[ Upstream commit 247c8d2f98 ]
A couple of functions from fs/pipe.c are used both internally
and for the watch queue code, but the declaration is only
visible when the latter is enabled:
fs/pipe.c:1254:5: error: no previous prototype for 'pipe_resize_ring'
fs/pipe.c:758:15: error: no previous prototype for 'account_pipe_buffers'
fs/pipe.c:764:6: error: no previous prototype for 'too_many_pipe_buffers_soft'
fs/pipe.c:771:6: error: no previous prototype for 'too_many_pipe_buffers_hard'
fs/pipe.c:777:6: error: no previous prototype for 'pipe_is_unprivileged_user'
Make the visible unconditionally to avoid these warnings.
Fixes: c73be61ced ("pipe: Add general notification queue support")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Message-Id: <20230516195629.551602-1-arnd@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 2820433be2 upstream.
[Why & How]
In the commit 32953485c5 ("drm/amd/display: Do not update DRR while
BW optimizations pending"), a modification was added to avoid adjusting
DRR if optimized bandwidth is set. This change was only intended for
DCN, but one part of the patch changed the code path for DCE devices and
caused regressions to the kms_vrr test. To address this problem, this
commit adds a modification in which dc_stream_adjust_vmin_vmax will be
fully executed in DCE devices.
Fixes: 32953485c5 ("drm/amd/display: Do not update DRR while BW optimizations pending")
Reviewed-by: Aric Cyr <Aric.Cyr@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit e1b37563ca upstream.
gtags considers any file outside of its current working directory
"outside the source tree" and refuses to index it. For O= kernel builds,
or when "make" is invoked from a directory other then the kernel source
tree, gtags ignores the entire kernel source and generates an empty
index.
Force-set gtags current working directory to the kernel source tree.
Due to commit 9da0763bdd ("kbuild: Use relative path when building in
a subdir of the source tree"), if the kernel build is done in a
sub-directory of the kernel source tree, the kernel Makefile will set
the kernel's $srctree to ".." for shorter compile-time and run-time
warnings. Consequently, the list of files to be indexed will be in the
"../*" form, rendering all such paths invalid once gtags switches to the
kernel source tree as its current working directory.
If gtags indexing is requested and the build directory is not the kernel
source tree, index all files in absolute-path form.
Note, indexing in absolute-path form will not affect the generated
index, as paths in gtags indices are always relative to the gtags "root
directory" anyway (as evidenced by "gtags --dump").
Signed-off-by: Ahmed S. Darwish <darwi@linutronix.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 1c24956542 upstream.
This problem was encountered on an arm64 system with a lot of memory.
Without kernel debug symbols installed, and with both kcore and kallsyms
available, perf managed to get confused and returned "unknown" for all
of the kernel symbols that it tried to look up.
On this system, stext fell within the vmalloc segment. The kcore symbol
matching code tries to find the first segment that contains stext and
uses that to replace the segment generated from just the kallsyms
information. In this case, however, there were two: a very large
vmalloc segment, and the text segment. This caused perf to get confused
because multiple overlapping segments were inserted into the RB tree
that holds the discovered segments. However, that alone wasn't
sufficient to cause the problem. Even when we could find the segment,
the offsets were adjusted in such a way that the newly generated symbols
didn't line up with the instruction addresses in the trace. The most
obvious solution would be to consult which segment type is text from
kcore, but this information is not exposed to users.
Instead, select the smallest matching segment that contains stext
instead of the first matching segment. This allows us to match the text
segment instead of vmalloc, if one is contained within the other.
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Krister Johansen <kjlx@templeofstupid.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: David Reaver <me@davidreaver.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20230125183418.GD1963@templeofstupid.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Krister Johansen <kjlx@templeofstupid.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit f66066bc51 upstream.
While our user stacks can grow either down (all common architectures) or
up (parisc and the ia64 register stack), the initial stack setup when we
copy the argument and environment strings to the new stack at execve()
time is always done by extending the stack downwards.
But it turns out that in commit 8d7071af89 ("mm: always expand the
stack with the mmap write lock held"), as part of making the stack
growing code more robust, 'expand_downwards()' was now made to actually
check the vma flags:
if (!(vma->vm_flags & VM_GROWSDOWN))
return -EFAULT;
and that meant that this execve-time stack expansion started failing on
parisc, because on that architecture, the stack flags do not contain the
VM_GROWSDOWN bit.
At the same time the new check in expand_downwards() is clearly correct,
and simplified the callers, so let's not remove it.
