clk: bcm: rpi: Set a default minimum rate

The M2MC clock provides the state machine clock for both HDMI
controllers.

However, if no HDMI monitor is plugged in at boot, its clock rate will
be left at 0 by the firmware and will make any register access end up in
a CPU stall, even though the clock was enabled.

We had some code in the HDMI controller to deal with this before, but it
makes more sense to have it in the clock driver. Move it there.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: bcm: rpi: Add variant structure

We only export a bunch of firmware clocks, and some of them require
special treatment.

This has been do so far using some tests on the clock id in various
places, but this is fairly hard to extend and doesn't scale very well.

Since we'll need some more cases in the next patches, let's switch to a
variant structure that defines the behaviour we need to have for a given
clock.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Test the clock pointer in clk_hw_get_name()

Unlike __clk_get_name(), clk_hw_get_name() doesn't test wether passed
clk_hw pointer is NULL or not and dereferences it directly. This can
then lead to NULL pointer dereference.

Let's make sure the pointer isn't NULL before dereferencing it.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Zero the clk_rate_request structure

In order to make sure we don't carry anything over from an already
existing clk_rate_request pointer we would pass to
clk_core_init_rate_req(), let's zero the entire structure before
initializing it.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Stop forwarding clk_rate_requests to the parent

If the clock cannot modify its rate and has CLK_SET_RATE_PARENT,
clk_mux_determine_rate_flags() and clk_core_round_rate_nolock() will
call clk_core_round_rate_nolock() with its parent clock but use the
request of the child node either directly (clk_core_round_rate_nolock())
or by copying it (clk_mux_determine_rate_flags()).

Both cases are problematic since the parent will now have a request with
the best parent fields of the child (so pointing to itself) and the
boundaries of the child as well.

clk_core_round_rate_nolock() is even worse since we would directly
modify the caller structure if the parent was ever to modify its own
parent or its parent rate, then returning to the caller a best parent
that isn't a parent of the clock we just called clk_determine_rate()
onto.

Let's create a new function that will create a new request to forward to
the parent, clk_core_forward_rate_req() and update the relevant call
sites to that new function.

Let's also add a test to make sure we avoid regressions there.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Introduce clk_core_has_parent()

We will need to know if a clk_core pointer has a given parent in other
functions, so let's create a clk_core_has_parent() function that
clk_has_parent() will call into.

For good measure, let's add some unit tests as well to make sure it
works properly.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Switch from __clk_determine_rate to clk_core_round_rate_nolock

clk_mux_determine_rate_flags() will call into __clk_determine_rate()
with a clk_hw pointer, while it has access to the clk_core pointer
already.

This leads to back and forth between clk_hw and clk_core, while
__clk_determine_rate will only call clk_core_round_rate_nolock() with
the clk_core pointer it retrieved from the clk_hw.

Let's simplify things a bit by calling into clk_core_round_rate_nolock
directly.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Remove redundant clk_core_init_rate_req() call

Since all the users of clk_core_round_rate_nolock() will now properly
initialize, there's no need for it to initialize the request itself.

This is even dangerous, as if the clock cannot change its rate by itself
and has CLK_SET_RATE_PARENT, clk_core_round_rate_nolock() will call
itself with the parent clock but the client clk_rate_request structure.

We will then reinitialize the child request with the parent context
(parent, boundaries, etc.), which is an issue if the parent ever changes
its own parent or parent rate.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Add missing clk_core_init_rate_req calls

Some callers of clk_core_round_rate_nolock() will initialize the
clk_rate_request structure by hand, missing a few parameters that leads
to inconsistencies in what drivers can expect from that structure.

Let's use clk_core_init_rate_req() everywhere to make sure it's built in
a consistent manner.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Introduce clk_hw_init_rate_request()

Some drivers (at91, imx, qcom) use __clk_determine_rate directly, and
thus will need to initialise a clk_rate_request structure.

However, drivers don't have access to the function that the core uses to
initialize that structure which could prove to be useful.

Let's create a function for clock providers that will initialize a
clk_rate_request for a given clk_hw pointer and a rate.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Change clk_core_init_rate_req prototype

The expectation is that a clk_rate_request structure is supposed to be
initialized using clk_core_init_rate_req(), yet the rate we want to
request still needs to be set by hand.

Let's just pass the rate as a function argument so that callers don't
have any extra work to do.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Add our request boundaries in clk_core_init_rate_req

The expectation is that a new clk_rate_request is initialized through a
call to clk_core_init_rate_req().

However, at the moment it only fills the parent rate and clk_hw pointer,
but omits the other fields such as the clock rate boundaries.

Some users of that function will update them after calling it, but most
don't.

As we are passed the clk_core pointer, we have access to those
boundaries in clk_core_init_rate_req() however, so let's just fill it
there and remove it from the few callers that do it right.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Skip set_rate_range if our clock is orphan

clk_set_rate_range will now force a clk_set_rate() call to
core->req_rate. However, if our clock is orphan, req_rate will be 0 and
we will thus end up calling a set_rate to 0 on potentially all that
clock subtree.

This can be fairly invasive and result in poor decisions. In such a
case, let's just store the new range but bail out and skip the set_rate.

When that clock is no longer orphan though, we should be enforcing the
new range but we don't. Let's add a skipped test to make sure we don't
forget about that corner case.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Set req_rate on reparenting

If a non-rate clock started by default with a parent that never
registered, core->req_rate will be 0. The expectation is that whenever
the parent will be registered, req_rate will be updated with the new
value that has just been computed.

However, if that clock is a mux, clk_set_parent() can also make that
clock no longer orphan. In this case however, we never update req_rate.
Let's make sure it's the case for the newly unorphan clock and all its
children.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Fix clk_get_parent() documentation

The clk_get_parent() documentation in the header states that it will
return a valid pointer, or an error pointer on failure.

However, the documentation in the source file, and the code itself, will
return also return NULL if there isn't any parent for that clock. Let's
mention it.

An orphan clock should return NULL too, so let's add a test for it.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Take into account uncached clocks in clk_set_rate_range()

clk_set_rate_range() will use the last requested rate for the clock when
it calls into the driver set_rate hook.

However, if CLK_GET_RATE_NOCACHE is set on that clock, the last
requested rate might not be matching the current rate of the clock. In
such a case, let's read out the rate from the hardware and use that in
our set_rate instead.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: tests: Add some tests for orphan with multiple parents

Let's leverage the dummy mux with multiple parents we have to create a
mux whose default parent will never be registered, and thus will always
be orphan by default.

We can then create some tests to make sure that the clock API behaves
properly in such a case, and that the transition to a non-orphan clock
when we change the parent is done properly.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: tests: Add tests for mux with multiple parents

We'll need to test a few corner cases that occur when we have a mux
clock whose default parent is missing.

For now, let's create the context structure and the trivial ops, along
with a test suite that just tests trivial things for now, without
considering the orphan case.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: tests: Add tests for single parent mux

We have a few tests for a mux with a single parent, testing the case
where it used to be orphan.

Let's leverage most of the code but register the clock properly to test
a few trivial things.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: tests: Add tests for uncached clock

The clock framework supports clocks that can have their rate changed
without the kernel knowing about it using the CLK_GET_RATE_NOCACHE flag.

As its name suggests, this flag turns off the rate caching in the clock
framework, reading out the rate from the hardware any time we need to
read it.

Let's add a couple of tests to make sure it works as intended.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: tests: Add reference to the orphan mux bug report

Some more context might be useful for unit-tests covering a previously
reported bug, so let's add a link to the discussion for that bug.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: tests: Add test suites description

We start to have a few test suites, and we'll add more, so it will get
pretty confusing to figure out what is supposed to be tested in what
suite.

Let's add some comments to explain what setup they create, and what we
should be testing in every suite.

Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Drop the rate range on clk_put()

When clk_put() is called we don't make another clk_set_rate() call to
re-evaluate the rate boundaries. This is unlike clk_set_rate_range()
that evaluates the rate again each time it is called.

However, clk_put() is essentially equivalent to clk_set_rate_range()
since after clk_put() completes the consumer's boundaries shouldn't be
enforced anymore.

Let's add a call to clk_set_rate_range() in clk_put() to make sure those
rate boundaries are dropped and the clock provider drivers can react. In
order to be as non-intrusive as possible, we'll just make that call if
the clock had non-default boundaries.

Also add a few tests to make sure this case is covered.

Fixes: c80ac50cbb37 ("clk: Always set the rate on clk_set_range_rate")
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com> # imx8mp
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> # exynos4210, meson g12b
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: test: Test clk_set_rate_range on orphan mux

A bug recently affected the Tegra30 where calling clk_set_rate_range()
on a clock would make it change its rate to the minimum.

This was due to the clock in question being a mux that was orphan at
registration, which lead to the clk_core req_rate being 0, and the
clk_set_rate_range() function then calling clk_set_rate() with req_rate,
effectively making that clock running at the minimum rate allowed, even
though the initial rate was within that range.

Make a test suite to create a mux initially orphan, and then make sure
that if our clock rate was initially within a given range, then
enforcing that range won't affect it.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Add clk_drop_range

In order to reset the range on a clock, we need to call
clk_set_rate_range with a minimum of 0 and a maximum of ULONG_MAX. Since
it's fairly inconvenient, let's introduce a clk_drop_range() function
that will do just this.

Suggested-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Always set the rate on clk_set_range_rate

When we change a clock minimum or maximum using clk_set_rate_range(),
clk_set_min_rate() or clk_set_max_rate(), the current code will only
trigger a new rate change if the rate is outside of the new boundaries.

However, a clock driver might want to always keep the clock rate to
one of its boundary, for example the minimum to keep the power
consumption as low as possible.

Since they don't always get called though, clock providers don't have the
opportunity to implement this behaviour.

Let's trigger a clk_set_rate() on the previous requested rate every time
clk_set_rate_range() is called. That way, providers that care about the
new boundaries have a chance to adjust the rate, while providers that
don't care about those new boundaries will return the same rate than
before, which will be ignored by clk_set_rate() and won't result in a
new rate change.

Suggested-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Use clamp instead of open-coding our own

The code in clk_set_rate_range() will, if the current rate is outside of
the new range, force it to the minimum or maximum.

Since it's running under the condition that the rate is either lower
than the minimum, or higher than the maximum, this is equivalent to
using clamp, while being less readable. Let's switch to using clamp
instead.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Always clamp the rounded rate

The current core while setting the min and max rate properly in the
clk_request structure will not make sure that the requested rate is
within these boundaries, leaving it to each and every driver to make
sure it is.

It's not clear if this was on purpose or not, but this introduces some
inconsistencies within the API.

