Now it's possible to control supported hashflows.
Added hashflow set/get callbacks.
Also, disabling RXH_IP_SRC/DST for TCP would disable then for UDP.
TCP and UDP supports only:
ethtool -U eth0 rx-flow-hash tcp4 sd
RXH_IP_SRC + RXH_IP_DST
ethtool -U eth0 rx-flow-hash tcp4 sdfn
RXH_IP_SRC + RXH_IP_DST + RXH_L4_B_0_1 + RXH_L4_B_2_3
Disabling happens because VirtioNET hashtype for IP doesn't check L4 proto,
it works for all IP packets(TCP, UDP, ICMP, etc.).
For TCP and UDP, it's possible to set IP+PORT hashes.
But disabling IP hashes will disable them for TCP and UDP simultaneously.
It's possible to set IP+PORT for TCP/UDP and disable/enable IP
for everything else(UDP, ICMP, etc.).
Signed-off-by: Andrew Melnychenko <andrew@daynix.com>
Link: https://lore.kernel.org/r/20220328175336.10802-5-andrew@daynix.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Added features for RSS.
Added initialization, RXHASH feature and ethtool ops.
By default RSS/RXHASH is disabled.
Virtio RSS "IPv6 extensions" hashes disabled.
Added ethtools ops to set key and indirection table.
Signed-off-by: Andrew Melnychenko <andrew@daynix.com>
Link: https://lore.kernel.org/r/20220328175336.10802-3-andrew@daynix.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
After waking up a suspended VM, the kernel prints the following trace
for virtio drivers which do not directly call virtio_device_ready() in
the .restore:
PM: suspend exit
irq 22: nobody cared (try booting with the "irqpoll" option)
Call Trace:
<IRQ>
dump_stack_lvl+0x38/0x49
dump_stack+0x10/0x12
__report_bad_irq+0x3a/0xaf
note_interrupt.cold+0xb/0x60
handle_irq_event+0x71/0x80
handle_fasteoi_irq+0x95/0x1e0
__common_interrupt+0x6b/0x110
common_interrupt+0x63/0xe0
asm_common_interrupt+0x1e/0x40
? __do_softirq+0x75/0x2f3
irq_exit_rcu+0x93/0xe0
sysvec_apic_timer_interrupt+0xac/0xd0
</IRQ>
<TASK>
asm_sysvec_apic_timer_interrupt+0x12/0x20
arch_cpu_idle+0x12/0x20
default_idle_call+0x39/0xf0
do_idle+0x1b5/0x210
cpu_startup_entry+0x20/0x30
start_secondary+0xf3/0x100
secondary_startup_64_no_verify+0xc3/0xcb
</TASK>
handlers:
[<000000008f9bac49>] vp_interrupt
[<000000008f9bac49>] vp_interrupt
Disabling IRQ #22
This happens because we don't invoke .enable_cbs callback in
virtio_device_restore(). That callback is used by some transports
(e.g. virtio-pci) to enable interrupts.
Let's fix it, by calling virtio_device_ready() as we do in
virtio_dev_probe(). This function calls .enable_cts callback and sets
DRIVER_OK status bit.
This fix also avoids setting DRIVER_OK twice for those drivers that
call virtio_device_ready() in the .restore.
Fixes: d50497eb4e ("virtio_config: introduce a new .enable_cbs method")
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Link: https://lore.kernel.org/r/20220322114313.116516-1-sgarzare@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
When using pthreads, one has to compile and link with -lpthread,
otherwise e.g. glibc is not guaranteed to be reentrant.
This replaces -lpthread.
Reported-by: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Allow an admin creating a vdpa device to specify the max MTU for the
net device.
For example, to create a device with max MTU of 1000, the following
command can be used:
$ vdpa dev add name vdpa-a mgmtdev auxiliary/mlx5_core.sf.1 mtu 1000
This configuration mechanism assumes that vdpa is the sole real user of
the function. mlx5_core could theoretically change the mtu of the
function using the ip command on the mlx5_core net device but this
should not be done.
