks() completes execution on CPU1.
Segcblist len for CPU2 becomes 0.
5. The loop iteration on CPU0, checks rcu_segcblist_n_cbs(&rdp->cblist)
for CPU2 and completes the loop iteration after setting
->barrier_seq_snap.
6. As there isn't any ->barrier_head callback entrained; at
this point, rcu_barrier() in CPU0 returns.
7. The callbacks, which migrated from CPU2 to CPU1, execute.
Straightforward per-CPU locking is also subject to the following race
condition noted by Boqun Feng:
1. CPU0 calls rcu_barrier(), starting a new barrier sequence by invoking
rcu_seq_start() and init_completion(), but does not yet initialize
rcu_state.barrier_cpu_count.
2. CPU1 starts offlining for CPU2, calling rcutree_migrate_callbacks(),
which in turn calls rcu_barrier_entrain() holding CPU2's.
rdp->barrier_lock. It then entrains ->barrier_head for CPU2
and atomically increments rcu_state.barrier_cpu_count, which is
unfortunately not yet initialized to the value 2.
3. The just-entrained RCU callback is invoked. It atomically
decrements rcu_state.barrier_cpu_count and sees that it is
now zero. This callback therefore invokes complete().
4. CPU0 continues executing rcu_barrier(), but is not blocked
by its call to wait_for_completion(). This results in rcu_barrier()
returning before all pre-existing callbacks have been invoked,
which is a bug.
Therefore, synchronization is provided by rcu_state.barrier_lock,
which is also held across the initialization sequence, especially the
rcu_seq_start() and the atomic_set() that sets rcu_state.barrier_cpu_count
to the value 2. In addition, this lock is held when entraining the
rcu_barrier() callback, when deciding whether or not a CPU has callbacks
that rcu_barrier() must wait on, when setting the ->qsmaskinitnext for
incoming CPUs, and when migrating callbacks from a CPU that is going
offline.
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Co-developed-by: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Signed-off-by: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
This commit reworks rcu_barrier() and callback-migration logic to
permit allowing rcu_barrier() to run concurrently with CPU-hotplug
operations. The key trick is for callback migration to check to see if
an rcu_barrier() is in flight, and, if so, enqueue the ->barrier_head
callback on its behalf.
This commit adds synchronization with RCU's CPU-hotplug notifiers. Taken
together, this will permit a later commit to remove the cpus_read_lock()
and cpus_read_unlock() calls from rcu_barrier().
[ paulmck: Updated per kbuild test robot feedback. ]
[ paulmck: Updated per reviews session with Neeraj, Frederic, Uladzislau, and Boqun. ]
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
This commit saves a few lines by checking first for an empty callback
list. If the callback list is empty, then that CPU is taken care of,
regardless of its online or nocb state. Also simplify tracing accordingly
and fold a few lines together.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
If we allow architectures to bring APs online in parallel, then we end
up requiring rcu_cpu_starting() to be reentrant. But currently, the
manipulation of rnp->ofl_seq is not thread-safe.
However, rnp->ofl_seq is also fairly much pointless anyway since both
rcu_cpu_starting() and rcu_report_dead() hold rcu_state.ofl_lock for
fairly much the whole time that rnp->ofl_seq is set to an odd number
to indicate that an operation is in progress.
So drop rnp->ofl_seq completely, and use only rcu_state.ofl_lock.
This has a couple of minor complexities: lockdep will complain when we
take rcu_state.ofl_lock, and currently accepts the 'excuse' of having
an odd value in rnp->ofl_seq. So switch it to an arch_spinlock_t to
avoid that false positive complaint. Since we're killing rnp->ofl_seq
of course that 'excuse' has to be changed too, so make it check for
arch_spin_is_locked(rcu_state.ofl_lock).
There's no arch_spin_lock_irqsave() so we have to manually save and
restore local interrupts around the locking.
At Paul's request based on Neeraj's analysis, make rcu_gp_init not just
wait but *exclude* any CPU online/offline activity, which was fairly
much true already by virtue of it holding rcu_state.ofl_lock.
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
* kvm-arm64/pmu-bl:
: .
: Improve PMU support on heterogeneous systems, courtesy of Alexandru Elisei
: .
KVM: arm64: Refuse to run VCPU if the PMU doesn't match the physical CPU
KVM: arm64: Add KVM_ARM_VCPU_PMU_V3_SET_PMU attribute
KVM: arm64: Keep a list of probed PMUs
KVM: arm64: Keep a per-VM pointer to the default PMU
perf: Fix wrong name in comment for struct perf_cpu_context
KVM: arm64: Do not change the PMU event filter after a VCPU has run
Signed-off-by: Marc Zyngier <maz@kernel.org>
Userspace can assign a PMU to a VCPU with the KVM_ARM_VCPU_PMU_V3_SET_PMU
device ioctl. If the VCPU is scheduled on a physical CPU which has a
different PMU, the perf events needed to emulate a guest PMU won't be
scheduled in and the guest performance counters will stop counting. Treat
it as an userspace error and refuse to run the VCPU in this situation.
Suggested-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Alexandru Elisei <alexandru.elisei@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20220127161759.53553-7-alexandru.elisei@arm.com
When KVM creates an event and there are more than one PMUs present on the
system, perf_init_event() will go through the list of available PMUs and
will choose the first one that can create the event. The order of the PMUs
in this list depends on the probe order, which can change under various
circumstances, for example if the order of the PMU nodes change in the DTB
or if asynchronous driver probing is enabled on the kernel command line
(with the driver_async_probe=armv8-pmu option).
Another consequence of this approach is that on heteregeneous systems all
virtual machines that KVM creates will use the same PMU. This might cause
unexpected behaviour for userspace: when a VCPU is executing on the
physical CPU that uses this default PMU, PMU events in the guest work
correctly; but when the same VCPU executes on another CPU, PMU events in
the guest will suddenly stop counting.
Fortunately, perf core allows user to specify on which PMU to create an
event by using the perf_event_attr->type field, which is used by
perf_init_event() as an index in the radix tree of available PMUs.
Add the KVM_ARM_VCPU_PMU_V3_CTRL(KVM_ARM_VCPU_PMU_V3_SET_PMU) VCPU
attribute to allow userspace to specify the arm_pmu that KVM will use when
creating events for that VCPU. KVM will make no attempt to run the VCPU on
the physical CPUs that share the PMU, leaving it up to userspace to manage
the VCPU threads' affinity accordingly.
To ensure that KVM doesn't expose an asymmetric system to the guest, the
PMU set for one VCPU will be used by all other VCPUs. Once a VCPU has run,
the PMU cannot be changed in order to avoid changing the list of available
events for a VCPU, or to change the semantics of existing events.
Signed-off-by: Alexandru Elisei <alexandru.elisei@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20220127161759.53553-6-alexandru.elisei@arm.com
The ARM PMU driver calls kvm_host_pmu_init() after probing to tell KVM that
a hardware PMU is available for guest emulation. Heterogeneous systems can
have more than one PMU present, and the callback gets called multiple
times, once for each of them. Keep track of all the PMUs available to KVM,
as they're going to be needed later.
Reviewed-by: Reiji Watanabe <reijiw@google.com>
Signed-off-by: Alexandru Elisei <alexandru.elisei@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20220127161759.53553-5-alexandru.elisei@arm.com
As we are about to allow selection of the PMU exposed to a guest, start by
keeping track of the default one instead of only the PMU version.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Alexandru Elisei <alexandru.elisei@arm.com>
Link: https://lore.kernel.org/r/20220127161759.53553-4-alexandru.elisei@arm.com
