The same benchmark is running in group C & F, no other tasks are
running, the benchmark is capable to consumed all the CPUs.
We suppose the group C will win more CPU resources since it could
enjoy all the shares of group A, but it's F who wins much more.
The reason is because we have group B with shares as 2, since
A->cfs_rq.load.weight == B->se.load.weight == B->shares/nr_cpus,
so A->cfs_rq.load.weight become very small.
And in calc_group_shares() we calculate shares as:
Since the 'cfs_rq->load.weight' is too small, the load become 0
after scale down, although 'tg_shares' is 102400, shares of the se
which stand for group A on root cfs_rq become 2.
While the se of D on root cfs_rq is far more bigger than 2, so it
wins the battle.
Thus when scale_load_down() scale real weight down to 0, it's no
longer telling the real story, the caller will have the wrong
information and the calculation will be buggy.
This patch add check in scale_load_down(), so the real weight will
be >= MIN_SHARES after scale, after applied the group C wins as
expected.
Suggested-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Michael Wang <yun.wang@linux.alibaba.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org> Link: https://lkml.kernel.org/r/38e8e212-59a1-64b2-b247-b6d0b52d8dc1@linux.alibaba.com Signed-off-by: Sasha Levin <sashal@kernel.org>
To minimize latency, PREEMPT_RT kernels expires hrtimers in preemptible
softirq context by default. This can be overriden by marking the timer's
expiry with HRTIMER_MODE_HARD.
sched_clock_timer is missing this annotation: if its callback is preempted
and the duration of the preemption exceeds the wrap around time of the
underlying clocksource, sched clock will get out of sync.
Mark the sched_clock_timer for expiry in hard interrupt context.
Enclose the chained handler with chained_irq_{enter,exit}(), so that the
muxed interrupts get properly acked.
This patch also fixes a reboot bug on OX820 SoC, where the jiffies timer
interrupt is never acked. The kernel waits a clock tick forever in
calibrate_delay_converge(), which leads to a boot hang.
When the module is being removed, the module state is set to
MODULE_STATE_GOING. At this point, try_module_get() fails.
And when {full/open}_proxy_open() is being called,
it calls try_module_get() to try to hold module reference count.
If it fails, it warns about the possibility of debugfs file leak.
If {full/open}_proxy_open() is called while the module is being removed,
it fails to hold the module.
So, It warns about debugfs file leak. But it is not the debugfs file
leak case. So, this patch just adds module state checking routine
in the {full/open}_proxy_open().
Test commands:
#SHELL1
while :
do
modprobe netdevsim
echo 1 > /sys/bus/netdevsim/new_device
modprobe -rv netdevsim
done
#SHELL2
while :
do
cat /sys/kernel/debug/netdevsim/netdevsim1/ports/0/ipsec
done
Field bdi->io_pages added in commit 22cebe959493 ("mm: don't cap request
size based on read-ahead setting") removes unneeded split of read requests.
Stacked drivers do not call blk_queue_max_hw_sectors(). Instead they set
limits of their devices by blk_set_stacking_limits() + disk_stack_limits().
Field bio->io_pages stays zero until user set max_sectors_kb via sysfs.
This patch updates io_pages after merging limits in disk_stack_limits().
Commit 6fe5e94f5cac ("dm: do not allow readahead to limit IO size") fixed
the same problem for device-mapper devices, this one fixes MD RAIDs.
Fixes: 22cebe959493 ("mm: don't cap request size based on read-ahead setting") Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de> Reviewed-by: Bob Liu <bob.liu@oracle.com> Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Signed-off-by: Song Liu <songliubraving@fb.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When dma_mmap_coherent() sets up a mapping to unencrypted coherent memory
under SEV encryption and sometimes under SME encryption, it will actually
set up an encrypted mapping rather than an unencrypted, causing devices
that DMAs from that memory to read encrypted contents. Fix this.
When force_dma_unencrypted() returns true, the linear kernel map of the
coherent pages have had the encryption bit explicitly cleared and the
page content is unencrypted. Make sure that any additional PTEs we set
up to these pages also have the encryption bit cleared by having
dma_pgprot() return a protection with the encryption bit cleared in this
case.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Acked-by: Tom Lendacky <thomas.lendacky@amd.com> Link: https://lkml.kernel.org/r/20200304114527.3636-3-thomas_os@shipmail.org Signed-off-by: Sasha Levin <sashal@kernel.org>
When SEV or SME is enabled and active, vm_get_page_prot() typically
returns with the encryption bit set. This means that users of
pgprot_modify(, vm_get_page_prot()) (mprotect_fixup(), do_mmap()) end up
with a value of vma->vm_pg_prot that is not consistent with the intended
protection of the PTEs.
This is also important for fault handlers that rely on the VMA
vm_page_prot to set the page protection. Fix this by not allowing
pgprot_modify() to change the encryption bit, similar to how it's done
for PAT bits.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com> Acked-by: Tom Lendacky <thomas.lendacky@amd.com> Link: https://lkml.kernel.org/r/20200304114527.3636-2-thomas_os@shipmail.org Signed-off-by: Sasha Levin <sashal@kernel.org>
The reason for clearing boot_ec_is_ecdt in acpi_ec_add() (if a
PNP0C09 device object matching the ECDT boot EC had been found in
the namespace) was to cause acpi_ec_ecdt_start() to return early,
but since the latter does not look at boot_ec_is_ecdt any more,
acpi_ec_add() need not clear it.
Moreover, doing that may be confusing as it may cause "DSDT" to be
printed instead of "ECDT" in the EC initialization completion
message, so stop doing it.
While at it, split the EC initialization completion message into
two messages, one regarding the boot EC and another one printed
regardless of whether or not the EC at hand is the boot one.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Bail out early if the xHC host needs to be reset at resume
but driver can't access xHC PCI registers.
If xhci driver already fails to reset the controller then there
is no point in attempting to free, re-initialize, re-allocate and
re-start the host. If failure to access the host is detected later,
failing the resume, xhci interrupts will be double freed
when remove is called.
Commit c5c3a48061b5 ("media: imx: Try colorimetry at both sink and
source pads") reworked the way that formats are set on the sink pad of
the CSI subdevice, and accidentally removed video field handling.
Restore it by defaulting to V4L2_FIELD_NONE if the field value isn't
supported, with the only two supported value being V4L2_FIELD_NONE and
V4L2_FIELD_INTERLACED.
Fixes: c5c3a48061b5 ("media: imx: Try colorimetry at both sink and source pads") Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Reviewed-by: Rui Miguel Silva <rmfrfs@gmail.com> 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>
It will start failing randomly including IO to unrelated zones because of
->error "reuse". Trigger can be partition detection as well if test is not
run immediately which is even more entertaining.
The fix is of course to clear ->error where necessary.
If null_add_dev() fails then null_del_dev() is called with a NULL argument.
Make null_del_dev() handle this scenario correctly. This patch fixes the
following KASAN complaint:
null-ptr-deref in null_del_dev+0x28/0x280 [null_blk]
Read of size 8 at addr 0000000000000000 by task find/1062
SDEI has private events that must be registered on each CPU. When
CPUs come and go they must re-register and re-enable their private
events. Each event has flags to indicate whether this should happen
to protect against an event being registered on a CPU coming online,
while all the others are unregistering the event.
