Currently, bfq can't handle sync io concurrently as long as they
are not issued from root group. This is because
'bfqd->num_groups_with_pending_reqs > 0' is always true in
bfq_asymmetric_scenario().
The way that bfqg is counted into 'num_groups_with_pending_reqs':
Before this patch:
1) root group will never be counted.
2) Count if bfqg or it's child bfqgs have pending requests.
3) Don't count if bfqg and it's child bfqgs complete all the requests.
After this patch:
1) root group is counted.
2) Count if bfqg have pending requests.
3) Don't count if bfqg complete all the requests.
With this change, the occasion that only one group is activated can be
detected, and next patch will support concurrent sync io in the
occasion.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Paolo Valente <paolo.valente@linaro.org>
Link: https://lore.kernel.org/r/20220916071942.214222-4-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The calling relationship in blk_mq_destroy_queue() is as follows:
blk_mq_destroy_queue()
...
-> blk_queue_start_drain()
-> blk_freeze_queue_start() <- called
...
-> blk_freeze_queue()
-> blk_freeze_queue_start() <- called again
-> blk_mq_freeze_queue_wait()
...
So there is a redundant call to blk_freeze_queue_start().
Replace blk_freeze_queue() with blk_mq_freeze_queue_wait() to avoid the
redundant call.
Signed-off-by: Jinlong Chen <nickyc975@zju.edu.cn>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20221030083212.1251255-1-nickyc975@zju.edu.cn
Signed-off-by: Jens Axboe <axboe@kernel.dk>
David Jeffery found one double ->queue_rq() issue, so far it can
be triggered in VM use case because of long vmexit latency or preempt
latency of vCPU pthread or long page fault in vCPU pthread, then block
IO req could be timed out before queuing the request to hardware but after
calling blk_mq_start_request() during ->queue_rq(), then timeout handler
may handle it by requeue, then double ->queue_rq() is caused, and kernel
panic.
So far, it is driver's responsibility to cover the race between timeout
and completion, so it seems supposed to be solved in driver in theory,
given driver has enough knowledge.
But it is really one common problem, lots of driver could have similar
issue, and could be hard to fix all affected drivers, even it isn't easy
for driver to handle the race. So David suggests this patch by draining
in-progress ->queue_rq() for solving this issue.
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Keith Busch <kbusch@kernel.org>
Cc: virtualization@lists.linux-foundation.org
Cc: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: David Jeffery <djeffery@redhat.com>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20221026051957.358818-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The fact that blk_mq_destroy_queue also drops a queue reference leads
to various places having to grab an extra reference. Move the call to
blk_put_queue into the callers to allow removing the extra references.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Keith Busch <kbusch@kernel.org>
Link: https://lore.kernel.org/r/20221018135720.670094-2-hch@lst.de
[axboe: fix fabrics_q vs admin_q conflict in nvme core.c]
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The current reference management logic of io scheduler modules contains
refcnt problems. For example, blk_mq_init_sched may fail before or after
the calling of e->ops.init_sched. If it fails before the calling, it does
nothing to the reference to the io scheduler module. But if it fails after
the calling, it releases the reference by calling kobject_put(&eq->kobj).
As the callers of blk_mq_init_sched can't know exactly where the failure
happens, they can't handle the reference to the io scheduler module
properly: releasing the reference on failure results in double-release if
blk_mq_init_sched has released it, and not releasing the reference results
in ghost reference if blk_mq_init_sched did not release it either.
The same problem also exists in io schedulers' init_sched implementations.
We can address the problem by adding releasing statements to the error
handling procedures of blk_mq_init_sched and init_sched implementations.
But that is counterintuitive and requires modifications to existing io
schedulers.
Instead, We make elevator_alloc get the io scheduler module references
that will be released by elevator_release. And then, we match each
elevator_get with an elevator_put. Therefore, each reference to an io
scheduler module explicitly has its own getter and releaser, and we no
longer need to worry about the refcnt problems.
