platform/kernel/linux-rpi.git
2 years agoPCI/MSI: Fix pci_irq_vector()/pci_irq_get_affinity()
Thomas Gleixner [Mon, 6 Dec 2021 22:27:26 +0000 (23:27 +0100)]
PCI/MSI: Fix pci_irq_vector()/pci_irq_get_affinity()

[ Upstream commit 29bbc35e29d9b6347780dcacde2deb4b39344167 ]

pci_irq_vector() and pci_irq_get_affinity() use the list position to find the
MSI-X descriptor at a given index. That's correct for the normal case where
the entry number is the same as the list position.

But it's wrong for cases where MSI-X was allocated with an entries array
describing sparse entry numbers into the hardware message descriptor
table. That's inconsistent at best.

Make it always check the entry number because that's what the zero base
index really means. This change won't break existing users which use a
sparse entries array for allocation because these users retrieve the Linux
interrupt number from the entries array after allocation and none of them
uses pci_irq_vector() or pci_irq_get_affinity().

Fixes: aff171641d18 ("PCI: Provide sensible IRQ vector alloc/free routines")
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Link: https://lore.kernel.org/r/20211206210223.929792157@linutronix.de
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoRDMA/qedr: Fix reporting max_{send/recv}_wr attrs
Kamal Heib [Mon, 6 Dec 2021 20:13:14 +0000 (22:13 +0200)]
RDMA/qedr: Fix reporting max_{send/recv}_wr attrs

[ Upstream commit b1a4da64bfc189510e08df1ccb1c589e667dc7a3 ]

Fix the wrongly reported max_send_wr and max_recv_wr attributes for user
QP by making sure to save their valuse on QP creation, so when query QP is
called the attributes will be reported correctly.

Fixes: cecbcddf6461 ("qedr: Add support for QP verbs")
Link: https://lore.kernel.org/r/20211206201314.124947-1-kamalheib1@gmail.com
Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Acked-by: Michal KalderonĀ <michal.kalderon@marvell.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoscsi: ufs: Fix race conditions related to driver data
Bart Van Assche [Fri, 3 Dec 2021 23:19:39 +0000 (15:19 -0800)]
scsi: ufs: Fix race conditions related to driver data

[ Upstream commit 21ad0e49085deb22c094f91f9da57319a97188e4 ]

The driver data pointer must be set before any callbacks are registered
that use that pointer. Hence move the initialization of that pointer from
after the ufshcd_init() call to inside ufshcd_init().

Link: https://lore.kernel.org/r/20211203231950.193369-7-bvanassche@acm.org
Fixes: 3b1d05807a9a ("[SCSI] ufs: Segregate PCI Specific Code")
Reported-by: Alexey Dobriyan <adobriyan@gmail.com>
Tested-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoscsi: core: Fix scsi_device_max_queue_depth()
Bart Van Assche [Fri, 3 Dec 2021 23:19:34 +0000 (15:19 -0800)]
scsi: core: Fix scsi_device_max_queue_depth()

[ Upstream commit 4bc3bffc1a885eb5cb259e4a25146a4c7b1034e3 ]

The comment above scsi_device_max_queue_depth() and also the description of
commit ca4453213951 ("scsi: core: Make sure sdev->queue_depth is <=
max(shost->can_queue, 1024)") contradict the implementation of the function
scsi_device_max_queue_depth(). Additionally, the maximum queue depth of a
SCSI LUN never exceeds host->can_queue. Fix scsi_device_max_queue_depth()
by changing max_t() into min_t().

Link: https://lore.kernel.org/r/20211203231950.193369-2-bvanassche@acm.org
Fixes: ca4453213951 ("scsi: core: Make sure sdev->queue_depth is <= max(shost->can_queue, 1024)")
Cc: Hannes Reinecke <hare@suse.de>
Cc: Sumanesh Samanta <sumanesh.samanta@broadcom.com>
Tested-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoiommu/io-pgtable-arm: Fix table descriptor paddr formatting
Hector Martin [Sat, 20 Nov 2021 03:13:43 +0000 (12:13 +0900)]
iommu/io-pgtable-arm: Fix table descriptor paddr formatting

[ Upstream commit 9abe2ac834851a7d0b0756e295cf7a292c45ca53 ]

Table descriptors were being installed without properly formatting the
address using paddr_to_iopte, which does not match up with the
iopte_deref in __arm_lpae_map. This is incorrect for the LPAE pte
format, as it does not handle the high bits properly.

This was found on Apple T6000 DARTs, which require a new pte format
(different shift); adding support for that to
paddr_to_iopte/iopte_to_paddr caused it to break badly, as even <48-bit
addresses would end up incorrect in that case.

Fixes: 6c89928ff7a0 ("iommu/io-pgtable-arm: Support 52-bit physical address")
Acked-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
Link: https://lore.kernel.org/r/20211120031343.88034-1-marcan@marcan.st
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoiommu: Extend mutex lock scope in iommu_probe_device()
Lu Baolu [Mon, 8 Nov 2021 06:13:49 +0000 (14:13 +0800)]
iommu: Extend mutex lock scope in iommu_probe_device()

[ Upstream commit 556f99ac886635e8da15528995f06d1d7028cfca ]

Extend the scope of holding group->mutex so that it can cover the default
domain check/attachment and direct mappings of reserved regions.

Cc: Ashish Mhetre <amhetre@nvidia.com>
Fixes: 211ff31b3d33b ("iommu: Fix race condition during default domain allocation")
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Link: https://lore.kernel.org/r/20211108061349.1985579-1-baolu.lu@linux.intel.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoopenrisc: Add clone3 ABI wrapper
Stafford Horne [Fri, 3 Dec 2021 22:10:18 +0000 (07:10 +0900)]
openrisc: Add clone3 ABI wrapper

[ Upstream commit 433fe39f674d58bc7a3e8254a5d2ffc290b7e04e ]

Like fork and clone the clone3 syscall needs a wrapper to save callee
saved registers, which is required by the OpenRISC ABI.  This came up
after auditing code following a discussion with Rob Landley and Arnd
Bergmann [0].

Tested with the clone3 kselftests and there were no issues.

[0] https://lore.kernel.org/all/41206fc7-f8ce-98aa-3718-ba3e1431e320@landley.net/T/#m9c0cdb2703813b9df4da04cf6b30de1f1aa89944

Fixes: 07e83dfbe16c ("openrisc: Enable the clone3 syscall")
Cc: Rob Landley <rob@landley.net>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Stafford Horne <shorne@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobinder: avoid potential data leakage when copying txn
Todd Kjos [Tue, 30 Nov 2021 18:51:50 +0000 (10:51 -0800)]
binder: avoid potential data leakage when copying txn

[ Upstream commit 6d98eb95b450a75adb4516a1d33652dc78d2b20c ]

Transactions are copied from the sender to the target
first and objects like BINDER_TYPE_PTR and BINDER_TYPE_FDA
are then fixed up. This means there is a short period where
the sender's version of these objects are visible to the
target prior to the fixups.

Instead of copying all of the data first, copy data only
after any needed fixups have been applied.

Fixes: 457b9a6f09f0 ("Staging: android: add binder driver")
Reviewed-by: Martijn Coenen <maco@android.com>
Acked-by: Christian Brauner <christian.brauner@ubuntu.com>
Signed-off-by: Todd Kjos <tkjos@google.com>
Link: https://lore.kernel.org/r/20211130185152.437403-3-tkjos@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobinder: fix handling of error during copy
Todd Kjos [Tue, 30 Nov 2021 18:51:49 +0000 (10:51 -0800)]
binder: fix handling of error during copy

[ Upstream commit fe6b1869243f23a485a106c214bcfdc7aa0ed593 ]

If a memory copy function fails to copy the whole buffer,
a positive integar with the remaining bytes is returned.
In binder_translate_fd_array() this can result in an fd being
skipped due to the failed copy, but the loop continues
processing fds since the early return condition expects a
negative integer on error.

Fix by returning "ret > 0 ? -EINVAL : ret" to handle this case.

Fixes: bb4a2e48d510 ("binder: return errors from buffer copy functions")
Suggested-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Christian Brauner <christian.brauner@ubuntu.com>
Signed-off-by: Todd Kjos <tkjos@google.com>
Link: https://lore.kernel.org/r/20211130185152.437403-2-tkjos@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agochar/mwave: Adjust io port register size
Kees Cook [Fri, 3 Dec 2021 08:42:06 +0000 (00:42 -0800)]
char/mwave: Adjust io port register size

[ Upstream commit f5912cc19acd7c24b2dbf65a6340bf194244f085 ]

Using MKWORD() on a byte-sized variable results in OOB read. Expand the
size of the reserved area so both MKWORD and MKBYTE continue to work
without overflow. Silences this warning on a -Warray-bounds build:

drivers/char/mwave/3780i.h:346:22: error: array subscript 'short unsigned int[0]' is partly outside array bounds of 'DSP_ISA_SLAVE_CONTROL[1]' [-Werror=array-bounds]
  346 | #define MKWORD(var) (*((unsigned short *)(&var)))
      |                     ~^~~~~~~~~~~~~~~~~~~~~~~~~~~~
drivers/char/mwave/3780i.h:356:40: note: in definition of macro 'OutWordDsp'
  356 | #define OutWordDsp(index,value)   outw(value,usDspBaseIO+index)
      |                                        ^~~~~
drivers/char/mwave/3780i.c:373:41: note: in expansion of macro 'MKWORD'
  373 |         OutWordDsp(DSP_IsaSlaveControl, MKWORD(rSlaveControl));
      |                                         ^~~~~~
drivers/char/mwave/3780i.c:358:31: note: while referencing 'rSlaveControl'
  358 |         DSP_ISA_SLAVE_CONTROL rSlaveControl;
      |                               ^~~~~~~~~~~~~

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20211203084206.3104326-1-keescook@chromium.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agomisc: at25: Make driver OF independent again
Andy Shevchenko [Thu, 25 Nov 2021 21:27:27 +0000 (23:27 +0200)]
misc: at25: Make driver OF independent again

[ Upstream commit 5b557298d7d09cce04e0565a535fbca63661724a ]

The commit f60e7074902a ("misc: at25: Make use of device property API")
made a good job by enabling the driver for non-OF platforms, but the
recent commit 604288bc6196 ("nvmem: eeprom: at25: fix type compiler warnings")
brought that back.

Restore greatness of the driver once again.

Fixes: eab61fb1cc2e ("nvmem: eeprom: at25: fram discovery simplification")
Fixes: fd307a4ad332 ("nvmem: prepare basics for FRAM support")
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20211125212729.86585-2-andriy.shevchenko@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoALSA: usb-audio: Drop superfluous '0' in Presonus Studio 1810c's ID
Takashi Iwai [Thu, 2 Dec 2021 08:38:33 +0000 (09:38 +0100)]
ALSA: usb-audio: Drop superfluous '0' in Presonus Studio 1810c's ID

[ Upstream commit 1e583aef12aa74afd37c1418255cc4b74e023236 ]

The vendor ID of Presonus Studio 1810c had a superfluous '0' in its
USB ID.  Drop it.

Fixes: 8dc5efe3d17c ("ALSA: usb-audio: Add support for Presonus Studio 1810c")
Link: https://lore.kernel.org/r/20211202083833.17784-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoALSA: oss: fix compile error when OSS_DEBUG is enabled
Bixuan Cui [Wed, 1 Dec 2021 08:58:54 +0000 (16:58 +0800)]
ALSA: oss: fix compile error when OSS_DEBUG is enabled

[ Upstream commit 8e7daf318d97f25e18b2fc7eb5909e34cd903575 ]

Fix compile error when OSS_DEBUG is enabled:
    sound/core/oss/pcm_oss.c: In function 'snd_pcm_oss_set_trigger':
    sound/core/oss/pcm_oss.c:2055:10: error: 'substream' undeclared (first
    use in this function); did you mean 'csubstream'?
      pcm_dbg(substream->pcm, "pcm_oss: trigger = 0x%x\n", trigger);
              ^

Fixes: 61efcee8608c ("ALSA: oss: Use standard printk helpers")
Signed-off-by: Bixuan Cui <cuibixuan@linux.alibaba.com>
Link: https://lore.kernel.org/r/1638349134-110369-1-git-send-email-cuibixuan@linux.alibaba.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoclocksource: Avoid accidental unstable marking of clocksources
Waiman Long [Thu, 18 Nov 2021 19:14:36 +0000 (14:14 -0500)]
clocksource: Avoid accidental unstable marking of clocksources

[ Upstream commit c86ff8c55b8ae68837b2fa59dc0c203907e9a15f ]

Since commit db3a34e17433 ("clocksource: Retry clock read if long delays
detected") and commit 2e27e793e280 ("clocksource: Reduce clocksource-skew
threshold"), it is found that tsc clocksource fallback to hpet can
sometimes happen on both Intel and AMD systems especially when they are
running stressful benchmarking workloads. Of the 23 systems tested with
a v5.14 kernel, 10 of them have switched to hpet clock source during
the test run.

The result of falling back to hpet is a drastic reduction of performance
when running benchmarks. For example, the fio performance tests can
drop up to 70% whereas the iperf3 performance can drop up to 80%.