Reviewed-by: Si-Wei Liu<si-wei.liu@oracle.com>
Signed-off-by: Eli Cohen <elic@nvidia.com>
Link: https://lore.kernel.org/r/20220221121927.194728-1-elic@nvidia.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
On some platforms/devices, there may not be enough MSI vectors
allocated for the virtqueues and config changes. In such a case,
the interrupt sources(virtqueues, config changes) must share
an IRQ/vector, to avoid initialization failures, keep
the device functional.
This commit handles three cases:
(1) number of the allocated vectors == the number of virtqueues + 1
(config changes), every virtqueue and the config interrupt has
a separated vector/IRQ, the best and the most likely case.
(2) number of the allocated vectors is less than the best case, but
greater than 1. In this case, all virtqueues share a vector/IRQ,
the config interrupt has a separated vector/IRQ
(3) only one vector is allocated, in this case, the virtqueues and
the config interrupt share a vector/IRQ. The worst and most
unlikely case.
Otherwise, it needs to fail.
This commit introduces some helper functions:
ifcvf_set_vq_vector() and ifcvf_set_config_vector() sets virtqueue
vector and config vector in the device config space, so that
the device can send interrupt DMA.
Signed-off-by: Zhu Lingshan <lingshan.zhu@intel.com>
Link: https://lore.kernel.org/r/20220222115428.998334-5-lingshan.zhu@intel.com
Signed-off-by: Tom Rix <trix@redhat.com>
Link: https://lore.kernel.org/r/20220315124130.1710030-1-trix@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
In vhost_enable_notify() we enable the notifications and we read
the avail index to check if new buffers have become available in
the meantime.
We are not caching the avail index, so when the device will call
vhost_get_vq_desc(), it will find the old value in the cache and
it will read the avail index again.
It would be better to refresh the cache every time we read avail
index, so let's change vhost_enable_notify() caching the value in
`avail_idx` and compare it with `last_avail_idx` to check if there
are new buffers available.
We don't expect a significant performance boost because
the above path is not very common, indeed vhost_enable_notify()
is often called with unlikely(), expecting that avail index has
not been updated.
We ran virtio-test/vhost-test and noticed minimal improvement as
expected. To stress the patch more, we modified vhost_test.ko to
call vhost_enable_notify()/vhost_disable_notify() on every cycle
when calling vhost_get_vq_desc(); in this case we observed a more
evident improvement, with a reduction of the test execution time
of about 3.7%.
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Link: https://lore.kernel.org/r/20220121153108.187291-1-sgarzare@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Pull kvm fix from Paolo Bonzini:
"Fix for the SLS mitigation, which makes a 'SETcc/RET' pair grow
to 'SETcc/RET/INT3'.
This doesn't fit in 4 bytes any more, so the alignment has to
change to 8 for this case"
* tag 'for-linus-5.17' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
kvm/emulate: Fix SETcc emulation function offsets with SLS
Pull input fixes from Dmitry Torokhov:
"Two driver fixes:
- a fix for zinitix touchscreen to properly report contacts
- a fix for aiptek tablet driver to be more resilient to devices with
incorrect descriptors"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: aiptek - properly check endpoint type
Input: zinitix - do not report shadow fingers
The commit in Fixes started adding INT3 after RETs as a mitigation
against straight-line speculation.
The fastop SETcc implementation in kvm's insn emulator uses macro magic
to generate all possible SETcc functions and to jump to them when
emulating the respective instruction.
