dev_coredumpm() creates a devcoredump device and adds it
to the core kernel framework which eventually end up
sending uevent to the user space and later creates a
symbolic link to the failed device. An application
running in userspace may be interested in this symbolic
link to get the name of the failed device.
In a issue scenario, once uevent sent to the user space
it start reading '/sys/class/devcoredump/devcdX/failing_device'
to get the actual name of the device which might not been
created and it is in its path of creation.
To fix this, suppress sending uevent till the failing device
symbolic link gets created and send uevent once symbolic
link is created successfully.
In following scenario(diagram), when one thread X running dev_coredumpm()
adds devcd device to the framework which sends uevent notification to
userspace and another thread Y reads this uevent and call to
devcd_data_write() which eventually try to delete the queued timer that
is not initialized/queued yet.
So, debug object reports some warning and in the meantime, timer is
initialized and queued from X path. and from Y path, it gets reinitialized
again and timer->entry.pprev=NULL and try_to_grab_pending() stucks.
To fix this, introduce mutex and a boolean flag to serialize the behaviour.
cpu0(X) cpu1(Y)
dev_coredump() uevent sent to user space
device_add() ======================> user space process Y reads the
uevents writes to devcd fd
which results into writes to
devcd_data_write()
mod_delayed_work()
try_to_grab_pending()
del_timer()
debug_assert_init()
INIT_DELAYED_WORK()
schedule_delayed_work()
debug_object_fixup()
timer_fixup_assert_init()
timer_setup()
do_init_timer()
/*
Above call reinitializes
the timer to
timer->entry.pprev=NULL
and this will be checked
later in timer_pending() call.
*/
timer_pending()
!hlist_unhashed_lockless(&timer->entry)
!h->pprev
/*
del_timer() checks h->pprev and finds
it to be NULL due to which
try_to_grab_pending() stucks.
*/
Deduplication isn't supported on cifs, but cifs doesn't reject it, instead
treating it as extent duplication/cloning. This can cause generic/304 to go
silly and run for hours on end.
Fix cifs to indicate EOPNOTSUPP if REMAP_FILE_DEDUP is set in
->remap_file_range().
Note that it's unclear whether or not commit b073a08016a1 is meant to cause
cifs to return an error if REMAP_FILE_DEDUP.
Fixes: b073a08016a1 ("cifs: fix that return -EINVAL when do dedupe operation") Cc: stable@vger.kernel.org Suggested-by: Dave Chinner <david@fromorbit.com>
cc: Xiaoli Feng <fengxiaoli0714@gmail.com>
cc: Shyam Prasad N <nspmangalore@gmail.com>
cc: Rohith Surabattula <rohiths.msft@gmail.com>
cc: Jeff Layton <jlayton@kernel.org>
cc: Darrick Wong <darrick.wong@oracle.com>
cc: fstests@vger.kernel.org
cc: linux-cifs@vger.kernel.org
cc: linux-fsdevel@vger.kernel.org Link: https://lore.kernel.org/r/3876191.1701555260@warthog.procyon.org.uk/ Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Steve French <stfrench@microsoft.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
As pointed out by many, the disk_super structure is NOT initialized
before it is dereferenced in the function
fs/btrfs/disk-io.c:open_ctree() that this commit adds, so something went
wrong here.
vgabios is passed from firmware to kernel on Loongson64 systems.
Sane firmware will keep this pointer in reserved memory space
passed from the firmware but insane firmware keeps it in low
memory before kernel entry that is not reserved.
Previously kernel won't try to allocate memory from low memory
before kernel entry on boot, but after converting to memblock
it will do that.
Fix by resversing those memory on early boot.
Cc: stable@vger.kernel.org Fixes: a94e4f24ec83 ("MIPS: init: Drop boot_mem_map") Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com> Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In general, activating long mode involves setting the EFER_LME bit in
the EFER register and then enabling the X86_CR0_PG bit in the CR0
register. At this point, the EFER_LMA bit will be set automatically by
hardware.
In the case of SVM/SEV guests where writes to CR0 are intercepted, it's
necessary for the host to set EFER_LMA on behalf of the guest since
hardware does not see the actual CR0 write.
In the case of SEV-ES guests where writes to CR0 are trapped instead of
intercepted, the hardware *does* see/record the write to CR0 before
exiting and passing the value on to the host, so as part of enabling
SEV-ES support commit f1c6366e3043 ("KVM: SVM: Add required changes to
support intercepts under SEV-ES") dropped special handling of the
EFER_LMA bit with the understanding that it would be set automatically.
However, since the guest never explicitly sets the EFER_LMA bit, the
host never becomes aware that it has been set. This becomes problematic
when userspace tries to get/set the EFER values via
KVM_GET_SREGS/KVM_SET_SREGS, since the EFER contents tracked by the host
will be missing the EFER_LMA bit, and when userspace attempts to pass
the EFER value back via KVM_SET_SREGS it will fail a sanity check that
asserts that EFER_LMA should always be set when X86_CR0_PG and EFER_LME
are set.
Fix this by always inferring the value of EFER_LMA based on X86_CR0_PG
and EFER_LME, regardless of whether or not SEV-ES is enabled.
Fixes: f1c6366e3043 ("KVM: SVM: Add required changes to support intercepts under SEV-ES") Reported-by: Peter Gonda <pgonda@google.com> Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20210507165947.2502412-2-seanjc@google.com>
[A two year old patch that was revived after we noticed the failure in
KVM_SET_SREGS and a similar patch was posted by Michael Roth. This is
Sean's patch, but with Michael's more complete commit message. - Paolo] Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the CMMA state needs to be reset, the no-dat bit also needs to be
reset. Failure to do so could cause issues in the guest, since the
guest expects the bit to be cleared after a reset.
Starting RX DMA on THRI interrupt is too early because TX may not have
finished yet.
This change is inspired by commit 90b8596ac460 ("serial: 8250: Prevent
starting up DMA Rx on THRI interrupt") and fixes DMA issues I had with
an AM62 SoC that is using the 8250 OMAP variant.
Cc: stable@vger.kernel.org Fixes: c26389f998a8 ("serial: 8250: 8250_omap: Add DMA support for UARTs on K3 SoCs") Signed-off-by: Ronald Wahl <ronald.wahl@raritan.com> Reviewed-by: Vignesh Raghavendra <vigneshr@ti.com> Link: https://lore.kernel.org/r/20231101171431.16495-1-rwahl@gmx.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This fixes commit 439c7183e5b9 ("serial: 8250: 8250_omap: Disable RX
interrupt after DMA enable") which unfortunately set the
UART_HAS_RHR_IT_DIS bit in the UART_OMAP_IER2 register and never
cleared it.