For example, a user setting a range and then calling clk_round_rate()
with a value outside of that range will get the same value back
(ignoring any driver adjustements), effectively ignoring the range that
was just set.

Another one, arguably worse, is that it also makes clk_round_rate() and
clk_set_rate() behave differently if there's a range and the rate being
used for both is outside that range. As we have seen, the rate will be
returned unchanged by clk_round_rate(), but clk_set_rate() will error
out returning -EINVAL.

Let's make sure the framework will always clamp the rate to the current
range found on the clock, which will fix both these inconsistencies.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Introduce Kunit Tests for the framework

Let's test various parts of the rate-related clock API with the kunit
testing framework.

Cc: kunit-dev@googlegroups.com
Tested-by: Daniel Latypov <dlatypov@google.com>
Suggested-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

config: Enable the NFT_SYNPROXY module

The NFT_SYNPROXY module is apparently useful for port scan protection,
and at 11kB barely changes the size of the downloads.

See: https://github.com/raspberrypi/linux/issues/4993

Signed-off-by: Phil Elwell <phil@raspberrypi.com>

configs: (Re)Enable CONFIG_IR_TOY

Somehow or other, CONFIG_IR_TOY=m got dropped from the standard Pi
defconfigs around 5.13. Restore it.

See: https://github.com/raspberrypi/linux/issues/4997

Signed-off-by: Phil Elwell <phil@raspberrypi.com>

Merge remote-tracking branch 'stable/linux-5.15.y' into rpi-5.15.y

Linux 5.15.34

Link: https://lore.kernel.org/r/20220412062942.022903016@linuxfoundation.org
Link: https://lore.kernel.org/r/20220412173836.126811734@linuxfoundation.org
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Fox Chen <foxhlchen@gmail.com>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Ron Economos <re@w6rz.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

stacktrace: move filter_irq_stacks() to kernel/stacktrace.c

commit f39f21b3ddc7fc0f87eb6dc75ddc81b5bbfb7672 upstream.

filter_irq_stacks() has little to do with the stackdepot implementation,
except that it is usually used by users (such as KASAN) of stackdepot to
reduce the stack trace.

However, filter_irq_stacks() itself is not useful without a stack trace
as obtained by stack_trace_save() and friends.

Therefore, move filter_irq_stacks() to kernel/stacktrace.c, so that new
users of filter_irq_stacks() do not have to start depending on
STACKDEPOT only for filter_irq_stacks().

Link: https://lkml.kernel.org/r/20210923104803.2620285-1-elver@google.com
Signed-off-by: Marco Elver <elver@google.com>
Acked-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: Aleksandr Nogikh <nogikh@google.com>
Cc: Taras Madan <tarasmadan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

powerpc: Fix virt_addr_valid() for 64-bit Book3E & 32-bit

commit ffa0b64e3be58519ae472ea29a1a1ad681e32f48 upstream.

mpe: On 64-bit Book3E vmalloc space starts at 0x8000000000000000.

Because of the way __pa() works we have:
  __pa(0x8000000000000000) == 0, and therefore
  virt_to_pfn(0x8000000000000000) == 0, and therefore
  virt_addr_valid(0x8000000000000000) == true

Which is wrong, virt_addr_valid() should be false for vmalloc space.
In fact all vmalloc addresses that alias with a valid PFN will return
true from virt_addr_valid(). That can cause bugs with hardened usercopy
as described below by Kefeng Wang:

  When running ethtool eth0 on 64-bit Book3E, a BUG occurred:

    usercopy: Kernel memory exposure attempt detected from SLUB object not in SLUB page?! (offset 0, size 1048)!
    kernel BUG at mm/usercopy.c:99
    ...
    usercopy_abort+0x64/0xa0 (unreliable)
    __check_heap_object+0x168/0x190
    __check_object_size+0x1a0/0x200
    dev_ethtool+0x2494/0x2b20
    dev_ioctl+0x5d0/0x770
    sock_do_ioctl+0xf0/0x1d0
    sock_ioctl+0x3ec/0x5a0
    __se_sys_ioctl+0xf0/0x160
    system_call_exception+0xfc/0x1f0
    system_call_common+0xf8/0x200

  The code shows below,

    data = vzalloc(array_size(gstrings.len, ETH_GSTRING_LEN));
    copy_to_user(useraddr, data, gstrings.len * ETH_GSTRING_LEN))

  The data is alloced by vmalloc(), virt_addr_valid(ptr) will return true
  on 64-bit Book3E, which leads to the panic.

  As commit 4dd7554a6456 ("powerpc/64: Add VIRTUAL_BUG_ON checks for __va
  and __pa addresses") does, make sure the virt addr above PAGE_OFFSET in
  the virt_addr_valid() for 64-bit, also add upper limit check to make
  sure the virt is below high_memory.

  Meanwhile, for 32-bit PAGE_OFFSET is the virtual address of the start
  of lowmem, high_memory is the upper low virtual address, the check is
  suitable for 32-bit, this will fix the issue mentioned in commit
  602946ec2f90 ("powerpc: Set max_mapnr correctly") too.

On 32-bit there is a similar problem with high memory, that was fixed in
commit 602946ec2f90 ("powerpc: Set max_mapnr correctly"), but that
commit breaks highmem and needs to be reverted.

We can't easily fix __pa(), we have code that relies on its current
behaviour. So for now add extra checks to virt_addr_valid().

For 64-bit Book3S the extra checks are not necessary, the combination of
virt_to_pfn() and pfn_valid() should yield the correct result, but they
are harmless.

Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
[mpe: Add additional change log detail]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220406145802.538416-1-mpe@ellerman.id.au
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

static_call: Don't make __static_call_return0 static

commit 8fd4ddda2f49a66bf5dd3d0c01966c4b1971308b upstream.

System.map shows that vmlinux contains several instances of
__static_call_return0():

c0004fc0 t __static_call_return0
c0011518 t __static_call_return0
c00d8160 t __static_call_return0

arch_static_call_transform() uses the middle one to check whether we are
setting a call to __static_call_return0 or not:

c0011520 <arch_static_call_transform>:
c0011520:       3d 20 c0 01     lis     r9,-16383 <== r9 =  0xc001 << 16
c0011524:       39 29 15 18     addi    r9,r9,5400 <== r9 += 0x1518
c0011528:       7c 05 48 00     cmpw    r5,r9 <== r9 has value 0xc0011518 here

So if static_call_update() is called with one of the other instances of
__static_call_return0(), arch_static_call_transform() won't recognise it.

In order to work properly, global single instance of __static_call_return0() is required.

Fixes: 3f2a8fc4b15d ("static_call/x86: Add __static_call_return0()")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Link: https://lkml.kernel.org/r/30821468a0e7d28251954b578e5051dc09300d04.1647258493.git.christophe.leroy@csgroup.eu
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/sparsemem: fix 'mem_section' will never be NULL gcc 12 warning

commit a431dbbc540532b7465eae4fc8b56a85a9fc7d17 upstream.

The gcc 12 compiler reports a "'mem_section' will never be NULL" warning
on the following code:

    static inline struct mem_section *__nr_to_section(unsigned long nr)
    {
    #ifdef CONFIG_SPARSEMEM_EXTREME
        if (!mem_section)
                return NULL;
    #endif
        if (!mem_section[SECTION_NR_TO_ROOT(nr)])
                return NULL;
       :

It happens with CONFIG_SPARSEMEM_EXTREME off.  The mem_section definition
is

    #ifdef CONFIG_SPARSEMEM_EXTREME
    extern struct mem_section **mem_section;
    #else
    extern struct mem_section mem_section[NR_SECTION_ROOTS][SECTIONS_PER_ROOT];
    #endif

In the !CONFIG_SPARSEMEM_EXTREME case, mem_section is a static
2-dimensional array and so the check "!mem_section[SECTION_NR_TO_ROOT(nr)]"
doesn't make sense.

Fix this warning by moving the "!mem_section[SECTION_NR_TO_ROOT(nr)]"
check up inside the CONFIG_SPARSEMEM_EXTREME block and adding an
explicit NR_SECTION_ROOTS check to make sure that there is no
out-of-bound array access.

Link: https://lkml.kernel.org/r/20220331180246.2746210-1-longman@redhat.com
Fixes: 3e347261a80b ("sparsemem extreme implementation")
Signed-off-by: Waiman Long <longman@redhat.com>
Reported-by: Justin Forbes <jforbes@redhat.com>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

irqchip/gic, gic-v3: Prevent GSI to SGI translations

commit 544808f7e21cb9ccdb8f3aa7de594c05b1419061 upstream.

At the moment the GIC IRQ domain translation routine happily converts
ACPI table GSI numbers below 16 to GIC SGIs (Software Generated
Interrupts aka IPIs). On the Devicetree side we explicitly forbid this
translation, actually the function will never return HWIRQs below 16 when
using a DT based domain translation.

We expect SGIs to be handled in the first part of the function, and any
further occurrence should be treated as a firmware bug, so add a check
and print to report this explicitly and avoid lengthy debug sessions.

Fixes: 64b499d8df40 ("irqchip/gic-v3: Configure SGIs as standard interrupts")
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20220404110842.2882446-1-andre.przywara@arm.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

powerpc/64: Fix build failure with allyesconfig in book3s_64_entry.S

commit af41d2866f7d75bbb38d487f6ec7770425d70e45 upstream.

Using conditional branches between two files is hasardous,
they may get linked too far from each other.

  arch/powerpc/kvm/book3s_64_entry.o:(.text+0x3ec): relocation truncated
  to fit: R_PPC64_REL14 (stub) against symbol `system_reset_common'
  defined in .text section in arch/powerpc/kernel/head_64.o

Reorganise the code to use non conditional branches.

Fixes: 89d35b239101 ("KVM: PPC: Book3S HV P9: Implement the rest of the P9 path in C")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
[mpe: Avoid odd-looking bne ., use named local labels]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/89cf27bf43ee07a0b2879b9e8e2f5cd6386a3645.1648366338.git.christophe.leroy@csgroup.eu
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

irqchip/gic-v4: Wait for GICR_VPENDBASER.Dirty to clear before descheduling

commit af27e41612ec7e5b4783f589b753a7c31a37aac8 upstream.

The way KVM drives GICv4.{0,1} is as follows:
- vcpu_load() makes the VPE resident, instructing the RD to start
  scanning for interrupts
- just before entering the guest, we check that the RD has finished
  scanning and that we can start running the vcpu
- on preemption, we deschedule the VPE by making it invalid on
  the RD

However, we are preemptible between the first two steps. If it so
happens *and* that the RD was still scanning, we nonetheless write
to the GICR_VPENDBASER register while Dirty is set, and bad things
happen (we're in UNPRED land).