The solution is instead to just codify the fact that yes, during
execve(), the stack grows down. This not only matches reality, it ends
up being particularly simple: we already have special execve-time flags
for the stack (VM_STACK_INCOMPLETE_SETUP) and use those flags to avoid
page migration during this setup time (see vma_is_temporary_stack() and
invalid_migration_vma()).
So just add VM_GROWSDOWN to that set of temporary flags, and now our
stack flags automatically match reality, and the parisc stack expansion
works again.
Note that the VM_STACK_INCOMPLETE_SETUP bits will be cleared when the
stack is finalized, so we only add the extra VM_GROWSDOWN bit on
CONFIG_STACK_GROWSUP architectures (ie parisc) rather than adding it in
general.
Link: https://lore.kernel.org/all/612eaa53-6904-6e16-67fc-394f4faa0e16@bell.net/
Link: https://lore.kernel.org/all/5fd98a09-4792-1433-752d-029ae3545168@gmx.de/
Fixes: 8d7071af89 ("mm: always expand the stack with the mmap write lock held")
Reported-by: John David Anglin <dave.anglin@bell.net>
Reported-and-tested-by: Helge Deller <deller@gmx.de>
Reported-and-tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 112a7f9c8e upstream.
ACPI r6.5, sec 6.5.4, describes how AML is unable to access an
OperationRegion unless _REG has been called to connect a handler:
The OS runs _REG control methods to inform AML code of a change in the
availability of an operation region. When an operation region handler is
unavailable, AML cannot access data fields in that region. (Operation
region writes will be ignored and reads will return indeterminate data.)
The PCI core does not call _REG at any time, leading to the undefined
behavior mentioned in the spec.
The spec explains that _REG should be executed to indicate whether a
given region can be accessed:
Once _REG has been executed for a particular operation region, indicating
that the operation region handler is ready, a control method can access
fields in the operation region. Conversely, control methods must not
access fields in operation regions when _REG method execution has not
indicated that the operation region handler is ready.
An example included in the spec demonstrates calling _REG when devices are
turned off: "when the host controller or bridge controller is turned off
or disabled, PCI Config Space Operation Regions for child devices are
no longer available. As such, ETH0’s _REG method will be run when it
is turned off and will again be run when PCI1 is turned off."
It is reported that ASMedia PCIe GPIO controllers fail functional tests
after the system has returning from suspend (S3 or s2idle). This is because
the BIOS checks whether the OSPM has called the _REG method to determine
whether it can interact with the OperationRegion assigned to the device as
part of the other AML called for the device.
To fix this issue, call acpi_evaluate_reg() when devices are transitioning
to D3cold or D0.
[bhelgaas: split pci_power_t checking to preliminary patch]
Link: https://uefi.org/specs/ACPI/6.5/06_Device_Configuration.html#reg-region
Link: https://lore.kernel.org/r/20230620140451.21007-1-mario.limonciello@amd.com
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Rafael J. Wysocki <rafael@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 5557b62634 upstream.
Previously acpi_pci_set_power_state() assumed the requested power state was
valid (PCI_D0 ... PCI_D3cold). If a caller supplied something else, we
could index outside the state_conv[] array and pass junk to
acpi_device_set_power().
Validate the pci_power_t parameter and return -EINVAL if it's invalid.
Link: https://lore.kernel.org/r/20230621222857.GA122930@bhelgaas
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 03f889378f upstream.
MMU version of lock_mm_and_find_vma releases the mm lock before
returning when VMA is not found. Do the same in noMMU version.
This fixes hang on an attempt to handle protection fault.
Fixes: d85a143b69 ("xtensa: fix NOMMU build with lock_mm_and_find_vma() conversion")
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit d85a143b69 upstream.
It turns out that xtensa has a really odd configuration situation: you
can do a no-MMU config, but still have the page fault code enabled.
Which doesn't sound all that sensible, but it turns out that xtensa can
have protection faults even without the MMU, and we have this:
config PFAULT
bool "Handle protection faults" if EXPERT && !MMU
default y
help
Handle protection faults. MMU configurations must enable it.
noMMU configurations may disable it if used memory map never
generates protection faults or faults are always fatal.
If unsure, say Y.
which completely violated my expectations of the page fault handling.
End result: Guenter reports that the xtensa no-MMU builds all fail with
arch/xtensa/mm/fault.c: In function ‘do_page_fault’:
arch/xtensa/mm/fault.c:133:8: error: implicit declaration of function ‘lock_mm_and_find_vma’
because I never exposed the new lock_mm_and_find_vma() function for the
no-MMU case.
