commit 46b5c4fb87 upstream.
If the calls to device_reset() or devm_spi_register_controller() fail on
probe of the MediaTek MT7621 SPI driver, the spi_controller struct is
erroneously not freed. Fix by switching over to the new
devm_spi_alloc_master() helper.
Additionally, there's an ordering issue in mt7621_spi_remove() wherein
the spi_controller is unregistered after disabling the SYS clock.
The correct order is to call spi_unregister_controller() *before* this
teardown step because bus accesses may still be ongoing until that
function returns.
All of these bugs have existed since the driver was first introduced,
so it seems fair to fix them together in a single commit.
Fixes: 1ab7f2a435 ("staging: mt7621-spi: add mt7621 support")
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Reviewed-by: Stefan Roese <sr@denx.de>
Cc: <stable@vger.kernel.org> # v4.17+: 5e844cc37a: spi: Introduce device-managed SPI controller allocation
Cc: <stable@vger.kernel.org> # v4.17+
Link: https://lore.kernel.org/r/72b680796149f5fcda0b3f530ffb7ee73b04f224.1607286887.git.lukas@wunner.de
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 8f96c434df upstream.
spi_geni_remove() accesses the driver's private data after calling
spi_unregister_master() even though that function releases the last
reference on the spi_master and thereby frees the private data.
Moreover, since commit 1a9e489e61 ("spi: spi-geni-qcom: Use OPP API to
set clk/perf state"), spi_geni_probe() leaks the spi_master allocation
if the calls to dev_pm_opp_set_clkname() or dev_pm_opp_of_add_table()
fail.
Fix by switching over to the new devm_spi_alloc_master() helper which
keeps the private data accessible until the driver has unbound and also
avoids the spi_master leak on probe.
Fixes: 561de45f72 ("spi: spi-geni-qcom: Add SPI driver support for GENI based QUP")
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Cc: <stable@vger.kernel.org> # v4.20+: 5e844cc37a: spi: Introduce device-managed SPI controller allocation
Cc: <stable@vger.kernel.org> # v4.20+
Cc: Rajendra Nayak <rnayak@codeaurora.org>
Cc: Girish Mahadevan <girishm@codeaurora.org>
Link: https://lore.kernel.org/r/dfa1d8c41b8acdfad87ec8654cd124e6e3cb3f31.1607286887.git.lukas@wunner.de
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 7174dc655e upstream.
If the call to devm_spi_register_master() fails on probe of the GPIO SPI
driver, the spi_master struct is erroneously not freed:
After allocating the spi_master, its reference count is 1. The driver
unconditionally decrements the reference count on unbind using a devm
action. Before calling devm_spi_register_master(), the driver
unconditionally increments the reference count because on success,
that function will decrement the reference count on unbind. However on
failure, devm_spi_register_master() does *not* decrement the reference
count, so the spi_master is leaked.
The issue was introduced by commits 8b797490b4 ("spi: gpio: Make sure
spi_master_put() is called in every error path") and 79567c1a32 ("spi:
gpio: Use devm_spi_register_master()"), which sought to plug leaks
introduced by 9b00bc7b90 ("spi: spi-gpio: Rewrite to use GPIO
descriptors") but missed this remaining leak.
The situation was later aggravated by commit d3b0ffa1d7 ("spi: gpio:
prevent memory leak in spi_gpio_probe"), which introduced a
use-after-free because it releases a reference on the spi_master if
devm_add_action_or_reset() fails even though the function already
does that.
Fix by switching over to the new devm_spi_alloc_master() helper.
Fixes: 9b00bc7b90 ("spi: spi-gpio: Rewrite to use GPIO descriptors")
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Cc: <stable@vger.kernel.org> # v4.17+: 5e844cc37a: spi: Introduce device-managed SPI controller allocation
Cc: <stable@vger.kernel.org> # v5.1-: 8b797490b4: spi: gpio: Make sure spi_master_put() is called in every error path
Cc: <stable@vger.kernel.org> # v5.1-: 45beec3519: spi: bitbang: Introduce spi_bitbang_init()
Cc: <stable@vger.kernel.org> # v5.1-: 79567c1a32: spi: gpio: Use devm_spi_register_master()
Cc: <stable@vger.kernel.org> # v5.4-: d3b0ffa1d7: spi: gpio: prevent memory leak in spi_gpio_probe
Cc: <stable@vger.kernel.org> # v4.17+
Cc: Navid Emamdoost <navid.emamdoost@gmail.com>
Cc: Andrey Smirnov <andrew.smirnov@gmail.com>
Link: https://lore.kernel.org/r/86eaed27431c3d709e3748eb76ceecbfc790dd37.1607286887.git.lukas@wunner.de
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 236924ee53 upstream.