These flags are protected by the sdei_list_lock spinlock, because
the cpuhp callbacks can't take the mutex.
Hibernate needs to unregister all events, but keep the in-memory
re-register and re-enable as they are. sdei_unregister_shared()
takes the spinlock to walk the list, then calls _sdei_event_unregister()
on each shared event. _sdei_event_unregister() tries to take the
same spinlock to update re-register and re-enable. This doesn't go
so well.
Push the re-register and re-enable updates out to their callers.
sdei_unregister_shared() doesn't want these values updated, so
doesn't need to do anything.
This also fixes shared events getting lost over hibernate as this
path made them look unregistered.
Fixes: ae98c65b7eff ("firmware: arm_sdei: Add support for CPU and system power states") Reported-by: Liguang Zhang <zhangliguang@linux.alibaba.com> Signed-off-by: James Morse <james.morse@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Some older MSM8916 Venus firmware versions also seem to indicate
support for encoding HEVC, even though they really can't.
This will lead to errors later because hfi_session_init() fails
in this case.
HEVC is already ignored for "dec_codecs", so add the same for
"enc_codecs" to make these old firmware versions work correctly.
Possible double unlocking of 'wilc->hif_cs' mutex was identified by
smatch [1]. Removed the extra call to release_bus() in
wilc_wlan_handle_txq() which was missed in earlier commit 1b8df37e6687
("staging: wilc1000: support suspend/resume functionality").
imx6ul_opp_check_speed_grading is called for both i.MX6UL and i.MX6ULL.
Since the i.MX6ULL was introduced to a separate ocotp compatible node
later, it is possible that the i.MX6ULL has also dtbs with
"fsl,imx6ull-ocotp". On a system without nvmem-cell speed grade a
missing check on this node causes a driver fail without considering
the cpu speed grade.
This patch prevents unwanted cpu overclocking on i.MX6ULL with compatible
node "fsl,imx6ull-ocotp" in old dtbs without nvmem-cell speed grade.
Fixes: 282bd77311d7 ("cpufreq: imx6q: read OCOTP through nvmem for imx6ul/imx6ull") Signed-off-by: Christoph Niedermaier <cniedermaier@dh-electronics.com> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When cfg80211_update_assoc_bss_entry() is called, there is a
verification that the BSS channel actually changed. As some APs use
CSA also for bandwidth changes, this would result with a kernel
warning.
There is one one corner case at dma_fence_signal_locked
which will raise the NULL pointer problem just like below.
->dma_fence_signal
->dma_fence_signal_locked
->test_and_set_bit
here trigger dma_fence_release happen due to the zero of fence refcount.
->dma_fence_put
->dma_fence_release
->drm_sched_fence_release_scheduled
->call_rcu
here make the union fled “cb_list” at finished fence
to NULL because struct rcu_head contains two pointer
which is same as struct list_head cb_list
Therefore, to hold the reference of finished fence at drm_sched_process_job
to prevent the null pointer during finished fence dma_fence_signal
v2: hold the finished fence at drm_sched_process_job instead of
amdgpu_fence_process
v3: resume the blank line
Signed-off-by: Yintian Tao <yttao@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Many systems build/test up-to-date kernels with older libcs, and
an older glibc (2.17) lacks the definition of SOL_DCCP in
/usr/include/bits/socket.h (it was added in the 4.6 timeframe).
Adding the definition to the test program avoids a compilation
failure that gets in the way of building tools/testing/selftests/net.
The test itself will work once the definition is added; either
skipping due to DCCP not being configured in the kernel under test
or passing, so there are no other more up-to-date glibc dependencies
here it seems beyond that missing definition.
Fixes: 2704251c070f ("selftests: net: reuseport_addr_any: add DCCP") Signed-off-by: Alan Maguire <alan.maguire@oracle.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
add read barrier in driver code to keep from reading other fileds
in dma memory which is writable for hw until we have verified the
memory is valid for driver
Signed-off-by: Luo bin <luobin9@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
should disable eq irq before freeing it, must clear event queue
depth in hw before freeing relevant memory to avoid illegal
memory access and update consumer idx to avoid invalid interrupt
Signed-off-by: Luo bin <luobin9@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
printk in macro vxge_debug_ll uses __VA_ARGS__ without "##" prefix,
it causes a build error when there is no variable
arguments(e.g. only fmt is specified.).
Signed-off-by: Zheng Wei <wei.zheng@vivo.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Not every stmmac based platform makes use of the eth_wake_irq or eth_lpi
interrupts. Use the platform_get_irq_byname_optional variant for these
interrupts, so no error message is displayed, if they can't be found.
Rather print an information to hint something might be wrong to assist
debugging on platforms which use these interrupts.
Signed-off-by: Markus Fuchs <mklntf@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
Fix the interruptibility of kernel-initiated client calls so that they're
either only interruptible when they're waiting for a call slot to come
available or they're not interruptible at all. Either way, they're not
interruptible during transmission.
This should help prevent StoreData calls from being interrupted when
writeback is in progress. It doesn't, however, handle interruption during
the receive phase.
Userspace-initiated calls are still interruptable. After the signal has
been handled, sendmsg() will return the amount of data copied out of the
buffer and userspace can perform another sendmsg() call to continue
transmission.
Fixes: 8d0c6fe295af ("rxrpc: Use MSG_WAITALL to tell sendmsg() to temporarily ignore signals") Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The dpio irqs must be registered when you can actually
receive interrupts, ie when the dpios are created.
Kernel goes through NULL pointer dereference errors
followed by kernel panic [1] because the dpio irqs are
enabled before the dpio is created.