This affects both the 4.0 and 4.1 implementations.

Make sure Dirty is cleared before performing the deschedule,
meaning that its_clear_vpend_valid() becomes a sort of full VPE
residency barrier.

Reported-by: Jingyi Wang <wangjingyi11@huawei.com>
Tested-by: Nianyao Tang <tangnianyao@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Fixes: 57e3cebd022f ("KVM: arm64: Delay the polling of the GICR_VPENDBASER.Dirty bit")
Link: https://lore.kernel.org/r/4aae10ba-b39a-5f84-754b-69c2eb0a2c03@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86,static_call: Fix __static_call_return0 for i386

commit 1cd5f059d956e6f614ba6666ecdbcf95db05d5f5 upstream.

Paolo reported that the instruction sequence that is used to replace:

    call __static_call_return0

namely:

    66 66 48 31 c0 data16 data16 xor %rax,%rax

decodes to something else on i386, namely:

    66 66 48 data16 dec %ax
    31 c0 xor    %eax,%eax

Which is a nonsensical sequence that happens to have the same outcome.
*However* an important distinction is that it consists of 2
instructions which is a problem when the thing needs to be overwriten
with a regular call instruction again.

As such, replace the instruction with something that decodes the same
on both i386 and x86_64.

Fixes: 3f2a8fc4b15d ("static_call/x86: Add __static_call_return0()")
Reported-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220318204419.GT8939@worktop.programming.kicks-ass.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

sched: Teach the forced-newidle balancer about CPU affinity limitation.

commit 386ef214c3c6ab111d05e1790e79475363abaa05 upstream.

try_steal_cookie() looks at task_struct::cpus_mask to decide if the
task could be moved to `this' CPU. It ignores that the task might be in
a migration disabled section while not on the CPU. In this case the task
must not be moved otherwise per-CPU assumption are broken.

Use is_cpu_allowed(), as suggested by Peter Zijlstra, to decide if the a
task can be moved.

Fixes: d2dfa17bc7de6 ("sched: Trivial forced-newidle balancer")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/YjNK9El+3fzGmswf@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/bug: Prevent shadowing in __WARN_FLAGS

commit 9ce02f0fc68326dd1f87a0a3a4c6ae7fdd39e6f6 upstream.

The macro __WARN_FLAGS() uses a local variable named "f". This being a
common name, there is a risk of shadowing other variables.

For example, GCC would yield:

| In file included from ./include/linux/bug.h:5,
|                  from ./include/linux/cpumask.h:14,
|                  from ./arch/x86/include/asm/cpumask.h:5,
|                  from ./arch/x86/include/asm/msr.h:11,
|                  from ./arch/x86/include/asm/processor.h:22,
|                  from ./arch/x86/include/asm/timex.h:5,
|                  from ./include/linux/timex.h:65,
|                  from ./include/linux/time32.h:13,
|                  from ./include/linux/time.h:60,
|                  from ./include/linux/stat.h:19,
|                  from ./include/linux/module.h:13,
|                  from virt/lib/irqbypass.mod.c:1:
| ./include/linux/rcupdate.h: In function 'rcu_head_after_call_rcu':
| ./arch/x86/include/asm/bug.h:80:21: warning: declaration of 'f' shadows a parameter [-Wshadow]
|    80 |         __auto_type f = BUGFLAG_WARNING|(flags);                \
|       |                     ^
| ./include/asm-generic/bug.h:106:17: note: in expansion of macro '__WARN_FLAGS'
|   106 |                 __WARN_FLAGS(BUGFLAG_ONCE |                     \
|       |                 ^~~~~~~~~~~~
| ./include/linux/rcupdate.h:1007:9: note: in expansion of macro 'WARN_ON_ONCE'
|  1007 |         WARN_ON_ONCE(func != (rcu_callback_t)~0L);
|       |         ^~~~~~~~~~~~
| In file included from ./include/linux/rbtree.h:24,
|                  from ./include/linux/mm_types.h:11,
|                  from ./include/linux/buildid.h:5,
|                  from ./include/linux/module.h:14,
|                  from virt/lib/irqbypass.mod.c:1:
| ./include/linux/rcupdate.h:1001:62: note: shadowed declaration is here
|  1001 | rcu_head_after_call_rcu(struct rcu_head *rhp, rcu_callback_t f)
|       |                                               ~~~~~~~~~~~~~~~^

For reference, sparse also warns about it, c.f. [1].

This patch renames the variable from f to __flags (with two underscore
prefixes as suggested in the Linux kernel coding style [2]) in order
to prevent collisions.

[1] https://lore.kernel.org/all/CAFGhKbyifH1a+nAMCvWM88TK6fpNPdzFtUXPmRGnnQeePV+1sw@mail.gmail.com/

[2] Linux kernel coding style, section 12) Macros, Enums and RTL,
paragraph 5) namespace collisions when defining local variables in
macros resembling functions
https://www.kernel.org/doc/html/latest/process/coding-style.html#macros-enums-and-rtl

Fixes: bfb1a7c91fb7 ("x86/bug: Merge annotate_reachable() into_BUG_FLAGS() asm")
Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Link: https://lkml.kernel.org/r/20220324023742.106546-1-mailhol.vincent@wanadoo.fr
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Drivers: hv: vmbus: Replace smp_store_mb() with virt_store_mb()

commit eaa03d34535872d29004cb5cf77dc9dec1ba9a25 upstream.

Following the recommendation in Documentation/memory-barriers.txt for
virtual machine guests.

Fixes: 8b6a877c060ed ("Drivers: hv: vmbus: Replace the per-CPU channel lists with a global array of channels")
Signed-off-by: Andrea Parri (Microsoft) <parri.andrea@gmail.com>
Link: https://lore.kernel.org/r/20220328154457.100872-1-parri.andrea@gmail.com
Signed-off-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm: don't skip swap entry even if zap_details specified

commit 5abfd71d936a8aefd9f9ccd299dea7a164a5d455 upstream.

Patch series "mm: Rework zap ptes on swap entries", v5.

Patch 1 should fix a long standing bug for zap_pte_range() on
zap_details usage.  The risk is we could have some swap entries skipped
while we should have zapped them.

Migration entries are not the major concern because file backed memory
always zap in the pattern that "first time without page lock, then
re-zap with page lock" hence the 2nd zap will always make sure all
migration entries are already recovered.

However there can be issues with real swap entries got skipped
errornoously.  There's a reproducer provided in commit message of patch
1 for that.

Patch 2-4 are cleanups that are based on patch 1.  After the whole
patchset applied, we should have a very clean view of zap_pte_range().

Only patch 1 needs to be backported to stable if necessary.

This patch (of 4):

The "details" pointer shouldn't be the token to decide whether we should
skip swap entries.

For example, when the callers specified details->zap_mapping==NULL, it
means the user wants to zap all the pages (including COWed pages), then
we need to look into swap entries because there can be private COWed
pages that was swapped out.

Skipping some swap entries when details is non-NULL may lead to wrongly
leaving some of the swap entries while we should have zapped them.

A reproducer of the problem:

===8<===
        #define _GNU_SOURCE         /* See feature_test_macros(7) */
        #include <stdio.h>
        #include <assert.h>
        #include <unistd.h>
        #include <sys/mman.h>
        #include <sys/types.h>

        int page_size;
        int shmem_fd;
        char *buffer;

        void main(void)
        {
                int ret;
                char val;

                page_size = getpagesize();
                shmem_fd = memfd_create("test", 0);
                assert(shmem_fd >= 0);

                ret = ftruncate(shmem_fd, page_size * 2);
                assert(ret == 0);

                buffer = mmap(NULL, page_size * 2, PROT_READ | PROT_WRITE,
                                MAP_PRIVATE, shmem_fd, 0);
                assert(buffer != MAP_FAILED);

                /* Write private page, swap it out */
                buffer[page_size] = 1;
                madvise(buffer, page_size * 2, MADV_PAGEOUT);

                /* This should drop private buffer[page_size] already */
                ret = ftruncate(shmem_fd, page_size);
                assert(ret == 0);
                /* Recover the size */
                ret = ftruncate(shmem_fd, page_size * 2);
                assert(ret == 0);

                /* Re-read the data, it should be all zero */
                val = buffer[page_size];
                if (val == 0)
                        printf("Good\n");
                else
                        printf("BUG\n");
        }
===8<===

We don't need to touch up the pmd path, because pmd never had a issue with
swap entries.  For example, shmem pmd migration will always be split into
pte level, and same to swapping on anonymous.

Add another helper should_zap_cows() so that we can also check whether we
should zap private mappings when there's no page pointer specified.

This patch drops that trick, so we handle swap ptes coherently.  Meanwhile
we should do the same check upon migration entry, hwpoison entry and
genuine swap entries too.

To be explicit, we should still remember to keep the private entries if
even_cows==false, and always zap them when even_cows==true.

The issue seems to exist starting from the initial commit of git.

[peterx@redhat.com: comment tweaks]
Link: https://lkml.kernel.org/r/20220217060746.71256-2-peterx@redhat.com
Link: https://lkml.kernel.org/r/20220217060746.71256-1-peterx@redhat.com
Link: https://lkml.kernel.org/r/20220216094810.60572-1-peterx@redhat.com
Link: https://lkml.kernel.org/r/20220216094810.60572-2-peterx@redhat.com
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: John Hubbard <jhubbard@nvidia.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Kirill A . Shutemov" <kirill@shutemov.name>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yang Shi <shy828301@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests: cgroup: Test open-time cgroup namespace usage for migration checks

commit bf35a7879f1dfb0d050fe779168bcf25c7de66f5 upstream.

When a task is writing to an fd opened by a different task, the perm check
should use the cgroup namespace of the latter task. Add a test for it.

Tested-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests: cgroup: Test open-time credential usage for migration checks

commit 613e040e4dc285367bff0f8f75ea59839bc10947 upstream.

When a task is writing to an fd opened by a different task, the perm check
should use the credentials of the latter task. Add a test for it.

Tested-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests: cgroup: Make cg_create() use 0755 for permission instead of 0644

commit b09c2baa56347ae65795350dfcc633dedb1c2970 upstream.

0644 is an odd perm to create a cgroup which is a directory. Use the regular
0755 instead. This is necessary for euid switching test case.

Reviewed-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Ovidiu Panait <ovidiu.panait@windriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ubsan: remove CONFIG_UBSAN_OBJECT_SIZE

commit 69d0db01e210e07fe915e5da91b54a867cda040f upstream.