4 hpet fallbacks happened during bootup. They were:

  [    8.749399] clocksource: timekeeping watchdog on CPU13: hpet read-back delay of 263750ns, attempt 4, marking unstable
  [   12.044610] clocksource: timekeeping watchdog on CPU19: hpet read-back delay of 186166ns, attempt 4, marking unstable
  [   17.336941] clocksource: timekeeping watchdog on CPU28: hpet read-back delay of 182291ns, attempt 4, marking unstable
  [   17.518565] clocksource: timekeeping watchdog on CPU34: hpet read-back delay of 252196ns, attempt 4, marking unstable

Other fallbacks happen when the systems were running stressful
benchmarks. For example:

  [ 2685.867873] clocksource: timekeeping watchdog on CPU117: hpet read-back delay of 57269ns, attempt 4, marking unstable
  [46215.471228] clocksource: timekeeping watchdog on CPU8: hpet read-back delay of 61460ns, attempt 4, marking unstable

Commit 2e27e793e280 ("clocksource: Reduce clocksource-skew threshold"),
changed the skew margin from 100us to 50us. I think this is too small
and can easily be exceeded when running some stressful workloads on a
thermally stressed system.  So it is switched back to 100us.

Even a maximum skew margin of 100us may be too small in for some systems
when booting up especially if those systems are under thermal stress. To
eliminate the case that the large skew is due to the system being too
busy slowing down the reading of both the watchdog and the clocksource,
an extra consecutive read of watchdog clock is being done to check this.

The consecutive watchdog read delay is compared against
WATCHDOG_MAX_SKEW/2. If the delay exceeds the limit, we assume that
the system is just too busy. A warning will be printed to the console
and the clock skew check is skipped for this round.

Fixes: db3a34e17433 ("clocksource: Retry clock read if long delays detected")
Fixes: 2e27e793e280 ("clocksource: Reduce clocksource-skew threshold")
Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agopowerpc/32s: Fix shift-out-of-bounds in KASAN init
Christophe Leroy [Tue, 30 Nov 2021 08:42:37 +0000 (09:42 +0100)]
powerpc/32s: Fix shift-out-of-bounds in KASAN init

[ Upstream commit af11dee4361b3519981fa04d014873f9d9edd6ac ]

================================================================================
UBSAN: shift-out-of-bounds in arch/powerpc/mm/kasan/book3s_32.c:22:23
shift exponent -1 is negative
CPU: 0 PID: 0 Comm: swapper Not tainted 5.15.5-gentoo-PowerMacG4 #9
Call Trace:
[c214be60] [c0ba0048] dump_stack_lvl+0x80/0xb0 (unreliable)
[c214be80] [c0b99288] ubsan_epilogue+0x10/0x5c
[c214be90] [c0b98fe0] __ubsan_handle_shift_out_of_bounds+0x94/0x138
[c214bf00] [c1c0f010] kasan_init_region+0xd8/0x26c
[c214bf30] [c1c0ed84] kasan_init+0xc0/0x198
[c214bf70] [c1c08024] setup_arch+0x18/0x54c
[c214bfc0] [c1c037f0] start_kernel+0x90/0x33c
[c214bff0] [00003610] 0x3610
================================================================================

This happens when the directly mapped memory is a power of 2.

Fix it by checking the shift and set the result to 0 when shift is -1

Fixes: 7974c4732642 ("powerpc/32s: Implement dedicated kasan_init_region()")
Reported-by: Erhard Furtner <erhard_f@mailbox.org>
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=215169
Link: https://lore.kernel.org/r/15cbc3439d4ad988b225e2119ec99502a5cc6ad3.1638261744.git.christophe.leroy@csgroup.eu
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agopowerpc/modules: Don't WARN on first module allocation attempt
Christophe Leroy [Tue, 30 Nov 2021 10:10:43 +0000 (11:10 +0100)]
powerpc/modules: Don't WARN on first module allocation attempt

[ Upstream commit f1797e4de1146009c888bcf8b6bb6648d55394f1 ]

module_alloc() first tries to allocate module text within 24 bits direct
jump from kernel text, and tries a wider allocation if first one fails.

When first allocation fails the following is observed in kernel logs:

  vmap allocation for size 2400256 failed: use vmalloc=<size> to increase size
  systemd-udevd: vmalloc error: size 2395133, vm_struct allocation failed, mode:0xcc0(GFP_KERNEL), nodemask=(null)
  CPU: 0 PID: 127 Comm: systemd-udevd Tainted: G        W         5.15.5-gentoo-PowerMacG4 #9
  Call Trace:
  [e2a53a50] [c0ba0048] dump_stack_lvl+0x80/0xb0 (unreliable)
  [e2a53a70] [c0540128] warn_alloc+0x11c/0x2b4
  [e2a53b50] [c0531be8] __vmalloc_node_range+0xd8/0x64c
  [e2a53c10] [c00338c0] module_alloc+0xa0/0xac
  [e2a53c40] [c027a368] load_module+0x2ae0/0x8148
  [e2a53e30] [c027fc78] sys_finit_module+0xfc/0x130
  [e2a53f30] [c0035098] ret_from_syscall+0x0/0x28
  ...

Add __GFP_NOWARN flag to first allocation so that no warning appears
when it fails.

Reported-by: Erhard Furtner <erhard_f@mailbox.org>
Fixes: 2ec13df16704 ("powerpc/modules: Load modules closer to kernel text")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/93c9b84d6ec76aaf7b4f03468e22433a6d308674.1638267035.git.christophe.leroy@csgroup.eu
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agopowerpc/perf: Fix PMU callbacks to clear pending PMI before resetting an overflown PMC
Athira Rajeev [Wed, 21 Jul 2021 05:48:29 +0000 (01:48 -0400)]
powerpc/perf: Fix PMU callbacks to clear pending PMI before resetting an overflown PMC

[ Upstream commit 2c9ac51b850d84ee496b0a5d832ce66d411ae552 ]

Running perf fuzzer showed below in dmesg logs:
  "Can't find PMC that caused IRQ"

This means a PMU exception happened, but none of the PMC's (Performance
Monitor Counter) were found to be overflown. There are some corner cases
that clears the PMCs after PMI gets masked. In such cases, the perf
interrupt handler will not find the active PMC values that had caused
the overflow and thus leads to this message while replaying.

Case 1: PMU Interrupt happens during replay of other interrupts and
counter values gets cleared by PMU callbacks before replay:

During replay of interrupts like timer, __do_irq() and doorbell
exception, we conditionally enable interrupts via may_hard_irq_enable().
This could potentially create a window to generate a PMI. Since irq soft
mask is set to ALL_DISABLED, the PMI will get masked here. We could get
IPIs run before perf interrupt is replayed and the PMU events could
be deleted or stopped. This will change the PMU SPR values and resets
the counters. Snippet of ftrace log showing PMU callbacks invoked in
__do_irq():

  <idle>-0 [051] dns. 132025441306354: __do_irq <-call_do_irq
  <idle>-0 [051] dns. 132025441306430: irq_enter <-__do_irq
  <idle>-0 [051] dns. 132025441306503: irq_enter_rcu <-__do_irq
  <idle>-0 [051] dnH. 132025441306599: xive_get_irq <-__do_irq
  <<>>
  <idle>-0 [051] dnH. 132025441307770: generic_smp_call_function_single_interrupt <-smp_ipi_demux_relaxed
  <idle>-0 [051] dnH. 132025441307839: flush_smp_call_function_queue <-smp_ipi_demux_relaxed
  <idle>-0 [051] dnH. 132025441308057: _raw_spin_lock <-event_function
  <idle>-0 [051] dnH. 132025441308206: power_pmu_disable <-perf_pmu_disable
  <idle>-0 [051] dnH. 132025441308337: power_pmu_del <-event_sched_out
  <idle>-0 [051] dnH. 132025441308407: power_pmu_read <-power_pmu_del
  <idle>-0 [051] dnH. 132025441308477: read_pmc <-power_pmu_read
  <idle>-0 [051] dnH. 132025441308590: isa207_disable_pmc <-power_pmu_del
  <idle>-0 [051] dnH. 132025441308663: write_pmc <-power_pmu_del
  <idle>-0 [051] dnH. 132025441308787: power_pmu_event_idx <-perf_event_update_userpage
  <idle>-0 [051] dnH. 132025441308859: rcu_read_unlock_strict <-perf_event_update_userpage
  <idle>-0 [051] dnH. 132025441308975: power_pmu_enable <-perf_pmu_enable
  <<>>
  <idle>-0 [051] dnH. 132025441311108: irq_exit <-__do_irq
  <idle>-0 [051] dns. 132025441311319: performance_monitor_exception <-replay_soft_interrupts

Case 2: PMI's masked during local_* operations, example local_add(). If
the local_add() operation happens within a local_irq_save(), replay of
PMI will be during local_irq_restore(). Similar to case 1, this could
also create a window before replay where PMU events gets deleted or
stopped.

Fix it by updating the PMU callback function power_pmu_disable() to
check for pending perf interrupt. If there is an overflown PMC and
pending perf interrupt indicated in paca, clear the PMI bit in paca to
drop that sample. Clearing of PMI bit is done in power_pmu_disable()
since disable is invoked before any event gets deleted/stopped. With
this fix, if there are more than one event running in the PMU, there is
a chance that we clear the PMI bit for the event which is not getting
deleted/stopped. The other events may still remain active. Hence to make
sure we don't drop valid sample in such cases, another check is added in
power_pmu_enable. This checks if there is an overflown PMC found among
the active events and if so enable back the PMI bit. Two new helper
functions are introduced to clear/set the PMI, ie
clear_pmi_irq_pending() and set_pmi_irq_pending(). Helper function
pmi_irq_pending() is introduced to give a warning if there is pending
PMI bit in paca, but no PMC is overflown.

Also there are corner cases which result in performance monitor
interrupts being triggered during power_pmu_disable(). This happens
since PMXE bit is not cleared along with disabling of other MMCR0 bits
in the pmu_disable. Such PMI's could leave the PMU running and could
trigger PMI again which will set MMCR0 PMAO bit. This could lead to
spurious interrupts in some corner cases. Example, a timer after
power_pmu_del() which will re-enable interrupts and triggers a PMI again
since PMAO bit is still set. But fails to find valid overflow since PMC
was cleared in power_pmu_del(). Fix that by disabling PMXE along with
disabling of other MMCR0 bits in power_pmu_disable().

We can't just replay PMI any time. Hence this approach is preferred
rather than replaying PMI before resetting overflown PMC. Patch also
documents core-book3s on a race condition which can trigger these PMC
messages during idle path in PowerNV.

Fixes: f442d004806e ("powerpc/64s: Add support to mask perf interrupts and replay them")
Reported-by: Nageswara R Sastry <nasastry@in.ibm.com>
Suggested-by: Nicholas Piggin <npiggin@gmail.com>
Suggested-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Tested-by: Nageswara R Sastry <rnsastry@linux.ibm.com>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Make pmi_irq_pending() return bool, reflow/reword some comments]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/1626846509-1350-2-git-send-email-atrajeev@linux.vnet.ibm.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agodt-bindings: thermal: Fix definition of cooling-maps contribution property
Niklas Sƶderlund [Tue, 9 Nov 2021 10:30:45 +0000 (11:30 +0100)]
dt-bindings: thermal: Fix definition of cooling-maps contribution property

[ Upstream commit 49bcb1506f2e095262c01bda7fd1c0db524c91e2 ]

When converting the thermal-zones bindings to yaml the definition of the
contribution property changed. The intention is the same, an integer
value expressing a ratio of a sum on how much cooling is provided by the
device to the zone. But after the conversion the integer value is
limited to the range 0 to 100 and expressed as a percentage.

This is problematic for two reasons.

- This do not match how the binding is used. Out of the 18 files that
  make use of the property only two (ste-dbx5x0.dtsi and
  ste-hrefv60plus.dtsi) sets it at a value that satisfy the binding,
  100. The remaining 16 files set the value higher and fail to validate.

- Expressing the value as a percentage instead of a ratio of the sum is
  confusing as there is nothing to enforce the sum in the zone is not
  greater then 100.

This patch restore the pre yaml conversion description and removes the
value limitation allowing the usage of the bindings to validate.

Fixes: 1202a442a31fd2e5 ("dt-bindings: thermal: Add yaml bindings for thermal zones")
Reported-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Niklas Sƶderlund <niklas.soderlund+renesas@ragnatech.se>
Link: https://lore.kernel.org/r/20211109103045.1403686-1-niklas.soderlund+renesas@ragnatech.se
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoALSA: hda: Make proper use of timecounter
Thomas Gleixner [Wed, 24 Nov 2021 22:40:01 +0000 (23:40 +0100)]
ALSA: hda: Make proper use of timecounter

[ Upstream commit 6dd21ad81bf96478db3403b1bbe251c0612d0431 ]

HDA uses a timecounter to read a hardware clock running at 24 MHz. The
conversion factor is set with a mult value of 125 and a shift value of 0,
which is not converting the hardware clock to nanoseconds, it is converting
to 1/3 nanoseconds because the conversion factor from 24Mhz to nanoseconds
is 125/3. The usage sites divide the "nanoseconds" value returned by
timecounter_read() by 3 to get a real nanoseconds value.

There is a lengthy comment in azx_timecounter_init() explaining this
choice. That comment makes blatantly wrong assumptions about how
timecounters work and what can overflow.

The comment says:

     * Applying the 1/3 factor as part of the multiplication
     * requires at least 20 bits for a decent precision, however
     * overflows occur after about 4 hours or less, not a option.

timecounters operate on time deltas between two readouts of a clock and use
the mult/shift pair to calculate a precise nanoseconds value:

    delta_nsec = (delta_clock * mult) >> shift;

The fractional part is also taken into account and preserved to prevent
accumulated rounding errors. For details see cyclecounter_cyc2ns().

The mult/shift pair has to be chosen so that the multiplication of the
maximum expected delta value does not result in a 64bit overflow. As the
counter wraps around on 32bit, the maximum observable delta between two
reads is (1 << 32) - 1 which is about 178.9 seconds.