Doing so is simple enough, and fixes the problem.
Reported-and-tested-by: Guenter Roeck <linux@roeck-us.net>
Fixes: a050ba1e74 ("mm/fault: convert remaining simple cases to lock_mm_and_find_vma()")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit e55e5df193 upstream.
As already mentioned in my merge message for the 'expand-stack' branch,
we have something like 24 different versions of the page fault path for
all our different architectures, all just _slightly_ different due to
various historical reasons (usually related to exactly when they
branched off the original i386 version, and the details of the other
architectures they had in their history).
And a few of them had some silly mistake in the conversion.
Most of the architectures call the faulting address 'address' in the
fault path. But not all. Some just call it 'addr'. And if you end up
doing a bit too much copy-and-paste, you end up with the wrong version
in the places that do it differently.
In this case it was csky.
Fixes: a050ba1e74 ("mm/fault: convert remaining simple cases to lock_mm_and_find_vma()")
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit ea3f827287 upstream.
In commit 8d7071af89 ("mm: always expand the stack with the mmap write
lock held") I tried to deal with the remaining odd page fault handling
cases. The oddest one is ia64, which has stacks that grow both up and
down. And because ia64 was _so_ odd, I asked people to verify the end
result.
But a close second oddity is parisc, which is the only one that has a
main stack growing up (our "CONFIG_STACK_GROWSUP" config option). But
it looked obvious enough that I didn't worry about it.
I should have worried a bit more. Not because it was particularly
complex, but because I just used the wrong variable name.
The previous vma isn't called "prev", it's called "prev_vma". Blush.
Fixes: 8d7071af89 ("mm: always expand the stack with the mmap write lock held")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 0b26eadbf2 upstream.
The sparc32 conversion to lock_mm_and_find_vma() in commit a050ba1e74
("mm/fault: convert remaining simple cases to lock_mm_and_find_vma()")
missed the fact that we didn't actually have a 'regs' pointer available
in the 'force_user_fault()' case.
It's there in the regular page fault path ("do_sparc_fault()"), but not
the window underflow/overflow paths.
Which is all fine - we can just pass in a NULL pointer. The register
state is only used to avoid deadlock with kernel faults, which is not
the case for any of these register window faults.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Fixes: a050ba1e74 ("mm/fault: convert remaining simple cases to lock_mm_and_find_vma()")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Naresh Kamboju <naresh.kamboju@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 8d7071af89 upstream
This finishes the job of always holding the mmap write lock when
extending the user stack vma, and removes the 'write_locked' argument
from the vm helper functions again.
For some cases, we just avoid expanding the stack at all: drivers and
page pinning really shouldn't be extending any stacks. Let's see if any
strange users really wanted that.
It's worth noting that architectures that weren't converted to the new
lock_mm_and_find_vma() helper function are left using the legacy
"expand_stack()" function, but it has been changed to drop the mmap_lock
and take it for writing while expanding the vma. This makes it fairly
straightforward to convert the remaining architectures.
As a result of dropping and re-taking the lock, the calling conventions
for this function have also changed, since the old vma may no longer be
valid. So it will now return the new vma if successful, and NULL - and
the lock dropped - if the area could not be extended.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[6.1: Patch drivers/iommu/io-pgfault.c instead]
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit f313c51d26 upstream.
This is a small step towards a model where GUP itself would not expand
the stack, and any user that needs GUP to not look up existing mappings,
but actually expand on them, would have to do so manually before-hand,
and with the mm lock held for writing.
It turns out that execve() already did almost exactly that, except it
didn't take the mm lock at all (it's single-threaded so no locking
technically needed, but it could cause lockdep errors). And it only did
it for the CONFIG_STACK_GROWSUP case, since in that case GUP has
obviously never expanded the stack downwards.
So just make that CONFIG_STACK_GROWSUP case do the right thing with
locking, and enable it generally. This will eventually help GUP, and in
the meantime avoids a special case and the lockdep issue.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[6.1 Minor context from still having FOLL_FORCE flags set]
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit f440fa1ac9 upstream.
Make calls to extend_vma() and find_extend_vma() fail if the write lock
is required.
To avoid making this a flag-day event, this still allows the old
read-locking case for the trivial situations, and passes in a flag to
say "is it write-locked". That way write-lockers can say "yes, I'm
being careful", and legacy users will continue to work in all the common
cases until they have been fully converted to the new world order.
Co-Developed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 2cd76c50d0 upstream.