The object-size sanitizer is redundant to -Warray-bounds, and
inappropriately performs its checks at run-time when all information
needed for the evaluation is available at compile-time, making it quite
difficult to use:

  https://bugzilla.kernel.org/show_bug.cgi?id=214861

With -Warray-bounds almost enabled globally, it doesn't make sense to
keep this around.

Link: https://lkml.kernel.org/r/20211203235346.110809-1-keescook@chromium.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Marco Elver <elver@google.com>
Cc: Masahiro Yamada <masahiroy@kernel.org>
Cc: Michal Marek <michal.lkml@markovi.net>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: "Peter Zijlstra (Intel)" <peterz@infradead.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Tadeusz Struk <tadeusz.struk@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "net/mlx5: Accept devlink user input after driver initialization complete"

This reverts commit 9cc25e8529d567e08da98d11f092b21449763144 which is
commit 64ea2d0e7263b67d8efc93fa1baace041ed36d1e upstream.

This patch was shown to introduce a regression:

  # devlink dev param show pci/0000:24:00.0 name flow_steering_mode
  pci/0000:24:00.0:
    name flow_steering_mode type driver-specific
      values:

  (flow steering mode description is missing beneath "values:")

  # devlink dev param set pci/0000:24:00.0 name flow_steering_mode value smfs cmode runtime
  Segmentation fault (core dumped)

  and also with upstream iproute
  # ./iproute2/devlink/devlink dev param set pci/0000:24:00.0 name flow_steering_mode value smfs cmode runtime
  Configuration mode not supported

Note: Instead of reverting, we could instead also backport commit cf530217408e
("devlink: Notify users when objects are accessible"). However, that makes
changes to core devlink code that I'm not sure are suitable for a stable
backport.

Cc: Leon Romanovsky <leonro@nvidia.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Sasha Levin <sashal@kernel.org>
Signed-off-by: dann frazier <dann.frazier@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

KVM: avoid NULL pointer dereference in kvm_dirty_ring_push

commit 5593473a1e6c743764b08e3b6071cb43b5cfa6c4 upstream.

kvm_vcpu_release() will call kvm_dirty_ring_free(), freeing
ring->dirty_gfns and setting it to NULL.  Afterwards, it calls
kvm_arch_vcpu_destroy().

However, if closing the file descriptor races with KVM_RUN in such away
that vcpu->arch.st.preempted == 0, the following call stack leads to a
NULL pointer dereference in kvm_dirty_run_push():

 mark_page_dirty_in_slot+0x192/0x270 arch/x86/kvm/../../../virt/kvm/kvm_main.c:3171
 kvm_steal_time_set_preempted arch/x86/kvm/x86.c:4600 [inline]
 kvm_arch_vcpu_put+0x34e/0x5b0 arch/x86/kvm/x86.c:4618
 vcpu_put+0x1b/0x70 arch/x86/kvm/../../../virt/kvm/kvm_main.c:211
 vmx_free_vcpu+0xcb/0x130 arch/x86/kvm/vmx/vmx.c:6985
 kvm_arch_vcpu_destroy+0x76/0x290 arch/x86/kvm/x86.c:11219
 kvm_vcpu_destroy arch/x86/kvm/../../../virt/kvm/kvm_main.c:441 [inline]

The fix is to release the dirty page ring after kvm_arch_vcpu_destroy
has run.

Reported-by: Qiuhao Li <qiuhao@sysec.org>
Reported-by: Gaoning Pan <pgn@zju.edu.cn>
Reported-by: Yongkang Jia <kangel@zju.edu.cn>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

dmaengine: Revert "dmaengine: shdma: Fix runtime PM imbalance on error"

commit d143f939a95696d38ff800ada14402fa50ebbd6c upstream.

This reverts commit 455896c53d5b ("dmaengine: shdma: Fix runtime PM
imbalance on error") as the patch wrongly reduced the count on error and
did not bail out. So drop the count by reverting the patch .

Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tools build: Use $(shell ) instead of `` to get embedded libperl's ccopts

commit 541f695cbcb6932c22638b06e0cbe1d56177e2e9 upstream.

Just like its done for ldopts and for both in tools/perf/Makefile.config.

Using `` to initialize PERL_EMBED_CCOPTS somehow precludes using:

  $(filter-out SOMETHING_TO_FILTER,$(PERL_EMBED_CCOPTS))

And we need to do it to allow for building with versions of clang where
some gcc options selected by distros are not available.

Tested-by: Sedat Dilek <sedat.dilek@gmail.com> # Debian/Selfmade LLVM-14 (x86-64)
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Keeping <john@metanate.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Link: http://lore.kernel.org/lkml/YktYX2OnLtyobRYD@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tools build: Filter out options and warnings not supported by clang

commit 41caff459a5b956b3e23ba9ca759dd0629ad3dda upstream.

These make the feature check fail when using clang, so remove them just
like is done in tools/perf/Makefile.config to build perf itself.

Adding -Wno-compound-token-split-by-macro to tools/perf/Makefile.config
when building with clang is also necessary to avoid these warnings
turned into errors (-Werror):

    CC      /tmp/build/perf/util/scripting-engines/trace-event-perl.o
  In file included from util/scripting-engines/trace-event-perl.c:35:
  In file included from /usr/lib64/perl5/CORE/perl.h:4085:
  In file included from /usr/lib64/perl5/CORE/hv.h:659:
  In file included from /usr/lib64/perl5/CORE/hv_func.h:34:
  In file included from /usr/lib64/perl5/CORE/sbox32_hash.h:4:
  /usr/lib64/perl5/CORE/zaphod32_hash.h:150:5: error: '(' and '{' tokens introducing statement expression appear in different macro expansion contexts [-Werror,-Wcompound-token-split-by-macro]
      ZAPHOD32_SCRAMBLE32(state[0],0x9fade23b);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  /usr/lib64/perl5/CORE/zaphod32_hash.h:80:38: note: expanded from macro 'ZAPHOD32_SCRAMBLE32'
  #define ZAPHOD32_SCRAMBLE32(v,prime) STMT_START {  \
                                       ^~~~~~~~~~
  /usr/lib64/perl5/CORE/perl.h:737:29: note: expanded from macro 'STMT_START'
  #   define STMT_START   (void)( /* gcc supports "({ STATEMENTS; })" */
                                ^
  /usr/lib64/perl5/CORE/zaphod32_hash.h:150:5: note: '{' token is here
      ZAPHOD32_SCRAMBLE32(state[0],0x9fade23b);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  /usr/lib64/perl5/CORE/zaphod32_hash.h:80:49: note: expanded from macro 'ZAPHOD32_SCRAMBLE32'
  #define ZAPHOD32_SCRAMBLE32(v,prime) STMT_START {  \
                                                  ^
  /usr/lib64/perl5/CORE/zaphod32_hash.h:150:5: error: '}' and ')' tokens terminating statement expression appear in different macro expansion contexts [-Werror,-Wcompound-token-split-by-macro]
      ZAPHOD32_SCRAMBLE32(state[0],0x9fade23b);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  /usr/lib64/perl5/CORE/zaphod32_hash.h:87:41: note: expanded from macro 'ZAPHOD32_SCRAMBLE32'
      v ^= (v>>23);                       \
                                          ^
  /usr/lib64/perl5/CORE/zaphod32_hash.h:150:5: note: ')' token is here
      ZAPHOD32_SCRAMBLE32(state[0],0x9fade23b);
      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  /usr/lib64/perl5/CORE/zaphod32_hash.h:88:3: note: expanded from macro 'ZAPHOD32_SCRAMBLE32'
  } STMT_END
    ^~~~~~~~
  /usr/lib64/perl5/CORE/perl.h:738:21: note: expanded from macro 'STMT_END'
  #   define STMT_END     )
                          ^

Please refer to the discussion on the Link: tag below, where Nathan
clarifies the situation:

<quote>
acme> And then get to the problems at the end of this message, which seem
acme> similar to the problem described here:
acme>
acme> From  Nathan Chancellor <>
acme> Subject [PATCH] mwifiex: Remove unnecessary braces from HostCmd_SET_SEQ_NO_BSS_INFO
acme>
acme> https://lkml.org/lkml/2020/9/1/135
acme>
acme> So perhaps in this case its better to disable that
acme> -Werror,-Wcompound-token-split-by-macro when building with clang?

Yes, I think that is probably the best solution. As far as I can tell,
at least in this file and context, the warning appears harmless, as the
"create a GNU C statement expression from two different macros" is very
much intentional, based on the presence of PERL_USE_GCC_BRACE_GROUPS.
The warning is fixed in upstream Perl by just avoiding creating GNU C
statement expressions using STMT_START and STMT_END:

  https://github.com/Perl/perl5/issues/18780
  https://github.com/Perl/perl5/pull/18984

If I am reading the source code correctly, an alternative to disabling
the warning would be specifying -DPERL_GCC_BRACE_GROUPS_FORBIDDEN but it
seems like that might end up impacting more than just this site,
according to the issue discussion above.
</quote>

Based-on-a-patch-by: Sedat Dilek <sedat.dilek@gmail.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com> # Debian/Selfmade LLVM-14 (x86-64)
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Keeping <john@metanate.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Link: http://lore.kernel.org/lkml/YkxWcYzph5pC1EK8@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf python: Fix probing for some clang command line options

commit dd6e1fe91cdd52774ca642d1da75b58a86356b56 upstream.

The clang compiler complains about some options even without a source
file being available, while others require one, so use the simple
tools/build/feature/test-hello.c file.

Then check for the "is not supported" string in its output, in addition
to the "unknown argument" already being looked for.

This was noticed when building with clang-13 where -ffat-lto-objects
isn't supported and since we were looking just for "unknown argument"
and not providing a source code to clang, was mistakenly assumed as
being available and not being filtered to set of command line options
provided to clang, leading to a build failure.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Keeping <john@metanate.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Sedat Dilek <sedat.dilek@gmail.com>
Link: http://lore.kernel.org/lkml/
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf build: Don't use -ffat-lto-objects in the python feature test when building with clang-13

commit 3a8a0475861a443f02e3a9b57d044fe2a0a99291 upstream.

Using -ffat-lto-objects in the python feature test when building with
clang-13 results in:

  clang-13: error: optimization flag '-ffat-lto-objects' is not supported [-Werror,-Wignored-optimization-argument]
  error: command '/usr/sbin/clang' failed with exit code 1
  cp: cannot stat '/tmp/build/perf/python_ext_build/lib/perf*.so': No such file or directory
  make[2]: *** [Makefile.perf:639: /tmp/build/perf/python/perf.so] Error 1

Noticed when building on a docker.io/library/archlinux:base container.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Keeping <john@metanate.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "nbd: fix possible overflow on 'first_minor' in nbd_dev_add()"

commit 7198bfc2017644c6b92d2ecef9b8b8e0363bb5fd upstream.