The bugs and the patch can be validated with tools here:
https://github.com/nickyc975/linux_elv_refcnt_bug.git
[hch: split out a few bits into separate patches, use a non-try
module_get in elevator_alloc]
Signed-off-by: Jinlong Chen <nickyc975@zju.edu.cn>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20221020064819.1469928-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Make sure we have helpers for all relevant module refcount operations on
the elevator_type in elevator.h, and use them. Move the call to the get
helper in blk_mq_elv_switch_none a bit so that it is obvious with a less
verbose comment.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20221020064819.1469928-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Commit b5dc5d4d1f ("block,bfq: Disable writeback throttling") tries to
disable wbt for bfq, it's done by calling wbt_disable_default() in
bfq_init_queue(). However, wbt is still enabled if default elevator is
bfq:
device_add_disk
elevator_init_mq
bfq_init_queue
wbt_disable_default -> done nothing
blk_register_queue
wbt_enable_default -> wbt is enabled
Fix the problem by adding a new flag ELEVATOR_FLAG_DISBALE_WBT, bfq
will set the flag in bfq_init_queue, and following wbt_enable_default()
won't enable wbt while the flag is set.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20221019121518.3865235-7-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This won't cause any severe problem currently, however, this doesn't
seems appropriate:
1) 'ioc->params' is read from multiple places without holding
'ioc->lock', unexpected value might be read if writing it concurrently.
2) If configuration is changed while io is throttling, the functionality
might be affected. For example, if module params is updated and cost
becomes smaller, waiting for timer that is caculated under old
configuration is not appropriate.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20221012094035.390056-4-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
ioc_qos_write() and ioc_cost_model_write() are the same:
1) hold lock to read 'ioc->params' to local variable;
2) update params to local variable without lock;
3) hold lock to write local variable to 'ioc->params';
In theroy, if user updates params concurrenty, the params might be lost:
t1: update params a t2: update params b
spin_lock_irq(&ioc->lock);
memcpy(qos, ioc->params.qos, sizeof(qos))
spin_unlock_irq(&ioc->lock);
qos[a] = xxx;
spin_lock_irq(&ioc->lock);
memcpy(qos, ioc->params.qos, sizeof(qos))
spin_unlock_irq(&ioc->lock);
qos[b] = xxx;
spin_lock_irq(&ioc->lock);
memcpy(ioc->params.qos, qos, sizeof(qos));
ioc_refresh_params(ioc, true);
spin_unlock_irq(&ioc->lock);
spin_lock_irq(&ioc->lock);
// updates of a will be lost
memcpy(ioc->params.qos, qos, sizeof(qos));
ioc_refresh_params(ioc, true);
spin_unlock_irq(&ioc->lock);
Althrough this is not common case, the problem can by fixed easily by
holding the lock through the read, update, write process.
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://lore.kernel.org/r/20221012094035.390056-3-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pull kvm fixes from Paolo Bonzini:
"RISC-V:
- Fix compilation without RISCV_ISA_ZICBOM
- Fix kvm_riscv_vcpu_timer_pending() for Sstc
ARM:
- Fix a bug preventing restoring an ITS containing mappings for very
large and very sparse device topology
- Work around a relocation handling error when compiling the nVHE
object with profile optimisation
- Fix for stage-2 invalidation holding the VM MMU lock for too long
by limiting the walk to the largest block mapping size
- Enable stack protection and branch profiling for VHE
- Two selftest fixes
x86:
- add compat implementation for KVM_X86_SET_MSR_FILTER ioctl
selftests:
- synchronize includes between include/uapi and tools/include/uapi"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
tools: include: sync include/api/linux/kvm.h
KVM: x86: Add compat handler for KVM_X86_SET_MSR_FILTER
KVM: x86: Copy filter arg outside kvm_vm_ioctl_set_msr_filter()
kvm: Add support for arch compat vm ioctls
RISC-V: KVM: Fix kvm_riscv_vcpu_timer_pending() for Sstc
RISC-V: Fix compilation without RISCV_ISA_ZICBOM
KVM: arm64: vgic: Fix exit condition in scan_its_table()
KVM: arm64: nvhe: Fix build with profile optimization
KVM: selftests: Fix number of pages for memory slot in memslot_modification_stress_test
KVM: arm64: selftests: Fix multiple versions of GIC creation
KVM: arm64: Enable stack protection and branch profiling for VHE
KVM: arm64: Limit stage2_apply_range() batch size to largest block
KVM: arm64: Work out supported block level at compile time
This reverts commit 72a9585972.