If the call to devm_spi_register_controller() fails on probe of the
Qualcomm Atheros AR934x/QCA95xx SPI driver, the spi_controller struct is
erroneously not freed. Fix by switching over to the new
devm_spi_alloc_master() helper.
Moreover, the controller's clock is enabled on probe but not disabled if
any of the subsequent probe steps fail.
Finally, there's an ordering issue in ar934x_spi_remove() wherein the
clock is disabled even though the controller is not yet unregistered.
It is unregistered after ar934x_spi_remove() by the devres framework.
As long as it is not unregistered, SPI transfers may still be ongoing
and disabling the clock may break them. It is not possible to use
devm_spi_register_controller() in this case, so move to the non-devm
variant.
All of these bugs have existed since the driver was first introduced,
so it seems fair to fix them together in a single commit.
Fixes: 047980c582 ("spi: add driver for ar934x spi controller")
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Cc: <stable@vger.kernel.org> # v5.7+: 5e844cc37a: spi: Introduce device-managed SPI controller allocation
Cc: <stable@vger.kernel.org> # v5.7+
Cc: Chuanhong Guo <gch981213@gmail.com>
Link: https://lore.kernel.org/r/1d58367d74d55741e0c2730a51a2b65012c8ab33.1607286887.git.lukas@wunner.de
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 0fb6ee8d0b upstream.
Use a heap allocated memory for the SPI transfer buffer. Using stack memory
can corrupt stack memory when using DMA on some systems.
This change moves the buffer from the stack of the trigger handler call to
the heap of the buffer of the state struct. The size increases takes into
account the alignment for the timestamp, which is 8 bytes.
The 'data' buffer is split into 'tx_buf' and 'rx_buf', to make a clearer
separation of which part of the buffer should be used for TX & RX.
Fixes: af3008485e ("iio:adc: Add common code for ADI Sigma Delta devices")
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Link: https://lore.kernel.org/r/20201124123807.19717-1-alexandru.ardelean@analog.com
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 0e53656ad8 upstream.
When inserting a VMA, we restrict the placement to the low 4G unless the
caller opts into using the full range. This was done to allow usersapce
the opportunity to transition slowly from a 32b address space, and to
avoid breaking inherent 32b assumptions of some commands.
However, for insert we limited ourselves to 4G-4K, but on verification
we allowed the full 4G. This causes some attempts to bind a new buffer
to sporadically fail with -ENOSPC, but at other times be bound
successfully.
commit 48ea1e32c3 ("drm/i915/gen9: Set PIN_ZONE_4G end to 4GB - 1
page") suggests that there is a genuine problem with stateless addressing
that cannot utilize the last page in 4G and so we purposefully excluded
it. This means that the quick pin pass may cause us to utilize a buggy
placement.
Reported-by: CQ Tang <cq.tang@intel.com>
Testcase: igt/gem_exec_params/larger-than-life-batch
Fixes: 48ea1e32c3 ("drm/i915/gen9: Set PIN_ZONE_4G end to 4GB - 1 page")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: CQ Tang <cq.tang@intel.com>
Reviewed-by: CQ Tang <cq.tang@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Cc: <stable@vger.kernel.org> # v4.5+
Link: https://patchwork.freedesktop.org/patch/msgid/20201216092951.7124-1-chris@chris-wilson.co.uk
(cherry picked from commit 5f22cc0b13)
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit eca6ba20f3 upstream.
The only reference to the mlxplat_mlxcpld_psu[] array got removed,
so there is now a warning from clang:
drivers/platform/x86/mlx-platform.c:322:30: error: variable 'mlxplat_mlxcpld_psu' is not needed and will not be emitted [-Werror,-Wunneeded-internal-declaration]
static struct i2c_board_info mlxplat_mlxcpld_psu[] = {
Remove the array as well and adapt the ARRAY_SIZE() call
accordingly.