[1]
Unable to handle kernel NULL pointer dereference at virtual address 0040
fsl_mc_dpio dpio.14: probed
fsl_mc_dpio dpio.13: Adding to iommu group 11
ISV = 0, ISS = 0x00000004
Unable to handle kernel NULL pointer dereference at virtual address 0040
Mem abort info:
ESR = 0x96000004
EC = 0x25: DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
Data abort info:
ISV = 0, ISS = 0x00000004
CM = 0, WnR = 0
[0000000000000040] user address but active_mm is swapper
Internal error: Oops: 96000004 [#1] PREEMPT SMP
Modules linked in:
CPU: 2 PID: 151 Comm: kworker/2:1 Not tainted 5.6.0-rc4-next-20200304 #1
Hardware name: NXP Layerscape LX2160ARDB (DT)
Workqueue: events deferred_probe_work_func
pstate: 00000085 (nzcv daIf -PAN -UAO)
pc : dpaa2_io_irq+0x18/0xe0
lr : dpio_irq_handler+0x1c/0x28
sp : ffff800010013e20
x29: ffff800010013e20 x28: ffff0026d9b4c140
x27: ffffa1d38a142018 x26: ffff0026d2953400
x25: ffffa1d38a142018 x24: ffffa1d38a7ba1d8
x23: ffff800010013f24 x22: 0000000000000000
x21: 0000000000000072 x20: ffff0026d2953400
x19: ffff0026d2a68b80 x18: 0000000000000001
x17: 000000002fb37f3d x16: 0000000035eafadd
x15: ffff0026d9b4c5b8 x14: ffffffffffffffff
x13: ff00000000000000 x12: 0000000000000038
x11: 0101010101010101 x10: 0000000000000040
x9 : ffffa1d388db11e4 x8 : ffffa1d38a7e40f0
x7 : ffff0026da414f38 x6 : 0000000000000000
x5 : ffff0026da414d80 x4 : ffff5e5353d0c000
x3 : ffff800010013f60 x2 : ffffa1d388db11c8
x1 : ffff0026d2a67c00 x0 : 0000000000000000
Call trace:
dpaa2_io_irq+0x18/0xe0
dpio_irq_handler+0x1c/0x28
__handle_irq_event_percpu+0x78/0x2c0
handle_irq_event_percpu+0x38/0x90
handle_irq_event+0x4c/0xd0
handle_fasteoi_irq+0xbc/0x168
generic_handle_irq+0x2c/0x40
__handle_domain_irq+0x68/0xc0
gic_handle_irq+0x64/0x150
el1_irq+0xb8/0x180
_raw_spin_unlock_irqrestore+0x14/0x48
irq_set_affinity_hint+0x6c/0xa0
dpaa2_dpio_probe+0x2a4/0x518
fsl_mc_driver_probe+0x28/0x70
really_probe+0xdc/0x320
driver_probe_device+0x5c/0xf0
__device_attach_driver+0x88/0xc0
bus_for_each_drv+0x7c/0xc8
__device_attach+0xe4/0x140
device_initial_probe+0x18/0x20
bus_probe_device+0x98/0xa0
device_add+0x41c/0x758
fsl_mc_device_add+0x184/0x530
dprc_scan_objects+0x280/0x370
dprc_probe+0x124/0x3b0
fsl_mc_driver_probe+0x28/0x70
really_probe+0xdc/0x320
driver_probe_device+0x5c/0xf0
__device_attach_driver+0x88/0xc0
bus_for_each_drv+0x7c/0xc8
__device_attach+0xe4/0x140
device_initial_probe+0x18/0x20
bus_probe_device+0x98/0xa0
deferred_probe_work_func+0x74/0xa8
process_one_work+0x1c8/0x470
worker_thread+0x1f8/0x428
kthread+0x124/0x128
ret_from_fork+0x10/0x18
Code: a9bc7bfd910003fda9025bf5a90363f7 (f9402015)
---[ end trace 38298e1a29e7a570 ]---
Kernel panic - not syncing: Fatal exception in interrupt
SMP: stopping secondary CPUs
Mem abort info:
ESR = 0x96000004
CM = 0, WnR = 0
EC = 0x25: DABT (current EL), IL = 32 bits
[0000000000000040] user address but active_mm is swapper
SET = 0, FnV = 0
EA = 0, S1PTW = 0
Data abort info:
ISV = 0, ISS = 0x00000004
CM = 0, WnR = 0
[0000000000000040] user address but active_mm is swapper
SMP: failed to stop secondary CPUs 0-2
Kernel Offset: 0x21d378600000 from 0xffff800010000000
PHYS_OFFSET: 0xffffe92180000000
CPU features: 0x10002,21806008
Memory Limit: none
---[ end Kernel panic - not syncing: Fatal exception in interrupt ]---
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com> Signed-off-by: Grigore Popescu <grigore.popescu@nxp.com> Signed-off-by: Li Yang <leoyang.li@nxp.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
The TLC configuration did not take into consideration the station's
SMPS configuration, and thus configured rates for 2 NSS even if
static SMPS was reported by the station. Fix this.
The above call stack is actually very similar to an issue
reported by Commit 1310d82b01c0 ("bpf/stackmap: Fix deadlock with
rq_lock in bpf_get_stack()") by Song Liu. The only difference is
bpf_send_signal() helper instead of bpf_get_stack() helper.
The above deadlock is triggered with a perf_sw_event.
Similar to Commit 1310d82b01c0, the below almost identical reproducer
used tracepoint point sched/sched_switch so the issue can be easily caught.
/* stress_test.c */
#include <stdio.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <pthread.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
int main(int argc, char *argv[])
{
void *ptr;
int i;
pthread_t threads[THREAD_COUNT];
if (argc < 2)
return 0;
filename = argv[1];
for (i = 0; i < THREAD_COUNT; i++) {
if (pthread_create(threads + i, NULL, worker, NULL)) {
fprintf(stderr, "Error creating thread\n");
return 0;
}
}
for (i = 0; i < THREAD_COUNT; i++)
pthread_join(threads[i], NULL);
return 0;
}
and the following command:
1. run `stress_test /bin/ls` in one windown
2. hack bcc trace.py with the following change:
# --- a/tools/trace.py
# +++ b/tools/trace.py
@@ -513,6 +513,7 @@ BPF_PERF_OUTPUT(%s);
__data.tgid = __tgid;
__data.pid = __pid;
bpf_get_current_comm(&__data.comm, sizeof(__data.comm));
+ bpf_send_signal(10);
%s
%s
%s.perf_submit(%s, &__data, sizeof(__data));
3. in a different window run
./trace.py -p $(pidof stress_test) t:sched:sched_switch
The deadlock can be reproduced in our production system.
Similar to Song's fix, the fix is to delay sending signal if
irqs is disabled to avoid deadlocks involving with rq_lock.
With this change, my above stress-test in our production system
won't cause deadlock any more.
I also implemented a scale-down version of reproducer in the
selftest (a subsequent commit). With latest bpf-next,
it complains for the following potential deadlock.
[ 32.832450] -> #1 (&p->pi_lock){-.-.}:
[ 32.833100] _raw_spin_lock_irqsave+0x44/0x80
[ 32.833696] task_rq_lock+0x2c/0xa0
[ 32.834182] task_sched_runtime+0x59/0xd0
[ 32.834721] thread_group_cputime+0x250/0x270
[ 32.835304] thread_group_cputime_adjusted+0x2e/0x70
[ 32.835959] do_task_stat+0x8a7/0xb80
[ 32.836461] proc_single_show+0x51/0xb0
...
[ 32.839512] -> #0 (&(&sighand->siglock)->rlock){....}:
[ 32.840275] __lock_acquire+0x1358/0x1a20
[ 32.840826] lock_acquire+0xc7/0x1d0
[ 32.841309] _raw_spin_lock_irqsave+0x44/0x80
[ 32.841916] __lock_task_sighand+0x79/0x160
[ 32.842465] do_send_sig_info+0x35/0x90
[ 32.842977] bpf_send_signal+0xa/0x10
[ 32.843464] bpf_prog_bc13ed9e4d3163e3_send_signal_tp_sched+0x465/0x1000
[ 32.844301] trace_call_bpf+0x115/0x270
[ 32.844809] perf_trace_run_bpf_submit+0x4a/0xc0
[ 32.845411] perf_trace_sched_switch+0x10f/0x180
[ 32.846014] __schedule+0x45d/0x880
[ 32.846483] schedule+0x5f/0xd0
...