That in turn means the maximum multiplication factor which fits into an u32
will not cause a 64bit overflow ever because it's guaranteed that:

     ((1 << 32) - 1) ^ 2 < (1 << 64)

The resulting correct multiplication factor is 2796202667 and the shift
value is 26, i.e. 26 bit precision. The overflow of the multiplication
would happen exactly at a clock readout delta of 6597069765 which is way
after the wrap around of the hardware clock at around 274.8 seconds which
is off from the claimed 4 hours by more than an order of magnitude.

If the counter ever wraps around the last read value then the calculation
is off by the number of wrap arounds times 178.9 seconds because the
overflow cannot be observed.

Use clocks_calc_mult_shift(), which calculates the most accurate mult/shift
pair based on the given clock frequency, and remove the bogus comment along
with the divisions at the readout sites.

Fixes: 5d890f591d15 ("ALSA: hda: support for wallclock timestamps")
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/871r35kwji.ffs@tglx
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoRDMA/rtrs-clt: Fix the initial value of min_latency
Jack Wang [Wed, 24 Nov 2021 08:10:40 +0000 (09:10 +0100)]
RDMA/rtrs-clt: Fix the initial value of min_latency

[ Upstream commit 925cac6358677d3d64f9b25f205eeb3d31c9f7f8 ]

The type of min_latency is ktime_t, so use KTIME_MAX to initialize the
initial value.

Fixes: dc3b66a0ce70 ("RDMA/rtrs-clt: Add a minimum latency multipath policy")
Link: https://lore.kernel.org/r/20211124081040.19533-1-jinpu.wang@ionos.com
Signed-off-by: Jack Wang <jinpu.wang@ionos.com>
Reviewed-by: Guoqing Jiang <Guoqing.Jiang@linux.dev>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoASoC: codecs: wcd938x: add SND_SOC_WCD938_SDW to codec list instead
Lukas Bulwahn [Thu, 25 Nov 2021 09:51:58 +0000 (10:51 +0100)]
ASoC: codecs: wcd938x: add SND_SOC_WCD938_SDW to codec list instead

[ Upstream commit 2039cc1da4bee1fd0df644e26b28ed769cd32a81 ]

Commit 045442228868 ("ASoC: codecs: wcd938x: add audio routing and
Kconfig") adds SND_SOC_WCD937X, which does not exist, and
SND_SOC_WCD938X, which seems not really to be the intended config to be
selected, but only a supporting config symbol to the actual config
SND_SOC_WCD938X_SDW for the codec.

Add SND_SOC_WCD938_SDW to the list instead of SND_SOC_WCD93{7,8}X.

The issue was identified with ./scripts/checkkconfigsymbols.py.

Fixes: 045442228868 ("ASoC: codecs: wcd938x: add audio routing and Kconfig")
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Link: https://lore.kernel.org/r/20211125095158.8394-3-lukas.bulwahn@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoASoC: uniphier: drop selecting non-existing SND_SOC_UNIPHIER_AIO_DMA
Lukas Bulwahn [Thu, 25 Nov 2021 09:51:57 +0000 (10:51 +0100)]
ASoC: uniphier: drop selecting non-existing SND_SOC_UNIPHIER_AIO_DMA

[ Upstream commit 49f893253ab43566e34332a969324531fea463f6 ]

Commit f37fe2f9987b ("ASoC: uniphier: add support for UniPhier AIO common
driver") adds configs SND_SOC_UNIPHIER_{LD11,PXS2}, which select the
non-existing config SND_SOC_UNIPHIER_AIO_DMA.

Hence, ./scripts/checkkconfigsymbols.py warns:

  SND_SOC_UNIPHIER_AIO_DMA
  Referencing files: sound/soc/uniphier/Kconfig

Probably, there is actually no further config intended to be selected
here. So, just drop selecting the non-existing config.

Fixes: f37fe2f9987b ("ASoC: uniphier: add support for UniPhier AIO common driver")
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Link: https://lore.kernel.org/r/20211125095158.8394-2-lukas.bulwahn@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agopowerpc/prom_init: Fix improper check of prom_getprop()
Peiwei Hu [Fri, 19 Nov 2021 09:12:18 +0000 (17:12 +0800)]
powerpc/prom_init: Fix improper check of prom_getprop()

[ Upstream commit 869fb7e5aecbc163003f93f36dcc26d0554319f6 ]

prom_getprop() can return PROM_ERROR. Binary operator can not identify
it.

Fixes: 94d2dde738a5 ("[POWERPC] Efika: prune fixups and make them more carefull")
Signed-off-by: Peiwei Hu <jlu.hpw@foxmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/tencent_BA28CC6897B7C95A92EB8C580B5D18589105@qq.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoclk: imx8mn: Fix imx8mn_clko1_sels
Adam Ford [Wed, 17 Nov 2021 13:32:02 +0000 (07:32 -0600)]
clk: imx8mn: Fix imx8mn_clko1_sels

[ Upstream commit 570727e9acfac1c2330a01dd5e1272e9c3acec08 ]

When attempting to use sys_pll1_80m as the parent for clko1, the
system hangs.  This is due to the fact that the source select
for sys_pll1_80m was incorrectly pointing to m7_alt_pll_clk, which
doesn't yet exist.

According to Rev 3 of the TRM, The imx8mn_clko1_sels also incorrectly
references an osc_27m which does not exist, nor does an entry for
source select bits 010b.  Fix both by inserting a dummy clock into
the missing space in the table and renaming the incorrectly name clock
with dummy.

Fixes: 96d6392b54db ("clk: imx: Add support for i.MX8MN clock driver")
Signed-off-by: Adam Ford <aford173@gmail.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
Link: https://lore.kernel.org/r/20211117133202.775633-1-aford173@gmail.com
Signed-off-by: Abel Vesa <abel.vesa@nxp.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoclk: renesas: rzg2l: propagate return value of_genpd_add_provider_simple()
Lad Prabhakar [Wed, 17 Nov 2021 11:50:59 +0000 (11:50 +0000)]
clk: renesas: rzg2l: propagate return value of_genpd_add_provider_simple()

[ Upstream commit 33748744f15a110a233b6ae0380f476006e770f0 ]

of_genpd_add_provider_simple() might fail, this patch makes sure we check
the return value of of_genpd_add_provider_simple() by propagating the
return value to the caller of rzg2l_cpg_add_clk_domain().

Fixes: ef3c613ccd68a ("clk: renesas: Add CPG core wrapper for RZ/G2L SoC")
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Link: https://lore.kernel.org/r/20211117115101.28281-3-prabhakar.mahadev-lad.rj@bp.renesas.com
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoclk: renesas: rzg2l: Check return value of pm_genpd_init()
Lad Prabhakar [Wed, 17 Nov 2021 11:50:58 +0000 (11:50 +0000)]
clk: renesas: rzg2l: Check return value of pm_genpd_init()

[ Upstream commit 27527a3d3b162e4512798c058c0e8a216c721187 ]

Make sure we check the return value of pm_genpd_init() which might fail.
Also add a devres action to remove the power-domain in-case the probe
callback fails further down in the code flow.

Fixes: ef3c613ccd68a ("clk: renesas: Add CPG core wrapper for RZ/G2L SoC")
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Link: https://lore.kernel.org/r/20211117115101.28281-2-prabhakar.mahadev-lad.rj@bp.renesas.com
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoscsi: pm80xx: Update WARN_ON check in pm8001_mpi_build_cmd()
Igor Pylypiv [Mon, 1 Nov 2021 23:28:24 +0000 (16:28 -0700)]
scsi: pm80xx: Update WARN_ON check in pm8001_mpi_build_cmd()

[ Upstream commit 606c54ae975ad3af540b505b46b55a687501711f ]

Starting from commit 05c6c029a44d ("scsi: pm80xx: Increase number of
supported queues") driver initializes only max_q_num queues.  Do not use an
invalid queue if the WARN_ON condition is true.

Link: https://lore.kernel.org/r/20211101232825.2350233-4-ipylypiv@google.com
Fixes: 7640e1eb8c5d ("scsi: pm80xx: Make mpi_build_cmd locking consistent")
Reviewed-by: Vishakha Channapattan <vishakhavc@google.com>
Acked-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Igor Pylypiv <ipylypiv@google.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoRDMA/hns: Validate the pkey index
Kamal Heib [Wed, 17 Nov 2021 14:59:54 +0000 (16:59 +0200)]
RDMA/hns: Validate the pkey index

[ Upstream commit 2a67fcfa0db6b4075515bd23497750849b88850f ]

Before query pkey, make sure that the queried index is valid.

Fixes: 9a4435375cd1 ("IB/hns: Add driver files for hns RoCE driver")
Link: https://lore.kernel.org/r/20211117145954.123893-1-kamalheib1@gmail.com
Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoRDMA/bnxt_re: Scan the whole bitmap when checking if "disabling RCFW with pending...
Christophe JAILLET [Fri, 12 Nov 2021 08:59:04 +0000 (09:59 +0100)]
RDMA/bnxt_re: Scan the whole bitmap when checking if "disabling RCFW with pending cmd-bit"

[ Upstream commit a917dfb66c0a1fa1caacf3d71edcafcab48e6ff0 ]

The 'cmdq->cmdq_bitmap' bitmap is 'rcfw->cmdq_depth' bits long.  The size
stored in 'cmdq->bmap_size' is the size of the bitmap in bytes.

Remove this erroneous 'bmap_size' and use 'rcfw->cmdq_depth' directly in
'bnxt_qplib_disable_rcfw_channel()'. Otherwise some error messages may be
missing.

Other uses of 'cmdq_bitmap' already take into account 'rcfw->cmdq_depth'
directly.

Fixes: 1ac5a4047975 ("RDMA/bnxt_re: Add bnxt_re RoCE driver")
Link: https://lore.kernel.org/r/47ed717c3070a1d0f53e7b4c768a4fd11caf365d.1636707421.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Acked-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoALSA: hda: Fix potential deadlock at codec unbinding
Takashi Iwai [Tue, 16 Nov 2021 07:24:59 +0000 (08:24 +0100)]
ALSA: hda: Fix potential deadlock at codec unbinding

[ Upstream commit 7206998f578d5553989bc01ea2e544b622e79539 ]

When a codec is unbound dynamically via sysfs while its stream is in
use, we may face a potential deadlock at the proc remove or a UAF.
This happens since the hda_pcm is managed by a linked list, as it
handles the hda_pcm object release via kref.

When a PCM is opened at the unbinding time, the release of hda_pcm
gets delayed and it ends up with the close of the PCM stream releasing
the associated hda_pcm object of its own.  The hda_pcm destructor
contains the PCM device release that includes the removal of procfs
entries.  And, this removal has the sync of the close of all in-use
files -- which would never finish because it's called from the PCM
file descriptor itself, i.e. it's trying to shoot its foot.

For addressing the deadlock above, this patch changes the way to
manage and release the hda_pcm object.  The kref of hda_pcm is
dropped, and instead a simple refcount is introduced in hda_codec for
keeping the track of the active PCM streams, and at each PCM open and
close, this refcount is adjusted accordingly.  At unbinding, the
driver calls snd_device_disconnect() for each PCM stream, then
synchronizes with the refcount finish, and finally releases the object
resources.

Fixes: bbbc7e8502c9 ("ALSA: hda - Allocate hda_pcm objects dynamically")
Link: https://lore.kernel.org/r/20211116072459.18930-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoALSA: hda: Add missing rwsem around snd_ctl_remove() calls
Takashi Iwai [Tue, 16 Nov 2021 07:13:14 +0000 (08:13 +0100)]
ALSA: hda: Add missing rwsem around snd_ctl_remove() calls

[ Upstream commit 80bd64af75b4bb11c0329bc66c35da2ddfb66d88 ]

snd_ctl_remove() has to be called with card->controls_rwsem held (when
called after the card instantiation).  This patch add the missing
rwsem calls around it.

Fixes: d13bd412dce2 ("ALSA: hda - Manage kcontrol lists")
Link: https://lore.kernel.org/r/20211116071314.15065-3-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoALSA: PCM: Add missing rwsem around snd_ctl_remove() calls
Takashi Iwai [Tue, 16 Nov 2021 07:13:13 +0000 (08:13 +0100)]
ALSA: PCM: Add missing rwsem around snd_ctl_remove() calls

[ Upstream commit 5471e9762e1af4b7df057a96bfd46cc250979b88 ]

snd_ctl_remove() has to be called with card->controls_rwsem held (when
called after the card instantiation).  This patch add the missing
rwsem calls around it.

Fixes: a8ff48cb7083 ("ALSA: pcm: Free chmap at PCM free callback, too")
Link: https://lore.kernel.org/r/20211116071314.15065-2-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoALSA: jack: Add missing rwsem around snd_ctl_remove() calls
Takashi Iwai [Tue, 16 Nov 2021 07:13:12 +0000 (08:13 +0100)]
ALSA: jack: Add missing rwsem around snd_ctl_remove() calls

[ Upstream commit 06764dc931848c3a9bc01a63bbf76a605408bb54 ]

snd_ctl_remove() has to be called with card->controls_rwsem held (when
called after the card instantiation).  This patch add the missing
rwsem calls around it.

Fixes: 9058cbe1eed2 ("ALSA: jack: implement kctl creating for jack devices")
Link: https://lore.kernel.org/r/20211116071314.15065-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoASoC: Intel: sof_sdw: fix jack detection on HP Spectre x360 convertible
Pierre-Louis Bossart [Wed, 27 Oct 2021 02:18:16 +0000 (10:18 +0800)]
ASoC: Intel: sof_sdw: fix jack detection on HP Spectre x360 convertible

[ Upstream commit 0527b19fa4f390a6054612e1fa1dd4f8efc96739 ]

Tests on device show the JD2 mode does not work at all, the 'Headphone
Jack' and 'Headset Mic Jack' are shown as 'on' always.