This device has a silicon bug that makes it report a timeout interrupt
but no data in the FIFO.
The datasheet states the following in the errata section 18.1.4:
"If the host reads the receive FIFO at the same time as a
time-out interrupt condition happens, the host might read 0xCC
(time-out) in the Interrupt Indication Register (IIR), but bit 0
of the Line Status Register (LSR) is not set (means there is no
data in the receive FIFO)."
The errata description seems to indicate it concerns only polled mode of
operation when reading bit 0 of the LSR register. However, tests have
shown and NXP has confirmed that the RXLVL register also yields 0 when
the bug is triggered, and hence the IRQ driven implementation in this
driver is equally affected.
This bug has hit us on production units and when it does, sc16is7xx_irq()
would spin forever because sc16is7xx_port_irq() keeps seeing an
interrupt in the IIR register that is not cleared because the driver
does not call into sc16is7xx_handle_rx() unless the RXLVL register
reports at least one byte in the FIFO.
Fix this by always reading one byte from the FIFO when this condition
is detected in order to clear the interrupt. This approach was
confirmed to be correct by NXP through their support channels.
Tested by: Hugo Villeneuve <hvilleneuve@dimonoff.com>
Since there is no guarantee that the memory returned by
dma_alloc_coherent() is associated with a 'struct page', using the
architecture specific phys_to_page() is wrong, but using
virt_to_page() would be as well.
Stop using sg lists altogether and just use the *_single() functions
instead. This also simplifies the code a bit since the scatterlists in
this driver always have only one entry anyway.
https://lore.kernel.org/lkml/86db0fe5-930d-4cbb-bd7d-03367da38951@app.fastmail.com/
Use consistent names for dma buffers
gc: Add a commit log from the initial thread:
https://lore.kernel.org/lkml/86db0fe5-930d-4cbb-bd7d-03367da38951@app.fastmail.com/
Use consistent names for dma buffers
When typec_altmode_put_partner is called by a plug altmode upon release,
the port altmode the plug belongs to will not remove its reference to the
plug. The check to see if the altmode being released evaluates against the
released altmode's partner instead of the calling altmode itself, so change
adev in typec_altmode_put_partner to properly refer to the altmode being
released.
typec_altmode_set_partner is not run for port altmodes, so also add a check
in typec_altmode_release to prevent typec_altmode_put_partner() calls on
port altmode release.
Fixes: 8a37d87d72f0 ("usb: typec: Bus type for alternate modes") Cc: stable@vger.kernel.org Signed-off-by: RD Babiera <rdbabiera@google.com> Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com> Link: https://lore.kernel.org/r/20231129192349.1773623-2-rdbabiera@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Enabling runtime pm as default for all AMD xHC 1.1 controllers caused
regression. An initial attempt to fix those was done in commit a5d6264b638e
("xhci: Enable RPM on controllers that support low-power states") but new
issues are still seen.
Revert this to get those AMD xHC 1.1 systems working
This patch went to stable an needs to be reverted from there as well.
The commit 89ff3dfac604 ("usb: gadget: f_hid: fix f_hidg lifetime vs
cdev") has introduced a bug that leads to hid device corruption after
the replug operation.
Reverse device managed memory allocation for the report descriptor
to fix the issue.
Tested:
This change was tested on the AMD EthanolX CRB server with the BMC
based on the OpenBMC distribution. The BMC provides KVM functionality
via the USB gadget device:
- before: KVM page refresh results in a broken USB device,
- after: KVM page refresh works without any issues.
WARNING: Block comments use * on subsequent lines
WARNING: Block comments use a trailing */ on a separate line
WARNING: suspect code indent for conditional statements (8, 12)
WARNING: braces {} are not necessary for single statement blocks
Cc: Christian König <christian.koenig@amd.com> Cc: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Srinivasan Shanmugam <srinivasan.shanmugam@amd.com> Reviewed-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Stable-dep-of: c6df7f313794 ("drm/amdgpu: correct the amdgpu runtime dereference usage count") Signed-off-by: Sasha Levin <sashal@kernel.org>
If gpio_set_transitory() fails, we should free the GPIO again. Most
notably, the flag FLAG_REQUESTED has previously been set in
gpiod_request_commit(), and should be reset on failure.
To my knowledge, this does not affect any current users, since the
gpio_set_transitory() mainly returns 0 and -ENOTSUPP, which is converted
to 0. However the gpio_set_transitory() function calles the .set_config()
function of the corresponding GPIO chip and there are some GPIO drivers in
which some (unlikely) branches return other values like -EPROBE_DEFER,
and -EINVAL. In these cases, the above mentioned FLAG_REQUESTED would not
be reset, which results in the pin being blocked until the next reboot.
Budimir noted that perf_event_validate_size() only checks the size of
the newly added event, even though the sizes of all existing events
can also change due to not all events having the same read_format.
When we attach the new event, perf_group_attach(), we do re-compute
the size for all events.
Sometimes we want to know an accurate number of samples even if it's
lost. Currenlty PERF_RECORD_LOST is generated for a ring-buffer which
might be shared with other events. So it's hard to know per-event
lost count.
Add event->lost_samples field and PERF_FORMAT_LOST to retrieve it from
userspace.
Original-patch-by: Jiri Olsa <jolsa@redhat.com> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lkml.kernel.org/r/20220616180623.1358843-1-namhyung@kernel.org
Stable-dep-of: 382c27f4ed28 ("perf: Fix perf_event_validate_size()") Signed-off-by: Sasha Levin <sashal@kernel.org>
Align the name of thermal zone node to dtschema to fix warnings like:
arch/arm64/boot/dts/mediatek/mt8173-elm.dt.yaml:
thermal-zones: 'cpu_thermal' does not match any of the regexes: '^[a-zA-Z][a-zA-Z0-9\\-]{1,12}-thermal$', 'pinctrl-[0-9]+'
Commit 1ea0d3b46798 ("platform/x86: asus-wmi: Simplify tablet-mode-switch
handling") unified the asus-wmi tablet-switch handling, but it did not take
into account that the value returned for the kbd_dock_devid WMI method is
inverted where as the other ones are not inverted.
This causes asus-wmi to report an inverted tablet-switch state for devices
which use the kbd_dock_devid, which causes libinput to ignore touchpad
events while the affected T10x model 2-in-1s are docked.
Add inverting of the return value in the kbd_dock_devid case to fix this.
Otherwise set elements can be deactivated twice which will cause a crash.