Deadlock happens on CPU0 when it tries to acquire &sighand->siglock
but it has been held by CPU1 and CPU1 tries to grab &rq->lock
and cannot get it.
This is not exactly the callstack in our production environment,
but sympotom is similar and both locks are using spin_lock_irqsave()
to acquire the lock, and both involves rq_lock. The fix to delay
sending signal when irq is disabled also fixed this issue.
Signed-off-by: Yonghong Song <yhs@fb.com> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Cc: Song Liu <songliubraving@fb.com> Link: https://lore.kernel.org/bpf/20200304191104.2796501-1-yhs@fb.com Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When doing a 16-bit read that returns data in the MSB byte, the
RSB_DATA register will keep the MSB byte unchanged when doing
the following 8-bit read. sunxi_rsb_read() will then return
a result that contains high byte from 16-bit read mixed with
the 8-bit result.
The consequence is that after this happens the PMIC's regmap will
look like this: (0x33 is the high byte from the 16-bit read)
Fix this by masking the result of the read with the correct mask
based on the size of the read. There are no 16-bit users in the
mainline kernel, so this doesn't need to get into the stable tree.
Signed-off-by: Ondrej Jirman <megous@megous.com> Acked-by: Chen-Yu Tsai <wens@csie.org> Signed-off-by: Maxime Ripard <maxime@cerno.tech> Signed-off-by: Sasha Levin <sashal@kernel.org>
Since commit 093e746ced027 ("intel-iommu: Prevent devices with
RMRRs from being placed into SI Domain"), the Intel IOMMU driver
doesn't allow any devices with RMRR locked to use the identity
domain. This was added to to fix the issue where the RMRR info
for devices being placed in and out of the identity domain gets
lost. This identity maps all RMRRs when setting up the identity
domain, so that devices with RMRRs could also use it.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com> Signed-off-by: Joerg Roedel <jroedel@suse.de> Cc: Alan Stern <stern@rowland.harvard.edu> Cc: John Donnelly <john.p.donnelly@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Sultan Alsawaf [Tue, 7 Apr 2020 20:32:22 +0000 (13:32 -0700)]
drm/i915: Fix ref->mutex deadlock in i915_active_wait()
The following deadlock exists in i915_active_wait() due to a double lock
on ref->mutex (call chain listed in order from top to bottom):
i915_active_wait();
mutex_lock_interruptible(&ref->mutex); <-- ref->mutex first acquired
i915_active_request_retire();
node_retire();
active_retire();
mutex_lock_nested(&ref->mutex, SINGLE_DEPTH_NESTING); <-- DEADLOCK
Fix the deadlock by skipping the second ref->mutex lock when
active_retire() is called through i915_active_request_retire().
Note that this bug only affects 5.4 and has since been fixed in 5.5.
Normally, a backport of the fix from 5.5 would be in order, but the
patch set that fixes this deadlock involves massive changes that are
neither feasible nor desirable for backporting [1][2][3]. Therefore,
this small patch was made to address the deadlock specifically for 5.4.
[1] fd6853792ac5 ("drm/i915: Push the i915_active.retire into a worker")
[2] 21d560702947 ("drm/i915: Split i915_active.mutex into an irq-safe spinlock for the rbtree")
[3] 5146cee74d8d ("drm/i915: Serialise with remote retirement")
Fixes: 221628386a33 ("drm/i915: Provide an i915_active.acquire callback") Cc: <stable@vger.kernel.org> # 5.4.x Signed-off-by: Sultan Alsawaf <sultan@kerneltoast.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
blk_mq_map_queues() and multiple .map_queues() implementations expect that
set->map[HCTX_TYPE_DEFAULT].nr_queues is set to the number of hardware
queues. Hence set .nr_queues before calling these functions. This patch
fixes the following kernel warning:
Holding the rtnl_lock while iterating a devices interface address list
potentially causes deadlocks with the cma_netdev_callback. While this was
implemented to limit the scope of a wildcard listen to addresses of the
current device only, a better solution limits the scope of the socket to
the device. This completely avoiding locking, and also results in
significant code simplification.
Fixes: 7739a189c7c3 ("RDMA/siw: Add missing rtnl_lock around access to ifa") Link: https://lore.kernel.org/r/20200228173534.26815-1-bmt@zurich.ibm.com Reported-by: syzbot+55de90ab5f44172b0c90@syzkaller.appspotmail.com Suggested-by: Jason Gunthorpe <jgg@ziepe.ca> Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This lock ordering only happens when bonding is enabled and a certain
bonding related event fires. However, since it can happen this is a global
restriction on lock ordering.
Teach lockdep about the order directly and unconditionally so bugs here
are found quickly.
This appears to be a bug in the design, as it does have lots of locking
that seems like it should allow concurrency. However, when it is all said
and done every single place that uses the cma_exch() scheme is broken, and
all the unlocked reads from the ucma of the cm_id data are wrong too.
syzkaller has been finding endless bugs related to this.
Fixing this in any elegant way is some enormous amount of work. Take a
very big hammer and put a mutex around everything to do with the
ucma_context at the top of every syscall.
QEMU has a funny new build error message when I use the upstream kernel
headers:
CC block/file-posix.o
In file included from /home/cborntra/REPOS/qemu/include/qemu/timer.h:4,
from /home/cborntra/REPOS/qemu/include/qemu/timed-average.h:29,
from /home/cborntra/REPOS/qemu/include/block/accounting.h:28,
from /home/cborntra/REPOS/qemu/include/block/block_int.h:27,
from /home/cborntra/REPOS/qemu/block/file-posix.c:30:
/usr/include/linux/swab.h: In function `__swab':
/home/cborntra/REPOS/qemu/include/qemu/bitops.h:20:34: error: "sizeof" is not defined, evaluates to 0 [-Werror=undef]
20 | #define BITS_PER_LONG (sizeof (unsigned long) * BITS_PER_BYTE)
| ^~~~~~
/home/cborntra/REPOS/qemu/include/qemu/bitops.h:20:41: error: missing binary operator before token "("
20 | #define BITS_PER_LONG (sizeof (unsigned long) * BITS_PER_BYTE)
| ^
cc1: all warnings being treated as errors
make: *** [/home/cborntra/REPOS/qemu/rules.mak:69: block/file-posix.o] Error 1
rm tests/qemu-iotests/socket_scm_helper.o
This was triggered by commit 30f08e5ce30 ("uapi: rename ext2_swab() to
swab() and share globally in swab.h"). That patch is doing
#include <asm/bitsperlong.h>
but it uses BITS_PER_LONG.
The kernel file asm/bitsperlong.h provide only __BITS_PER_LONG.
Let us use the __ variant in swap.h
Link: http://lkml.kernel.org/r/20200213142147.17604-1-borntraeger@de.ibm.com Fixes: 30f08e5ce30 ("uapi: rename ext2_swab() to swab() and share globally in swab.h") Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Yury Norov <yury.norov@gmail.com> Cc: Allison Randal <allison@lohutok.net> Cc: Joe Perches <joe@perches.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: William Breathitt Gray <vilhelm.gray@gmail.com> Cc: Torsten Hilbrich <torsten.hilbrich@secunet.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
syzbot reported that 458797c4c569 ("ceph: remove the extra slashes in
the server path") had caused a regression where an allocation could be
done under a spinlock -- compare_mount_options() is called by sget_fc()
with sb_lock held.