JD1 seems to be the better option, with at least a change between the
two cases.

Jack not plugged-in:
[root@fedora ~]# amixer -Dhw:0 cget numid=12
numid=12,iface=CARD,name='Headphone Jack'
  ; type=BOOLEAN,access=r-------,values=1
  : values=off
[root@fedora ~]# amixer -Dhw:0 cget numid=13
numid=13,iface=CARD,name='Headset Mic Jack'
  ; type=BOOLEAN,access=r-------,values=1
  : values=off

Jack plugged-in:
[root@fedora ~]# amixer -Dhw:0 cget numid=13
numid=13,iface=CARD,name='Headset Mic Jack'
  ; type=BOOLEAN,access=r-------,values=1
  : values=on
[root@fedora ~]# amixer -Dhw:0 cget numid=13
numid=13,iface=CARD,name='Headset Mic Jack'
  ; type=BOOLEAN,access=r-------,values=1
  : values=on

The 'Headset Mic Jack' is updated with a delay which seems normal with
additional calibration needed.

Fixes: d92e279dee56 ('ASoC: Intel: sof_sdw: add quirk for HP Spectre x360 convertible')
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20211027021824.24776-3-yung-chuan.liao@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoext4: avoid trim error on fs with small groups
Jan Kara [Fri, 12 Nov 2021 15:22:02 +0000 (16:22 +0100)]
ext4: avoid trim error on fs with small groups

[ Upstream commit 173b6e383d2a204c9921ffc1eca3b87aa2106c33 ]

A user reported FITRIM ioctl failing for him on ext4 on some devices
without apparent reason.  After some debugging we've found out that
these devices (being LVM volumes) report rather large discard
granularity of 42MB and the filesystem had 1k blocksize and thus group
size of 8MB. Because ext4 FITRIM implementation puts discard
granularity into minlen, ext4_trim_fs() declared the trim request as
invalid. However just silently doing nothing seems to be a more
appropriate reaction to such combination of parameters since user did
not specify anything wrong.

CC: Lukas Czerner <lczerner@redhat.com>
Fixes: 5c2ed62fd447 ("ext4: Adjust minlen with discard_granularity in the FITRIM ioctl")
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20211112152202.26614-1-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet: mcs7830: handle usb read errors properly
Pavel Skripkin [Thu, 6 Jan 2022 22:57:16 +0000 (01:57 +0300)]
net: mcs7830: handle usb read errors properly

[ Upstream commit d668769eb9c52b150753f1653f7f5a0aeb8239d2 ]

Syzbot reported uninit value in mcs7830_bind(). The problem was in
missing validation check for bytes read via usbnet_read_cmd().

usbnet_read_cmd() internally calls usb_control_msg(), that returns
number of bytes read. Code should validate that requested number of bytes
was actually read.

So, this patch adds missing size validation check inside
mcs7830_get_reg() to prevent uninit value bugs

Reported-and-tested-by: syzbot+003c0a286b9af5412510@syzkaller.appspotmail.com
Fixes: 2a36d7083438 ("USB: driver for mcs7830 (aka DeLOCK) USB ethernet adapter")
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20220106225716.7425-1-paskripkin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobnxt_en: use firmware provided max timeout for messages
Edwin Peer [Sun, 9 Jan 2022 23:54:44 +0000 (18:54 -0500)]
bnxt_en: use firmware provided max timeout for messages

[ Upstream commit bce9a0b7900836df223ab638090df0cb8430d9e8 ]

Some older devices cannot accommodate the 40 seconds timeout
cap for long running commands (such as NVRAM commands) due to
hardware limitations. Allow these devices to request more time for
these long running commands, but print a warning, since the longer
timeout may cause the hung task watchdog to trigger. In the case of a
firmware update operation, this is preferable to failing outright.

v2: Use bp->hwrm_cmd_max_timeout directly without the constants.

Fixes: 881d8353b05e ("bnxt_en: Add an upper bound for all firmware command timeouts.")
Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobnxt_en: move coredump functions into dedicated file
Edwin Peer [Fri, 29 Oct 2021 07:47:48 +0000 (03:47 -0400)]
bnxt_en: move coredump functions into dedicated file

[ Upstream commit b032228e58ea2477955058ad4d70a636ce1dec51 ]

Change bnxt_get_coredump() and bnxt_get_coredump_length() to non-static
functions.

Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobnxt_en: Refactor coredump functions
Edwin Peer [Fri, 29 Oct 2021 07:47:47 +0000 (03:47 -0400)]
bnxt_en: Refactor coredump functions

[ Upstream commit 9a575c8c25ae2372112db6d6b3e553cd90e9f02b ]

The coredump functionality will be used by devlink health. Refactor
these functions that get coredump and coredump length. There is no
functional change, but the following checkpatch warnings were
addressed:

  - strscpy is preferred over strlcpy.
  - sscanf results should be checked, with an additional warning.

Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoiwlwifi: mvm: Use div_s64 instead of do_div in iwl_mvm_ftm_rtt_smoothing()
Nathan Chancellor [Mon, 27 Dec 2021 19:17:57 +0000 (12:17 -0700)]
iwlwifi: mvm: Use div_s64 instead of do_div in iwl_mvm_ftm_rtt_smoothing()

[ Upstream commit 4ccdcc8ffd955490feec05380223db6a48961eb5 ]

When building ARCH=arm allmodconfig:

drivers/net/wireless/intel/iwlwifi/mvm/ftm-initiator.c: In function ā€˜iwl_mvm_ftm_rtt_smoothingā€™:
./include/asm-generic/div64.h:222:35: error: comparison of distinct pointer types lacks a cast [-Werror]
  222 |         (void)(((typeof((n)) *)0) == ((uint64_t *)0));  \
      |                                   ^~
drivers/net/wireless/intel/iwlwifi/mvm/ftm-initiator.c:1070:9: note: in expansion of macro ā€˜do_divā€™
 1070 |         do_div(rtt_avg, 100);
      |         ^~~~~~

do_div() has to be used with an unsigned 64-bit integer dividend but
rtt_avg is a signed 64-bit integer.

div_s64() expects a signed 64-bit integer dividend and signed 32-bit
divisor, which fits this scenario, so use that function here to fix the
warning.

Fixes: 8b0f92549f2c ("iwlwifi: mvm: fix 32-bit build in FTM")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Link: https://lore.kernel.org/r/20211227191757.2354329-1-nathan@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet: openvswitch: Fix ct_state nat flags for conns arriving from tc
Paul Blakey [Thu, 6 Jan 2022 15:38:04 +0000 (17:38 +0200)]
net: openvswitch: Fix ct_state nat flags for conns arriving from tc

[ Upstream commit 6f022c2ddbcefaee79502ce5386dfe351d457070 ]

Netfilter conntrack maintains NAT flags per connection indicating
whether NAT was configured for the connection. Openvswitch maintains
NAT flags on the per packet flow key ct_state field, indicating
whether NAT was actually executed on the packet.

When a packet misses from tc to ovs the conntrack NAT flags are set.
However, NAT was not necessarily executed on the packet because the
connection's state might still be in NEW state. As such, openvswitch
wrongly assumes that NAT was executed and sets an incorrect flow key
NAT flags.

Fix this, by flagging to openvswitch which NAT was actually done in
act_ct via tc_skb_ext and tc_skb_cb to the openvswitch module, so
the packet flow key NAT flags will be correctly set.

Fixes: b57dc7c13ea9 ("net/sched: Introduce action ct")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://lore.kernel.org/r/20220106153804.26451-1-paulb@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet: openvswitch: Fix matching zone id for invalid conns arriving from tc
Paul Blakey [Tue, 14 Dec 2021 17:24:35 +0000 (19:24 +0200)]
net: openvswitch: Fix matching zone id for invalid conns arriving from tc

[ Upstream commit 635d448a1cce4b4ebee52b351052c70434fa90ea ]

Zone id is not restored if we passed ct and ct rejected the connection,
as there is no ct info on the skb.

Save the zone from tc skb cb to tc skb extension and pass it on to
ovs, use that info to restore the zone id for invalid connections.

Fixes: d29334c15d33 ("net/sched: act_api: fix miss set post_ct for ovs after do conntrack in act_ct")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/sched: flow_dissector: Fix matching on zone id for invalid conns
Paul Blakey [Tue, 14 Dec 2021 17:24:34 +0000 (19:24 +0200)]
net/sched: flow_dissector: Fix matching on zone id for invalid conns

[ Upstream commit 3849595866166b23bf6a0cb9ff87e06423167f67 ]

If ct rejects a flow, it removes the conntrack info from the skb.
act_ct sets the post_ct variable so the dissector will see this case
as an +tracked +invalid state, but the zone id is lost with the
conntrack info.

To restore the zone id on such cases, set the last executed zone,
via the tc control block, when passing ct, and read it back in the
dissector if there is no ct info on the skb (invalid connection).

Fixes: 7baf2429a1a9 ("net/sched: cls_flower add CT_FLAGS_INVALID flag support")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agopcmcia: fix setting of kthread task states
Dominik Brodowski [Sun, 9 Jan 2022 09:02:51 +0000 (10:02 +0100)]
pcmcia: fix setting of kthread task states

[ Upstream commit fbb3485f1f931102d8ba606f1c28123f5b48afa3 ]

We need to set TASK_INTERRUPTIBLE before calling kthread_should_stop().
Otherwise, kthread_stop() might see that the pccardd thread is still
in TASK_RUNNING state and fail to wake it up.

Additionally, we only need to set the state back to TASK_RUNNING if
kthread_should_stop() breaks the loop.

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Fixes: d3046ba809ce ("pcmcia: fix a boot time warning in pcmcia cs code")
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agocan: rcar_canfd: rcar_canfd_channel_probe(): make sure we free CAN network device
Lad Prabhakar [Thu, 6 Jan 2022 11:48:00 +0000 (11:48 +0000)]
can: rcar_canfd: rcar_canfd_channel_probe(): make sure we free CAN network device

[ Upstream commit 72b1e360572f9fa7d08ee554f1da29abce23f288 ]

Make sure we free CAN network device in the error path. There are
several jumps to fail label after allocating the CAN network device
successfully. This patch places the free_candev() under fail label so
that in failure path a jump to fail label frees the CAN network
device.

Fixes: 76e9353a80e9 ("can: rcar_canfd: Add support for RZ/G2L family")
Link: https://lore.kernel.org/all/20220106114801.20563-1-prabhakar.mahadev-lad.rj@bp.renesas.com
Reported-by: Pavel Machek <pavel@denx.de>
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agocan: xilinx_can: xcan_probe(): check for error irq
Jiasheng Jiang [Fri, 24 Dec 2021 02:13:24 +0000 (10:13 +0800)]
can: xilinx_can: xcan_probe(): check for error irq

[ Upstream commit c6564c13dae25cd7f8e1de5127b4da4500ee5844 ]

For the possible failure of the platform_get_irq(), the returned irq
could be error number and will finally cause the failure of the
request_irq().

Consider that platform_get_irq() can now in certain cases return
-EPROBE_DEFER, and the consequences of letting request_irq()
effectively convert that into -EINVAL, even at probe time rather than
later on. So it might be better to check just now.

Fixes: b1201e44f50b ("can: xilinx CAN controller support")
Link: https://lore.kernel.org/all/20211224021324.1447494-1-jiasheng@iscas.ac.cn
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agocan: softing: softing_startstop(): fix set but not used variable warning
Marc Kleine-Budde [Sat, 8 Jan 2022 20:57:51 +0000 (21:57 +0100)]
can: softing: softing_startstop(): fix set but not used variable warning

[ Upstream commit 370d988cc529598ebaec6487d4f84c2115dc696b ]

In the function softing_startstop() the variable error_reporting is
assigned but not used. The code that uses this variable is commented
out. Its stated that the functionality is not finally verified.

To fix the warning:

| drivers/net/can/softing/softing_fw.c:424:9: error: variable 'error_reporting' set but not used [-Werror,-Wunused-but-set-variable]

remove the comment, activate the code, but add a "0 &&" to the if
expression and rely on the optimizer rather than the preprocessor to
remove the code.

Link: https://lore.kernel.org/all/20220109103126.1872833-1-mkl@pengutronix.de
Fixes: 03fd3cf5a179 ("can: add driver for Softing card")
Cc: Kurt Van Dijck <dev.kurt@vandijck-laurijssen.be>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agotpm_tis: Fix an error handling path in 'tpm_tis_core_init()'
Christophe Jaillet [Sat, 6 Nov 2021 16:42:04 +0000 (17:42 +0100)]
tpm_tis: Fix an error handling path in 'tpm_tis_core_init()'

[ Upstream commit e96d52822f5ac0a25de78f95cd23421bcbc93584 ]

Commit 79ca6f74dae0 ("tpm: fix Atmel TPM crash caused by too frequent
queries") has moved some code around without updating the error handling
path.

This is now pointless to 'goto out_err' when neither 'clk_enable()' nor
'ioremap()' have been called yet.

Make a direct return instead to avoid undoing things that have not been
done.

Fixes: 79ca6f74dae0 ("tpm: fix Atmel TPM crash caused by too frequent queries")
Signed-off-by: Christophe Jaillet <christophe.jaillet@wanadoo.fr>
Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agotpm: add request_locality before write TPM_INT_ENABLE
Chen Jun [Wed, 13 Oct 2021 06:25:56 +0000 (06:25 +0000)]
tpm: add request_locality before write TPM_INT_ENABLE

[ Upstream commit 0ef333f5ba7f24f5d8478425c163d3097f1c7afd ]

Locality is not appropriately requested before writing the int mask.
Add the missing boilerplate.