This is one of the simple cases, except there's no pt_regs pointer.
Which is fine, as lock_mm_and_find_vma() is set up to work fine with a
NULL pt_regs.
Powerpc already enabled LOCK_MM_AND_FIND_VMA for the main CPU faulting,
so we can just use the helper without any extra work.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit a050ba1e74 upstream.
This does the simple pattern conversion of alpha, arc, csky, hexagon,
loongarch, nios2, sh, sparc32, and xtensa to the lock_mm_and_find_vma()
helper. They all have the regular fault handling pattern without odd
special cases.
The remaining architectures all have something that keeps us from a
straightforward conversion: ia64 and parisc have stacks that can grow
both up as well as down (and ia64 has special address region checks).
And m68k, microblaze, openrisc, sparc64, and um end up having extra
rules about only expanding the stack down a limited amount below the
user space stack pointer. That is something that x86 used to do too
(long long ago), and it probably could just be skipped, but it still
makes the conversion less than trivial.
Note that this conversion was done manually and with the exception of
alpha without any build testing, because I have a fairly limited cross-
building environment. The cases are all simple, and I went through the
changes several times, but...
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 8b35ca3e45 upstream.
arm has an additional check for address < FIRST_USER_ADDRESS before
expanding the stack. Since FIRST_USER_ADDRESS is defined everywhere
(generally as 0), move that check to the generic expand_downwards().
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit ae870a68b5 upstream.
This converts arm64 to use the new page fault helper. It was very
straightforward, but still needed a fix for the "obvious" conversion I
initially did. Thanks to Suren for the fix and testing.
Fixed-and-tested-by: Suren Baghdasaryan <surenb@google.com>
Unnecessary-code-removal-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[6.1: Ignore CONFIG_PER_VMA_LOCK context]
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit eda0047296 upstream.
This is done as a separate patch from introducing the new
lock_mm_and_find_vma() helper, because while it's an obvious change,
it's not what x86 used to do in this area.
We already abort the page fault on fatal signals anyway, so why should
we wait for the mmap lock only to then abort later? With the new helper
function that returns without the lock held on failure anyway, this is
particularly easy and straightforward.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c2508ec5a5 upstream.
.. and make x86 use it.
This basically extracts the existing x86 "find and expand faulting vma"
code, but extends it to also take the mmap lock for writing in case we
actually do need to expand the vma.
We've historically short-circuited that case, and have some rather ugly
special logic to serialize the stack segment expansion (since we only
hold the mmap lock for reading) that doesn't match the normal VM
locking.
That slight violation of locking worked well, right up until it didn't:
the maple tree code really does want proper locking even for simple
extension of an existing vma.
So extract the code for "look up the vma of the fault" from x86, fix it
up to do the necessary write locking, and make it available as a helper
function for other architectures that can use the common helper.
Note: I say "common helper", but it really only handles the normal
stack-grows-down case. Which is all architectures except for PA-RISC
and IA64. So some rare architectures can't use the helper, but if they
care they'll just need to open-code this logic.
It's also worth pointing out that this code really would like to have an
optimistic "mmap_upgrade_trylock()" to make it quicker to go from a
read-lock (for the common case) to taking the write lock (for having to
extend the vma) in the normal single-threaded situation where there is
no other locking activity.
But that _is_ all the very uncommon special case, so while it would be
nice to have such an operation, it probably doesn't matter in reality.
I did put in the skeleton code for such a possible future expansion,
even if it only acts as pseudo-documentation for what we're doing.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[6.1: Ignore CONFIG_PER_VMA_LOCK context]
Signed-off-by: Samuel Mendoza-Jonas <samjonas@amazon.com>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit cd00dd2585 upstream.
Check the write offset end bounds before using it as the offset into the
pivot array. This avoids a possible out-of-bounds access on the pivot
array if the write extends to the last slot in the node, in which case the
node maximum should be used as the end pivot.
akpm: this doesn't affect any current callers, but new users of mapletree
may encounter this problem if backported into earlier kernels, so let's
fix it in -stable kernels in case of this.
Link: https://lkml.kernel.org/r/20230506024752.2550-1-zhangpeng.00@bytedance.com
Fixes: 54a611b605 ("Maple Tree: add new data structure")
Signed-off-by: Peng Zhang <zhangpeng.00@bytedance.com>
Reviewed-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit e38910c007 upstream.
With commit d674a8f123 ("can: isotp: isotp_sendmsg(): fix return
error on FC timeout on TX path") the missing correct return value in
the case of a protocol error was introduced.