This reverts commit 6d35d04a9e18990040e87d2bbf72689252669d54.

Both Gabriel and Borislav report that this commit casues a regression
with nbd:

sysfs: cannot create duplicate filename '/dev/block/43:0'

Revert it before 5.18-rc1 and we'll investigage this separately in
due time.

Link: https://lore.kernel.org/all/YkiJTnFOt9bTv6A2@zn.tnic/
Reported-by: Gabriel L. Somlo <somlo@cmu.edu>
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

SUNRPC: Don't call connect() more than once on a TCP socket

commit 89f42494f92f448747bd8a7ab1ae8b5d5520577d upstream.

Avoid socket state races due to repeated calls to ->connect() using the
same socket. If connect() returns 0 due to the connection having
completed, but we are in fact in a closing state, then we may leave the
XPRT_CONNECTING flag set on the transport.

Reported-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
Fixes: 3be232f11a3c ("SUNRPC: Prevent immediate close+reconnect")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

rtc: mc146818-lib: fix signedness bug in mc146818_get_time()

commit 7372971c1be5b7d4fdd8ad237798bdc1d1d54162 upstream.

The mc146818_get_time() function returns zero on success or negative
a error code on failure.  It needs to be type int.

Fixes: d35786b3a28d ("rtc: mc146818-lib: change return values of mc146818_get_time()")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Mateusz Jończyk <mat.jonczyk@o2.pl>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Link: https://lore.kernel.org/r/20220111071922.GE11243@kili
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests/bpf: Fix u8 narrow load checks for bpf_sk_lookup remote_port

commit 3c69611b8926f8e74fcf76bd97ae0e5dafbeb26a upstream.

In commit 9a69e2b385f4 ("bpf: Make remote_port field in struct
bpf_sk_lookup 16-bit wide") ->remote_port field changed from __u32 to
__be16.

However, narrow load tests which exercise 1-byte sized loads from
offsetof(struct bpf_sk_lookup, remote_port) were not adopted to reflect the
change.

As a result, on little-endian we continue testing loads from addresses:

 - (__u8 *)&ctx->remote_port + 3
 - (__u8 *)&ctx->remote_port + 4

which map to the zero padding following the remote_port field, and don't
break the tests because there is no observable change.

While on big-endian, we observe breakage because tests expect to see zeros
for values loaded from:

 - (__u8 *)&ctx->remote_port - 1
 - (__u8 *)&ctx->remote_port - 2

Above addresses map to ->remote_ip6 field, which precedes ->remote_port,
and are populated during the bpf_sk_lookup IPv6 tests.

Unsurprisingly, on s390x we observe:

  #136/38 sk_lookup/narrow access to ctx v4:OK
  #136/39 sk_lookup/narrow access to ctx v6:FAIL

Fix it by removing the checks for 1-byte loads from offsets outside of the
->remote_port field.

Fixes: 9a69e2b385f4 ("bpf: Make remote_port field in struct bpf_sk_lookup 16-bit wide")
Suggested-by: Ilya Leoshkevich <iii@linux.ibm.com>
Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Link: https://lore.kernel.org/bpf/20220319183356.233666-3-jakub@cloudflare.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bpf: Make remote_port field in struct bpf_sk_lookup 16-bit wide

commit 9a69e2b385f443f244a7e8b8bcafe5ccfb0866b4 upstream.

remote_port is another case of a BPF context field documented as a 32-bit
value in network byte order for which the BPF context access converter
generates a load of a zero-padded 16-bit integer in network byte order.

First such case was dst_port in bpf_sock which got addressed in commit
4421a582718a ("bpf: Make dst_port field in struct bpf_sock 16-bit wide").

Loading 4-bytes from the remote_port offset and converting the value with
bpf_ntohl() leads to surprising results, as the expected value is shifted
by 16 bits.

Reduce the confusion by splitting the field in two - a 16-bit field holding
a big-endian integer, and a 16-bit zero-padding anonymous field that
follows it.

Suggested-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20220209184333.654927-2-jakub@cloudflare.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "selftests: net: Add tls config dependency for tls selftests"

commit 20695e9a9fd39103d1b0669470ae74030b7aa196 upstream.

This reverts commit d9142e1cf3bbdaf21337767114ecab26fe702d47.

The test is supposed to run cleanly with TLS is disabled,
to test compatibility with TCP behavior. I can't repro
the failure [1], the problem should be debugged rather
than papered over.

Link: https://lore.kernel.org/all/20220325161203.7000698c@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com/
Fixes: d9142e1cf3bb ("selftests: net: Add tls config dependency for tls selftests")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/r/20220328212904.2685395-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net/smc: send directly on setting TCP_NODELAY

commit b70a5cc045197aad9c159042621baf3c015f6cc7 upstream.

In commit ea785a1a573b("net/smc: Send directly when
TCP_CORK is cleared"), we don't use delayed work
to implement cork.

This patch use the same algorithm, removes the
delayed work when setting TCP_NODELAY and send
directly in setsockopt(). This also makes the
TCP_NODELAY the same as TCP.

Cc: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: Dust Li <dust.li@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

KVM: SVM: Allow AVIC support on system w/ physical APIC ID > 255

commit 4a204f7895878363ca8211f50ec610408c8c70aa upstream.

Expand KVM's mask for the AVIC host physical ID to the full 12 bits defined
by the architecture.  The number of bits consumed by hardware is model
specific, e.g. early CPUs ignored bits 11:8, but there is no way for KVM
to enumerate the "true" size.  So, KVM must allow using all bits, else it
risks rejecting completely legal x2APIC IDs on newer CPUs.

This means KVM relies on hardware to not assign x2APIC IDs that exceed the
"true" width of the field, but presumably hardware is smart enough to tie
the width to the max x2APIC ID.  KVM also relies on hardware to support at
least 8 bits, as the legacy xAPIC ID is writable by software.  But, those
assumptions are unavoidable due to the lack of any way to enumerate the
"true" width.

Cc: stable@vger.kernel.org
Cc: Maxim Levitsky <mlevitsk@redhat.com>
Suggested-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Sean Christopherson <seanjc@google.com>
Fixes: 44a95dae1d22 ("KVM: x86: Detect and Initialize AVIC support")
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Message-Id: <20220211000851.185799-1-suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[modified due to the conflict caused by the commit 391503528257 ("KVM:
x86: SVM: move avic definitions from AMD's spec to svm.h")]
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amdgpu: don't use BACO for reset in S3

commit ebc002e3ee78409c42156e62e4e27ad1d09c5a75 upstream.

Seems to cause a reboots or hangs on some systems.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1924
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1953
Fixes: daf8de0874ab5b ("drm/amdgpu: always reset the asic in suspend (v2)")
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amdkfd: Create file descriptor after client is added to smi_clients list

commit e79a2398e1b2d47060474dca291542368183bc0f upstream.

This ensures userspace cannot prematurely clean-up the client before
it is fully initialised which has been proven to cause issues in the
past.

Cc: Felix Kuehling <Felix.Kuehling@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: "Pan, Xinhui" <Xinhui.Pan@amd.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: amd-gfx@lists.freedesktop.org
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/nouveau/pmu: Add missing callbacks for Tegra devices

commit 38d4e5cf5b08798f093374e53c2f4609d5382dd5 upstream.

Fixes a crash booting on those platforms with nouveau.

Fixes: 4cdd2450bf73 ("drm/nouveau/pmu/gm200-: use alternate falcon reset sequence")
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Karol Herbst <kherbst@redhat.com>
Cc: dri-devel@lists.freedesktop.org
Cc: nouveau@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v5.17+
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220322124800.2605463-1-kherbst@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amdgpu/vcn: Fix the register setting for vcn1

commit 02fc996d5098f4c3f65bdf6cdb6b28e3f29ba789 upstream.

Correct the code error for setting register UVD_GFX10_ADDR_CONFIG.
Need to use inst_idx, or it only will set VCN0.

Signed-off-by: Emily Deng <Emily.Deng@amd.com>
Reviewed-by: James Zhu <James.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amdgpu/smu10: fix SoC/fclk units in auto mode

commit 2f25d8ce09b7ba5d769c132ba3d4eb84a941d2cb upstream.

SMU takes clock limits in Mhz units.  socclk and fclk were
using 10 khz units in some cases.  Switch to Mhz units.
Fixes higher than required SoC clocks.

Fixes: 97cf32996c46d9 ("drm/amd/pm: Removed fixed clock in auto mode DPM")
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amdgpu/display: change pipe policy for DCN 2.1

commit 879791ad8bf3dc5453061cad74776a617b6e3319 upstream.

Fixes crash on MST Hub disconnect.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1849
Fixes: ee2698cf79cc ("drm/amd/display: Changed pipe split policy to allow for multi-display pipe split")
Signed-off-by: Benjamin Marty <info@benjaminmarty.ch>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/panel: ili9341: fix optional regulator handling

commit d14eb80e27795b7b20060f7b151cdfe39722a813 upstream.

If the optional regulator lookup fails, reset the pointer to NULL.
Other functions such as mipi_dbi_poweron_reset_conditional() only do
a NULL pointer check and will otherwise dereference the error pointer.

Fixes: 5a04227326b04c15 ("drm/panel: Add ilitek ili9341 panel driver")
Signed-off-by: Daniel Mack <daniel@zonque.org>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20220317225537.826302-1-daniel@zonque.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

SUNRPC: Prevent immediate close+reconnect

commit 3be232f11a3cc9b0ef0795e39fa11bdb8e422a06 upstream.

If we have already set up the socket and are waiting for it to connect,
then don't immediately close and retry.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

amd/display: set backlight only if required

commit 4052287a75eb3fc0f487fcc5f768a38bede455c8 upstream.

[Why]
comparing pwm bl values (coverted) with user brightness(converted)
levels in commit_tail leads to continuous setting of backlight via dmub
as they don't to match.
This leads overdrive in queuing of commands to DMCU that sometimes lead
to depending on load on DMCU fw:

"[drm:dc_dmub_srv_wait_idle] *ERROR* Error waiting for DMUB idle: status=3"

[How]
Store last successfully set backlight value and compare with it instead
of pwm reads which is not what we should compare with.

Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fbdev: Fix unregistering of framebuffers without device

commit 0f525289ff0ddeb380813bd81e0f9bdaaa1c9078 upstream.