However, it hardcodes the size and alignment of those functions to 4: a
three-byte SETcc insn and a single-byte RET. BUT, with SLS, there's an
INT3 that gets slapped after the RET, which brings the whole scheme out
of alignment:
15: 0f 90 c0 seto %al
18: c3 ret
19: cc int3
1a: 0f 1f 00 nopl (%rax)
1d: 0f 91 c0 setno %al
20: c3 ret
21: cc int3
22: 0f 1f 00 nopl (%rax)
25: 0f 92 c0 setb %al
28: c3 ret
29: cc int3
and this explodes like this:
int3: 0000 [#1] PREEMPT SMP PTI
CPU: 0 PID: 2435 Comm: qemu-system-x86 Not tainted 5.17.0-rc8-sls #1
Hardware name: Dell Inc. Precision WorkStation T3400 /0TP412, BIOS A14 04/30/2012
RIP: 0010:setc+0x5/0x8 [kvm]
Code: 00 00 0f 1f 00 0f b6 05 43 24 06 00 c3 cc 0f 1f 80 00 00 00 00 0f 90 c0 c3 cc 0f \
1f 00 0f 91 c0 c3 cc 0f 1f 00 0f 92 c0 c3 cc <0f> 1f 00 0f 93 c0 c3 cc 0f 1f 00 \
0f 94 c0 c3 cc 0f 1f 00 0f 95 c0
Call Trace:
<TASK>
? x86_emulate_insn [kvm]
? x86_emulate_instruction [kvm]
? vmx_handle_exit [kvm_intel]
? kvm_arch_vcpu_ioctl_run [kvm]
? kvm_vcpu_ioctl [kvm]
? __x64_sys_ioctl
? do_syscall_64
? entry_SYSCALL_64_after_hwframe
</TASK>
Raise the alignment value when SLS is enabled and use a macro for that
instead of hard-coding naked numbers.
Fixes: e463a09af2 ("x86: Add straight-line-speculation mitigation")
Reported-by: Jamie Heilman <jamie@audible.transient.net>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Jamie Heilman <jamie@audible.transient.net>
Link: https://lore.kernel.org/r/YjGzJwjrvxg5YZ0Z@audible.transient.net
[Add a comment and a bit of safety checking, since this is going to be changed
again for IBT support. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Pull ARM SoC fix from Arnd Bergmann:
"Here is one last regression fix for 5.17, reverting a patch that went
into 5.16 as a cleanup that ended up breaking external interrupts on
Layerscape chips.
The revert makes it work again, but also reintroduces a build time
warning about the nonstandard DT binding that will have to be dealt
with in the future"
* tag 'soc-fixes-5.17-4' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
Revert "arm64: dts: freescale: Fix 'interrupt-map' parent address cells"
Pull SCSI fixes from James Bottomley:
"Two small(ish) fixes, both in drivers"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: fnic: Finish scsi_cmnd before dropping the spinlock
scsi: mpt3sas: Page fault in reply q processing
Pull perf tools fixes from Arnaldo Carvalho de Melo:
- Avoid iterating empty evlist, fixing a segfault with 'perf stat --null'
- Ignore case in topdown.slots check, fixing issue with Intel Icelake
JSON metrics.
- Fix symbol size calculation condition for fixing up corner case
symbol end address obtained from Kallsyms.
* tag 'perf-tools-fixes-for-v5.17-2022-03-19' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux:
perf parse-events: Ignore case in topdown.slots check
perf evlist: Avoid iteration for empty evlist.
perf symbols: Fix symbol size calculation condition
Pull char/misc driver fix from Greg KH:
"Here is a single driver fix for 5.17-final that has been submitted
many times but I somehow missed it in my patch queue:
- fix for counter sysfs code for reported problem
This has been in linux-next all week with no reported issues"
* tag 'char-misc-5.17-final' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
counter: Stop using dev_get_drvdata() to get the counter device
Pull USB fixes from Greg KH:
"Here are some small remaining USB fixes for 5.17-final.
They include:
- two USB gadget driver fixes for reported problems
- usbtmc driver fix for syzbot found issues
- musb patch partial revert to resolve a reported regression.