Fixes: 912b341585 ("platform/x86: mlx-platform: Remove PSU EEPROM from MSN274x platform configuration")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Vadim Pasternak <vadimp@nvidia.com>
Link: https://lore.kernel.org/r/20201203223105.1195709-1-arnd@kernel.org
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Cc: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 5bc5cc2819 upstream.
We've fixed many races in panfrost_job_timedout() but some remain.
Instead of trying to fix it again, let's simplify the logic and move
the reset bits to a separate work scheduled when one of the queue
reports a timeout.
v5:
- Simplify panfrost_scheduler_stop() (Steven Price)
- Always restart the queue in panfrost_scheduler_start() even if
the status is corrupted (Steven Price)
v4:
- Rework the logic to prevent a race between drm_sched_start()
(reset work) and drm_sched_job_timedout() (timeout work)
- Drop Steven's R-b
- Add dma_fence annotation to the panfrost_reset() function (Daniel Vetter)
v3:
- Replace the atomic_cmpxchg() by an atomic_xchg() (Robin Murphy)
- Add Steven's R-b
v2:
- Use atomic_cmpxchg() to conditionally schedule the reset work
(Steven Price)
Fixes: 1a11a88cfd ("drm/panfrost: Fix job timeout handling")
Cc: <stable@vger.kernel.org>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201105151704.2010667-1-boris.brezillon@collabora.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 1a11a88cfd upstream.
If more than two jobs end up timeout-ing concurrently, only one of them
(the one attached to the scheduler acquiring the lock) is fully handled.
The other one remains in a dangling state where it's no longer part of
the scheduling queue, but still blocks something in scheduler, leading
to repetitive timeouts when new jobs are queued.
Let's make sure all bad jobs are properly handled by the thread
acquiring the lock.
v3:
- Add Steven's R-b
- Don't take the sched_lock when stopping the schedulers
v2:
- Fix the subject prefix
- Stop the scheduler before returning from panfrost_job_timedout()
- Call cancel_delayed_work_sync() after drm_sched_stop() to make sure
no timeout handlers are in flight when we reset the GPU (Steven Price)
- Make sure we release the reset lock before restarting the
schedulers (Steven Price)
Fixes: f3ba91228e ("drm/panfrost: Add initial panfrost driver")
Cc: <stable@vger.kernel.org>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201002122506.1374183-1-boris.brezillon@collabora.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit c61b3e4839 upstream.
Bounds checking tools can flag a bug in dbAdjTree() for an array index
out of bounds in dmt_stree. Since dmt_stree can refer to the stree in
both structures dmaptree and dmapctl, use the larger array to eliminate
the false positive.
Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Reported-by: butt3rflyh4ck <butterflyhuangxx@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit fecc455978 upstream.
fsnotify_parent() used to send two separate events to backends when a
parent inode is watching children and the child inode is also watching.
In an attempt to avoid duplicate events in fanotify, we unified the two
backend callbacks to a single callback and handled the reporting of the
two separate events for the relevant backends (inotify and dnotify).
However the handling is buggy and can result in inotify and dnotify
listeners receiving events of the type they never asked for or spurious
events.
The problem is the unified event callback with two inode marks (parent and
child) is called when any of the parent and child inodes are watched and
interested in the event, but the parent inode's mark that is interested
in the event on the child is not necessarily the one we are currently
reporting to (it could belong to a different group).
So before reporting the parent or child event flavor to backend we need
to check that the mark is really interested in that event flavor.
The semantics of INODE and CHILD marks were hard to follow and made the
logic more complicated than it should have been. Replace it with INODE
and PARENT marks semantics to hopefully make the logic more clear.
Thanks to Hugh Dickins for spotting a bug in the earlier version of this
patch.
Fixes: 497b0c5a7c ("fsnotify: send event to parent and child with single callback")
CC: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20201202120713.702387-4-amir73il@gmail.com
Reported-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 950cc0d2be upstream.
The handle_inode_event() interface was added as (quoting comment):
"a simple variant of handle_event() for groups that only have inode
marks and don't have ignore mask".
In other words, all backends except fanotify. The inotify backend
also falls under this category, but because it required extra arguments
it was left out of the initial pass of backends conversion to the
simple interface.
This results in code duplication between the generic helper
fsnotify_handle_event() and the inotify_handle_event() callback
which also happen to be buggy code.
Generalize the handle_inode_event() arguments and add the check for
FS_EXCL_UNLINK flag to the generic helper, so inotify backend could
be converted to use the simple interface.