Reported-by: Xingyuan Mo <hdthky0@gmail.com> Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges") Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
File reference cycles have caused lots of problems for io_uring
in the past, and it still doesn't work exactly right and races with
unix_stream_read_generic(). The safest fix would be to completely
disallow sending io_uring files via sockets via SCM_RIGHT, so there
are no possible cycles invloving registered files and thus rendering
SCM accounting on the io_uring side unnecessary.
syzbot reports oops in lockdep's __lock_acquire(), called from
__pte_offset_map_lock() called from filemap_map_pages(); or when I run the
repro, the oops comes in pmd_install(), called from filemap_map_pmd()
called from filemap_map_pages(), just before the __pte_offset_map_lock().
The problem is that filemap_map_pmd() has been assuming that when it finds
pmd_none(), a page table has already been prepared in prealloc_pte; and
indeed do_fault_around() has been careful to preallocate one there, when
it finds pmd_none(): but what if *pmd became none in between?
My 6.6 mods in mm/khugepaged.c, avoiding mmap_lock for write, have made it
easy for *pmd to be cleared while servicing a page fault; but even before
those, a huge *pmd might be zapped while a fault is serviced.
The difference in symptomatic stack traces comes from the "memory model"
in use: pmd_install() uses pmd_populate() uses page_to_pfn(): in some
models that is strict, and will oops on the NULL prealloc_pte; in other
models, it will construct a bogus value to be populated into *pmd, then
__pte_offset_map_lock() oops when trying to access split ptlock pointer
(or some other symptom in normal case of ptlock embedded not pointer).
When FIFO reaches near full state, device will issue pause frame.
If pause slot is enabled(set to 1), in this time, device will issue
pause frame only once. But if pause slot is disabled(set to 0), device
will keep sending pause frames until FIFO reaches near empty state.
When pause slot is disabled, if there is no one to handle receive
packets, device FIFO will reach near full state and keep sending
pause frames. That will impact entire local area network.
This issue can be reproduced in Chromebox (not Chromebook) in
developer mode running a test image (and v5.10 kernel):
1) ping -f $CHROMEBOX (from workstation on same local network)
2) run "powerd_dbus_suspend" from command line on the $CHROMEBOX
3) ping $ROUTER (wait until ping fails from workstation)
Takes about ~20-30 seconds after step 2 for the local network to
stop working.
Fix this issue by enabling pause slot to only send pause frame once
when FIFO reaches near full state.
Fixes: f1bce4ad2f1c ("r8169: add support for RTL8125") Reported-by: Grant Grundler <grundler@chromium.org> Tested-by: Grant Grundler <grundler@chromium.org> Cc: stable@vger.kernel.org Signed-off-by: ChunHao Lin <hau@realtek.com> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Reviewed-by: Heiner Kallweit <hkallweit1@gmail.com> Link: https://lore.kernel.org/r/20231129155350.5843-1-hau@realtek.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
When the ring buffer is being resized, it can cause side effects to the
running tracer. For instance, there's a race with irqsoff tracer that
swaps individual per cpu buffers between the main buffer and the snapshot
buffer. The resize operation modifies the main buffer and then the
snapshot buffer. If a swap happens in between those two operations it will
break the tracer.
Simply stop the running tracer before resizing the buffers and enable it
again when finished.
Link: https://lkml.kernel.org/r/20231205220010.748996423@goodmis.org Cc: stable@vger.kernel.org Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Cc: Andrew Morton <akpm@linux-foundation.org> Fixes: 3928a8a2d9808 ("ftrace: make work with new ring buffer") Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently we can resize trace ringbuffer by writing a value into file
'buffer_size_kb', then by reading the file, we get the value that is
usually what we wrote. However, this value may be not actual size of
trace ring buffer because of the round up when doing resize in kernel,
and the actual size would be more useful.
There's a race where if an event is discarded from the ring buffer and an
interrupt were to happen at that time and insert an event, the time stamp
is still used from the discarded event as an offset. This can screw up the
timings.
If the event is going to be discarded, set the "before_stamp" to zero.
When a new event comes in, it compares the "before_stamp" with the
"write_stamp" and if they are not equal, it will insert an absolute
timestamp. This will prevent the timings from getting out of sync due to
the discarded event.
Link: https://lore.kernel.org/linux-trace-kernel/20231206100244.5130f9b3@gandalf.local.home Cc: stable@vger.kernel.org Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Fixes: 6f6be606e763f ("ring-buffer: Force before_stamp and write_stamp to be different on discard") Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Clang static analyzer complains that value stored to 'rets' is never
read.Let 'buf_len = -EOVERFLOW' to make sure we can return '-EOVERFLOW'.
Fixes: 8c8d964ce90f ("mei: move hbuf_depth from the mei device to the hw modules") Signed-off-by: Su Hui <suhui@nfschina.com> Link: https://lore.kernel.org/r/20231120095523.178385-2-suhui@nfschina.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
etm4_platform_driver (which lives in ".data" contains a reference to
etm4_remove_platform_dev(). So the latter must not be marked with __exit
which results in the function being discarded for a build with
CONFIG_CORESIGHT_SOURCE_ETM4X=y which in turn makes the remove pointer
contain invalid data.
etm4x_amba_driver referencing etm4_remove_amba() has the same issue.
Drop the __exit annotations for the two affected functions and a third
one that is called by the other two.
For reasons I don't understand this isn't catched by building with
CONFIG_DEBUG_SECTION_MISMATCH=y.
etm4_remove_dev() returned zero unconditionally. Make it return void
instead, which makes it clear in the callers that there is no error to
handle. Simplify etm4_remove_platform_dev() accordingly.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Link: https://lore.kernel.org/r/20230518201629.260672-1-u.kleine-koenig@pengutronix.de
Stable-dep-of: 348ddab81f7b ("coresight: etm4x: Remove bogous __exit annotation for some functions") Signed-off-by: Sasha Levin <sashal@kernel.org>
A transaction complete work is allocated and queued for each
transaction. Under certain conditions the work->type might be marked as
BINDER_WORK_TRANSACTION_ONEWAY_SPAM_SUSPECT to notify userspace about
potential spamming threads or as BINDER_WORK_TRANSACTION_PENDING when
the target is currently frozen.
However, these work types are not being handled in binder_release_work()
so they will leak during a cleanup. This was reported by syzkaller with
the following kmemleak dump:
The reserved memory for scp had node name "scp_mem_region" and also
without unit-address: change the name to "memory@(address)".
This fixes a unit_address_vs_reg warning.
Fix a unit_address_vs_reg warning for the USB VBUS fixed regulators
by renaming the regulator nodes from regulator@{0,1} to regulator-usb-p0
and regulator-usb-p1.
dtbs_check throws a warning at the dsi node:
Warning (avoid_unnecessary_addr_size): /soc/dsi@14014000: unnecessary #address-cells/#size-cells without "ranges" or child "reg" property
Other DTS have a panel child node with a reg, so the parent dtsi
must have the address-cells and size-cells, however this specific DT
has the panel removed, but not the cells, hence the warning above.