All of these have been in linux-next this week with no reported
problems"
* tag 'usb-5.17-final' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
usb: gadget: Fix use-after-free bug by not setting udc->dev.driver
usb: usbtmc: Fix bug in pipe direction for control transfers
partially Revert "usb: musb: Set the DT node on the child device"
usb: gadget: rndis: prevent integer overflow in rndis_set_response()
Before this patch, the symbol end address fixup to be called, needed two
conditions being met:
if (prev->end == prev->start && prev->end != curr->start)
Where
"prev->end == prev->start" means that prev is zero-long
(and thus needs a fixup)
and
"prev->end != curr->start" means that fixup hasn't been applied yet
However, this logic is incorrect in the following situation:
*curr = {rb_node = {__rb_parent_color = 278218928,
rb_right = 0x0, rb_left = 0x0},
start = 0xc000000000062354,
end = 0xc000000000062354, namelen = 40, type = 2 '\002',
binding = 0 '\000', idle = 0 '\000', ignore = 0 '\000',
inlined = 0 '\000', arch_sym = 0 '\000', annotate2 = false,
name = 0x1159739e "kprobe_optinsn_page\t[__builtin__kprobes]"}
*prev = {rb_node = {__rb_parent_color = 278219041,
rb_right = 0x109548b0, rb_left = 0x109547c0},
start = 0xc000000000062354,
end = 0xc000000000062354, namelen = 12, type = 2 '\002',
binding = 1 '\001', idle = 0 '\000', ignore = 0 '\000',
inlined = 0 '\000', arch_sym = 0 '\000', annotate2 = false,
name = 0x1095486e "optinsn_slot"}
In this case, prev->start == prev->end == curr->start == curr->end,
thus the condition above thinks that "we need a fixup due to zero
length of prev symbol, but it has been probably done, since the
prev->end == curr->start", which is wrong.
After the patch, the execution path proceeds to arch__symbols__fixup_end
function which fixes up the size of prev symbol by adding page_size to
its end offset.
Fixes: 3b01a413c1 ("perf symbols: Improve kallsyms symbol end addr calculation")
Signed-off-by: Michael Petlan <mpetlan@redhat.com>
Cc: Athira Jajeev <atrajeev@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Link: http://lore.kernel.org/lkml/20220317135536.805-1-mpetlan@redhat.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Pull arm64 fixes from Catalin Marinas:
"Fix two compiler warnings introduced by recent commits: pointer
arithmetic and double initialisation of struct field"
* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
arm64: errata: avoid duplicate field initializer
arm64: fix clang warning about TRAMP_VALIAS
Pull cifs fix from Steve French:
"Small fix for regression in multiuser mounts.
The additional improvements suggested by Ronnie to make the server and
session status handling code easier to read can wait for the 5.18
merge window."
* tag '5.17-rc8-smb3-fix' of git://git.samba.org/sfrench/cifs-2.6:
smb3: fix incorrect session setup check for multiuser mounts
Pull block fixes from Jens Axboe:
- Revert of a nvme target feature (Hannes)
- Fix a memory leak with rq-qos (Ming)
* tag 'block-5.17-2022-03-18' of git://git.kernel.dk/linux-block:
nvmet: revert "nvmet: make discovery NQN configurable"
block: release rq qos structures for queue without disk
Pull drm fixes from Dave Airlie:
"A few minor changes to finish things off, one mgag200 regression, imx
fix and couple of panel changes.