Fixes: e6aef069b6e9 ("tpm_tis: convert to using locality callbacks")
Signed-off-by: Chen Jun <chenjun102@huawei.com>
Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agocan: mcp251xfd: add missing newline to printed strings
Marc Kleine-Budde [Tue, 19 Oct 2021 15:05:25 +0000 (17:05 +0200)]
can: mcp251xfd: add missing newline to printed strings

[ Upstream commit 3bd9d8ce6f8c5c43ee2f1106021db0f98882cc75 ]

This patch adds the missing newline to printed strings.

Fixes: 55e5b97f003e ("can: mcp25xxfd: add driver for Microchip MCP25xxFD SPI CAN")
Link: https://lore.kernel.org/all/20220105154300.1258636-4-mkl@pengutronix.de
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet: mscc: ocelot: fix incorrect balancing with down LAG ports
Vladimir Oltean [Fri, 7 Jan 2022 16:43:32 +0000 (18:43 +0200)]
net: mscc: ocelot: fix incorrect balancing with down LAG ports

[ Upstream commit a14e6b69f393d651913edcbe4ec0dec27b8b4b40 ]

Assuming the test setup described here:
https://patchwork.kernel.org/project/netdevbpf/cover/20210205130240.4072854-1-vladimir.oltean@nxp.com/
(swp1 and swp2 are in bond0, and bond0 is in a bridge with swp0)

it can be seen that when swp1 goes down (on either board A or B), then
traffic that should go through that port isn't forwarded anywhere.

A dump of the PGID table shows the following:

PGID_DST[0] = ports 0
PGID_DST[1] = ports 1
PGID_DST[2] = ports 2
PGID_DST[3] = ports 3
PGID_DST[4] = ports 4
PGID_DST[5] = ports 5
PGID_DST[6] = no ports
PGID_AGGR[0] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[1] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[2] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[3] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[4] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[5] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[6] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[7] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[8] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[9] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[10] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[11] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[12] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[13] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[14] = ports 0, 1, 2, 3, 4, 5
PGID_AGGR[15] = ports 0, 1, 2, 3, 4, 5
PGID_SRC[0] = ports 1, 2
PGID_SRC[1] = ports 0
PGID_SRC[2] = ports 0
PGID_SRC[3] = no ports
PGID_SRC[4] = no ports
PGID_SRC[5] = no ports
PGID_SRC[6] = ports 0, 1, 2, 3, 4, 5

Whereas a "good" PGID configuration for that setup should have looked
like this:

PGID_DST[0] = ports 0
PGID_DST[1] = ports 1, 2
PGID_DST[2] = ports 1, 2
PGID_DST[3] = ports 3
PGID_DST[4] = ports 4
PGID_DST[5] = ports 5
PGID_DST[6] = no ports
PGID_AGGR[0] = ports 0, 2, 3, 4, 5
PGID_AGGR[1] = ports 0, 2, 3, 4, 5
PGID_AGGR[2] = ports 0, 2, 3, 4, 5
PGID_AGGR[3] = ports 0, 2, 3, 4, 5
PGID_AGGR[4] = ports 0, 2, 3, 4, 5
PGID_AGGR[5] = ports 0, 2, 3, 4, 5
PGID_AGGR[6] = ports 0, 2, 3, 4, 5
PGID_AGGR[7] = ports 0, 2, 3, 4, 5
PGID_AGGR[8] = ports 0, 2, 3, 4, 5
PGID_AGGR[9] = ports 0, 2, 3, 4, 5
PGID_AGGR[10] = ports 0, 2, 3, 4, 5
PGID_AGGR[11] = ports 0, 2, 3, 4, 5
PGID_AGGR[12] = ports 0, 2, 3, 4, 5
PGID_AGGR[13] = ports 0, 2, 3, 4, 5
PGID_AGGR[14] = ports 0, 2, 3, 4, 5
PGID_AGGR[15] = ports 0, 2, 3, 4, 5
PGID_SRC[0] = ports 1, 2
PGID_SRC[1] = ports 0
PGID_SRC[2] = ports 0
PGID_SRC[3] = no ports
PGID_SRC[4] = no ports
PGID_SRC[5] = no ports
PGID_SRC[6] = ports 0, 1, 2, 3, 4, 5

In other words, in the "bad" configuration, the attempt is to remove the
inactive swp1 from the destination ports via PGID_DST. But when a MAC
table entry is learned, it is learned towards PGID_DST 1, because that
is the logical port id of the LAG itself (it is equal to the lowest
numbered member port). So when swp1 becomes inactive, if we set
PGID_DST[1] to contain just swp1 and not swp2, the packet will not have
any chance to reach the destination via swp2.

The "correct" way to remove swp1 as a destination is via PGID_AGGR
(remove swp1 from the aggregation port groups for all aggregation
codes). This means that PGID_DST[1] and PGID_DST[2] must still contain
both swp1 and swp2. This makes the MAC table still treat packets
destined towards the single-port LAG as "multicast", and the inactive
ports are removed via the aggregation code tables.

The change presented here is a design one: the ocelot_get_bond_mask()
function used to take an "only_active_ports" argument. We don't need
that. The only call site that specifies only_active_ports=true,
ocelot_set_aggr_pgids(), must retrieve the entire bonding mask, because
it must program that into PGID_DST. Additionally, it must also clear the
inactive ports from the bond mask here, which it can't do if bond_mask
just contains the active ports:

ac = ocelot_read_rix(ocelot, ANA_PGID_PGID, i);
ac &= ~bond_mask;  <---- here
/* Don't do division by zero if there was no active
 * port. Just make all aggregation codes zero.
 */
if (num_active_ports)
ac |= BIT(aggr_idx[i % num_active_ports]);
ocelot_write_rix(ocelot, ac, ANA_PGID_PGID, i);

So it becomes the responsibility of ocelot_set_aggr_pgids() to take
ocelot_port->lag_tx_active into consideration when populating the
aggr_idx array.

Fixes: 23ca3b727ee6 ("net: mscc: ocelot: rebalance LAGs on link up/down events")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://lore.kernel.org/r/20220107164332.402133-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoregmap: Call regmap_debugfs_exit() prior to _init()
Fabio Estevam [Fri, 7 Jan 2022 16:33:07 +0000 (13:33 -0300)]
regmap: Call regmap_debugfs_exit() prior to _init()

[ Upstream commit 530792efa6cb86f5612ff093333fec735793b582 ]

Since commit cffa4b2122f5 ("regmap: debugfs: Fix a memory leak when
calling regmap_attach_dev"), the following debugfs error is seen
on i.MX boards:

debugfs: Directory 'dummy-iomuxc-gpr@20e0000' with parent 'regmap' already present!

In the attempt to fix the memory leak, the above commit added a NULL check
for map->debugfs_name. For the first debufs entry, map->debugfs_name is NULL
and then the new name is allocated via kasprintf().

For the second debugfs entry, map->debugfs_name() is no longer NULL, so
it will keep using the old entry name and the duplicate name error is seen.

Quoting Mark Brown:

"That means that if the device gets freed we'll end up with the old debugfs
file hanging around pointing at nothing.
...
To be more explicit this means we need a call to regmap_debugfs_exit()
which will clean up all the existing debugfs stuff before we loose
references to it."

Call regmap_debugfs_exit() prior to regmap_debugfs_init() to fix
the problem.

Tested on i.MX6Q and i.MX6SX boards.

Fixes: cffa4b2122f5 ("regmap: debugfs: Fix a memory leak when calling regmap_attach_dev")
Suggested-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Fabio Estevam <festevam@denx.de>
Link: https://lore.kernel.org/r/20220107163307.335404-1-festevam@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonetrom: fix api breakage in nr_setsockopt()
Dan Carpenter [Fri, 7 Jan 2022 07:12:10 +0000 (10:12 +0300)]
netrom: fix api breakage in nr_setsockopt()

[ Upstream commit dc35616e6c2907b0c0c391a205802d8880f7fd85 ]

This needs to copy an unsigned int from user space instead of a long to
avoid breaking user space with an API change.

I have updated all the integer overflow checks from ULONG to UINT as
well.  This is a slight API change but I do not expect it to affect
anything in real life.

Fixes: 3087a6f36ee0 ("netrom: fix copying in user data in nr_setsockopt")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoax25: uninitialized variable in ax25_setsockopt()
Dan Carpenter [Fri, 7 Jan 2022 07:13:12 +0000 (10:13 +0300)]
ax25: uninitialized variable in ax25_setsockopt()

[ Upstream commit 9371937092d5fd502032c1bb4475b36b39b1f1b3 ]

The "opt" variable is unsigned long but we only copy 4 bytes from
the user so the lower 4 bytes are uninitialized.

I have changed the integer overflow checks from ULONG to UINT as well.
This is a slight API change but I don't expect it to break anything.

Fixes: a7b75c5a8c41 ("net: pass a sockptr_t into ->setsockopt")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoocteontx2-af: Increment ptp refcount before use
Subbaraya Sundeep [Fri, 7 Jan 2022 06:30:29 +0000 (12:00 +0530)]
octeontx2-af: Increment ptp refcount before use

[ Upstream commit 93440f4888cf049dbd22b41aaf94d2e2153b3eb8 ]

Before using the ptp pci device by AF driver increment
the reference count of it.

Fixes: a8b90c9d26d6 ("octeontx2-af: Add PTP device id for CN10K and 95O silcons")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agospi: spi-meson-spifc: Add missing pm_runtime_disable() in meson_spifc_probe
Miaoqian Lin [Fri, 7 Jan 2022 07:54:24 +0000 (07:54 +0000)]
spi: spi-meson-spifc: Add missing pm_runtime_disable() in meson_spifc_probe

[ Upstream commit 69c1b87516e327a60b39f96b778fe683259408bf ]

If the probe fails, we should use pm_runtime_disable() to balance
pm_runtime_enable().
Add missing pm_runtime_disable() for meson_spifc_probe.

Fixes: c3e4bc5434d2 ("spi: meson: Add support for Amlogic Meson SPIFC")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220107075424.7774-1-linmq006@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agomptcp: fix a DSS option writing error
Geliang Tang [Thu, 6 Jan 2022 22:06:37 +0000 (14:06 -0800)]
mptcp: fix a DSS option writing error

[ Upstream commit 110b6d1fe98fd7af9893992459b651594d789293 ]

'ptr += 1;' was omitted in the original code.

If the DSS is the last option -- which is what we have most of the
time -- that's not an issue. But it is if we need to send something else
after like a RM_ADDR or an MP_PRIO.

Fixes: 1bff1e43a30e ("mptcp: optimize out option generation")
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Geliang Tang <geliang.tang@suse.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agomptcp: fix opt size when sending DSS + MP_FAIL
Matthieu Baerts [Thu, 6 Jan 2022 22:06:36 +0000 (14:06 -0800)]
mptcp: fix opt size when sending DSS + MP_FAIL

[ Upstream commit 04fac2cae9422a3401c172571afbcfdd58fa5c7e ]

When these two options had to be sent -- which is not common -- the DSS
size was not being taken into account in the remaining size.

Additionally in this situation, the reported size was only the one of
the MP_FAIL which can cause issue if at the end, we need to write more
in the TCP options than previously said.

Here we use a dedicated variable for MP_FAIL size to keep the
WARN_ON_ONCE() just after.

Fixes: c25aeb4e0953 ("mptcp: MP_FAIL suboption sending")
Acked-and-tested-by: Geliang Tang <geliang.tang@suse.com>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agomptcp: fix per socket endpoint accounting
Paolo Abeni [Fri, 7 Jan 2022 00:20:20 +0000 (16:20 -0800)]
mptcp: fix per socket endpoint accounting

[ Upstream commit f7d6a237d7422809d458d754016de2844017cb4d ]

Since full-mesh endpoint support, the reception of a single ADD_ADDR
option can cause multiple subflows creation. When such option is
accepted we increment 'add_addr_accepted' by one. When we received
a paired RM_ADDR option, we deleted all the relevant subflows,
decrementing 'add_addr_accepted' by one for each of them.

We have a similar issue for 'local_addr_used'

Fix them moving the pm endpoint accounting outside the subflow
traversal.

Fixes: 1a0d6136c5f0 ("mptcp: local addresses fullmesh")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoBluetooth: L2CAP: uninitialized variables in l2cap_sock_setsockopt()
Dan Carpenter [Fri, 7 Jan 2022 07:16:44 +0000 (10:16 +0300)]
Bluetooth: L2CAP: uninitialized variables in l2cap_sock_setsockopt()

[ Upstream commit 2b70d4f9b20635ac328836e50d183632e1930f94 ]

The "opt" variable is a u32, but on some paths only the top bytes
were initialized and the others contained random stack data.

Fixes: a7b75c5a8c41 ("net: pass a sockptr_t into ->setsockopt")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agolib/mpi: Add the return value check of kcalloc()
Zizhuang Deng [Thu, 30 Dec 2021 07:03:31 +0000 (15:03 +0800)]
lib/mpi: Add the return value check of kcalloc()

[ Upstream commit dd827abe296fe4249b2f8c9b95f72f814ea8348c ]

Add the return value check of kcalloc() to avoid potential
NULL ptr dereference.

Fixes: a8ea8bdd9df9 ("lib/mpi: Extend the MPI library")
Signed-off-by: Zizhuang Deng <sunsetdzz@gmail.com>
Reviewed-by: Tianjia Zhang <tianjia.zhang@linux.alibaba.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/mlx5: Set command entry semaphore up once got index free
Moshe Shemesh [Sun, 5 Dec 2021 10:07:49 +0000 (12:07 +0200)]
net/mlx5: Set command entry semaphore up once got index free

[ Upstream commit 8e715cd613a1e872b9d918e912d90b399785761a ]

Avoid a race where command work handler may fail to allocate command
entry index, by holding the command semaphore down till command entry
index is being freed.