If panel is deleted then the cells must also be deleted since they are
tied together, as the child node in this DT does not have a reg.
Cc: stable@vger.kernel.org Fixes: cabc71b08eb5 ("arm64: dts: mt8183: Add kukui-jacuzzi-damu board") Signed-off-by: Eugen Hristev <eugen.hristev@collabora.com> Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Link: https://lore.kernel.org/r/20230814071053.5459-1-eugen.hristev@collabora.com Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Serdev recv_buf() callback is supposed to return the amount of bytes
consumed, therefore an int in between 0 and count.
Do not return negative number in case of issue, when
ssam_controller_receive_buf() returns ESHUTDOWN just returns 0, e.g. no
bytes consumed, this keep the exact same behavior as it was before.
This fixes a potential WARN in serdev-ttyport.c:ttyport_receive_buf().
Since commit 0ec7731655de ("regmap: Ensure range selector registers
are updated after cache sync") opening pcm512x based soundcards fail
with EINVAL and dmesg shows sync cache and pm_runtime_get errors:
[ 228.794676] pcm512x 1-004c: Failed to sync cache: -22
[ 228.794740] pcm512x 1-004c: ASoC: error at snd_soc_pcm_component_pm_runtime_get on pcm512x.1-004c: -22
This is caused by the cache check result leaking out into the
regcache_sync return value.
Fix this by making the check local-only, as the comment above the
regcache_read call states a non-zero return value means there's
nothing to do so the return value should not be altered.
Fixes: 0ec7731655de ("regmap: Ensure range selector registers are updated after cache sync") Cc: stable@vger.kernel.org Signed-off-by: Matthias Reichl <hias@horus.com> Link: https://lore.kernel.org/r/20231203222216.96547-1-hias@horus.com Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In some potential instances the reference count on struct packet_sock
could be saturated and cause overflows which gets the kernel a bit
confused. To prevent this, move to a 64-bit atomic reference count on
64-bit architectures to prevent the possibility of this type to overflow.
Because we can not handle saturation, using refcount_t is not possible
in this place. Maybe someday in the future if it changes it could be
used. Also, instead of using plain atomic64_t, use atomic_long_t instead.
32-bit machines tend to be memory-limited (i.e. anything that increases
a reference uses so much memory that you can't actually get to 2**32
references). 32-bit architectures also tend to have serious problems
with 64-bit atomics. Hence, atomic_long_t is the more natural solution.
Reported-by: "The UK's National Cyber Security Centre (NCSC)" <security@ncsc.gov.uk> Co-developed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: stable@kernel.org Reviewed-by: Willem de Bruijn <willemb@google.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Link: https://lore.kernel.org/r/20231201131021.19999-1-daniel@iogearbox.net Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Function trace_buffered_event_disable() is responsible for freeing pages
backing buffered events and this process can run concurrently with
trace_event_buffer_lock_reserve().
The following race is currently possible:
* Function trace_buffered_event_disable() is called on CPU 0. It
increments trace_buffered_event_cnt on each CPU and waits via
synchronize_rcu() for each user of trace_buffered_event to complete.
* After synchronize_rcu() is finished, function
trace_buffered_event_disable() has the exclusive access to
trace_buffered_event. All counters trace_buffered_event_cnt are at 1
and all pointers trace_buffered_event are still valid.
* At this point, on a different CPU 1, the execution reaches
trace_event_buffer_lock_reserve(). The function calls
preempt_disable_notrace() and only now enters an RCU read-side
critical section. The function proceeds and reads a still valid
pointer from trace_buffered_event[CPU1] into the local variable
"entry". However, it doesn't yet read trace_buffered_event_cnt[CPU1]
which happens later.
* Function trace_buffered_event_disable() continues. It frees
trace_buffered_event[CPU1] and decrements
trace_buffered_event_cnt[CPU1] back to 0.
* Function trace_event_buffer_lock_reserve() continues. It reads and
increments trace_buffered_event_cnt[CPU1] from 0 to 1. This makes it
believe that it can use the "entry" that it already obtained but the
pointer is now invalid and any access results in a use-after-free.
Fix the problem by making a second synchronize_rcu() call after all
trace_buffered_event values are set to NULL. This waits on all potential
users in trace_event_buffer_lock_reserve() that still read a previous
pointer from trace_buffered_event.
For instance, it can be triggered by running these two commands in
parallel:
$ while true; do
echo hist:key=id.syscall:val=hitcount > \
/sys/kernel/debug/tracing/events/raw_syscalls/sys_enter/trigger;
done
$ stress-ng --sysinfo $(nproc)
The warning indicates that the current ring_buffer_per_cpu is not in the
committing state. It happens because the active ring_buffer_event
doesn't actually come from the ring_buffer_per_cpu but is allocated from
trace_buffered_event.
The bug is in function trace_buffered_event_disable() where the
following normally happens:
* The code invokes disable_trace_buffered_event() via
smp_call_function_many() and follows it by synchronize_rcu(). This
increments the per-CPU variable trace_buffered_event_cnt on each
target CPU and grants trace_buffered_event_disable() the exclusive
access to the per-CPU variable trace_buffered_event.
* Maintenance is performed on trace_buffered_event, all per-CPU event
buffers get freed.
* The code invokes enable_trace_buffered_event() via
smp_call_function_many(). This decrements trace_buffered_event_cnt and
releases the access to trace_buffered_event.
A problem is that smp_call_function_many() runs a given function on all
target CPUs except on the current one. The following can then occur:
* Task X executing trace_buffered_event_disable() runs on CPU 0.
* The control reaches synchronize_rcu() and the task gets rescheduled on
another CPU 1.
* The RCU synchronization finishes. At this point,
trace_buffered_event_disable() has the exclusive access to all
trace_buffered_event variables except trace_buffered_event[CPU0]
because trace_buffered_event_cnt[CPU0] is never incremented and if the
buffer is currently unused, remains set to 0.
* A different task Y is scheduled on CPU 0 and hits a trace event. The
code in trace_event_buffer_lock_reserve() sees that
trace_buffered_event_cnt[CPU0] is set to 0 and decides the use the
buffer provided by trace_buffered_event[CPU0].
* Task X continues its execution in trace_buffered_event_disable(). The
code incorrectly frees the event buffer pointed by
trace_buffered_event[CPU0] and resets the variable to NULL.
* Task Y writes event data to the now freed buffer and later detects the
created inconsistency.
The issue is observable since commit dea499781a11 ("tracing: Fix warning
in trace_buffered_event_disable()") which moved the call of
trace_buffered_event_disable() in __ftrace_event_enable_disable()
earlier, prior to invoking call->class->reg(.. TRACE_REG_UNREGISTER ..).