imx:
- Don't test bus flags in atomic check
mgag200:
- Fix PLL setup on some models
panel:
- Fix bpp settings on Innolux G070Y2-L01
- Fix DRM_PANEL_EDP Kconfig dependencies"
* tag 'drm-fixes-2022-03-18' of git://anongit.freedesktop.org/drm/drm:
drm: Don't make DRM_PANEL_BRIDGE dependent on DRM_KMS_HELPERS
drm/panel: simple: Fix Innolux G070Y2-L01 BPP settings
drm/imx: parallel-display: Remove bus flags check in imx_pd_bridge_atomic_check()
drm/mgag200: Fix PLL setup for g200wb and g200ew
The '.type' field is initialized both in place and in the macro
as reported by this W=1 warning:
arch/arm64/include/asm/cpufeature.h:281:9: error: initialized field overwritten [-Werror=override-init]
281 | (ARM64_CPUCAP_SCOPE_LOCAL_CPU | ARM64_CPUCAP_OPTIONAL_FOR_LATE_CPU)
| ^
arch/arm64/kernel/cpu_errata.c:136:17: note: in expansion of macro 'ARM64_CPUCAP_LOCAL_CPU_ERRATUM'
136 | .type = ARM64_CPUCAP_LOCAL_CPU_ERRATUM, \
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
arch/arm64/kernel/cpu_errata.c:145:9: note: in expansion of macro 'ERRATA_MIDR_RANGE'
145 | ERRATA_MIDR_RANGE(m, var, r_min, var, r_max)
| ^~~~~~~~~~~~~~~~~
arch/arm64/kernel/cpu_errata.c:613:17: note: in expansion of macro 'ERRATA_MIDR_REV_RANGE'
613 | ERRATA_MIDR_REV_RANGE(MIDR_CORTEX_A510, 0, 0, 2),
| ^~~~~~~~~~~~~~~~~~~~~
arch/arm64/include/asm/cpufeature.h:281:9: note: (near initialization for 'arm64_errata[18].type')
281 | (ARM64_CPUCAP_SCOPE_LOCAL_CPU | ARM64_CPUCAP_OPTIONAL_FOR_LATE_CPU)
| ^
Remove the extranous initializer.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 1dd498e5e2 ("KVM: arm64: Workaround Cortex-A510's single-step and PAC trap errata")
Link: https://lore.kernel.org/r/20220316183800.1546731-1-arnd@kernel.org
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The newly introduced TRAMP_VALIAS definition causes a build warning
with clang-14:
arch/arm64/include/asm/vectors.h:66:31: error: arithmetic on a null pointer treated as a cast from integer to pointer is a GNU extension [-Werror,-Wnull-pointer-arithmetic]
return (char *)TRAMP_VALIAS + SZ_2K * slot;
Change the addition to something clang does not complain about.
Fixes: bd09128d16 ("arm64: Add percpu vectors for EL1")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: James Morse <james.morse@arm.com>
Link: https://lore.kernel.org/r/20220316183833.1563139-1-arnd@kernel.org
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Pull networking fixes from Jakub Kicinski:
"Including fixes from netfilter, ipsec, and wireless.
A few last minute revert / disable and fix patches came down from our
sub-trees. We're not waiting for any fixes at this point.
Current release - regressions:
- Revert "netfilter: nat: force port remap to prevent shadowing
well-known ports", restore working conntrack on asymmetric paths
- Revert "ath10k: drop beacon and probe response which leak from
other channel", restore working AP and mesh mode on QCA9984
- eth: intel: fix hang during reboot/shutdown
Current release - new code bugs:
- netfilter: nf_tables: disable register tracking, it needs more work
to cover all corner cases
Previous releases - regressions:
- ipv6: fix skb_over_panic in __ip6_append_data when (admin-only)
extension headers get specified
- esp6: fix ESP over TCP/UDP, interpret ipv6_skip_exthdr's return
value more selectively
- bnx2x: fix driver load failure when FW not present in initrd
Previous releases - always broken:
- vsock: stop destroying unrelated sockets in nested virtualization
- packet: fix slab-out-of-bounds access in packet_recvmsg()
Misc:
- add Paolo Abeni to networking maintainers!"
* tag 'net-5.17-final' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (26 commits)
iavf: Fix hang during reboot/shutdown
net: mscc: ocelot: fix backwards compatibility with single-chain tc-flower offload
net: bcmgenet: skip invalid partial checksums
bnx2x: fix built-in kernel driver load failure
net: phy: mscc: Add MODULE_FIRMWARE macros
net: dsa: Add missing of_node_put() in dsa_port_parse_of
net: handle ARPHRD_PIMREG in dev_is_mac_header_xmit()
Revert "ath10k: drop beacon and probe response which leak from other channel"
hv_netvsc: Add check for kvmalloc_array
iavf: Fix double free in iavf_reset_task
ice: destroy flow director filter mutex after releasing VSIs
ice: fix NULL pointer dereference in ice_update_vsi_tx_ring_stats()
Add Paolo Abeni to networking maintainers
atm: eni: Add check for dma_map_single
net/packet: fix slab-out-of-bounds access in packet_recvmsg()
net: mdio: mscc-miim: fix duplicate debugfs entry
net: phy: marvell: Fix invalid comparison in the resume and suspend functions
esp6: fix check on ipv6_skip_exthdr's return value
net: dsa: microchip: add spi_device_id tables
netfilter: nf_tables: disable register tracking
...