Fixes: 410bd754cd73 ("net/mlx5: Add retry mechanism to the command entry index allocation")
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Eran Ben Elisha <eranbe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/mlx5e: Sync VXLAN udp ports during uplink representor profile change
Maor Dickman [Mon, 3 Jan 2022 13:04:18 +0000 (15:04 +0200)]
net/mlx5e: Sync VXLAN udp ports during uplink representor profile change

[ Upstream commit 07f6dc4024ea1d2314b9c8b81fd4e492864fcca1 ]

Currently during NIC profile disablement all VXLAN udp ports offloaded to the
HW are flushed and during its enablement the driver send notification to
the stack to inform the core that the entire UDP tunnel port state has been
lost, uplink representor doesn't have the same behavior which can cause
VXLAN udp ports offload to be in bad state while moving between modes while
VXLAN interface exist.

Fixed by aligning the uplink representor profile behavior to the NIC behavior.

Fixes: 84db66124714 ("net/mlx5e: Move set vxlan nic info to profile init")
Signed-off-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/mlx5: Fix access to sf_dev_table on allocation failure
Shay Drory [Thu, 30 Dec 2021 06:54:08 +0000 (08:54 +0200)]
net/mlx5: Fix access to sf_dev_table on allocation failure

[ Upstream commit a1c7c49c2091926962f8c1c866d386febffec5d8 ]

Even when SF devices are supported, the SF device table allocation
can still fail.
In such case mlx5_sf_dev_supported still reports true, but SF device
table is invalid. This can result in NULL table access.

Hence, fix it by adding NULL table check.

Fixes: 1958fc2f0712 ("net/mlx5: SF, Add auxiliary device driver")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/mlx5e: Fix matching on modified inner ip_ecn bits
Paul Blakey [Wed, 16 Jun 2021 07:55:56 +0000 (10:55 +0300)]
net/mlx5e: Fix matching on modified inner ip_ecn bits

[ Upstream commit b6dfff21a170af5c695ebaa153b7f5e297ddca03 ]

Tunnel device follows RFC 6040, and during decapsulation inner
ip_ecn might change depending on inner and outer ip_ecn as follows:

 +---------+----------------------------------------+
 |Arriving |         Arriving Outer Header          |
 |   Inner +---------+---------+---------+----------+
 |  Header | Not-ECT | ECT(0)  | ECT(1)  |   CE     |
 +---------+---------+---------+---------+----------+
 | Not-ECT | Not-ECT | Not-ECT | Not-ECT | <drop>   |
 |  ECT(0) |  ECT(0) | ECT(0)  | ECT(1)  |   CE*    |
 |  ECT(1) |  ECT(1) | ECT(1)  | ECT(1)* |   CE*    |
 |    CE   |   CE    |  CE     | CE      |   CE     |
 +---------+---------+---------+---------+----------+

Cells marked above are changed from original inner packet ip_ecn value.

Tc then matches on the modified inner ip_ecn, but hw offload which
matches the inner ip_ecn value before decap, will fail.

Fix that by mapping all the cases of outer and inner ip_ecn matching,
and only supporting cases where we know inner wouldn't be changed by
decap, or in the outer ip_ecn=CE case, inner ip_ecn didn't matter.

Fixes: bcef735c59f2 ("net/mlx5e: Offload TC matching on tos/ttl for ip tunnels")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Reviewed-by: Eli Cohen <elic@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoRevert "net/mlx5e: Block offload of outer header csum for GRE tunnel"
Aya Levin [Sun, 24 Oct 2021 13:52:23 +0000 (16:52 +0300)]
Revert "net/mlx5e: Block offload of outer header csum for GRE tunnel"

[ Upstream commit 01c3fd113ef50490ffd43f78f347ef6bb008510b ]

This reverts commit 54e1217b90486c94b26f24dcee1ee5ef5372f832.

Although the NIC doesn't support offload of outer header CSUM, using
gso_partial_features allows offloading the tunnel's segmentation. The
driver relies on the stack CSUM calculation of the outer header. For
this, NETIF_F_GSO_GRE_CSUM must be a member of the device's features.

Fixes: 54e1217b9048 ("net/mlx5e: Block offload of outer header csum for GRE tunnel")
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoRevert "net/mlx5e: Block offload of outer header csum for UDP tunnels"
Aya Levin [Sun, 24 Oct 2021 08:47:41 +0000 (11:47 +0300)]
Revert "net/mlx5e: Block offload of outer header csum for UDP tunnels"

[ Upstream commit 64050cdad0983ad8060e33c3f4b5aee2366bcebd ]

This reverts commit 6d6727dddc7f93fcc155cb8d0c49c29ae0e71122.

Although the NIC doesn't support offload of outer header CSUM, using
gso_partial_features allows offloading the tunnel's segmentation. The
driver relies on the stack CSUM calculation of the outer header. For
this, NETIF_F_GSO_UDP_TUNNEL_CSUM must be a member of the device's
features.

Fixes: 6d6727dddc7f ("net/mlx5e: Block offload of outer header csum for UDP tunnels")
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/mlx5e: Don't block routes with nexthop objects in SW
Maor Dickman [Thu, 30 Dec 2021 09:20:10 +0000 (11:20 +0200)]
net/mlx5e: Don't block routes with nexthop objects in SW

[ Upstream commit 9e72a55a3c9d54b38a704bb7292d984574a81d9d ]

Routes with nexthop objects is currently not supported by multipath offload
and any attempts to use it is blocked, however this also block adding SW
routes with nexthop.

Resolve this by returning NOTIFY_DONE instead of an error which will allow such
a route to be created in SW but not offloaded.

This fix also solve an issue which block adding such routes on different devices
due to missing check if the route FIB device is one of multipath devices.

Fixes: 6a87afc072c3 ("mlx5: Fail attempts to use routes with nexthop objects")
Signed-off-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/mlx5e: Fix wrong usage of fib_info_nh when routes with nexthop objects are used
Maor Dickman [Wed, 29 Dec 2021 14:10:41 +0000 (16:10 +0200)]
net/mlx5e: Fix wrong usage of fib_info_nh when routes with nexthop objects are used

[ Upstream commit 885751eb1b01d276e38f57d78c583e4ce006c5ed ]

Creating routes with nexthop objects while in switchdev mode leads to access to
un-allocated memory and trigger bellow call trace due to hitting WARN_ON.
This is caused due to illegal usage of fib_info_nh in TC tunnel FIB event handling to
resolve the FIB device while fib_info built in with nexthop.

Fixed by ignoring attempts to use nexthop objects with routes until support can be
properly added.

WARNING: CPU: 1 PID: 1724 at include/net/nexthop.h:468 mlx5e_tc_tun_fib_event+0x448/0x570 [mlx5_core]
CPU: 1 PID: 1724 Comm: ip Not tainted 5.15.0_for_upstream_min_debug_2021_11_09_02_04 #1
RIP: 0010:mlx5e_tc_tun_fib_event+0x448/0x570 [mlx5_core]
RSP: 0018:ffff8881349f7910 EFLAGS: 00010202
RAX: ffff8881492f1980 RBX: ffff8881349f79e8 RCX: 0000000000000000
RDX: ffff8881349f79e8 RSI: 0000000000000000 RDI: 0000000000000000
RBP: ffff8881349f7950 R08: 00000000000000fe R09: 0000000000000001
R10: 0000000000000000 R11: 0000000000000000 R12: ffff88811e9d0000
R13: ffff88810eb62000 R14: ffff888106710268 R15: 0000000000000018
FS:  00007f1d5ca6e800(0000) GS:ffff88852c880000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007ffedba44ff8 CR3: 0000000129808004 CR4: 0000000000370ea0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 <TASK>
 atomic_notifier_call_chain+0x42/0x60
 call_fib_notifiers+0x21/0x40
 fib_table_insert+0x479/0x6d0
 ? try_charge_memcg+0x480/0x6d0
 inet_rtm_newroute+0x65/0xb0
 rtnetlink_rcv_msg+0x2af/0x360
 ? page_add_file_rmap+0x13/0x130
 ? do_set_pte+0xcd/0x120
 ? rtnl_calcit.isra.0+0x120/0x120
 netlink_rcv_skb+0x4e/0xf0
 netlink_unicast+0x1ee/0x2b0
 netlink_sendmsg+0x22e/0x460
 sock_sendmsg+0x33/0x40
 ____sys_sendmsg+0x1d1/0x1f0
 ___sys_sendmsg+0xab/0xf0
 ? __mod_memcg_lruvec_state+0x40/0x60
 ? __mod_lruvec_page_state+0x95/0xd0
 ? page_add_new_anon_rmap+0x4e/0xf0
 ? __handle_mm_fault+0xec6/0x1470
 __sys_sendmsg+0x51/0x90
 ? internal_get_user_pages_fast+0x480/0xa10
 do_syscall_64+0x3d/0x90
 entry_SYSCALL_64_after_hwframe+0x44/0xae

Fixes: 8914add2c9e5 ("net/mlx5e: Handle FIB events to update tunnel endpoint device")
Signed-off-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/mlx5e: Fix page DMA map/unmap attributes
Aya Levin [Thu, 23 Dec 2021 12:38:28 +0000 (14:38 +0200)]
net/mlx5e: Fix page DMA map/unmap attributes

[ Upstream commit 0b7cfa4082fbf550595bc0e40f05614bd83bf0cd ]

Driver initiates DMA sync, hence it may skip CPU sync. Add
DMA_ATTR_SKIP_CPU_SYNC as input attribute both to dma_map_page and
dma_unmap_page to avoid redundant sync with the CPU.
When forcing the device to work with SWIOTLB, the extra sync might cause
data corruption. The driver unmaps the whole page while the hardware
used just a part of the bounce buffer. So syncing overrides the entire
page with bounce buffer that only partially contains real data.

Fixes: bc77b240b3c5 ("net/mlx5e: Add fragmented memory support for RX multi packet WQE")
Fixes: db05815b36cb ("net/mlx5e: Add XSK zero-copy support")
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agox86, sched: Fix undefined reference to init_freq_invariance_cppc() build error
Huang Rui [Thu, 6 Jan 2022 07:43:06 +0000 (15:43 +0800)]
x86, sched: Fix undefined reference to init_freq_invariance_cppc() build error

[ Upstream commit 6c4ab1b86dac3954d15c00c1a6396d60a1023fab ]

The init_freq_invariance_cppc function is implemented in smpboot and depends on
CONFIG_SMP.

  MODPOST vmlinux.symvers
  MODINFO modules.builtin.modinfo
  GEN     modules.builtin
  LD      .tmp_vmlinux.kallsyms1
ld: drivers/acpi/cppc_acpi.o: in function `acpi_cppc_processor_probe':
/home/ray/brahma3/linux/drivers/acpi/cppc_acpi.c:819: undefined reference to `init_freq_invariance_cppc'
make: *** [Makefile:1161: vmlinux] Error 1

See https://lore.kernel.org/lkml/484af487-7511-647e-5c5b-33d4429acdec@infradead.org/.

Fixes: 41ea667227ba ("x86, sched: Calculate frequency invariance for AMD systems")
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Huang Rui <ray.huang@amd.com>
[ rjw: Subject edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoserial: stm32: move tx dma terminate DMA to shutdown
Valentin Caron [Tue, 4 Jan 2022 18:24:42 +0000 (19:24 +0100)]
serial: stm32: move tx dma terminate DMA to shutdown

[ Upstream commit 56a23f9319e86e1d62a109896e2c7e52c414e67d ]

Terminate DMA transaction and clear CR3_DMAT when shutdown is requested,
instead of when remove is requested. If DMA transfer is not stopped in
shutdown ops, driver will fail to start a new DMA transfer after next
startup ops.

Fixes: 3489187204eb ("serial: stm32: adding dma support")
Signed-off-by: Erwan Le Ray <erwan.leray@foss.st.com>
Signed-off-by: Valentin Caron <valentin.caron@foss.st.com>
Link: https://lore.kernel.org/r/20220104182445.4195-2-valentin.caron@foss.st.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoserial: liteuart: fix MODULE_ALIAS
Alyssa Ross [Tue, 4 Jan 2022 13:10:28 +0000 (13:10 +0000)]
serial: liteuart: fix MODULE_ALIAS

[ Upstream commit 556172fabd226ba14b70c1740d0826a4717473dc ]

modprobe can't handle spaces in aliases.

Fixes: 1da81e5562fa ("drivers/tty/serial: add LiteUART driver")
Signed-off-by: Alyssa Ross <hi@alyssa.is>
Link: https://lore.kernel.org/r/20220104131030.1674733-1-hi@alyssa.is
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agodrivers/firmware: Add missing platform_device_put() in sysfb_create_simplefb
Miaoqian Lin [Fri, 31 Dec 2021 08:04:25 +0000 (08:04 +0000)]
drivers/firmware: Add missing platform_device_put() in sysfb_create_simplefb

[ Upstream commit 0589e8889dce8e0f0ea5bbf757f38865e2a469c1 ]

Add the missing platform_device_put() before return from
sysfb_create_simplefb() in the error handling case.

Fixes: 8633ef82f101 ("drivers/firmware: consolidate EFI framebuffer setup for all arches")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20211231080431.15385-1-linmq006@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agodebugfs: lockdown: Allow reading debugfs files that are not world readable
Michal Suchanek [Tue, 4 Jan 2022 17:05:05 +0000 (18:05 +0100)]
debugfs: lockdown: Allow reading debugfs files that are not world readable

[ Upstream commit 358fcf5ddbec4e6706405847d6a666f5933a6c25 ]

When the kernel is locked down the kernel allows reading only debugfs
files with mode 444. Mode 400 is also valid but is not allowed.