Link: https://lore.kernel.org/r/20201202120713.702387-2-amir73il@gmail.com
CC: stable@vger.kernel.org
Fixes: b9a1b97725 ("fsnotify: create method handle_inode_event() in fsnotify_operations")
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 08cd274f9b upstream.
The jffs2 mount options will be ignored when remounting jffs2.
It can be easily reproduced with the steps listed below.
1. mount -t jffs2 -o compr=none /dev/mtdblockx /mnt
2. mount -o remount compr=zlib /mnt
Since ec10a24f10, the option parsing happens before fill_super and
then pass fc, which contains the options parsing results, to function
jffs2_reconfigure during remounting. But function jffs2_reconfigure do
not update c->mount_opts.
This patch add a function jffs2_update_mount_opts to fix this problem.
By the way, I notice that tmpfs use the same way to update remounting
options. If it is necessary to unify them?
Cc: <stable@vger.kernel.org>
Fixes: ec10a24f10 ("vfs: Convert jffs2 to use the new mount API")
Signed-off-by: lizhe <lizhe67@huawei.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 9afc9a8a49 upstream.
The log of this problem is:
jffs2: Error garbage collecting node at 0x***!
jffs2: No space for garbage collection. Aborting GC thread
This is because GC believe that it do nothing, so it abort.
After going over the image of jffs2, I find a scene that
can trigger this problem stably.
The scene is: there is a normal dirent node at summary-area,
but abnormal at corresponding not-summary-area with error
name_crc.
The reason that GC exit abnormally is because it find that
abnormal dirent node to GC, but when it goes to function
jffs2_add_fd_to_list, it cannot meet the condition listed
below:
if ((*prev)->nhash == new->nhash && !strcmp((*prev)->name, new->name))
So no node is marked obsolete, statistical information of
erase_block do not change, which cause GC exit abnormally.
The root cause of this problem is: we do not check the
name_crc of the abnormal dirent node with summary is enabled.
Noticed that in function jffs2_scan_dirent_node, we use
function jffs2_scan_dirty_space to deal with the dirent
node with error name_crc. So this patch add a checking
code in function read_direntry to ensure the correctness
of dirent node. If checked failed, the dirent node will
be marked obsolete so GC will pass this node and this
problem will be fixed.
Cc: <stable@vger.kernel.org>
Signed-off-by: Zhe Li <lizhe67@huawei.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 20f1431160 upstream.
Write buffers use a kmalloc()'ed buffer, they can leak
up to seven bytes of kernel memory to flash if writes are not
aligned.
So use ubifs_pad() to fill these gaps with padding bytes.
This was never a problem while scanning because the scanner logic
manually aligns node lengths and skips over these gaps.
Cc: <stable@vger.kernel.org>
Fixes: 1e51764a3c ("UBIFS: add new flash file system")
Signed-off-by: Richard Weinberger <richard@nod.at>
Reviewed-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 7955f105af upstream.
In the negotiate protocol preauth context, the server is not required
to populate the salt (although it is done by most servers) so do
not warn on mount.
We retain the checks (warn) that the preauth context is the minimum
size and that the salt does not exceed DataLength of the SMB response.
Although we use the defaults in the case that the preauth context
response is invalid, these checks may be useful in the future
as servers add support for additional mechanisms.
CC: Stable <stable@vger.kernel.org>
Reviewed-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit bc7c4129d4 upstream.
Azure does not send an SPNEGO blob in the negotiate protocol response,
so we shouldn't assume that it is there when validating the location
of the first negotiate context. This avoids the potential confusing
mount warning:
CIFS: Invalid negotiate context offset
CC: Stable <stable@vger.kernel.org>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit ebcd6de987 upstream.
Mounts to Azure cause an unneeded warning message in dmesg
"CIFS: VFS: parse_server_interfaces: incomplete interface info"
Azure rounds up the size (by 8 additional bytes, to a
16 byte boundary) of the structure returned on the query
of the server interfaces at mount time. This is permissible
even though different than other servers so do not log a warning
if query network interfaces response is only rounded up by 8
bytes or fewer.
CC: Stable <stable@vger.kernel.org>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit e5cafce3ad upstream.
A NULL pointer dereference may occur in __ceph_remove_cap with some of the
callbacks used in ceph_iterate_session_caps, namely trim_caps_cb and
remove_session_caps_cb. Those callers hold the session->s_mutex, so they
are prevented from concurrent execution, but ceph_evict_inode does not.