It broke reboots on big-endian MIPS and MIPS64 malta QEMU instances,
which use the syscon driver. Little-endian is not effected, which means
likely it's important to handle regmap_get_val_endian() in this function
after all.
Fixes: 72a9585972 ("mfd: syscon: Remove repetition of the regmap_get_val_endian()")
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Lee Jones <lee@kernel.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull perf fixes from Borislav Petkov:
- Fix raw data handling when perf events are used in bpf
- Rework how SIGTRAPs get delivered to events to address a bunch of
problems with it. Add a selftest for that too
* tag 'perf_urgent_for_v6.1_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
bpf: Fix sample_flags for bpf_perf_event_output
selftests/perf_events: Add a SIGTRAP stress test with disables
perf: Fix missing SIGTRAPs
Pull scheduler fixes from Borislav Petkov:
- Adjust code to not trip up CFI
- Fix sched group cookie matching
* tag 'sched_urgent_for_v6.1_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched: Introduce struct balance_callback to avoid CFI mismatches
sched/core: Fix comparison in sched_group_cookie_match()
Pull objtool fix from Borislav Petkov:
- Fix ORC stack unwinding when GCOV is enabled
* tag 'objtool_urgent_for_v6.1_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/unwind/orc: Fix unreliable stack dump with gcov
Pull x86 fixes from Borislav Petkov:
"As usually the case, right after a major release, the tip urgent
branches accumulate a couple more fixes than normal. And here is the
x86, a bit bigger, urgent pile.
- Use the correct CPU capability clearing function on the error path
in Intel perf LBR
- A CFI fix to ftrace along with a simplification
- Adjust handling of zero capacity bit mask for resctrl cache
allocation on AMD
- A fix to the AMD microcode loader to attempt patch application on
every logical thread
- A couple of topology fixes to handle CPUID leaf 0x1f enumeration
info properly
- Drop a -mabi=ms compiler option check as both compilers support it
now anyway
- A couple of fixes to how the initial, statically allocated FPU
buffer state is setup and its interaction with dynamic states at
runtime"
* tag 'x86_urgent_for_v6.0_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/fpu: Fix copy_xstate_to_uabi() to copy init states correctly
perf/x86/intel/lbr: Use setup_clear_cpu_cap() instead of clear_cpu_cap()
ftrace,kcfi: Separate ftrace_stub() and ftrace_stub_graph()
x86/ftrace: Remove ftrace_epilogue()
x86/resctrl: Fix min_cbm_bits for AMD
x86/microcode/AMD: Apply the patch early on every logical thread
x86/topology: Fix duplicated core ID within a package
x86/topology: Fix multiple packages shown on a single-package system
hwmon/coretemp: Handle large core ID value
x86/Kconfig: Drop check for -mabi=ms for CONFIG_EFI_STUB
x86/fpu: Exclude dynamic states from init_fpstate
x86/fpu: Fix the init_fpstate size check with the actual size
x86/fpu: Configure init_fpstate attributes orderly
Pull io_uring follow-up from Jens Axboe:
"Currently the zero-copy has automatic fallback to normal transmit, and
it was decided that it'd be cleaner to return an error instead if the
socket type doesn't support it.