OF framebuffers do not have an underlying device in the Linux
device hierarchy. Do a regular unregister call instead of hot
unplugging such a non-existing device. Fixes a NULL dereference.
An example error message on ppc64le is shown below.

  BUG: Kernel NULL pointer dereference on read at 0x00000060
  Faulting instruction address: 0xc00000000080dfa4
  Oops: Kernel access of bad area, sig: 11 [#1]
  LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
  [...]
  CPU: 2 PID: 139 Comm: systemd-udevd Not tainted 5.17.0-ae085d7f9365 #1
  NIP:  c00000000080dfa4 LR: c00000000080df9c CTR: c000000000797430
  REGS: c000000004132fe0 TRAP: 0300   Not tainted  (5.17.0-ae085d7f9365)
  MSR:  8000000002009033 <SF,VEC,EE,ME,IR,DR,RI,LE>  CR: 28228282  XER: 20000000
  CFAR: c00000000000c80c DAR: 0000000000000060 DSISR: 40000000 IRQMASK: 0
  GPR00: c00000000080df9c c000000004133280 c00000000169d200 0000000000000029
  GPR04: 00000000ffffefff c000000004132f90 c000000004132f88 0000000000000000
  GPR08: c0000000015658f8 c0000000015cd200 c0000000014f57d0 0000000048228283
  GPR12: 0000000000000000 c00000003fffe300 0000000020000000 0000000000000000
  GPR16: 0000000000000000 0000000113fc4a40 0000000000000005 0000000113fcfb80
  GPR20: 000001000f7283b0 0000000000000000 c000000000e4a588 c000000000e4a5b0
  GPR24: 0000000000000001 00000000000a0000 c008000000db0168 c0000000021f6ec0
  GPR28: c0000000016d65a8 c000000004b36460 0000000000000000 c0000000016d64b0
  NIP [c00000000080dfa4] do_remove_conflicting_framebuffers+0x184/0x1d0
  [c000000004133280] [c00000000080df9c] do_remove_conflicting_framebuffers+0x17c/0x1d0 (unreliable)
  [c000000004133350] [c00000000080e4d0] remove_conflicting_framebuffers+0x60/0x150
  [c0000000041333a0] [c00000000080e6f4] remove_conflicting_pci_framebuffers+0x134/0x1b0
  [c000000004133450] [c008000000e70438] drm_aperture_remove_conflicting_pci_framebuffers+0x90/0x100 [drm]
  [c000000004133490] [c008000000da0ce4] bochs_pci_probe+0x6c/0xa64 [bochs]
  [...]
  [c000000004133db0] [c00000000002aaa0] system_call_exception+0x170/0x2d0
  [c000000004133e10] [c00000000000c3cc] system_call_common+0xec/0x250

The bug [1] was introduced by commit 27599aacbaef ("fbdev: Hot-unplug
firmware fb devices on forced removal"). Most firmware framebuffers
have an underlying platform device, which can be hot-unplugged
before loading the native graphics driver. OF framebuffers do not
(yet) have that device. Fix the code by unregistering the framebuffer
as before without a hot unplug.

Tested with 5.17 on qemu ppc64le emulation.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Fixes: 27599aacbaef ("fbdev: Hot-unplug firmware fb devices on forced removal")
Reported-by: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Tested-by: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk>
Cc: Zack Rusin <zackr@vmware.com>
Cc: Javier Martinez Canillas <javierm@redhat.com>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: stable@vger.kernel.org # v5.11+
Cc: Helge Deller <deller@gmx.de>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Zheyu Ma <zheyuma97@gmail.com>
Cc: Xiyu Yang <xiyuyang19@fudan.edu.cn>
Cc: Zhen Lei <thunder.leizhen@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: linux-fbdev@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Link: https://lore.kernel.org/all/YkHXO6LGHAN0p1pq@debian/
Link: https://patchwork.freedesktop.org/patch/msgid/20220404194402.29974-1-tzimmermann@suse.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

irqchip/gic-v3: Fix GICR_CTLR.RWP polling

commit 0df6664531a12cdd8fc873f0cac0dcb40243d3e9 upstream.

It turns out that our polling of RWP is totally wrong when checking
for it in the redistributors, as we test the *distributor* bit index,
whereas it is a different bit number in the RDs... Oopsie boo.

This is embarassing. Not only because it is wrong, but also because
it took *8 years* to notice the blunder...

Just fix the damn thing.

Fixes: 021f653791ad ("irqchip: gic-v3: Initial support for GICv3")
Signed-off-by: Marc Zyngier <maz@kernel.org>
Cc: stable@vger.kernel.org
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Link: https://lore.kernel.org/r/20220315165034.794482-2-maz@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf/core: Inherit event_caps

commit e3265a4386428d3d157d9565bb520aabff8b4bf0 upstream.

It was reported that some perf event setup can make fork failed on
ARM64.  It was the case of a group of mixed hw and sw events and it
failed in perf_event_init_task() due to armpmu_event_init().

The ARM PMU code checks if all the events in a group belong to the
same PMU except for software events.  But it didn't set the event_caps
of inherited events and no longer identify them as software events.
Therefore the test failed in a child process.

A simple reproducer is:

  $ perf stat -e '{cycles,cs,instructions}' perf bench sched messaging
  # Running 'sched/messaging' benchmark:
  perf: fork(): Invalid argument

The perf stat was fine but the perf bench failed in fork().  Let's
inherit the event caps from the parent.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: <stable@vger.kernel.org>
Link: https://lkml.kernel.org/r/20220328200112.457740-1-namhyung@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf: qcom_l2_pmu: fix an incorrect NULL check on list iterator

commit 2012a9e279013933885983cbe0a5fe828052563b upstream.

The bug is here:
return cluster;

The list iterator value 'cluster' will *always* be set and non-NULL
by list_for_each_entry(), so it is incorrect to assume that the
iterator value will be NULL if the list is empty or no element
is found.

To fix the bug, return 'cluster' when found, otherwise return NULL.

Cc: stable@vger.kernel.org
Fixes: 21bdbb7102ed ("perf: add qcom l2 cache perf events driver")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Link: https://lore.kernel.org/r/20220327055733.4070-1-xiam0nd.tong@gmail.com
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ata: sata_dwc_460ex: Fix crash due to OOB write

commit 7aa8104a554713b685db729e66511b93d989dd6a upstream.

the driver uses libata's "tag" values from in various arrays.
Since the mentioned patch bumped the ATA_TAG_INTERNAL to 32,
the value of the SATA_DWC_QCMD_MAX needs to account for that.

Otherwise ATA_TAG_INTERNAL usage cause similar crashes like
this as reported by Tice Rex on the OpenWrt Forum and
reproduced (with symbols) here:

| BUG: Kernel NULL pointer dereference at 0x00000000
| Faulting instruction address: 0xc03ed4b8
| Oops: Kernel access of bad area, sig: 11 [#1]
| BE PAGE_SIZE=4K PowerPC 44x Platform
| CPU: 0 PID: 362 Comm: scsi_eh_1 Not tainted 5.4.163 #0
| NIP:  c03ed4b8 LR: c03d27e8 CTR: c03ed36c
| REGS: cfa59950 TRAP: 0300   Not tainted  (5.4.163)
| MSR:  00021000 <CE,ME>  CR: 42000222  XER: 00000000
| DEAR: 00000000 ESR: 00000000
| GPR00: c03d27e8 cfa59a08 cfa55fe0 00000000 0fa46bc0 [...]
| [..]
| NIP [c03ed4b8] sata_dwc_qc_issue+0x14c/0x254
| LR [c03d27e8] ata_qc_issue+0x1c8/0x2dc
| Call Trace:
| [cfa59a08] [c003f4e0] __cancel_work_timer+0x124/0x194 (unreliable)
| [cfa59a78] [c03d27e8] ata_qc_issue+0x1c8/0x2dc
| [cfa59a98] [c03d2b3c] ata_exec_internal_sg+0x240/0x524
| [cfa59b08] [c03d2e98] ata_exec_internal+0x78/0xe0
| [cfa59b58] [c03d30fc] ata_read_log_page.part.38+0x1dc/0x204
| [cfa59bc8] [c03d324c] ata_identify_page_supported+0x68/0x130
| [...]

This is because sata_dwc_dma_xfer_complete() NULLs the
dma_pending's next neighbour "chan" (a *dma_chan struct) in
this '32' case right here (line ~735):
> hsdevp->dma_pending[tag] = SATA_DWC_DMA_PENDING_NONE;

Then the next time, a dma gets issued; dma_dwc_xfer_setup() passes
the NULL'd hsdevp->chan to the dmaengine_slave_config() which then
causes the crash.

With this patch, SATA_DWC_QCMD_MAX is now set to ATA_MAX_QUEUE + 1.
This avoids the OOB. But please note, there was a worthwhile discussion
on what ATA_TAG_INTERNAL and ATA_MAX_QUEUE is. And why there should not
be a "fake" 33 command-long queue size.

Ideally, the dw driver should account for the ATA_TAG_INTERNAL.
In Damien Le Moal's words: "... having looked at the driver, it
is a bigger change than just faking a 33rd "tag" that is in fact
not a command tag at all."

Fixes: 28361c403683c ("libata: add extra internal command")
Cc: stable@kernel.org # 4.18+
BugLink: https://github.com/openwrt/openwrt/issues/9505
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf/x86/intel: Don't extend the pseudo-encoding to GP counters

commit 4a263bf331c512849062805ef1b4ac40301a9829 upstream.

The INST_RETIRED.PREC_DIST event (0x0100) doesn't count on SPR.
perf stat -e cpu/event=0xc0,umask=0x0/,cpu/event=0x0,umask=0x1/ -C0

 Performance counter stats for 'CPU(s) 0':

           607,246      cpu/event=0xc0,umask=0x0/
                 0      cpu/event=0x0,umask=0x1/

The encoding for INST_RETIRED.PREC_DIST is pseudo-encoding, which
doesn't work on the generic counters. However, current perf extends its
mask to the generic counters.

The pseudo event-code for a fixed counter must be 0x00. Check and avoid
extending the mask for the fixed counter event which using the
pseudo-encoding, e.g., ref-cycles and PREC_DIST event.

With the patch,
perf stat -e cpu/event=0xc0,umask=0x0/,cpu/event=0x0,umask=0x1/ -C0

 Performance counter stats for 'CPU(s) 0':

           583,184      cpu/event=0xc0,umask=0x0/
           583,048      cpu/event=0x0,umask=0x1/

Fixes: 2de71ee153ef ("perf/x86/intel: Fix ICL/SPR INST_RETIRED.PREC_DIST encodings")
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/1648482543-14923-1-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/mm/tlb: Revert retpoline avoidance approach

commit d39268ad24c0fd0665d0c5cf55a7c1a0ebf94766 upstream.