The underlying problem in trace_buffered_event_disable() is however
present since the original implementation in commit 0fc1b09ff1ff
("tracing: Use temp buffer when filtering events").
Fix the problem by replacing the two smp_call_function_many() calls with
on_each_cpu_mask() which invokes a given callback on all CPUs.
It use to be that only the top level instance had a snapshot buffer (for
latency tracers like wakeup and irqsoff). When stopping a tracer in an
instance would not disable the snapshot buffer. This could have some
unintended consequences if the irqsoff tracer is enabled.
Consolidate the tracing_start/stop() with tracing_start/stop_tr() so that
all instances behave the same. The tracing_start/stop() functions will
just call their respective tracing_start/stop_tr() with the global_array
passed in.
Link: https://lkml.kernel.org/r/20231205220011.041220035@goodmis.org Cc: stable@vger.kernel.org Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Cc: Andrew Morton <akpm@linux-foundation.org> Fixes: 6d9b3fa5e7f6 ("tracing: Move tracing_max_latency into trace_array") Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It use to be that only the top level instance had a snapshot buffer (for
latency tracers like wakeup and irqsoff). The update of the ring buffer
size would check if the instance was the top level and if so, it would
also update the snapshot buffer as it needs to be the same as the main
buffer.
Now that lower level instances also has a snapshot buffer, they too need
to update their snapshot buffer sizes when the main buffer is changed,
otherwise the following can be triggered:
All addresses printed by checkstack have an extra incorrect 0 appended at
the end.
This was introduced with commit 677f1410e058 ("scripts/checkstack.pl: don't
display $dre as different entity"): since then the address is taken from
the line which contains the function name, instead of the line which
contains stack consumption. E.g. on s390:
So the used regex which matches spaces and hexadecimal numbers to extract
an address now matches a different substring. Subsequently replacing spaces
with 0 appends a zero at the and, instead of replacing leading spaces.
Fix this by using the proper regex, and simplify the code a bit.
If nilfs2 reads a disk image with corrupted segment usage metadata, and
its segment usage information is marked as an error for the segment at the
write location, nilfs_sufile_set_segment_usage() can trigger WARN_ONs
during log writing.
Segments newly allocated for writing with nilfs_sufile_alloc() will not
have this error flag set, but this unexpected situation will occur if the
segment indexed by either nilfs->ns_segnum or nilfs->ns_nextnum (active
segment) was marked in error.
Fix this issue by inserting a sanity check to treat it as a file system
corruption.
Since error returns are not allowed during the execution phase where
nilfs_sufile_set_segment_usage() is used, this inserts the sanity check
into nilfs_sufile_mark_dirty() which pre-reads the buffer containing the
segment usage record to be updated and sets it up in a dirty state for
writing.
In addition, nilfs_sufile_set_segment_usage() is also called when
canceling log writing and undoing segment usage update, so in order to
avoid issuing the same kernel warning in that case, in case of
cancellation, avoid checking the error flag in
nilfs_sufile_set_segment_usage().
When mounting a filesystem image with a block size larger than the page
size, nilfs2 repeatedly outputs long error messages with stack traces to
the kernel log, such as the following:
This overloads the system logger. And to make matters worse, it sometimes
crashes the kernel with a memory access violation.
This is because the return value of the sb_set_blocksize() call, which
should be checked for errors, is not checked.
The latter issue is due to out-of-buffer memory being accessed based on a
large block size that caused sb_set_blocksize() to fail for buffers read
with the initial minimum block size that remained unupdated in the
super_block structure.
Since nilfs2 mkfs tool does not accept block sizes larger than the system
page size, this has been overlooked. However, it is possible to create
this situation by intentionally modifying the tool or by passing a
filesystem image created on a system with a large page size to a system
with a smaller page size and mounting it.
Fix this issue by inserting the expected error handling for the call to
sb_set_blocksize().
The pcm state can be SNDRV_PCM_STATE_DISCONNECTED at disconnect
callback, and there is not an entry of SNDRV_PCM_STATE_DISCONNECTED
in snd_pcm_state_names.
This patch adds the missing entry to resolve this issue.
cat /proc/asound/card2/pcm0p/sub0/status
That results in stack traces like the following:
These values mirror those of the Pioneer DJM-250MK2 as the channel layout
appears identical based on my observations. This duplication could be removed in
later contributions if desired.
Signed-off-by: Sarah Grant <s@srd.tw> Cc: <stable@vger.kernel.org> Link: https://lore.kernel.org/r/20231201181654.5058-1-s@srd.tw Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Callers of mutex_unlock() have to make sure that the mutex stays alive
for the whole duration of the function call. For io_uring that means
that the following pattern is not valid unless we ensure that the
context outlives the mutex_unlock() call.
mutex_lock(&ctx->uring_lock);
req_put(req); // typically via io_req_task_submit()
mutex_unlock(&ctx->uring_lock);
Most contexts are fine: io-wq pins requests, syscalls hold the file,
task works are taking ctx references and so on. However, the task work
fallback path doesn't follow the rule.
Cc: <stable@vger.kernel.org> Fixes: 04fc6c802d ("io_uring: save ctx put/get for task_work submit") Reported-by: Jann Horn <jannh@google.com> Signed-off-by: Pavel Begunkov <asml.silence@gmail.com> Link: https://lore.kernel.org/io-uring/CAG48ez3xSoYb+45f1RLtktROJrpiDQ1otNvdR+YLQf7m+Krj5Q@mail.gmail.com/ Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It seems that the pointer-to-kretprobe "rp" within the kretprobe_holder is
RCU-managed, based on the (non-rethook) implementation of get_kretprobe().
The thought behind this patch is to make use of the RCU API where possible
when accessing this pointer so that the needed barriers are always in place
and to self-document the code.
The __rcu annotation to "rp" allows for sparse RCU checking. Plain writes
done to the "rp" pointer are changed to make use of the RCU macro for
assignment. For the single read, the implementation of get_kretprobe()
is simplified by making use of an RCU macro which accomplishes the same,
but note that the log warning text will be more generic.
I did find that there is a difference in assembly generated between the
usage of the RCU macros vs without. For example, on arm64, when using
rcu_assign_pointer(), the corresponding store instruction is a
store-release (STLR) which has an implicit barrier. When normal assignment
is done, a regular store (STR) is found. In the macro case, this seems to
be a result of rcu_assign_pointer() using smp_store_release() when the
value to write is not NULL.
If md_set_readonly() failed, the array could still be read-write, however
'MD_RECOVERY_FROZEN' could still be set, which leave the array in an
abnormal state that sync or recovery can't continue anymore.
Hence make sure the flag is cleared after md_set_readonly() returns.