We don't really need the supplied server path, so canonicalize it
in place and compare it directly. To make this work, the leading
slash is kept around and the logic in ceph_real_mount() to skip it
is restored. CEPH_MSG_CLIENT_SESSION now reports the same (i.e.
canonicalized) path, with the leading slash of course.
Fixes: 458797c4c569 ("ceph: remove the extra slashes in the server path") Reported-by: syzbot+98704a51af8e3d9425a9@syzkaller.appspotmail.com Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Reviewed-by: Jeff Layton <jlayton@kernel.org> Signed-off-by: Luis Henriques <lhenriques@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
And then when creating any new file or directory under the mount
point, we can hit the following BUG_ON in ceph_fill_trace():
BUG_ON(ceph_snap(dir) != dvino.snap);
Have the client ignore the extra slashes in the server path when
mounting. This will also canonicalize the path, so that identical mounts
can be consilidated.
Regardless of the internal treatment of these paths, the kernel still
stores the original string including the leading '/' for presentation
to userland.
URL: https://tracker.ceph.com/issues/42771 Signed-off-by: Xiubo Li <xiubli@redhat.com> Reviewed-by: Jeff Layton <jlayton@kernel.org> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Luis Henriques <lhenriques@suse.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The v7 ARM states that all cache and branch predictor maintenance
operations that do not specify an address execute, relative to
each other, in program order.
However, because of this erratum, an L2 set/way cache maintenance
operation can overtake an L1 set/way cache maintenance operation.
This ERRATA only affected the Cortex-A7 and present in r0p2, r0p3,
r0p4, r0p5.
i.MX6UL and i.MX7D have Cortex-A7 r0p5 inside, need to enable
ARM_ERRATA_814220 for proper workaround.
Until now the flex parser capability was used in ib_query_device() to
indicate tunnel_offloads_caps support for mpls_over_gre/mpls_over_udp.
Newer devices and firmware will have configurations with the flexparser
but without mpls support.
Testing for the flex parser capability was a mistake, the tunnel_stateless
capability was intended for detecting mpls and was introduced at the same
time as the flex parser capability.
Otherwise userspace will be incorrectly informed that a future device
supports MPLS when it does not.
Link: https://lore.kernel.org/r/20200305123841.196086-1-leon@kernel.org Cc: <stable@vger.kernel.org> # 4.17 Fixes: 125a114c8495 ("IB/mlx5: Expose MPLS related tunneling offloads") Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed-by: Ariel Levkovich <lariel@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
- Releasing dd->per_sdma[i].kobject in hfi1_unregister_sysfs().
- This will fix the memory leak.
- Calling kobject_put() to unwind operations only for those entries in
dd->per_sdma[] whose operations have succeeded (including the current
one that has just failed) in hfi1_verbs_register_sysfs().
Cc: <stable@vger.kernel.org> Fixes: 5676484ef423 ("IB/hfi1: Add sysfs interface for affinity setup") Link: https://lore.kernel.org/r/20200326163807.21129.27371.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When kobject_init_and_add() returns an error in the function
hfi1_create_port_files(), the function kobject_put() is not called for the
corresponding kobject, which potentially leads to memory leak.
This patch fixes the issue by calling kobject_put() even if
kobject_init_and_add() fails.
Cc: <stable@vger.kernel.org> Link: https://lore.kernel.org/r/20200326163813.21129.44280.stgit@awfm-01.aw.intel.com Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The Power Management Events (PMEs) the INT0002 driver listens for get
signalled by the Power Management Controller (PMC) using the same IRQ
as used for the ACPI SCI.
Since commit b9a2789137f0 ("ACPI: PM: s2idle: Prevent spurious SCIs from
waking up the system") the SCI triggering, without there being a wakeup
cause recognized by the ACPI sleep code, will no longer wakeup the system.
This breaks PMEs / wakeups signalled to the INT0002 driver, the system
never leaves the s2idle_loop() now.
Use acpi_register_wakeup_handler() to register a function which checks
the GPE0a_STS register for a PME and trigger a wakeup when a PME has
been signalled.
Fixes: b9a2789137f0 ("ACPI: PM: s2idle: Prevent spurious SCIs from waking up the system") Cc: 5.4+ <stable@vger.kernel.org> # 5.4+ Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Andy Shevchenko <andy.shevchenko@gmail.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit b9a2789137f0 ("ACPI: PM: s2idle: Prevent spurious SCIs from
waking up the system") the SCI triggering without there being a wakeup
cause recognized by the ACPI sleep code will no longer wakeup the system.
This works as intended, but this is a problem for devices where the SCI
is shared with another device which is also a wakeup source.
In the past these, from the pov of the ACPI sleep code, spurious SCIs
would still cause a wakeup so the wakeup from the device sharing the
interrupt would actually wakeup the system. This now no longer works.
This is a problem on e.g. Bay Trail-T and Cherry Trail devices where
some peripherals (typically the XHCI controller) can signal a
Power Management Event (PME) to the Power Management Controller (PMC)
to wakeup the system, this uses the same interrupt as the SCI.
These wakeups are handled through a special INT0002 ACPI device which
checks for events in the GPE0a_STS for this and takes care of acking
the PME so that the shared interrupt stops triggering.
The change to the ACPI sleep code to ignore the spurious SCI, causes
the system to no longer wakeup on these PME events. To make things
worse this means that the INT0002 device driver interrupt handler will
no longer run, causing the PME to not get cleared and resulting in the
system hanging. Trying to wakeup the system after such a PME through e.g.
the power button no longer works.
Add an acpi_register_wakeup_handler() function which registers
a handler to be called from acpi_s2idle_wake() and when the handler
returns true, return true from acpi_s2idle_wake().
The INT0002 driver will use this mechanism to check the GPE0a_STS
register from acpi_s2idle_wake() and to tell the system to wakeup
if a PME is signaled in the register.
Fixes: b9a2789137f0 ("ACPI: PM: s2idle: Prevent spurious SCIs from waking up the system") Cc: 5.4+ <stable@vger.kernel.org> # 5.4+ Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Make sure that the rngc interrupt is masked if the rngc self test fails.
Self test failure means that probe fails as well. Interrupts should be
masked in this case, regardless of the error.
Cc: stable@vger.kernel.org Fixes: 21958a749752 ("hwrng: mx-rngc - add a driver for Freescale RNGC") Reviewed-by: PrasannaKumar Muralidharan <prasannatsmkumar@gmail.com> Signed-off-by: Martin Kaiser <martin@kaiser.cx> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A recent change to the netlink code: b366e7295052 ("netlink: Relax attr
validation for fixed length types") logs a warning when programs send
messages with invalid attributes (e.g., wrong length for a u32). Yafang
reported this error message for tools/accounting/getdelays.c.
send_cmd() is wrongly adding 1 to the attribute length. As noted in
include/uapi/linux/netlink.h nla_len should be NLA_HDRLEN + payload
length, so drop the +1.