Zero-copy does work with UDP and TCP, it's more of a future proofing
kind of thing (eg for samba)"
* tag 'io_uring-6.1-2022-10-22' of git://git.kernel.dk/linux:
io_uring/net: fail zc sendmsg when unsupported by socket
io_uring/net: fail zc send when unsupported by socket
net: flag sockets supporting msghdr originated zerocopy
Pull hwmon fixes from Guenter Roeck:
- corsair-psu: Fix typo in USB id description, and add USB ID for new
PSU
- pwm-fan: Fix fan power handling when disabling fan control
* tag 'hwmon-for-v6.1-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
hwmon: (corsair-psu) Add USB id of the new HX1500i psu
hwmon: (pwm-fan) Explicitly switch off fan power when setting pwm1_enable to 0
hwmon: (corsair-psu) fix typo in USB id description
Pull i2c fixes from Wolfram Sang:
"RPM fix for qcom-cci, platform module alias for xiic, build warning
fix for mlxbf, typo fixes in comments"
* tag 'i2c-for-6.1-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
i2c: mlxbf: depend on ACPI; clean away ifdeffage
i2c: fix spelling typos in comments
i2c: qcom-cci: Fix ordering of pm_runtime_xx and i2c_add_adapter
i2c: xiic: Add platform module alias
Pull pci fixes from Bjorn Helgaas:
- Revert a simplification that broke pci-tegra due to a masking error
- Update MAINTAINERS for Kishon's email address change and TI
DRA7XX/J721E maintainer change
* tag 'pci-v6.1-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
MAINTAINERS: Update Kishon's email address in PCI endpoint subsystem
MAINTAINERS: Add Vignesh Raghavendra as maintainer of TI DRA7XX/J721E PCI driver
Revert "PCI: tegra: Use PCI_CONF1_EXT_ADDRESS() macro"
Pull missed media updates from Mauro Carvalho Chehab:
"It seems I screwed-up my previous pull request: it ends up that only
half of the media patches that were in linux-next got merged in -rc1.
The script which creates the signed tags silently failed due to
5.19->6.0 so it ended generating a tag with incomplete stuff.
So here are the missing parts:
- a DVB core security fix
- lots of fixes and cleanups for atomisp staging driver
- old drivers that are VB1 are being moved to staging to be
deprecated
- several driver updates - mostly for embedded systems, but there are
also some things addressing issues with some PC webcams, in the UVC
video driver"
* tag 'media/v6.1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (163 commits)
media: sun6i-csi: Move csi buffer definition to main header file
media: sun6i-csi: Introduce and use video helper functions
media: sun6i-csi: Add media ops with link notify callback
media: sun6i-csi: Remove controls handler from the driver
media: sun6i-csi: Register the media device after creation
media: sun6i-csi: Pass and store csi device directly in video code
media: sun6i-csi: Tidy up video code
media: sun6i-csi: Tidy up v4l2 code
media: sun6i-csi: Tidy up Kconfig
media: sun6i-csi: Use runtime pm for clocks and reset
media: sun6i-csi: Define and use variant to get module clock rate
media: sun6i-csi: Always set exclusive module clock rate
media: sun6i-csi: Tidy up platform code
media: sun6i-csi: Refactor main driver data structures
media: sun6i-csi: Define and use driver name and (reworked) description
media: cedrus: Add a Kconfig dependency on RESET_CONTROLLER
media: sun8i-rotate: Add a Kconfig dependency on RESET_CONTROLLER
media: sun8i-di: Add a Kconfig dependency on RESET_CONTROLLER
media: sun4i-csi: Add a Kconfig dependency on RESET_CONTROLLER
media: sun6i-csi: Add a Kconfig dependency on RESET_CONTROLLER
...
Provide a definition of KVM_CAP_DIRTY_LOG_RING_ACQ_REL.
Fixes: 17601bfed9 ("KVM: Add KVM_CAP_DIRTY_LOG_RING_ACQ_REL capability and config option")
Cc: Marc Zyngier <maz@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