Fixes: 88724bfa68be ("md: wait for pending superblock updates before switching to read-only") Signed-off-by: Yu Kuai <yukuai3@huawei.com> Acked-by: Xiao Ni <xni@redhat.com> Signed-off-by: Song Liu <song@kernel.org> Link: https://lore.kernel.org/r/20231205094215.1824240-3-yukuai1@huaweicloud.com Signed-off-by: Sasha Levin <sashal@kernel.org>
Introduce md_ro_state for mddev->ro, so it is easy to understand.
Signed-off-by: Ye Bin <yebin10@huawei.com> Signed-off-by: Song Liu <song@kernel.org>
Stable-dep-of: c9f7cb5b2bc9 ("md: don't leave 'MD_RECOVERY_FROZEN' in error path of md_set_readonly()") Signed-off-by: Sasha Levin <sashal@kernel.org>
The timer nodes declare compatibility with "fsl,imx6sx-gpt", which
itself is compatible with "fsl,imx6dl-gpt". Switch the fallback
compatible from "fsl,imx6sx-gpt" to "fsl,imx6dl-gpt".
devm_kasprintf() returns a pointer to dynamically allocated memory
which can be NULL upon failure. Ensure the allocation was successful
by checking the pointer validity.
Release the id allocated in 'mmdc_pmu_init' when 'devm_kasprintf'
return NULL
When an error occurs in the for loop of beiscsi_init_wrb_handle(), we
should free phwi_ctxt->be_wrbq before returning an error code to prevent
potential memleak.
Fixes: a7909b396ba7 ("[SCSI] be2iscsi: Fix dynamic CID allocation Mechanism in driver") Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn> Link: https://lore.kernel.org/r/20231123081941.24854-1-dinghao.liu@zju.edu.cn Reviewed-by: Mike Christie <michael.christie@oracle.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
Function trace_buffered_event_disable() produces an unexpected warning
when the previous call to trace_buffered_event_enable() fails to
allocate pages for buffered events.
The situation can occur as follows:
* The counter trace_buffered_event_ref is at 0.
* The soft mode gets enabled for some event and
trace_buffered_event_enable() is called. The function increments
trace_buffered_event_ref to 1 and starts allocating event pages.
* The allocation fails for some page and trace_buffered_event_disable()
is called for cleanup.
* Function trace_buffered_event_disable() decrements
trace_buffered_event_ref back to 0, recognizes that it was the last
use of buffered events and frees all allocated pages.
* The control goes back to trace_buffered_event_enable() which returns.
The caller of trace_buffered_event_enable() has no information that
the function actually failed.
* Some time later, the soft mode is disabled for the same event.
Function trace_buffered_event_disable() is called. It warns on
"WARN_ON_ONCE(!trace_buffered_event_ref)" and returns.
Buffered events are just an optimization and can handle failures. Make
trace_buffered_event_enable() exit on the first failure and left any
cleanup later to when trace_buffered_event_disable() is called.
Since commit c7e73b5051d6 ("ARM: imx: mach-imx6ul: remove 14x14 EVK
specific PHY fixup")thet Ethernet PHY is no longer configured via code
in board file.
This caused Ethernet to stop working.
Fix this problem by describing the clocks and clock-names to the
Ethernet PHY node so that the KSZ8081 chip can be clocked correctly.
The i.MX8MP and i.MX8MQ devices both use the same DWC3 controller and
are both affected by a known issue with the controller due to specific
behaviour when park mode is enabled in SuperSpeed host mode operation.
Under heavy USB traffic from multiple endpoints the controller will
sometimes incorrectly process transactions such that some transactions
are lost, or the controller may hang when processing transactions. When
the controller hangs it does not recover.
This issue is documented partially within the linux-imx vendor kernel
which references a Synopsys STAR number 9001415732 in commits [1] and
additional details in [2]. Those commits provide some additional
controller internal implementation specifics around the incorrect
behaviour of the SuperSpeed host controller operation when park mode is
enabled.
The summary of this issue is that the host controller can incorrectly
enter/exit park mode such that part of the controller is in a state
which behaves as if in park mode even though it is not. In this state
the controller incorrectly calculates the number of TRBs available which
results in incorrect access of the internal caches causing the overwrite
of pending requests in the cache which should have been processed but
are ignored. This can cause the controller to drop the requests or hang
waiting for the pending state of the dropped requests.
The workaround for this issue is to disable park mode for SuperSpeed
operation of the controller through the GUCTL1[17] bit. This is already
available as a quirk for the DWC3 controller and can be enabled via the
'snps,parkmode-disable-ss-quirk' device tree property.
It is possible to replicate this failure on an i.MX8MP EVK with a USB
Hub connecting 4 SuperSpeed USB flash drives. Performing continuous
small read operations (dd if=/dev/sd... of=/dev/null bs=16) on the block
devices will result in device errors initially and will eventually
result in the controller hanging.
[13240.896936] xhci-hcd xhci-hcd.0.auto: WARN Event TRB for slot 4 ep 2 with no TDs queued?
[13240.990708] usb 2-1.3: reset SuperSpeed USB device number 5 using xhci-hcd
[13241.015582] sd 2:0:0:0: [sdc] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x07 driverbyte=DRIVER_OK cmd_age=0s
[13241.025198] sd 2:0:0:0: [sdc] tag#0 CDB: opcode=0x28 28 00 00 00 03 e0 00 01 00 00
[13241.032949] I/O error, dev sdc, sector 992 op 0x0:(READ) flags 0x80700 phys_seg 25 prio class 2
[13272.150710] usb 2-1.2: reset SuperSpeed USB device number 4 using xhci-hcd
[13272.175469] sd 1:0:0:0: [sdb] tag#0 UNKNOWN(0x2003) Result: hostbyte=0x03 driverbyte=DRIVER_OK cmd_age=31s
[13272.185365] sd 1:0:0:0: [sdb] tag#0 CDB: opcode=0x28 28 00 00 00 03 e0 00 01 00 00
[13272.193385] I/O error, dev sdb, sector 992 op 0x0:(READ) flags 0x80700 phys_seg 18 prio class 2
[13434.846556] xhci-hcd xhci-hcd.0.auto: xHCI host not responding to stop endpoint command
[13434.854592] xhci-hcd xhci-hcd.0.auto: xHCI host controller not responding, assume dead
[13434.862553] xhci-hcd xhci-hcd.0.auto: HC died; cleaning up
When creating ceq_0 during probing irdma, cqp.sc_cqp will be sent as a
cqp_request to cqp->sc_cqp.sq_ring. If the request is pending when
removing the irdma driver or unplugging its aux device, cqp.sc_cqp will be
dereferenced as wrong struct in irdma_free_pending_cqp_request().