Under CONFIG_SLAB_FREELIST_HARDENED=y, the obfuscation was relatively weak
in that the ptr and ptr address were usually so close that the first XOR
would result in an almost entirely 0-byte value[1], leaving most of the
"secret" number ultimately being stored after the third XOR. A single
blind memory content exposure of the freelist was generally sufficient to
learn the secret.
Add a swab() call to mix bits a little more. This is a cheap way (1
cycle) to make attacks need more than a single exposure to learn the
secret (or to know _where_ the exposure is in memory).
When skipping TRBs, we need to account for wrapping around the ring
buffer and not modifying some invalid TRBs. Without this fix, dwc3 won't
be able to check for available TRBs.
It turns out that RDRAND is pretty slow. Comparing these two
constructions:
for (i = 0; i < CHACHA_BLOCK_SIZE; i += sizeof(ret))
arch_get_random_long(&ret);
and
long buf[CHACHA_BLOCK_SIZE / sizeof(long)];
extract_crng((u8 *)buf);
it amortizes out to 352 cycles per long for the top one and 107 cycles
per long for the bottom one, on Coffee Lake Refresh, Intel Core i9-9880H.
And importantly, the top one has the drawback of not benefiting from the
real rng, whereas the bottom one has all the nice benefits of using our
own chacha rng. As get_random_u{32,64} gets used in more places (perhaps
beyond what it was originally intended for when it was introduced as
get_random_{int,long} back in the md5 monstrosity era), it seems like it
might be a good thing to strengthen its posture a tiny bit. Doing this
should only be stronger and not any weaker because that pool is already
initialized with a bunch of rdrand data (when available). This way, we
get the benefits of the hardware rng as well as our own rng.
Another benefit of this is that we no longer hit pitfalls of the recent
stream of AMD bugs in RDRAND. One often used code pattern for various
things is:
do {
val = get_random_u32();
} while (hash_table_contains_key(val));
That recent AMD bug rendered that pattern useless, whereas we're really
very certain that chacha20 output will give pretty distributed numbers,
no matter what.
So, this simplification seems better both from a security perspective
and from a performance perspective.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Link: https://lore.kernel.org/r/20200221201037.30231-1-Jason@zx2c4.com Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When userspace executes a syscall or gets interrupted,
BEAR contains a kernel address when returning to userspace.
This make it pretty easy to figure out where the kernel is
mapped even with KASLR enabled. To fix this, add lpswe to
lowcore and always execute it there, so userspace sees only
the lowcore address of lpswe. For this we have to extend
both critical_cleanup and the SWITCH_ASYNC macro to also check
for lpswe addresses in lowcore.
Fixes: 2296e9899f6f ("s390/kernel: add support for kernel address space layout randomization (KASLR)") Cc: <stable@vger.kernel.org> # v5.2+ Reviewed-by: Gerald Schaefer <gerald.schaefer@de.ibm.com> Signed-off-by: Sven Schnelle <svens@linux.ibm.com> Signed-off-by: Vasily Gorbik <gor@linux.ibm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There has been a number of reports that using SG/TSO on different chip
versions results in tx timeouts. However for a lot of people SG/TSO
works fine. Therefore disable both features by default, but allow users
to enable them. Use at own risk!
The handler for FLOW_ACTION_VLAN_MANGLE ends by returning whatever the
lower-level function that it calls returns. If there are more actions lined
up after this action, those are never offloaded. Fix by only bailing out
when the called function returns an error.
Fixes: da819ac7406f ("mlxsw: spectrum: Add support for vlan modify TC action") Signed-off-by: Petr Machata <petrm@mellanox.com> Reviewed-by: Jiri Pirko <jiri@mellanox.com> Signed-off-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When an XDP program is installed, tun_build_skb() grabs a reference to
the current page fragment page if the program returns XDP_REDIRECT or
XDP_TX. However, since tun_xdp_act() passes through negative return
values from the XDP program, it is possible to trigger the error path by
mistake and accidentally drop a reference to the fragments page without
taking one, leading to a spurious free. This is believed to be the cause
of some KASAN use-after-free reports from syzbot [1], although without a
reproducer it is not possible to confirm whether this patch fixes the
problem.
Ensure that we only drop a reference to the fragments page if the XDP
transmit or redirect operations actually fail.
Cc: "David S. Miller" <davem@davemloft.net> Cc: Alexei Starovoitov <ast@kernel.org> Cc: Daniel Borkmann <daniel@iogearbox.net> CC: Eric Dumazet <edumazet@google.com> Acked-by: Jason Wang <jasowang@redhat.com> Fixes: 587e717e0586 ("tuntap: split out XDP logic") Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
struct can_frame contains some padding which is not explicitly zeroed in
slc_bump. This uninitialized data will then be transmitted if the stack
initialization hardening feature is not enabled (CONFIG_INIT_STACK_ALL).
This commit just zeroes the whole struct including the padding.
Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com> Fixes: 8df5b711a451 ("can: add slcan driver for serial/USB-serial CAN adapters") Reviewed-by: Kees Cook <keescook@chromium.org> Cc: linux-can@vger.kernel.org Cc: netdev@vger.kernel.org Cc: security@kernel.org Cc: wg@grandegger.com Cc: mkl@pengutronix.de Cc: davem@davemloft.net Acked-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit d27d3176050d ("net: stmmac: dwmac1000: Clear unused address
entries") cleared the unused mac address entries, but introduced an
out-of bounds mac address register programming bug -- After setting
the secondary unicast mac addresses, the "reg" value has reached
netdev_uc_count() + 1, thus we should only clear address entries
if (addr < perfect_addr_number)
Fixes: d27d3176050d ("net: stmmac: dwmac1000: Clear unused address entries") Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The initial refcnt of struct tcindex_data should be 1,
it is clear that I forgot to set it to 1 in tcindex_init().
This leads to a dec-after-zero warning.
Reported-by: syzbot+8325e509a1bf83ec741d@syzkaller.appspotmail.com Fixes: cee5dd3e5b3b ("net_sched: add a temporary refcnt for struct tcindex_data") Cc: Jamal Hadi Salim <jhs@mojatatu.com> Cc: Jiri Pirko <jiri@resnulli.us> Cc: Paul E. McKenney <paulmck@kernel.org> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Although we intentionally use an ordered workqueue for all tc
filter works, the ordering is not guaranteed by RCU work,
given that tcf_queue_work() is esstenially a call_rcu().
This problem is demostrated by Thomas:
CPU 0:
tcf_queue_work()
tcf_queue_work(&r->rwork, tcindex_destroy_rexts_work);
-> Migration to CPU 1
CPU 1:
tcf_queue_work(&p->rwork, tcindex_destroy_work);
so the 2nd work could be queued before the 1st one, which leads
to a free-after-free.
Enforcing this order in RCU work is hard as it requires to change
RCU code too. Fortunately we can workaround this problem in tcindex
filter by taking a temporary refcnt, we only refcnt it right before
we begin to destroy it. This simplifies the code a lot as a full
refcnt requires much more changes in tcindex_set_parms().