0day reported a regression on a microbenchmark which is intended to
stress the TLB flushing path:

https://lore.kernel.org/all/20220317090415.GE735@xsang-OptiPlex-9020/

It pointed at a commit from Nadav which intended to remove retpoline
overhead in the TLB flushing path by taking the 'cond'-ition in
on_each_cpu_cond_mask(), pre-calculating it, and incorporating it into
'cpumask'.  That allowed the code to use a bunch of earlier direct
calls instead of later indirect calls that need a retpoline.

But, in practice, threads can go idle (and into lazy TLB mode where
they don't need to flush their TLB) between the early and late calls.
It works in this direction and not in the other because TLB-flushing
threads tend to hold mmap_lock for write.  Contention on that lock
causes threads to _go_ idle right in this early/late window.

There was not any performance data in the original commit specific
to the retpoline overhead.  I did a few tests on a system with
retpolines:

https://lore.kernel.org/all/dd8be93c-ded6-b962-50d4-96b1c3afb2b7@intel.com/

which showed a possible small win.  But, that small win pales in
comparison with the bigger loss induced on non-retpoline systems.

Revert the patch that removed the retpolines.  This was not a
clean revert, but it was self-contained enough not to be too painful.

Fixes: 6035152d8eeb ("x86/mm/tlb: Open-code on_each_cpu_cond_mask() for tlb_is_not_lazy()")
Reported-by: kernel test robot <oliver.sang@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Nadav Amit <namit@vmware.com>
Cc: <stable@vger.kernel.org>
Link: https://lkml.kernel.org/r/164874672286.389.7021457716635788197.tip-bot2@tip-bot2
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/msi: Fix msi message data shadow struct

commit 59b18a1e65b7a2134814106d0860010e10babe18 upstream.

The x86 MSI message data is 32 bits in total and is either in
compatibility or remappable format, see Intel Virtualization Technology
for Directed I/O, section 5.1.2.

Fixes: 6285aa50736 ("x86/msi: Provide msi message shadow structs")
Co-developed-by: Adrian-Ken Rueegsegger <ken@codelabs.ch>
Signed-off-by: Adrian-Ken Rueegsegger <ken@codelabs.ch>
Signed-off-by: Reto Buerki <reet@codelabs.ch>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220407110647.67372-1-reet@codelabs.ch
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

gpio: Restrict usage of GPIO chip irq members before initialization

commit 5467801f1fcbdc46bc7298a84dbf3ca1ff2a7320 upstream.

GPIO chip irq members are exposed before they could be completely
initialized and this leads to race conditions.

One such issue was observed for the gc->irq.domain variable which
was accessed through the I2C interface in gpiochip_to_irq() before
it could be initialized by gpiochip_add_irqchip(). This resulted in
Kernel NULL pointer dereference.

Following are the logs for reference :-

kernel: Call Trace:
kernel:  gpiod_to_irq+0x53/0x70
kernel:  acpi_dev_gpio_irq_get_by+0x113/0x1f0
kernel:  i2c_acpi_get_irq+0xc0/0xd0
kernel:  i2c_device_probe+0x28a/0x2a0
kernel:  really_probe+0xf2/0x460
kernel: RIP: 0010:gpiochip_to_irq+0x47/0xc0

To avoid such scenarios, restrict usage of GPIO chip irq members before
they are completely initialized.

Signed-off-by: Shreeya Patel <shreeya.patel@collabora.com>
Cc: stable@vger.kernel.org
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Bartosz Golaszewski <brgl@bgdev.pl>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

RDMA/hfi1: Fix use-after-free bug for mm struct

commit 2bbac98d0930e8161b1957dc0ec99de39ade1b3c upstream.

Under certain conditions, such as MPI_Abort, the hfi1 cleanup code may
represent the last reference held on the task mm.
hfi1_mmu_rb_unregister() then drops the last reference and the mm is freed
before the final use in hfi1_release_user_pages().  A new task may
allocate the mm structure while it is still being used, resulting in
problems. One manifestation is corruption of the mmap_sem counter leading
to a hang in down_write().  Another is corruption of an mm struct that is
in use by another task.

Fixes: 3d2a9d642512 ("IB/hfi1: Ensure correct mm is used at all times")
Link: https://lore.kernel.org/r/20220408133523.122165.72975.stgit@awfm-01.cornelisnetworks.com
Cc: <stable@vger.kernel.org>
Signed-off-by: Douglas Miller <doug.miller@cornelisnetworks.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

arm64: patch_text: Fixup last cpu should be master

commit 31a099dbd91e69fcab55eef4be15ed7a8c984918 upstream.

These patch_text implementations are using stop_machine_cpuslocked
infrastructure with atomic cpu_count. The original idea: When the
master CPU patch_text, the others should wait for it. But current
implementation is using the first CPU as master, which couldn't
guarantee the remaining CPUs are waiting. This patch changes the
last CPU as the master to solve the potential risk.

Fixes: ae16480785de ("arm64: introduce interfaces to hotpatch kernel and module code")
Signed-off-by: Guo Ren <guoren@linux.alibaba.com>
Signed-off-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20220407073323.743224-2-guoren@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

spi: core: add dma_map_dev for __spi_unmap_msg()

commit 409543cec01a84610029d6440c480c3fdd7214fb upstream.

Commit b470e10eb43f ("spi: core: add dma_map_dev for dma device") added
dma_map_dev for _spi_map_msg() but missed to add for unmap routine,
__spi_unmap_msg(), so add it now.

Fixes: b470e10eb43f ("spi: core: add dma_map_dev for dma device")
Cc: stable@vger.kernel.org # v5.14+
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20220406132238.1029249-1-vkoul@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

btrfs: prevent subvol with swapfile from being deleted

commit 60021bd754c6ca0addc6817994f20290a321d8d6 upstream.

A subvolume with an active swapfile must not be deleted otherwise it
would not be possible to deactivate it.

After the subvolume is deleted, we cannot swapoff the swapfile in this
deleted subvolume because the path is unreachable.  The swapfile is
still active and holding references, the filesystem cannot be unmounted.

The test looks like this:

  mkfs.btrfs -f $dev > /dev/null
  mount $dev $mnt

  btrfs sub create $mnt/subvol
  touch $mnt/subvol/swapfile
  chmod 600 $mnt/subvol/swapfile
  chattr +C $mnt/subvol/swapfile
  dd if=/dev/zero of=$mnt/subvol/swapfile bs=1K count=4096
  mkswap $mnt/subvol/swapfile
  swapon $mnt/subvol/swapfile

  btrfs sub delete $mnt/subvol
  swapoff $mnt/subvol/swapfile  # failed: No such file or directory
  swapoff --all

  unmount $mnt                  # target is busy.

To prevent above issue, we simply check that whether the subvolume
contains any active swapfile, and stop the deleting process.  This
behavior is like snapshot ioctl dealing with a swapfile.

CC: stable@vger.kernel.org # 5.4+
Reviewed-by: Robbie Ko <robbieko@synology.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Kaiwen Hu <kevinhu@synology.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

btrfs: fix qgroup reserve overflow the qgroup limit

commit b642b52d0b50f4d398cb4293f64992d0eed2e2ce upstream.

We use extent_changeset->bytes_changed in qgroup_reserve_data() to record
how many bytes we set for EXTENT_QGROUP_RESERVED state. Currently the
bytes_changed is set as "unsigned int", and it will overflow if we try to
fallocate a range larger than 4GiB. The result is we reserve less bytes
and eventually break the qgroup limit.

Unlike regular buffered/direct write, which we use one changeset for
each ordered extent, which can never be larger than 256M.  For
fallocate, we use one changeset for the whole range, thus it no longer
respects the 256M per extent limit, and caused the problem.

The following example test script reproduces the problem:

  $ cat qgroup-overflow.sh
  #!/bin/bash

  DEV=/dev/sdj
  MNT=/mnt/sdj

  mkfs.btrfs -f $DEV
  mount $DEV $MNT

  # Set qgroup limit to 2GiB.
  btrfs quota enable $MNT
  btrfs qgroup limit 2G $MNT

  # Try to fallocate a 3GiB file. This should fail.
  echo
  echo "Try to fallocate a 3GiB file..."
  fallocate -l 3G $MNT/3G.file

  # Try to fallocate a 5GiB file.
  echo
  echo "Try to fallocate a 5GiB file..."
  fallocate -l 5G $MNT/5G.file

  # See we break the qgroup limit.
  echo
  sync
  btrfs qgroup show -r $MNT

  umount $MNT

When running the test:

  $ ./qgroup-overflow.sh
  (...)

  Try to fallocate a 3GiB file...
  fallocate: fallocate failed: Disk quota exceeded

  Try to fallocate a 5GiB file...

  qgroupid         rfer         excl     max_rfer
  --------         ----         ----     --------
  0/5           5.00GiB      5.00GiB      2.00GiB

Since we have no control of how bytes_changed is used, it's better to
set it to u64.

CC: stable@vger.kernel.org # 4.14+
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Ethan Lien <ethanlien@synology.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf/x86/intel: Update the FRONTEND MSR mask on Sapphire Rapids

commit e590928de7547454469693da9bc7ffd562e54b7e upstream.

On Sapphire Rapids, the FRONTEND_RETIRED.MS_FLOWS event requires the
FRONTEND MSR value 0x8. However, the current FRONTEND MSR mask doesn't
support it.

Update intel_spr_extra_regs[] to support it.

Fixes: 61b985e3e775 ("perf/x86/intel: Add perf core PMU support for Sapphire Rapids")
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/1648482543-14923-2-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/speculation: Restore speculation related MSRs during S3 resume

commit e2a1256b17b16f9b9adf1b6fea56819e7b68e463 upstream.

After resuming from suspend-to-RAM, the MSRs that control CPU's
speculative execution behavior are not being restored on the boot CPU.

These MSRs are used to mitigate speculative execution vulnerabilities.
Not restoring them correctly may leave the CPU vulnerable.  Secondary
CPU's MSRs are correctly being restored at S3 resume by
identify_secondary_cpu().

During S3 resume, restore these MSRs for boot CPU when restoring its
processor state.

Fixes: 772439717dbf ("x86/bugs/intel: Set proper CPU features and setup RDS")
Reported-by: Neelima Krishnan <neelima.krishnan@intel.com>
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Tested-by: Neelima Krishnan <neelima.krishnan@intel.com>
Acked-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/pm: Save the MSR validity status at context setup

commit 73924ec4d560257004d5b5116b22a3647661e364 upstream.