64k pages introduce the situation in this diagram when the HCA 4k page
size is being used:
+-------------------------------------------+ <--- 64k aligned VA
| |
| HCA 4k page |
| |
+-------------------------------------------+
| o |
| |
| o |
| |
| o |
+-------------------------------------------+
| |
| HCA 4k page |
| |
+-------------------------------------------+ <--- Live HCA page
|OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO| <--- offset
| | <--- VA
| MR data |
+-------------------------------------------+
| |
| HCA 4k page |
| |
+-------------------------------------------+
| o |
| |
| o |
| |
| o |
+-------------------------------------------+
| |
| HCA 4k page |
| |
+-------------------------------------------+
The VA addresses are coming from rdma-core in this diagram can be
arbitrary, but for 64k pages, the VA may be offset by some number of HCA
4k pages and followed by some number of HCA 4k pages.
The current iterator doesn't account for either the preceding 4k pages or
the following 4k pages.
Fix the issue by extending the ib_block_iter to contain the number of DMA
pages like comment [1] says and by using __sg_advance to start the
iterator at the first live HCA page.
The changes are contained in a parallel set of iterator start and next
functions that are umem aware and specific to umem since there is one user
of the rdma_for_each_block() without umem.
These two fixes prevents the extra pages before and after the user MR
data.
Fix the preceding pages by using the __sq_advance field to start at the
first 4k page containing MR data.
Fix the following pages by saving the number of pgsz blocks in the
iterator state and downcounting on each next.
This fix allows for the elimination of the small page crutch noted in the
Fixes.
Fixes: 10c75ccb54e4 ("RDMA/umem: Prevent small pages from being returned by ib_umem_find_best_pgsz()") Link: https://lore.kernel.org/r/20231129202143.1434-2-shiraz.saleem@intel.com Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Sasha Levin <sashal@kernel.org>
When wm_adsp_buffer_read() fails, we should free buf->regions.
Otherwise, the callers of wm_adsp_buffer_populate() will
directly free buf on failure, which makes buf->regions a leaked
memory.
Fixes: a792af69b08f ("ASoC: wm_adsp: Refactor compress stream initialisation") Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn> Reviewed-by: Richard Fitzgerald <rf@opensource.cirrus.com> Link: https://lore.kernel.org/r/20231204074158.12026-1-dinghao.liu@zju.edu.cn Signed-off-by: Mark Brown <broonie@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
"If an error occurs while obtaining the meter reading or if the value
is not available then an Integer with all bits set is returned"
Since the "integer" is 32 bits in case of the ACPI power meter,
userspace will get a power reading of 2^32 * 1000 miliwatts (~4.29 MW)
in case of such an error. This was discovered due to a lm_sensors
bugreport (https://github.com/lm-sensors/lm-sensors/issues/460).
Fix this by returning -ENODATA instead.
Tested-by: <urbinek@gmail.com> Fixes: de584afa5e18 ("hwmon driver for ACPI 4.0 power meters") Signed-off-by: Armin Wolf <W_Armin@gmx.de> Link: https://lore.kernel.org/r/20231124182747.13956-1-W_Armin@gmx.de Signed-off-by: Guenter Roeck <linux@roeck-us.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
As we chain the WR during write request: memory registration,
rdma write, local invalidate, if only the last WR fail to send due
to send queue overrun, the server can send back the reply, while
client mark the req->in_use to false in case of error in rtrs_clt_req
when error out from rtrs_post_rdma_write_sg.
Fixes: 6a98d71daea1 ("RDMA/rtrs: client: main functionality") Signed-off-by: Jack Wang <jinpu.wang@ionos.com> Reviewed-by: Md Haris Iqbal <haris.iqbal@ionos.com> Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com> Link: https://lore.kernel.org/r/20231120154146.920486-8-haris.iqbal@ionos.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
For each write request, we need Request, Response Memory Registration,
Local Invalidate.
Fixes: 6a98d71daea1 ("RDMA/rtrs: client: main functionality") Signed-off-by: Jack Wang <jinpu.wang@ionos.com> Reviewed-by: Md Haris Iqbal <haris.iqbal@ionos.com> Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com> Link: https://lore.kernel.org/r/20231120154146.920486-7-haris.iqbal@ionos.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Destroying path files may lead to the freeing of rdma_stats. This creates
the following race.
An IO is in-flight, or has just passed the session state check in
process_read/process_write. The close_work gets triggered and the function
rtrs_srv_close_work() starts and does destroy path which frees the
rdma_stats. After this the function process_read/process_write resumes and
tries to update the stats through the function rtrs_srv_update_rdma_stats
This commit solves the problem by moving the destroy path function to a
later point. This point makes sure any inflights are completed. This is
done by qp drain, and waiting for all in-flights through ops_id.
Fixes: 9cb837480424 ("RDMA/rtrs: server: main functionality") Signed-off-by: Md Haris Iqbal <haris.iqbal@ionos.com> Signed-off-by: Santosh Kumar Pradhan <santosh.pradhan@ionos.com> Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com> Link: https://lore.kernel.org/r/20231120154146.920486-6-haris.iqbal@ionos.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Since srv_mr->iu is allocated and used only when always_invalidate is
true, free it only when always_invalidate is true.
Fixes: 9cb837480424 ("RDMA/rtrs: server: main functionality") Signed-off-by: Md Haris Iqbal <haris.iqbal@ionos.com> Signed-off-by: Jack Wang <jinpu.wang@ionos.com> Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com> Link: https://lore.kernel.org/r/20231120154146.920486-5-haris.iqbal@ionos.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
While processing info request, it could so happen that the srv_path goes
to CLOSING state, cause of any of the error events from RDMA. That state
change should be picked up while trying to change the state in
process_info_req, by checking the return value. In case the state change
call in process_info_req fails, we fail the processing.
We should also check the return value for rtrs_srv_path_up, since it
sends a link event to the client above, and the client can fail for any
reason.
Fixes: 9cb837480424 ("RDMA/rtrs: server: main functionality") Signed-off-by: Md Haris Iqbal <haris.iqbal@ionos.com> Signed-off-by: Jack Wang <jinpu.wang@ionos.com> Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com> Link: https://lore.kernel.org/r/20231120154146.920486-4-haris.iqbal@ionos.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
If we start hb too early, it will confuse server side to close
the session.
Fixes: 6a98d71daea1 ("RDMA/rtrs: client: main functionality") Signed-off-by: Jack Wang <jinpu.wang@ionos.com> Reviewed-by: Md Haris Iqbal <haris.iqbal@ionos.com> Signed-off-by: Grzegorz Prajsner <grzegorz.prajsner@ionos.com> Link: https://lore.kernel.org/r/20231120154146.920486-3-haris.iqbal@ionos.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Expand the reg size for the vdec node to include cache/performance
registers the rkvdec driver writes to. Also add missing clocks to the
related power-domain.