Reported-by: syzbot+46f513c3033d592409d2@syzkaller.appspotmail.com Fixes: 25e38fda1eda ("net_sched: fix a race condition in tcindex_destroy()") Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Paul E. McKenney <paulmck@kernel.org> Cc: Jamal Hadi Salim <jhs@mojatatu.com> Cc: Jiri Pirko <jiri@resnulli.us> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com> Reviewed-by: Paul E. McKenney <paulmck@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After the power-down bit is cleared, the chip internally triggers a
global reset. According to the KSZ9031 documentation, we have to wait at
least 1ms for the reset to finish.
If the chip is accessed during reset, read will return 0xffff, while
write will be ignored. Depending on the system performance and MDIO bus
speed, we may or may not run in to this issue.
This bug was discovered on an iMX6QP system with KSZ9031 PHY and
attached PHY interrupt line. If IRQ was used, the link status update was
lost. In polling mode, the link status update was always correct.
The investigation showed, that during a read-modify-write access, the
read returned 0xffff (while the chip was still in reset) and
corresponding write hit the chip _after_ reset and triggered (due to the
0xffff) another reset in an undocumented bit (register 0x1f, bit 1),
resulting in the next write being lost due to the new reset cycle.
This patch fixes the issue by adding a 1...2 ms sleep after the
genphy_resume().
Fixes: b83246da8b7e ("net: phy: micrel: Add specific suspend") Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The 2nd gmac of mediatek soc ethernet may not be connected to a PHY
and a phy-handle isn't always available.
Unfortunately, mt7530 dsa driver assumes that the 2nd gmac is always
connected to switch port 5 and setup mt7530 according to phy address
of 2nd gmac node, causing null pointer dereferencing when phy-handle
isn't defined in dts.
This commit fix this setup code by checking return value of
of_parse_phandle before using it.
Fixes: 347b5aa9eb37 ("net: dsa: mt7530: Add support for port 5") Signed-off-by: Chuanhong Guo <gch981213@gmail.com> Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com> Reviewed-by: Florian Fainelli <f.fainelli@gmail.com> Tested-by: René van Dorst <opensource@vdorst.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the bcm_sf2 was converted into a proper platform device driver and
used the new dsa_register_switch() interface, we would still be parsing
the legacy DSA node that contained all the port information since the
platform firmware has intentionally maintained backward and forward
compatibility to client programs. Ensure that we do parse the correct
node, which is "ports" per the revised DSA binding.
Fixes: 778738bbbf05 ("net: dsa: bcm_sf2: Make it a real platform device driver") Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We were registering our slave MDIO bus with OF and doing so with
assigning the newly created slave_mii_bus of_node to the master MDIO bus
controller node. This is a bad thing to do for a number of reasons:
- we are completely lying about the slave MII bus is arranged and yet we
still want to control which MDIO devices it probes. It was attempted
before to play tricks with the bus_mask to perform that:
https://www.spinics.net/lists/netdev/msg429420.html but the approach
was rightfully rejected
- the device_node reference counting is messed up and we are effectively
doing a double probe on the devices we already probed using the
master, this messes up all resources reference counts (such as clocks)
The proper fix for this as indicated by David in his reply to the
thread above is to use a platform data style registration so as to
control exactly which devices we probe:
https://www.spinics.net/lists/netdev/msg430083.html
By using mdiobus_register(), our slave_mii_bus->phy_mask value is used
as intended, and all the PHY addresses that must be redirected towards
our slave MDIO bus is happening while other addresses get redirected
towards the master MDIO bus.
Fixes: 1d2c524a6dde ("net: dsa: bcm_sf2: Register our slave MDIO bus") Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Bonding slave and team port devices should not have link-local addresses
automatically added to them, as it can interfere with openvswitch being
able to properly add tc ingress.
Basic reproducer, courtesy of Marcelo:
$ ip link add name bond0 type bond
$ ip link set dev ens2f0np0 master bond0
$ ip link set dev ens2f1np2 master bond0
$ ip link set dev bond0 up
$ ip a s
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN
group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: ens2f0np0: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc
mq master bond0 state UP group default qlen 1000
link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
5: ens2f1np2: <NO-CARRIER,BROADCAST,MULTICAST,SLAVE,UP> mtu 1500 qdisc
mq master bond0 state DOWN group default qlen 1000
link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
11: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UP group default qlen 1000
link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
inet6 fe80::20f:53ff:fe2f:ea40/64 scope link
valid_lft forever preferred_lft forever
$ ip a l ens2f0np0
2: ens2f0np0: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc
mq master bond0 state UP group default qlen 1000
link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
inet6 fe80::20f:53ff:fe2f:ea40/64 scope link tentative
valid_lft forever preferred_lft forever
$ ip a l ens2f1np2
5: ens2f1np2: <NO-CARRIER,BROADCAST,MULTICAST,SLAVE,UP> mtu 1500 qdisc
mq master bond0 state DOWN group default qlen 1000
link/ether 00:0f:53:2f:ea:40 brd ff:ff:ff:ff:ff:ff
inet6 fe80::20f:53ff:fe2f:ea40/64 scope link tentative
valid_lft forever preferred_lft forever
Looks like addrconf_sysctl_addr_gen_mode() bypasses the original "is
this a slave interface?" check added by commit bc43c5cd188c, and
results in an address getting added, while w/the proposed patch added,
no address gets added. This simply adds the same gating check to another
code path, and thus should prevent the same devices from erroneously
obtaining an ipv6 link-local address.
Fixes: 87112f6f21f6 ("net/ipv6: allow sysctl to change link-local address generation mode") Reported-by: Moshe Levi <moshele@mellanox.com> CC: Stephen Hemminger <stephen@networkplumber.org> CC: Marcelo Ricardo Leitner <mleitner@redhat.com> CC: netdev@vger.kernel.org Signed-off-by: Jarod Wilson <jarod@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
cxgb4_update_mac_filt() earlier requests firmware to add a new MAC
address into MPS TCAM. The MPS TCAM index returned by firmware is
stored in pi->xact_addr_filt. However, the saved MPS TCAM index gets
overwritten again with the return value of cxgb4_update_mac_filt(),
which is wrong.
When trying to update to another MAC address later, the wrong MPS TCAM
index is sent to firmware, which causes firmware to return error,
because it's not the same MPS TCAM index that firmware had sent
earlier to driver.
So, fix by removing the wrong overwrite being done after call to
cxgb4_update_mac_filt().
Fixes: 77df08e02130 ("cxgb4/cxgb4vf: Program hash region for {t4/t4vf}_change_mac()") Signed-off-by: Herat Ramani <herat@chelsio.com> Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net: phy: realtek: fix handling of RTL8105e-integrated PHY
[ No applicable upstream commit ]
After the referenced fix it turned out that one particular RTL8168
chip version (RTL8105e) does not work on 5.4 because no dedicated PHY
driver exists. Adding this PHY driver was done for fixing a different
issue for versions from 5.5 already. I re-send the same change for 5.4
because the commit message differs.