Pull ACPI fix from Rafael Wysocki:
"Revert recent commit that caused multiple systems to misbehave due to
firmware issues"
* tag 'acpi-5.17-rc9' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
Revert "ACPI: scan: Do not add device IDs from _CID if _HID is not valid"
Merge misc fixes from Andrew Morton:
"Four patches.
Subsystems affected by this patch series: mm/swap, kconfig, ocfs2, and
selftests"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
selftests: vm: fix clang build error multiple output files
ocfs2: fix crash when initialize filecheck kobj fails
configs/debug: restore DEBUG_INFO=y for overriding
mm: swap: get rid of livelock in swapin readahead
When building the vm selftests using clang, some errors are seen due to
having headers in the compilation command:
clang -Wall -I ../../../../usr/include -no-pie gup_test.c ../../../../mm/gup_test.h -lrt -lpthread -o .../tools/testing/selftests/vm/gup_test
clang: error: cannot specify -o when generating multiple output files
make[1]: *** [../lib.mk:146: .../tools/testing/selftests/vm/gup_test] Error 1
Rework to add the header files to LOCAL_HDRS before including ../lib.mk,
since the dependency is evaluated in '$(OUTPUT)/%:%.c $(LOCAL_HDRS)' in
file lib.mk.
Link: https://lkml.kernel.org/r/20220304000645.1888133-1-yosryahmed@google.com
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In our testing, a livelock task was found. Through sysrq printing, same
stack was found every time, as follows:
__swap_duplicate+0x58/0x1a0
swapcache_prepare+0x24/0x30
__read_swap_cache_async+0xac/0x220
read_swap_cache_async+0x58/0xa0
swapin_readahead+0x24c/0x628
do_swap_page+0x374/0x8a0
__handle_mm_fault+0x598/0xd60
handle_mm_fault+0x114/0x200
do_page_fault+0x148/0x4d0
do_translation_fault+0xb0/0xd4
do_mem_abort+0x50/0xb0
The reason for the livelock is that swapcache_prepare() always returns
EEXIST, indicating that SWAP_HAS_CACHE has not been cleared, so that it
cannot jump out of the loop. We suspect that the task that clears the
SWAP_HAS_CACHE flag never gets a chance to run. We try to lower the
priority of the task stuck in a livelock so that the task that clears
the SWAP_HAS_CACHE flag will run. The results show that the system
returns to normal after the priority is lowered.
In our testing, multiple real-time tasks are bound to the same core, and
the task in the livelock is the highest priority task of the core, so
the livelocked task cannot be preempted.
Although cond_resched() is used by __read_swap_cache_async, it is an
empty function in the preemptive system and cannot achieve the purpose
of releasing the CPU. A high-priority task cannot release the CPU
unless preempted by a higher-priority task. But when this task is
already the highest priority task on this core, other tasks will not be
able to be scheduled. So we think we should replace cond_resched() with
schedule_timeout_uninterruptible(1), schedule_timeout_interruptible will
call set_current_state first to set the task state, so the task will be
removed from the running queue, so as to achieve the purpose of giving
up the CPU and prevent it from running in kernel mode for too long.
(akpm: ugly hack becomes uglier. But it fixes the issue in a
backportable-to-stable fashion while we hopefully work on something
better)
Link: https://lkml.kernel.org/r/20220221111749.1928222-1-cgel.zte@gmail.com
Signed-off-by: Guo Ziliang <guo.ziliang@zte.com.cn>
Reported-by: Zeal Robot <zealci@zte.com.cn>
Reviewed-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
Reviewed-by: Jiang Xuexin <jiang.xuexin@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Roger Quadros <rogerq@kernel.org>
Cc: Ziliang Guo <guo.ziliang@zte.com.cn>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