Make the 444 into a mask.

Fixes: 5496197f9b08 ("debugfs: Restrict debugfs when the kernel is locked down")
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Link: https://lore.kernel.org/r/20220104170505.10248-1-msuchanek@suse.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoHID: hid-uclogic-params: Invalid parameter check in uclogic_params_frame_init_v1_butt...
JosĆ© ExpĆ³sito [Wed, 5 Jan 2022 17:29:15 +0000 (18:29 +0100)]
HID: hid-uclogic-params: Invalid parameter check in uclogic_params_frame_init_v1_buttonpad

[ Upstream commit aa320fdbbbb482c19100f51461bd0069753ce3d7 ]

The function performs a check on the hdev input parameters, however, it
is used before the check.

Initialize the udev variable after the sanity check to avoid a
possible NULL pointer dereference.

Fixes: 9614219e9310e ("HID: uclogic: Extract tablet parameter discovery into a module")
Addresses-Coverity-ID: 1443763 ("Null pointer dereference")
Signed-off-by: JosĆ© ExpĆ³sito <jose.exposito89@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoHID: hid-uclogic-params: Invalid parameter check in uclogic_params_huion_init
JosĆ© ExpĆ³sito [Wed, 5 Jan 2022 17:29:14 +0000 (18:29 +0100)]
HID: hid-uclogic-params: Invalid parameter check in uclogic_params_huion_init

[ Upstream commit ff6b548afe4d9d1ff3a0f6ef79e8cbca25d8f905 ]

The function performs a check on its input parameters, however, the
hdev parameter is used before the check.

Initialize the stack variables after checking the input parameters to
avoid a possible NULL pointer dereference.

Fixes: 9614219e9310e ("HID: uclogic: Extract tablet parameter discovery into a module")
Addresses-Coverity-ID: 1443804 ("Null pointer dereference")
Signed-off-by: JosĆ© ExpĆ³sito <jose.exposito89@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoHID: hid-uclogic-params: Invalid parameter check in uclogic_params_get_str_desc
JosĆ© ExpĆ³sito [Wed, 5 Jan 2022 17:29:13 +0000 (18:29 +0100)]
HID: hid-uclogic-params: Invalid parameter check in uclogic_params_get_str_desc

[ Upstream commit 0a94131d6920916ccb6a357037c535533af08819 ]

The function performs a check on the hdev input parameters, however, it
is used before the check.

Initialize the udev variable after the sanity check to avoid a
possible NULL pointer dereference.

Fixes: 9614219e9310e ("HID: uclogic: Extract tablet parameter discovery into a module")
Addresses-Coverity-ID: 1443827 ("Null pointer dereference")
Signed-off-by: JosĆ© ExpĆ³sito <jose.exposito89@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoHID: hid-uclogic-params: Invalid parameter check in uclogic_params_init
JosĆ© ExpĆ³sito [Wed, 5 Jan 2022 17:29:12 +0000 (18:29 +0100)]
HID: hid-uclogic-params: Invalid parameter check in uclogic_params_init

[ Upstream commit f364c571a5c77e96de2d32062ff019d6b8d2e2bc ]

The function performs a check on its input parameters, however, the
hdev parameter is used before the check.

Initialize the stack variables after checking the input parameters to
avoid a possible NULL pointer dereference.

Fixes: 9614219e9310e ("HID: uclogic: Extract tablet parameter discovery into a module")
Addresses-Coverity-ID: 1443831 ("Null pointer dereference")
Signed-off-by: JosĆ© ExpĆ³sito <jose.exposito89@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agousb: gadget: u_audio: Subdevice 0 for capture ctls
Pavel Hofman [Wed, 5 Jan 2022 10:46:43 +0000 (11:46 +0100)]
usb: gadget: u_audio: Subdevice 0 for capture ctls

[ Upstream commit 601a5bc1aeef772ab1f47582fd322957799f5ab5 ]

Both capture and playback alsa devices use subdevice 0. Yet capture-side
ctls are defined for subdevice 1. The patch sets subdevice 0 for them.

Fixes: 02de698ca812 ("usb: gadget: u_audio: add bi-directional volume and mute support")
Signed-off-by: Pavel Hofman <pavel.hofman@ivitera.com>
Link: https://lore.kernel.org/r/20220105104643.90125-1-pavel.hofman@ivitera.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agousb: dwc2: gadget: initialize max_speed from params
John Keeping [Thu, 6 Jan 2022 11:57:31 +0000 (11:57 +0000)]
usb: dwc2: gadget: initialize max_speed from params

[ Upstream commit 92ef98a4caacad6d4a1490dda45d81ae5ccf5bc9 ]

DWC2 may be paired with a full-speed PHY which is not capable of
high-speed operation.  Report this correctly to the gadget core by
setting max_speed from the core parameters.

Prior to commit 5324bad66f09f ("usb: dwc2: gadget: implement
udc_set_speed()") this didn't cause the hardware to be configured
incorrectly, although the speed may have been reported incorrectly.  But
after that commit params.speed is updated based on a value passed in by
the gadget core which may set it to a faster speed than is supported by
the hardware.  Initialising the max_speed parameter ensures the speed
passed to dwc2_gadget_set_speed() will be one supported by the hardware.

Fixes: 5324bad66f09f ("usb: dwc2: gadget: implement udc_set_speed()")
Acked-by: Minas Harutyunyan <Minas.Harutyunyan@synopsys.com>
Signed-off-by: John Keeping <john@metanate.com>
Link: https://lore.kernel.org/r/20220106115731.1473909-1-john@metanate.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agousb: dwc2: do not gate off the hardware if it does not support clock gating
Dinh Nguyen [Tue, 4 Jan 2022 13:59:22 +0000 (07:59 -0600)]
usb: dwc2: do not gate off the hardware if it does not support clock gating

[ Upstream commit 34146c68083f1aef6709196b3dc888c1ceffd357 ]

We should not be clearing the HCD_FLAG_HW_ACCESSIBLE bit if the hardware
does not support clock gating.

Fixes: 50fb0c128b6e ("usb: dwc2: Add clock gating entering flow by system suspend")
Acked-by: Minas Harutyunyan <Minas.Harutyunyan@synopsys.com>
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
Link: https://lore.kernel.org/r/20220104135922.734776-1-dinguyen@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agousb: dwc3: qcom: Fix NULL vs IS_ERR checking in dwc3_qcom_probe
Miaoqian Lin [Wed, 22 Dec 2021 11:18:23 +0000 (11:18 +0000)]
usb: dwc3: qcom: Fix NULL vs IS_ERR checking in dwc3_qcom_probe

[ Upstream commit b52fe2dbb3e655eb1483000adfab68a219549e13 ]

Since the acpi_create_platform_device() function may return error
pointers, dwc3_qcom_create_urs_usb_platdev() function may return error
pointers too. Using IS_ERR_OR_NULL() to check the return value to fix this.

Fixes: c25c210f590e ("usb: dwc3: qcom: add URS Host support for sdm845 ACPI boot")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20211222111823.22887-1-linmq006@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet/smc: Reset conn->lgr when link group registration fails
Wen Gu [Thu, 6 Jan 2022 12:42:08 +0000 (20:42 +0800)]
net/smc: Reset conn->lgr when link group registration fails

[ Upstream commit 36595d8ad46d9e4c41cc7c48c4405b7c3322deac ]

SMC connections might fail to be registered in a link group due to
unable to find a usable link during its creation. As a result,
smc_conn_create() will return a failure and most resources related
to the connection won't be applied or initialized, such as
conn->abort_work or conn->lnk.

If smc_conn_free() is invoked later, it will try to access the
uninitialized resources related to the connection, thus causing
a warning or crash.

This patch tries to fix this by resetting conn->lgr to NULL if an
abnormal exit occurs in smc_lgr_register_conn(), thus avoiding the
access to uninitialized resources in smc_conn_free().

Meanwhile, the new created link group should be terminated if smc
connections can't be registered in it. So smc_lgr_cleanup_early() is
modified to take care of link group only and invoked to terminate
unusable link group by smc_conn_create(). The call to smc_conn_free()
is moved out from smc_lgr_cleanup_early() to smc_conn_abort().

Fixes: 56bc3b2094b4 ("net/smc: assign link to a new connection")
Suggested-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Acked-by: Karsten Graul <kgraul@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoBluetooth: hci_qca: Fix NULL vs IS_ERR_OR_NULL check in qca_serdev_probe
Miaoqian Lin [Fri, 24 Dec 2021 08:02:49 +0000 (08:02 +0000)]
Bluetooth: hci_qca: Fix NULL vs IS_ERR_OR_NULL check in qca_serdev_probe

[ Upstream commit 6845667146a28c09b5dfc401c1ad112374087944 ]

The function devm_gpiod_get_index() return error pointers on error.
Thus devm_gpiod_get_index_optional() could return NULL and error pointers.
The same as devm_gpiod_get_optional() function. Using IS_ERR_OR_NULL()
check to catch error pointers.

Fixes: 77131dfe ("Bluetooth: hci_qca: Replace devm_gpiod_get() with devm_gpiod_get_optional()")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoBluetooth: hci_bcm: Check for error irq
Jiasheng Jiang [Fri, 24 Dec 2021 02:53:18 +0000 (10:53 +0800)]
Bluetooth: hci_bcm: Check for error irq

[ Upstream commit b38cd3b42fba66cc538edb9cf77e07881f43f8e2 ]

For the possible failure of the platform_get_irq(), the returned irq
could be error number and will finally cause the failure of the
request_irq().
Consider that platform_get_irq() can now in certain cases return
-EPROBE_DEFER, and the consequences of letting request_irq() effectively
convert that into -EINVAL, even at probe time rather than later on.
So it might be better to check just now.

Fixes: 0395ffc1ee05 ("Bluetooth: hci_bcm: Add PM for BCM devices")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agofsl/fman: Check for null pointer after calling devm_ioremap
Jiasheng Jiang [Thu, 6 Jan 2022 10:04:10 +0000 (18:04 +0800)]
fsl/fman: Check for null pointer after calling devm_ioremap

[ Upstream commit d5a73ec96cc57cf67e51b12820fc2354e7ca46f8 ]

As the possible failure of the allocation, the devm_ioremap() may return
NULL pointer.
Take tgec_initialization() as an example.
If allocation fails, the params->base_addr will be NULL pointer and will
be assigned to tgec->regs in tgec_config().
Then it will cause the dereference of NULL pointer in set_mac_address(),
which is called by tgec_init().
Therefore, it should be better to add the sanity check after the calling
of the devm_ioremap().

Fixes: 3933961682a3 ("fsl/fman: Add FMan MAC driver")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agostaging: greybus: audio: Check null pointer
Jiasheng Jiang [Tue, 4 Jan 2022 15:06:28 +0000 (23:06 +0800)]
staging: greybus: audio: Check null pointer

[ Upstream commit 2e81948177d769106754085c3e03534e6cc1f623 ]

As the possible alloc failure of devm_kcalloc(), it could return null
pointer.
Therefore, 'strings' should be checked and return NULL if alloc fails to
prevent the dereference of the NULL pointer.
Also, the caller should also deal with the return value of the
gb_generate_enum_strings() and return -ENOMEM if returns NULL.
Moreover, because the memory allocated with devm_kzalloc() will be
freed automatically when the last reference to the device is dropped,
the 'gbe' in gbaudio_tplg_create_enum_kctl() and
gbaudio_tplg_create_enum_ctl() do not need to free manually.
But the 'control' in gbaudio_tplg_create_widget() and
gbaudio_tplg_process_kcontrols() has a specially error handle to
cleanup.
So it should be better to cleanup 'control' when fails.

Fixes: e65579e335da ("greybus: audio: topology: Enable enumerated control support")
Reviewed-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Link: https://lore.kernel.org/r/20220104150628.1987906-1-jiasheng@iscas.ac.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agorocker: fix a sleeping in atomic bug
Dan Carpenter [Thu, 6 Jan 2022 11:57:54 +0000 (14:57 +0300)]
rocker: fix a sleeping in atomic bug

[ Upstream commit 43d012123122cc69feacab55b71369f386c19566 ]

This code is holding the &ofdpa->flow_tbl_lock spinlock so it is not
allowed to sleep.  That means we have to pass the OFDPA_OP_FLAG_NOWAIT
flag to ofdpa_flow_tbl_del().

Fixes: 936bd486564a ("rocker: use FIB notifications instead of switchdev calls")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agoppp: ensure minimum packet size in ppp_write()
Eric Dumazet [Wed, 5 Jan 2022 11:48:42 +0000 (03:48 -0800)]
ppp: ensure minimum packet size in ppp_write()

[ Upstream commit 44073187990d5629804ce0627525f6ea5cfef171 ]

It seems pretty clear ppp layer assumed user space
would always be kind to provide enough data
in their write() to a ppp device.

This patch makes sure user provides at least
2 bytes.

It adds PPP_PROTO_LEN macro that could replace
in net-next many occurrences of hard-coded 2 value.

I replaced only one occurrence to ease backports
to stable kernels.