======================================================
WARNING: possible circular locking dependency detected
5.4.0-rc8-padata-cpuhp-v3+ #1 Not tainted
------------------------------------------------------
bash/205 is trying to acquire lock: ffffffff8286bcd0 (cpu_hotplug_lock.rw_sem){++++}, at: padata_set_cpumask+0x2b/0x120
but task is already holding lock: ffff8880001abfa0 (&pinst->lock){+.+.}, at: padata_set_cpumask+0x26/0x120
which lock already depends on the new lock.
padata doesn't take cpu_hotplug_lock and pinst->lock in a consistent
order. Which should be first? CPU hotplug calls into padata with
cpu_hotplug_lock already held, so it should have priority.
Fixes: 7d53f15c3057 ("padata: Use get_online_cpus/put_online_cpus") Signed-off-by: Daniel Jordan <daniel.m.jordan@oracle.com> Cc: Eric Biggers <ebiggers@kernel.org> Cc: Herbert Xu <herbert@gondor.apana.org.au> Cc: Steffen Klassert <steffen.klassert@secunet.com> Cc: linux-crypto@vger.kernel.org Cc: linux-kernel@vger.kernel.org Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fixes the lower and upper bounds when there are multiple TCs and
traffic is on the the same TC on the same device.
The lower bound is represented by 'qoffset' and the upper limit for
hash value is 'qcount + qoffset'. This gives a clean Rx to Tx queue
mapping when there are multiple TCs, as the queue indices for upper TCs
will be offset by 'qoffset'.
v2: Fixed commit description based on comments.
Fixes: 79cca800c7ef ("net: Revoke export for __skb_tx_hash, update it to just be static skb_tx_hash") Fixes: eef64f64be1b ("net: Add support for subordinate traffic classes to netdev_pick_tx") Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com> Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com> Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Martin noticed that nct6775 driver does not load properly on his system
in v5.4+ kernels. The issue was bisected to commit 08f7417baac7 ("i2c:
i801: Use iTCO version 6 in Cannon Lake PCH and beyond") but it is
likely not the culprit because the faulty code has been in the driver
already since commit 31fb4a912803 ("i2c: i801: Create iTCO device on
newer Intel PCHs"). So more likely some commit that added PCI IDs of
recent chipsets made the driver to create the iTCO_wdt device on Martins
system.
The issue was debugged to be PCI configuration access to the PMC device
that is not present. This returns all 1's when read and this caused the
iTCO_wdt driver to accidentally request resourses used by nct6775.
It turns out that the SMI resource is only required for some ancient
systems, not the ones supported by this driver. For this reason do not
populate the SMI resource at all and drop all the related code. The
driver now always populates the main I/O resource and only in case of SPT
(Intel Sunrisepoint) compatible devices it adds another resource for the
NO_REBOOT bit. These two resources are of different types so
platform_get_resource() used by the iTCO_wdt driver continues to find
the both resources at index 0.
Link: https://lore.kernel.org/linux-hwmon/CAM1AHpQ4196tyD=HhBu-2donSsuogabkfP03v1YF26Q7_BgvgA@mail.gmail.com/ Fixes: 31fb4a912803 ("i2c: i801: Create iTCO device on newer Intel PCHs")
[wsa: complete fix needs all of http://patchwork.ozlabs.org/project/linux-i2c/list/?series=160959&state=*] Reported-by: Martin Volf <martin.volf.42@gmail.com> Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Reviewed-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Wolfram Sang <wsa@the-dreams.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The iTCO_wdt driver only needs ICH_RES_IO_SMI I/O resource when either
turn_SMI_watchdog_clear_off module parameter is set to match ->iTCO_version
(or higher), and when legacy iTCO_vendorsupport is set. Modify the driver
so that ICH_RES_IO_SMI is optional if the two conditions are not met.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Reviewed-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Wolfram Sang <wsa@the-dreams.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In preparation for making ->smi_res optional the iTCO_wdt driver needs
to know whether vendorsupport is being set to non-zero. For this reason
export the variable.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Reviewed-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Wolfram Sang <wsa@the-dreams.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In a rare corner case the new logic for undo of SYNACK RTO could
result in triggering the warning in tcp_fastretrans_alert() that says:
WARN_ON(tp->retrans_out != 0);
The warning looked like:
WARNING: CPU: 1 PID: 1 at net/ipv4/tcp_input.c:2818 tcp_ack+0x13e0/0x3270
The sequence that tickles this bug is:
- Fast Open server receives TFO SYN with data, sends SYNACK
- (client receives SYNACK and sends ACK, but ACK is lost)
- server app sends some data packets
- (N of the first data packets are lost)
- server receives client ACK that has a TS ECR matching first SYNACK,
and also SACKs suggesting the first N data packets were lost
- server performs TS undo of SYNACK RTO, then immediately
enters recovery
- buggy behavior then performed a *second* undo that caused
the connection to be in CA_Open with retrans_out != 0
Basically, the incoming ACK packet with SACK blocks causes us to first
undo the cwnd reduction from the SYNACK RTO, but then immediately
enters fast recovery, which then makes us eligible for undo again. And
then tcp_rcv_synrecv_state_fastopen() accidentally performs an undo
using a "mash-up" of state from two different loss recovery phases: it
uses the timestamp info from the ACK of the original SYNACK, and the
undo_marker from the fast recovery.
This fix refines the logic to only invoke the tcp_try_undo_loss()
inside tcp_rcv_synrecv_state_fastopen() if the connection is still in
CA_Loss. If peer SACKs triggered fast recovery, then
tcp_rcv_synrecv_state_fastopen() can't safely undo.
Fixes: 254c39cefe6e ("tcp: undo cwnd on Fast Open spurious SYNACK retransmit") Signed-off-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: Yuchung Cheng <ycheng@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The issue happens when wait_event_interruptible_timeout() returns a value
<= 0.
In that case, the pq is left on the list. The code continues sending
packets and potentially can complete the current request with the pq still
on the dmawait list provided no descriptor shortage is seen.
If the pq is torn down in that state, the sdma interrupt handler could
find the now freed pq on the list with list corruption or memory
corruption resulting.
Fix by adding a flush routine to ensure that the pq is never on a list
after processing a request.
A follow-up patch series will address issues with seqlock surfaced in:
https://lore.kernel.org/r/20200320003129.GP20941@ziepe.ca
The seqlock use for sdma will then be converted to a spin lock since the
list_empty() doesn't need the protection afforded by the sequence lock
currently in use.
Fixes: b0778372358d ("staging/rdma/hfi1: Add page lock limit check for SDMA requests") Link: https://lore.kernel.org/r/20200320200200.23203.37777.stgit@awfm-01.aw.intel.com Reviewed-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix the handling of sendmsg() with MSG_WAITALL for userspace to round the
timeout for when a signal occurs up to at least two jiffies as a 1 jiffy
timeout may end up being effectively 0 if jiffies wraps at the wrong time.
Fixes: 8d0c6fe295af ("rxrpc: Use MSG_WAITALL to tell sendmsg() to temporarily ignore signals") Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the firmware is in a bad state or not initialized fully, sending
the DBGC_SUSPEND_RESUME command fails but we can still collect logs.
Instead of aborting the entire dump process, simply ignore the error.
By removing the last callpoint that was checking the return value, we
can also convert the function to return void.
projects / kernel.git / log