But the way the error value has been read and sent to the user space
does not follow the common scheme to clear the error after reading
which is provided by the sock_error() function. This leads to an error
report at the following write() attempt although everything should be
working.
Fixes: d674a8f123 ("can: isotp: isotp_sendmsg(): fix return error on FC timeout on TX path")
Reported-by: Carsten Schmidt <carsten.schmidt-achim@t-online.de>
Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net>
Link: https://lore.kernel.org/all/20230607072708.38809-1-socketcan@hartkopp.net
Cc: stable@vger.kernel.org
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit d7893093a7 upstream.
TLDR: It's a mess.
When kexec() is executed on a system with offline CPUs, which are parked in
mwait_play_dead() it can end up in a triple fault during the bootup of the
kexec kernel or cause hard to diagnose data corruption.
The reason is that kexec() eventually overwrites the previous kernel's text,
page tables, data and stack. If it writes to the cache line which is
monitored by a previously offlined CPU, MWAIT resumes execution and ends
up executing the wrong text, dereferencing overwritten page tables or
corrupting the kexec kernels data.
Cure this by bringing the offlined CPUs out of MWAIT into HLT.
Write to the monitored cache line of each offline CPU, which makes MWAIT
resume execution. The written control word tells the offlined CPUs to issue
HLT, which does not have the MWAIT problem.
That does not help, if a stray NMI, MCE or SMI hits the offlined CPUs as
those make it come out of HLT.
A follow up change will put them into INIT, which protects at least against
NMI and SMI.
Fixes: ea53069231 ("x86, hotplug: Use mwait to offline a processor, fix the legacy case")
Reported-by: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Ashok Raj <ashok.raj@intel.com>
Reviewed-by: Ashok Raj <ashok.raj@intel.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230615193330.492257119@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit f9c9987bf5 upstream.
Monitoring idletask::thread_info::flags in mwait_play_dead() has been an
obvious choice as all what is needed is a cache line which is not written
by other CPUs.
But there is a use case where a "dead" CPU needs to be brought out of
MWAIT: kexec().
This is required as kexec() can overwrite text, pagetables, stacks and the
monitored cacheline of the original kernel. The latter causes MWAIT to
resume execution which obviously causes havoc on the kexec kernel which
results usually in triple faults.
Use a dedicated per CPU storage to prepare for that.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Ashok Raj <ashok.raj@intel.com>
Reviewed-by: Borislav Petkov (AMD) <bp@alien8.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230615193330.434553750@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 1f5e7eb786 upstream.
Tony reported intermittent lockups on poweroff. His analysis identified the
wbinvd() in stop_this_cpu() as the culprit. This was added to ensure that
on SME enabled machines a kexec() does not leave any stale data in the
caches when switching from encrypted to non-encrypted mode or vice versa.
That wbinvd() is conditional on the SME feature bit which is read directly
from CPUID. But that readout does not check whether the CPUID leaf is
available or not. If it's not available the CPU will return the value of
the highest supported leaf instead. Depending on the content the "SME" bit
might be set or not.
That's incorrect but harmless. Making the CPUID readout conditional makes
the observed hangs go away, but it does not fix the underlying problem:
CPU0 CPU1
stop_other_cpus()
send_IPIs(REBOOT); stop_this_cpu()
while (num_online_cpus() > 1); set_online(false);
proceed... -> hang
wbinvd()
WBINVD is an expensive operation and if multiple CPUs issue it at the same
time the resulting delays are even larger.
But CPU0 already observed num_online_cpus() going down to 1 and proceeds
which causes the system to hang.
This issue exists independent of WBINVD, but the delays caused by WBINVD
make it more prominent.
Make this more robust by adding a cpumask which is initialized to the
online CPU mask before sending the IPIs and CPUs clear their bit in
stop_this_cpu() after the WBINVD completed. Check for that cpumask to
become empty in stop_other_cpus() instead of watching num_online_cpus().
The cpumask cannot plug all holes either, but it's better than a raw
counter and allows to restrict the NMI fallback IPI to be sent only the
CPUs which have not reported within the timeout window.
Fixes: 08f253ec37 ("x86/cpu: Clear SME feature flag when not in use")
Reported-by: Tony Battersby <tonyb@cybernetics.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Ashok Raj <ashok.raj@intel.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/3817d810-e0f1-8ef8-0bbd-663b919ca49b@cybernetics.com
Link: https://lore.kernel.org/r/87h6r770bv.ffs@tglx
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