The bug manifests in the following report:

BUG: KMSAN: uninit-value in ppp_send_frame+0x28d/0x27c0 drivers/net/ppp/ppp_generic.c:1740
 ppp_send_frame+0x28d/0x27c0 drivers/net/ppp/ppp_generic.c:1740
 __ppp_xmit_process+0x23e/0x4b0 drivers/net/ppp/ppp_generic.c:1640
 ppp_xmit_process+0x1fe/0x480 drivers/net/ppp/ppp_generic.c:1661
 ppp_write+0x5cb/0x5e0 drivers/net/ppp/ppp_generic.c:513
 do_iter_write+0xb0c/0x1500 fs/read_write.c:853
 vfs_writev fs/read_write.c:924 [inline]
 do_writev+0x645/0xe00 fs/read_write.c:967
 __do_sys_writev fs/read_write.c:1040 [inline]
 __se_sys_writev fs/read_write.c:1037 [inline]
 __x64_sys_writev+0xe5/0x120 fs/read_write.c:1037
 do_syscall_x64 arch/x86/entry/common.c:51 [inline]
 do_syscall_64+0x54/0xd0 arch/x86/entry/common.c:82
 entry_SYSCALL_64_after_hwframe+0x44/0xae

Uninit was created at:
 slab_post_alloc_hook mm/slab.h:524 [inline]
 slab_alloc_node mm/slub.c:3251 [inline]
 __kmalloc_node_track_caller+0xe0c/0x1510 mm/slub.c:4974
 kmalloc_reserve net/core/skbuff.c:354 [inline]
 __alloc_skb+0x545/0xf90 net/core/skbuff.c:426
 alloc_skb include/linux/skbuff.h:1126 [inline]
 ppp_write+0x11d/0x5e0 drivers/net/ppp/ppp_generic.c:501
 do_iter_write+0xb0c/0x1500 fs/read_write.c:853
 vfs_writev fs/read_write.c:924 [inline]
 do_writev+0x645/0xe00 fs/read_write.c:967
 __do_sys_writev fs/read_write.c:1040 [inline]
 __se_sys_writev fs/read_write.c:1037 [inline]
 __x64_sys_writev+0xe5/0x120 fs/read_write.c:1037
 do_syscall_x64 arch/x86/entry/common.c:51 [inline]
 do_syscall_64+0x54/0xd0 arch/x86/entry/common.c:82
 entry_SYSCALL_64_after_hwframe+0x44/0xae

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: linux-ppp@vger.kernel.org
Reported-by: syzbot <syzkaller@googlegroups.com>
Acked-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet: fix SOF_TIMESTAMPING_BIND_PHC to work with multiple sockets
Miroslav Lichvar [Wed, 5 Jan 2022 10:33:26 +0000 (11:33 +0100)]
net: fix SOF_TIMESTAMPING_BIND_PHC to work with multiple sockets

[ Upstream commit 007747a984ea5e895b7d8b056b24ebf431e1e71d ]

When multiple sockets using the SOF_TIMESTAMPING_BIND_PHC flag received
a packet with a hardware timestamp (e.g. multiple PTP instances in
different PTP domains using the UDPv4/v6 multicast or L2 transport),
the timestamps received on some sockets were corrupted due to repeated
conversion of the same timestamp (by the same or different vclocks).

Fix ptp_convert_timestamp() to not modify the shared skb timestamp
and return the converted timestamp as a ktime_t instead. If the
conversion fails, return 0 to not confuse the application with
timestamps corresponding to an unexpected PHC.

Fixes: d7c088265588 ("net: socket: support hardware timestamp conversion to PHC bound")
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Cc: Yangbo Lu <yangbo.lu@nxp.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonetfilter: nft_set_pipapo: allocate pcpu scratch maps on clone
Florian Westphal [Wed, 5 Jan 2022 13:19:54 +0000 (14:19 +0100)]
netfilter: nft_set_pipapo: allocate pcpu scratch maps on clone

[ Upstream commit 23c54263efd7cb605e2f7af72717a2a951999217 ]

This is needed in case a new transaction is made that doesn't insert any
new elements into an already existing set.

Else, after second 'nft -f ruleset.txt', lookups in such a set will fail
because ->lookup() encounters raw_cpu_ptr(m->scratch) == NULL.

For the initial rule load, insertion of elements takes care of the
allocation, but for rule reloads this isn't guaranteed: we might not
have additions to the set.

Fixes: 3c4287f62044a90e ("nf_tables: Add set type for arbitrary concatenation of ranges")
Reported-by: etkaar <lists.netfilter.org@prvy.eu>
Signed-off-by: Florian Westphal <fw@strlen.de>
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonetfilter: nft_payload: do not update layer 4 checksum when mangling fragments
Pablo Neira Ayuso [Wed, 5 Jan 2022 15:09:57 +0000 (16:09 +0100)]
netfilter: nft_payload: do not update layer 4 checksum when mangling fragments

[ Upstream commit 4e1860a3863707e8177329c006d10f9e37e097a8 ]

IP fragments do not come with the transport header, hence skip bogus
layer 4 checksum updates.

Fixes: 1814096980bb ("netfilter: nft_payload: layer 4 checksum adjustment for pseudoheader fields")
Reported-and-tested-by: Steffen Weinreich <steve@weinreich.org>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobpf: Fix SO_RCVBUF/SO_SNDBUF handling in _bpf_setsockopt().
Kuniyuki Iwashima [Tue, 4 Jan 2022 01:31:48 +0000 (10:31 +0900)]
bpf: Fix SO_RCVBUF/SO_SNDBUF handling in _bpf_setsockopt().

[ Upstream commit 04c350b1ae6bdb12b84009a4d0bf5ab4e621c47b ]

The commit 4057765f2dee ("sock: consistent handling of extreme
SO_SNDBUF/SO_RCVBUF values") added a change to prevent underflow
in setsockopt() around SO_SNDBUF/SO_RCVBUF.

This patch adds the same change to _bpf_setsockopt().

Fixes: 4057765f2dee ("sock: consistent handling of extreme SO_SNDBUF/SO_RCVBUF values")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.co.jp>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220104013153.97906-2-kuniyu@amazon.co.jp
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobpf: Fix verifier support for validation of async callbacks
Kris Van Hees [Wed, 5 Jan 2022 21:01:50 +0000 (16:01 -0500)]
bpf: Fix verifier support for validation of async callbacks

[ Upstream commit a5bebc4f00dee47113eed48098c68e88b5ba70e8 ]

Commit bfc6bb74e4f1 ("bpf: Implement verifier support for validation of async callbacks.")
added support for BPF_FUNC_timer_set_callback to
the __check_func_call() function.  The test in __check_func_call() is
flaweed because it can mis-interpret a regular BPF-to-BPF pseudo-call
as a BPF_FUNC_timer_set_callback callback call.

Consider the conditional in the code:

if (insn->code == (BPF_JMP | BPF_CALL) &&
    insn->imm == BPF_FUNC_timer_set_callback) {

The BPF_FUNC_timer_set_callback has value 170.  This means that if you
have a BPF program that contains a pseudo-call with an instruction delta
of 170, this conditional will be found to be true by the verifier, and
it will interpret the pseudo-call as a callback.  This leads to a mess
with the verification of the program because it makes the wrong
assumptions about the nature of this call.

Solution: include an explicit check to ensure that insn->src_reg == 0.
This ensures that calls cannot be mis-interpreted as an async callback
call.

Fixes: bfc6bb74e4f1 ("bpf: Implement verifier support for validation of async callbacks.")
Signed-off-by: Kris Van Hees <kris.van.hees@oracle.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220105210150.GH1559@oracle.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobpf: Don't promote bogus looking registers after null check.
Daniel Borkmann [Wed, 5 Jan 2022 19:35:13 +0000 (11:35 -0800)]
bpf: Don't promote bogus looking registers after null check.

[ Upstream commit e60b0d12a95dcf16a63225cead4541567f5cb517 ]

If we ever get to a point again where we convert a bogus looking <ptr>_or_null
typed register containing a non-zero fixed or variable offset, then lets not
reset these bounds to zero since they are not and also don't promote the register
to a <ptr> type, but instead leave it as <ptr>_or_null. Converting to a unknown
register could be an avenue as well, but then if we run into this case it would
allow to leak a kernel pointer this way.

Fixes: f1174f77b50c ("bpf/verifier: rework value tracking")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobpf, sockmap: Fix double bpf_prog_put on error case in map_link
John Fastabend [Tue, 4 Jan 2022 21:46:45 +0000 (13:46 -0800)]
bpf, sockmap: Fix double bpf_prog_put on error case in map_link

[ Upstream commit 218d747a4142f281a256687bb513a135c905867b ]

sock_map_link() is called to update a sockmap entry with a sk. But, if the
sock_map_init_proto() call fails then we return an error to the map_update
op against the sockmap. In the error path though we need to cleanup psock
and dec the refcnt on any programs associated with the map, because we
refcnt them early in the update process to ensure they are pinned for the
psock. (This avoids a race where user deletes programs while also updating
the map with new socks.)

In current code we do the prog refcnt dec explicitely by calling
bpf_prog_put() when the program was found in the map. But, after commit
'38207a5e81230' in this error path we've already done the prog to psock
assignment so the programs have a reference from the psock as well. This
then causes the psock tear down logic, invoked by sk_psock_put() in the
error path, to similarly call bpf_prog_put on the programs there.

To be explicit this logic does the prog->psock assignment:

  if (msg_*)
    psock_set_prog(...)

Then the error path under the out_progs label does a similar check and
dec with:

  if (msg_*)
     bpf_prog_put(...)

And the teardown logic sk_psock_put() does ...

  psock_set_prog(msg_*, NULL)

... triggering another bpf_prog_put(...). Then KASAN gives us this splat,
found by syzbot because we've created an inbalance between bpf_prog_inc and
bpf_prog_put calling put twice on the program.

  BUG: KASAN: vmalloc-out-of-bounds in __bpf_prog_put kernel/bpf/syscall.c:1812 [inline]
  BUG: KASAN: vmalloc-out-of-bounds in __bpf_prog_put kernel/bpf/syscall.c:1812 [inline] kernel/bpf/syscall.c:1829
  BUG: KASAN: vmalloc-out-of-bounds in bpf_prog_put+0x8c/0x4f0 kernel/bpf/syscall.c:1829 kernel/bpf/syscall.c:1829
  Read of size 8 at addr ffffc90000e76038 by task syz-executor020/3641

To fix clean up error path so it doesn't try to do the bpf_prog_put in the
error path once progs are assigned then it relies on the normal psock
tear down logic to do complete cleanup.

For completness we also cover the case whereh sk_psock_init_strp() fails,
but this is not expected because it indicates an incorrect socket type
and should be caught earlier.

Fixes: 38207a5e8123 ("bpf, sockmap: Attach map progs to psock early for feature probes")
Reported-by: syzbot+bb73e71cf4b8fd376a4f@syzkaller.appspotmail.com
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20220104214645.290900-1-john.fastabend@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agobpf, sockmap: Fix return codes from tcp_bpf_recvmsg_parser()
John Fastabend [Tue, 4 Jan 2022 20:59:18 +0000 (12:59 -0800)]
bpf, sockmap: Fix return codes from tcp_bpf_recvmsg_parser()

[ Upstream commit 5b2c5540b8110eea0d67a78fb0ddb9654c58daeb ]

Applications can be confused slightly because we do not always return the
same error code as expected, e.g. what the TCP stack normally returns. For
example on a sock err sk->sk_err instead of returning the sock_error we
return EAGAIN. This usually means the application will 'try again'
instead of aborting immediately. Another example, when a shutdown event
is received we should immediately abort instead of waiting for data when
the user provides a timeout.

These tend to not be fatal, applications usually recover, but introduces
bogus errors to the user or introduces unexpected latency. Before
'c5d2177a72a16' we fell back to the TCP stack when no data was available
so we managed to catch many of the cases here, although with the extra
latency cost of calling tcp_msg_wait_data() first.

To fix lets duplicate the error handling in TCP stack into tcp_bpf so
that we get the same error codes.

These were found in our CI tests that run applications against sockmap
and do longer lived testing, at least compared to test_sockmap that
does short-lived ping/pong tests, and in some of our test clusters
we deploy.

Its non-trivial to do these in a shorter form CI tests that would be
appropriate for BPF selftests, but we are looking into it so we can
ensure this keeps working going forward. As a preview one idea is to
pull in the packetdrill testing which catches some of this.

Fixes: c5d2177a72a16 ("bpf, sockmap: Fix race in ingress receive verdict with redirect to self")
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20220104205918.286416-1-john.fastabend@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonetfilter: ipt_CLUSTERIP: fix refcount leak in clusterip_tg_check()
Xin Xiong [Thu, 23 Dec 2021 02:48:12 +0000 (10:48 +0800)]
netfilter: ipt_CLUSTERIP: fix refcount leak in clusterip_tg_check()

[ Upstream commit d94a69cb2cfa77294921aae9afcfb866e723a2da ]

The issue takes place in one error path of clusterip_tg_check(). When
memcmp() returns nonzero, the function simply returns the error code,
forgetting to decrease the reference count of a clusterip_config
object, which is bumped earlier by clusterip_config_find_get(). This
may incur reference count leak.

Fix this issue by decrementing the refcount of the object in specific
error path.

Fixes: 06aa151ad1fc74 ("netfilter: ipt_CLUSTERIP: check MAC address when duplicate config is set")
Signed-off-by: Xin Xiong <xiongx18@fudan.edu.cn>
Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2 years agonet: dsa: fix incorrect function pointer check for MRP ring roles
Vladimir Oltean [Wed, 5 Jan 2022 13:18:11 +0000 (15:18 +0200)]
net: dsa: fix incorrect function pointer check for MRP ring roles

[ Upstream commit ff91e1b68490b97c18c649b769618815eb945f11 ]

The cross-chip notifier boilerplate code meant to check the presence of
ds->ops->port_mrp_add_ring_role before calling it, but checked
ds->ops->port_mrp_add instead, before calling
ds->ops->port_mrp_add_ring_role.

Therefore, a driver which implements one operation but not the other
would trigger a NULL pointer dereference.

There isn't any such driver in DSA yet, so there is no reason to
backport the change. Issue found through code inspection.

Cc: Horatiu Vultur <horatiu.vultur@microchip.com>
Fixes: c595c4330da0 ("net: dsa: add MRP support")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>