The mechanism to save/restore MSRs during S3 suspend/resume checks for
the MSR validity during suspend, and only restores the MSR if its a
valid MSR.  This is not optimal, as an invalid MSR will unnecessarily
throw an exception for every suspend cycle.  The more invalid MSRs,
higher the impact will be.

Check and save the MSR validity at setup.  This ensures that only valid
MSRs that are guaranteed to not throw an exception will be attempted
during suspend.

Fixes: 7a9c2dd08ead ("x86/pm: Introduce quirk framework to save/restore extra MSR registers around suspend/resume")
Suggested-by: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com>
Reviewed-by: Dave Hansen <dave.hansen@linux.intel.com>
Acked-by: Borislav Petkov <bp@suse.de>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring: fix race between timeout flush and removal

commit e677edbcabee849bfdd43f1602bccbecf736a646 upstream.

io_flush_timeouts() assumes the timeout isn't in progress of triggering
or being removed/canceled, so it unconditionally removes it from the
timeout list and attempts to cancel it.

Leave it on the list and let the normal timeout cancelation take care
of it.

Cc: stable@vger.kernel.org # 5.5+
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring: implement compat handling for IORING_REGISTER_IOWQ_AFF

commit 0f5e4b83b37a96e3643951588ed7176b9b187c0a upstream.

Similarly to the way it is done im mbind syscall.

Cc: stable@vger.kernel.org # 5.14
Fixes: fe76421d1da1dcdb ("io_uring: allow user configurable IO thread CPU affinity")
Signed-off-by: Eugene Syromiatnikov <esyr@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring: defer splice/tee file validity check until command issue

commit a3e4bc23d5470b2beb7cc42a86b6a3e75b704c15 upstream.

In preparation for not using the file at prep time, defer checking if this
file refers to a valid io_uring instance until issue time.

This also means we can get rid of the cleanup flag for splice and tee.

Cc: stable@vger.kernel.org # v5.15+
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring: don't check req->file in io_fsync_prep()

commit ec858afda857e361182ceafc3d2ba2b164b8e889 upstream.

This is a leftover from the really old days where we weren't able to
track and error early if we need a file and it wasn't assigned. Kill
the check.

Cc: stable@vger.kernel.org # v5.15+
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/mempolicy: fix mpol_new leak in shared_policy_replace

commit 4ad099559b00ac01c3726e5c95dc3108ef47d03e upstream.

If mpol_new is allocated but not used in restart loop, mpol_new will be
freed via mpol_put before returning to the caller.  But refcnt is not
initialized yet, so mpol_put could not do the right things and might
leak the unused mpol_new.  This would happen if mempolicy was updated on
the shared shmem file while the sp->lock has been dropped during the
memory allocation.

This issue could be triggered easily with the below code snippet if
there are many processes doing the below work at the same time:

  shmid = shmget((key_t)5566, 1024 * PAGE_SIZE, 0666|IPC_CREAT);
  shm = shmat(shmid, 0, 0);
  loop many times {
    mbind(shm, 1024 * PAGE_SIZE, MPOL_LOCAL, mask, maxnode, 0);
    mbind(shm + 128 * PAGE_SIZE, 128 * PAGE_SIZE, MPOL_DEFAULT, mask,
          maxnode, 0);
  }

Link: https://lkml.kernel.org/r/20220329111416.27954-1-linmiaohe@huawei.com
Fixes: 42288fe366c4 ("mm: mempolicy: Convert shared_policy mutex to spinlock")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: <stable@vger.kernel.org> [3.8]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mmmremap.c: avoid pointless invalidate_range_start/end on mremap(old_size=0)

commit 01e67e04c28170c47700c2c226d732bbfedb1ad0 upstream.

If an mremap() syscall with old_size=0 ends up in move_page_tables(), it
will call invalidate_range_start()/invalidate_range_end() unnecessarily,
i.e.  with an empty range.

This causes a WARN in KVM's mmu_notifier.  In the past, empty ranges
have been diagnosed to be off-by-one bugs, hence the WARNing.  Given the
low (so far) number of unique reports, the benefits of detecting more
buggy callers seem to outweigh the cost of having to fix cases such as
this one, where userspace is doing something silly.  In this particular
case, an early return from move_page_tables() is enough to fix the
issue.

Link: https://lkml.kernel.org/r/20220329173155.172439-1-pbonzini@redhat.com
Reported-by: syzbot+6bde52d89cfdf9f61425@syzkaller.appspotmail.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

highmem: fix checks in __kmap_local_sched_{in,out}

commit 66f133ceab7456c789f70a242991ed1b27ba1c3d upstream.

When CONFIG_DEBUG_KMAP_LOCAL is enabled __kmap_local_sched_{in,out} check
that even slots in the tsk->kmap_ctrl.pteval are unmapped.  The slots are
initialized with 0 value, but the check is done with pte_none.  0 pte
however does not necessarily mean that pte_none will return true.  e.g.
on xtensa it returns false, resulting in the following runtime warnings:

 WARNING: CPU: 0 PID: 101 at mm/highmem.c:627 __kmap_local_sched_out+0x51/0x108
 CPU: 0 PID: 101 Comm: touch Not tainted 5.17.0-rc7-00010-gd3a1cdde80d2-dirty #13
 Call Trace:
   dump_stack+0xc/0x40
   __warn+0x8f/0x174
   warn_slowpath_fmt+0x48/0xac
   __kmap_local_sched_out+0x51/0x108
   __schedule+0x71a/0x9c4
   preempt_schedule_irq+0xa0/0xe0
   common_exception_return+0x5c/0x93
   do_wp_page+0x30e/0x330
   handle_mm_fault+0xa70/0xc3c
   do_page_fault+0x1d8/0x3c4
   common_exception+0x7f/0x7f

 WARNING: CPU: 0 PID: 101 at mm/highmem.c:664 __kmap_local_sched_in+0x50/0xe0
 CPU: 0 PID: 101 Comm: touch Tainted: G        W         5.17.0-rc7-00010-gd3a1cdde80d2-dirty #13
 Call Trace:
   dump_stack+0xc/0x40
   __warn+0x8f/0x174
   warn_slowpath_fmt+0x48/0xac
   __kmap_local_sched_in+0x50/0xe0
   finish_task_switch$isra$0+0x1ce/0x2f8
   __schedule+0x86e/0x9c4
   preempt_schedule_irq+0xa0/0xe0
   common_exception_return+0x5c/0x93
   do_wp_page+0x30e/0x330
   handle_mm_fault+0xa70/0xc3c
   do_page_fault+0x1d8/0x3c4
   common_exception+0x7f/0x7f

Fix it by replacing !pte_none(pteval) with pte_val(pteval) != 0.

Link: https://lkml.kernel.org/r/20220403235159.3498065-1-jcmvbkbc@gmail.com
Fixes: 5fbda3ecd14a ("sched: highmem: Store local kmaps in task struct")
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Cc: "Peter Zijlstra (Intel)" <peterz@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

lz4: fix LZ4_decompress_safe_partial read out of bound

commit eafc0a02391b7b36617b36c97c4b5d6832cf5e24 upstream.

When partialDecoding, it is EOF if we've either filled the output buffer
or can't proceed with reading an offset for following match.

In some extreme corner cases when compressed data is suitably corrupted,
UAF will occur.  As reported by KASAN [1], LZ4_decompress_safe_partial
may lead to read out of bound problem during decoding.  lz4 upstream has
fixed it [2] and this issue has been disscussed here [3] before.

current decompression routine was ported from lz4 v1.8.3, bumping
lib/lz4 to v1.9.+ is certainly a huge work to be done later, so, we'd
better fix it first.

[1] https://lore.kernel.org/all/000000000000830d1205cf7f0477@google.com/
[2] https://github.com/lz4/lz4/commit/c5d6f8a8be3927c0bec91bcc58667a6cfad244ad#
[3] https://lore.kernel.org/all/CC666AE8-4CA4-4951-B6FB-A2EFDE3AC03B@fb.com/

Link: https://lkml.kernel.org/r/20211111105048.2006070-1-guoxuenan@huawei.com
Reported-by: syzbot+63d688f1d899c588fb71@syzkaller.appspotmail.com
Signed-off-by: Guo Xuenan <guoxuenan@huawei.com>
Reviewed-by: Nick Terrell <terrelln@fb.com>
Acked-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Cc: Yann Collet <cyan@fb.com>
Cc: Chengyang Fan <cy.fan@huawei.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mmc: core: Fixup support for writeback-cache for eMMC and SD

commit 08ebf903af57cda6d773f3dd1671b64f73b432b8 upstream.

During the card initialization process, the mmc core checks whether the
eMMC/SD card supports an internal writeback-cache and then enables it
inside the card.

Unfortunately, this isn't according to what the mmc core reports to the
upper block layer. Instead, the writeback-cache support with REQ_FLUSH and
REQ_FUA, are being enabled depending on whether the host supports the CMD23
(MMC_CAP_CMD23) and whether an eMMC supports the reliable-write command.

This is wrong and it may also sound awkward. In fact, it's a remnant
from when both eMMC/SD cards didn't have dedicated commands/support to
control the internal writeback-cache. In other words, it was the best we
could do at that point in time.

To fix the problem, but also without breaking backwards compatibility,
let's align the REQ_FLUSH support with whether the writeback-cache became
successfully enabled - for both eMMC and SD cards.

Cc: stable@kernel.org
Fixes: 881d1c25f765 ("mmc: core: Add cache control for eMMC4.5 device")
Fixes: 130206a615a9 ("mmc: core: Add support for cache ctrl for SD cards")
Depends-on: 97fce126e279 ("mmc: block: Issue a cache flush only when it's enabled")
Reviewed-by: Avri Altman <Avri.Altman@wdc.com>
Signed-off-by: Michael Wu <michael@allwinnertech.com>
Link: https://lore.kernel.org/r/20220331073223.106415-1-michael@allwinnertech.com
[Ulf: Re-wrote the commit message]
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mmc: renesas_sdhi: don't overwrite TAP settings when HS400 tuning is complete

commit 03e59b1e2f56245163b14c69e0a830c24b1a3a47 upstream.

When HS400 tuning is complete and HS400 is going to be activated, we
have to keep the current number of TAPs and should not overwrite them
with a hardcoded value. This was probably a copy&paste mistake when
upporting HS400 support from the BSP.

Fixes: 26eb2607fa28 ("mmc: renesas_sdhi: add eMMC HS400 mode support")
Reported-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220404114902.12175-1-wsa+renesas@sang-engineering.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>