Fixes: cbd7214402ec ("arm64: dts: rockchip: Define the rockchip Video Decoder node on rk3399") Signed-off-by: Alex Bee <knaerzche@gmail.com> Signed-off-by: Jonas Karlman <jonas@kwiboo.se> Link: https://lore.kernel.org/r/20231105233630.3927502-10-jonas@kwiboo.se Signed-off-by: Heiko Stuebner <heiko@sntech.de> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently, there is no wait for the QP suspend to complete on a modify
to SQD state. Add a wait, after the modify to SQD state, for the Suspend
Complete AE. While we are at it, update the suspend timeout value in
irdma_prep_tc_change to use IRDMA_EVENT_TIMEOUT_MS too.
Add a default congest control algorithm so that driver won't return
an error when the configured algorithm is invalid.
Fixes: f91696f2f053 ("RDMA/hns: Support congestion control type selection according to the FW") Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Link: https://lore.kernel.org/r/20231028093242.670325-1-huangjunxian6@hisilicon.com Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
Currently supplicant dependent optee device enumeration only registers
devices whenever tee-supplicant is invoked for the first time. But it
forgets to remove devices when tee-supplicant daemon stops running and
closes its context gracefully. This leads to following error for fTPM
driver during reboot/shutdown:
Fix this by adding an attribute for supplicant dependent devices so that
the user-space service can detect and detach supplicant devices before
closing the supplicant:
$ for dev in /sys/bus/tee/devices/*; do if [[ -f "$dev/need_supplicant" && -f "$dev/driver/unbind" ]]; \
then echo $(basename "$dev") > $dev/driver/unbind; fi done
Reported-by: Jan Kiszka <jan.kiszka@siemens.com> Closes: https://github.com/OP-TEE/optee_os/issues/6094 Fixes: 5f178bb71e3a ("optee: enable support for multi-stage bus enumeration") Signed-off-by: Sumit Garg <sumit.garg@linaro.org> Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org> Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
[jw: fixed up Date documentation] Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
The "NET_DM" generic netlink family notifies drop locations over the
"events" multicast group. This is problematic since by default generic
netlink allows non-root users to listen to these notifications.
Fix by adding a new field to the generic netlink multicast group
structure that when set prevents non-root users or root without the
'CAP_SYS_ADMIN' capability (in the user namespace owning the network
namespace) from joining the group. Set this field for the "events"
group. Use 'CAP_SYS_ADMIN' rather than 'CAP_NET_ADMIN' because of the
nature of the information that is shared over this group.
Note that the capability check in this case will always be performed
against the initial user namespace since the family is not netns aware
and only operates in the initial network namespace.
A new field is added to the structure rather than using the "flags"
field because the existing field uses uAPI flags and it is inappropriate
to add a new uAPI flag for an internal kernel check. In net-next we can
rework the "flags" field to use internal flags and fold the new field
into it. But for now, in order to reduce the amount of changes, add a
new field.
Since the information can only be consumed by root, mark the control
plane operations that start and stop the tracing as root-only using the
'GENL_ADMIN_PERM' flag.
The "psample" generic netlink family notifies sampled packets over the
"packets" multicast group. This is problematic since by default generic
netlink allows non-root users to listen to these notifications.
Fix by marking the group with the 'GENL_UNS_ADMIN_PERM' flag. This will
prevent non-root users or root without the 'CAP_NET_ADMIN' capability
(in the user namespace owning the network namespace) from joining the
group.
This patch is based on a detailed report and ideas from Yepeng Pan
and Christian Rossow.
ACK seq validation is currently following RFC 5961 5.2 guidelines:
The ACK value is considered acceptable only if
it is in the range of ((SND.UNA - MAX.SND.WND) <= SEG.ACK <=
SND.NXT). All incoming segments whose ACK value doesn't satisfy the
above condition MUST be discarded and an ACK sent back. It needs to
be noted that RFC 793 on page 72 (fifth check) says: "If the ACK is a
duplicate (SEG.ACK < SND.UNA), it can be ignored. If the ACK
acknowledges something not yet sent (SEG.ACK > SND.NXT) then send an
ACK, drop the segment, and return". The "ignored" above implies that
the processing of the incoming data segment continues, which means
the ACK value is treated as acceptable. This mitigation makes the
ACK check more stringent since any ACK < SND.UNA wouldn't be
accepted, instead only ACKs that are in the range ((SND.UNA -
MAX.SND.WND) <= SEG.ACK <= SND.NXT) get through.
This can be refined for new (and possibly spoofed) flows,
by not accepting ACK for bytes that were never sent.
This greatly improves TCP security at a little cost.
I added a Fixes: tag to make sure this patch will reach stable trees,
even if the 'blamed' patch was adhering to the RFC.
tp->bytes_acked was added in linux-4.2
Following packetdrill test (courtesy of Yepeng Pan) shows
the issue at hand:
// when window scale is set to 14 the window size can be extended to
// 65535 * (2^14) = 1073725440. Linux would accept an ACK packet
// with ack number in (Server_ISN+1-1073725440. Server_ISN+1)
// ,though this ack number acknowledges some data never
// sent by the server.
// For the established connection, we send an ACK packet,
// the ack packet uses ack number 1 - 1073725300 + 2^32,
// where 2^32 is used to wrap around.
// Note: we used 1073725300 instead of 1073725440 to avoid possible
// edge cases.
// 1 - 1073725300 + 2^32 = 3221241997
// Oops, old kernels happily accept this packet.
+0 < . 1:1001(1000) ack 3221241997 win 65535
// After the kernel fix the following will be replaced by a challenge ACK,
// and prior malicious frame would be dropped.
+0 > . 1:1(0) ack 1001
Fixes: 354e4aa391ed ("tcp: RFC 5961 5.2 Blind Data Injection Attack Mitigation") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Yepeng Pan <yepeng.pan@cispa.de> Reported-by: Christian Rossow <rossow@cispa.de> Acked-by: Neal Cardwell <ncardwell@google.com> Link: https://lore.kernel.org/r/20231205161841.2702925-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>
A concurrently running sock_orphan() may NULL the sk_socket pointer in
between check and deref. Follow other users (like nft_meta.c for
instance) and acquire sk_callback_lock before dereferencing sk_socket.
Fixes: 0265ab44bacc ("[NETFILTER]: merge ipt_owner/ip6t_owner in xt_owner") Reported-by: Jann Horn <jannh@google.com> Signed-off-by: Phil Sutter <phil@nwl.cc> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Sasha Levin <sashal@kernel.org>