Since the callers of this function hold the i_ceph_lock, the fix is simply
a matter of returning immediately if caps->ci is NULL.
Cc: stable@vger.kernel.org
URL: https://tracker.ceph.com/issues/43272
Suggested-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Luis Henriques <lhenriques@suse.de>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit ff9632d2a6 upstream.
Since the time-travel rework, basic time-travel mode hasn't worked
properly, but there's no longer a need for this WARN_ON() so just
remove it and thereby fix things.
Cc: stable@vger.kernel.org
Fixes: 4b786e24ca ("um: time-travel: Rewrite as an event scheduler")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 207cdd565d upstream.
Commit a408e4a86b ("ima: open a new file instance if no read
permissions") already introduced a second open to measure a file when the
original file descriptor does not allow it. However, it didn't remove the
existing method of changing the mode of the original file descriptor, which
is still necessary if the current process does not have enough privileges
to open a new one.
Changing the mode isn't really an option, as the filesystem might need to
do preliminary steps to make the read possible. Thus, this patch removes
the code and keeps the second open as the only option to measure a file
when it is unreadable with the original file descriptor.
Cc: <stable@vger.kernel.org> # 4.20.x: 0014cc04e8 ima: Set file->f_mode
Fixes: 2fe5d6def1 ("ima: integrity appraisal extension")
Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 89bdfaf93d upstream.
ovl_ioctl_set_flags() does a capability check using flags, but then the
real ioctl double-fetches flags and uses potentially different value.
The "Check the capability before cred override" comment misleading: user
can skip this check by presenting benign flags first and then overwriting
them to non-benign flags.
Just remove the cred override for now, hoping this doesn't cause a
regression.
The proper solution is to create a new setxflags i_op (patches are in the
works).
Xfstests don't show a regression.
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Fixes: dab5ca8fd9 ("ovl: add lsattr/chattr support")
Cc: <stable@vger.kernel.org> # v4.19
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit d6718941a2 upstream.
It's very easy to crash the kernel right now by simply trying to
enable memtrace concurrently, hammering on the "enable" interface
loop.sh:
#!/bin/bash
dmesg --console-off
while true; do
echo 0x40000000 > /sys/kernel/debug/powerpc/memtrace/enable
done
[root@localhost ~]# loop.sh &
[root@localhost ~]# loop.sh &
Resulting quickly in a kernel crash. Let's properly protect using a
mutex.
Fixes: 9d5171a8f2 ("powerpc/powernv: Enable removal of memory for in memory tracing")
Cc: stable@vger.kernel.org# v4.14+
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201111145322.15793-3-david@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit b1198a8823 upstream.
We execute certain NPU2 setup code (such as mapping an LPID to a device
in NPU2) unconditionally if an Nvlink bridge is detected. However this
cannot succeed on POWER8NVL machines and errors appear in dmesg. This is
harmless as skiboot returns an error and the only place we check it is
vfio-pci but that code does not get called on P8+ either.
This adds a check if pnv_npu2_xxx helpers are called on a machine with
NPU2 which initializes pnv_phb::npu in pnv_npu2_init();
pnv_phb::npu==NULL on POWER8/NVL (Naples).
While at this, fix NULL derefencing in pnv_npu_peers_take_ownership/
pnv_npu_peers_release_ownership which occurs when GPUs on mentioned P8s
cause EEH which happens if "vfio-pci" disables devices using
the D3 power state; the vfio-pci's disable_idle_d3 module parameter
controls this and must be set on Naples. The EEH handling clears
the entire pnv_ioda_pe struct in pnv_ioda_free_pe() hence
the NULL derefencing. We cannot recover from that but at least we stop
crashing.
Tested on
- POWER9 pvr=004e1201, Ubuntu 19.04 host, Ubuntu 18.04 vm,
NVIDIA GV100 10de:1db1 driver 418.39
- POWER8 pvr=004c0100, RHEL 7.6 host, Ubuntu 16.10 vm,
NVIDIA P100 10de:15f9 driver 396.47
Fixes: 1b785611e1 ("powerpc/powernv/npu: Add release_ownership hook")
Cc: stable@vger.kernel.org # 5.0
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20201122073828.15446-1-aik@ozlabs.ru
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
