OSDN Git Service

tomoyo/tomoyo-test1.git
2 years agovirtio_ring: validate used buffer length
Jason Wang [Wed, 27 Oct 2021 02:21:04 +0000 (10:21 +0800)]
virtio_ring: validate used buffer length

This patch validate the used buffer length provided by the device
before trying to use it. This is done by record the in buffer length
in a new field in desc_state structure during virtqueue_add(), then we
can fail the virtqueue_get_buf() when we find the device is trying to
give us a used buffer length which is greater than the in buffer
length.

Since some drivers have already done the validation by themselves,
this patch tries to makes the core validation optional. For the driver
that doesn't want the validation, it can set the
suppress_used_validation to be true (which could be overridden by
force_used_validation module parameter). To be more efficient, a
dedicate array is used for storing the validate used length, this
helps to eliminate the cache stress if validation is done by the
driver.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211027022107.14357-2-jasowang@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_blk: correct types for status handling
Michael S. Tsirkin [Mon, 25 Oct 2021 07:54:03 +0000 (03:54 -0400)]
virtio_blk: correct types for status handling

virtblk_setup_cmd returns blk_status_t in an int, callers then assign it
back to a blk_status_t variable. blk_status_t is either u32 or (more
typically) u8 so it works, but is inelegant and causes sparse warnings.

Pass the status in blk_status_t in a consistent way.

Reported-by: kernel test robot <lkp@intel.com>
Fixes: b2c5221fd074 ("virtio-blk: avoid preallocating big SGL for data")
Cc: Max Gurtovoy <mgurtovoy@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
2 years agovirtio_blk: allow 0 as num_request_queues
Michael S. Tsirkin [Sun, 24 Oct 2021 13:41:40 +0000 (09:41 -0400)]
virtio_blk: allow 0 as num_request_queues

The default value is 0 meaning "no limit". However if 0
is specified on the command line it is instead silently
converted to 1. Further, the value is already validated
at point of use, there's no point in duplicating code
validating the value when it is set.

Simplify the code while making the behaviour more consistent
by using plain module_param.

Fixes: 1a662cf6cb9a ("virtio-blk: add num_request_queues module parameter")
Cc: Max Gurtovoy <mgurtovoy@nvidia.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoi2c: virtio: Add support for zero-length requests
Viresh Kumar [Thu, 21 Oct 2021 09:47:49 +0000 (15:17 +0530)]
i2c: virtio: Add support for zero-length requests

The virtio specification received a new mandatory feature
(VIRTIO_I2C_F_ZERO_LENGTH_REQUEST) for zero length requests. Fail if the
feature isn't offered by the device.

For each read-request, set the VIRTIO_I2C_FLAGS_M_RD flag, as required
by the VIRTIO_I2C_F_ZERO_LENGTH_REQUEST feature.

This allows us to support zero length requests, like SMBUS Quick, where
the buffer need not be sent anymore.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Link: https://lore.kernel.org/r/7c58868cd26d2fc4bd82d0d8b0dfb55636380110.1634808714.git.viresh.kumar@linaro.org
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jie Deng <jie.deng@intel.com> # once the spec is merged
2 years agovirtio-blk: fixup coccinelle warnings
Ye Guojin [Thu, 21 Oct 2021 06:51:11 +0000 (06:51 +0000)]
virtio-blk: fixup coccinelle warnings

coccicheck complains about the use of snprintf() in sysfs show
functions:
WARNING  use scnprintf or sprintf

Use sysfs_emit instead of scnprintf or sprintf makes more sense.

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: Ye Guojin <ye.guojin@zte.com.cn>
Link: https://lore.kernel.org/r/20211021065111.1047824-1-ye.guojin@zte.com.cn
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
2 years agovirtio_ring: fix typos in vring_desc_extra
Jason Wang [Tue, 19 Oct 2021 07:01:48 +0000 (15:01 +0800)]
virtio_ring: fix typos in vring_desc_extra

We're actually tracking descriptor address and length instead of the
buffer.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211019070152.8236-7-jasowang@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-pci: harden INTX interrupts
Jason Wang [Tue, 19 Oct 2021 07:01:47 +0000 (15:01 +0800)]
virtio-pci: harden INTX interrupts

This patch tries to make sure the virtio interrupt handler for INTX
won't be called after a reset and before virtio_device_ready(). We
can't use IRQF_NO_AUTOEN since we're using shared interrupt
(IRQF_SHARED). So this patch tracks the INTX enabling status in a new
intx_soft_enabled variable and toggle it during in
vp_disable/enable_vectors(). The INTX interrupt handler will check
intx_soft_enabled before processing the actual interrupt.

Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211019070152.8236-6-jasowang@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_pci: harden MSI-X interrupts
Jason Wang [Tue, 19 Oct 2021 07:01:46 +0000 (15:01 +0800)]
virtio_pci: harden MSI-X interrupts

We used to synchronize pending MSI-X irq handlers via
synchronize_irq(), this may not work for the untrusted device which
may keep sending interrupts after reset which may lead unexpected
results. Similarly, we should not enable MSI-X interrupt until the
device is ready. So this patch fixes those two issues by:

1) switching to use disable_irq() to prevent the virtio interrupt
   handlers to be called after the device is reset.
2) using IRQF_NO_AUTOEN and enable the MSI-X irq during .ready()

This can make sure the virtio interrupt handler won't be called before
virtio_device_ready() and after reset.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211019070152.8236-5-jasowang@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_config: introduce a new .enable_cbs method
Jason Wang [Tue, 19 Oct 2021 07:01:45 +0000 (15:01 +0800)]
virtio_config: introduce a new .enable_cbs method

This patch introduces a new method to enable the callbacks for config
and virtqueues. This will be used for making sure the virtqueue
callbacks are only enabled after virtio_device_ready() if transport
implements this method.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211019070152.8236-4-jasowang@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_console: validate max_nr_ports before trying to use it
Jason Wang [Tue, 19 Oct 2021 07:01:44 +0000 (15:01 +0800)]
virtio_console: validate max_nr_ports before trying to use it

We calculate nr_ports based on the max_nr_ports:

nr_queues = use_multiport(portdev) ? (nr_ports + 1) * 2 : 2;

If the device advertises a large max_nr_ports, we will end up with a
integer overflow. Fixing this by validating the max_nr_ports and fail
the probe for invalid max_nr_ports in this case.

Cc: Amit Shah <amit@kernel.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211019070152.8236-3-jasowang@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_blk: Fix spelling mistake: "advertisted" -> "advertised"
Colin Ian King [Mon, 25 Oct 2021 10:22:40 +0000 (11:22 +0100)]
virtio_blk: Fix spelling mistake: "advertisted" -> "advertised"

There is a spelling mistake in a dev_err error message. Fix it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20211025102240.22801-1-colin.i.king@gmail.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Acked-by: Jason Wang <jasowang@redhat.com>
2 years agovirtio-blk: validate num_queues during probe
Jason Wang [Tue, 19 Oct 2021 07:01:43 +0000 (15:01 +0800)]
virtio-blk: validate num_queues during probe

If an untrusted device neogitates BLK_F_MQ but advertises a zero
num_queues, the driver may end up trying to allocating zero size
buffers where ZERO_SIZE_PTR is returned which may pass the checking
against the NULL. This will lead unexpected results.

Fixing this by failing the probe in this case.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211019070152.8236-2-jasowang@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
2 years agovirtio-pmem: add myself as virtio-pmem maintainer
Pankaj Gupta [Sat, 16 Oct 2021 09:06:46 +0000 (11:06 +0200)]
virtio-pmem: add myself as virtio-pmem maintainer

Adding myself as virtio-pmem maintainer and also adding virtualization
mailing list entry for virtio specific bits. Helps to get notified for
appropriate bug fixes & enhancements.

Signed-off-by: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Link: https://lore.kernel.org/r/20211016090646.371145-1-pankaj.gupta.linux@gmail.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoALSA: virtio: Replace zero-length array with flexible-array member
Gustavo A. R. Silva [Wed, 29 Sep 2021 19:15:04 +0000 (14:15 -0500)]
ALSA: virtio: Replace zero-length array with flexible-array member

There is a regular need in the kernel to provide a way to declare
having a dynamically sized set of trailing elements in a structure.
Kernel code should always use “flexible array members”[1] for these
cases. The older style of one-element or zero-length arrays should
no longer be used[2].

Also, make use of the struct_size() helper in kzalloc().

[1] https://en.wikipedia.org/wiki/Flexible_array_member
[2] https://www.kernel.org/doc/html/v5.10/process/deprecated.html#zero-length-and-one-element-arrays

Link: https://github.com/KSPP/linux/issues/78
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Link: https://lore.kernel.org/r/20210929191504.GA337268@embeddedor
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_ring: check desc == NULL when using indirect with packed
Xuan Zhuo [Wed, 20 Oct 2021 11:23:23 +0000 (19:23 +0800)]
virtio_ring: check desc == NULL when using indirect with packed

When using indirect with packed, we don't check for allocation failures.
This patch checks that and fall back on direct.

Fixes: 1ce9e6055fa0 ("virtio_ring: introduce packed ring support")
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://lore.kernel.org/r/20211020112323.67466-3-xuanzhuo@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_ring: make virtqueue_add_indirect_packed prettier
Xuan Zhuo [Wed, 20 Oct 2021 11:23:22 +0000 (19:23 +0800)]
virtio_ring: make virtqueue_add_indirect_packed prettier

Align the arguments of virtqueue_add_indirect_packed() to the open ( to
make it look prettier.

Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/20211020112323.67466-2-xuanzhuo@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohwrng: virtio - always add a pending request
Laurent Vivier [Thu, 28 Oct 2021 10:11:11 +0000 (12:11 +0200)]
hwrng: virtio - always add a pending request

If we ensure we have already some data available by enqueuing
again the buffer once data are exhausted, we can return what we
have without waiting for the device answer.

Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Link: https://lore.kernel.org/r/20211028101111.128049-5-lvivier@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohwrng: virtio - don't waste entropy
Laurent Vivier [Thu, 28 Oct 2021 10:11:10 +0000 (12:11 +0200)]
hwrng: virtio - don't waste entropy

if we don't use all the entropy available in the buffer, keep it
and use it later.

Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Link: https://lore.kernel.org/r/20211028101111.128049-4-lvivier@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohwrng: virtio - don't wait on cleanup
Laurent Vivier [Thu, 28 Oct 2021 10:11:09 +0000 (12:11 +0200)]
hwrng: virtio - don't wait on cleanup

When virtio-rng device was dropped by the hwrng core we were forced
to wait the buffer to come back from the device to not have
remaining ongoing operation that could spoil the buffer.

But now, as the buffer is internal to the virtio-rng we can release
the waiting loop immediately, the buffer will be retrieve and use
when the virtio-rng driver will be selected again.

This avoids to hang on an rng_current write command if the virtio-rng
device is blocked by a lack of entropy. This allows to select
another entropy source if the current one is empty.

Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Link: https://lore.kernel.org/r/20211028101111.128049-3-lvivier@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agohwrng: virtio - add an internal buffer
Laurent Vivier [Thu, 28 Oct 2021 10:11:08 +0000 (12:11 +0200)]
hwrng: virtio - add an internal buffer

hwrng core uses two buffers that can be mixed in the
virtio-rng queue.

If the buffer is provided with wait=0 it is enqueued in the
virtio-rng queue but unused by the caller.
On the next call, core provides another buffer but the
first one is filled instead and the new one queued.
And the caller reads the data from the new one that is not
updated, and the data in the first one are lost.

To avoid this mix, virtio-rng needs to use its own unique
internal buffer at a cost of a data copy to the caller buffer.

Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Link: https://lore.kernel.org/r/20211028101111.128049-2-lvivier@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovdpa/mlx5: Propagate link status from device to vdpa driver
Eli Cohen [Thu, 9 Sep 2021 12:36:35 +0000 (15:36 +0300)]
vdpa/mlx5: Propagate link status from device to vdpa driver

Add code to register to hardware asynchronous events. Use this
mechanism to track link status events coming from the device and update
the config struct.

After doing link status change, call the vdpa callback to notify of the
link status change.

Signed-off-by: Eli Cohen <elic@nvidia.com>
Link: https://lore.kernel.org/r/20210909123635.30884-4-elic@nvidia.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
2 years agovdpa/mlx5: Rename control VQ workqueue to vdpa wq
Eli Cohen [Thu, 9 Sep 2021 12:36:34 +0000 (15:36 +0300)]
vdpa/mlx5: Rename control VQ workqueue to vdpa wq

A subesequent patch will use the same workqueue for executing other
work not related to control VQ. Rename the workqueue and the work queue
entry used to convey information to the workqueue.

Signed-off-by: Eli Cohen <elic@nvidia.com>
Link: https://lore.kernel.org/r/20210909123635.30884-3-elic@nvidia.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
2 years agovdpa/mlx5: Remove mtu field from vdpa net device
Eli Cohen [Thu, 9 Sep 2021 12:36:33 +0000 (15:36 +0300)]
vdpa/mlx5: Remove mtu field from vdpa net device

No need to save the mtu int the net device struct. We can save it in the
config struct which cannot be modified.

Moreover, move the initialization to. mlx5_vdpa_set_features() callback
is not the right place to put it.

Signed-off-by: Eli Cohen <elic@nvidia.com>
Link: https://lore.kernel.org/r/20210909123635.30884-2-elic@nvidia.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoeni_vdpa: add vDPA driver for Alibaba ENI
Wu Zongyong [Fri, 29 Oct 2021 09:14:49 +0000 (17:14 +0800)]
eni_vdpa: add vDPA driver for Alibaba ENI

This patch adds a new vDPA driver for Alibaba ENI(Elastic Network
Interface) which is build upon virtio 0.9.5 specification.
And this driver is only enabled on X86 host currently.

Link: https://lore.kernel.org/r/6a9f32c00609af16bbb2ea32e633b3beb1cbf84b.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Link: https://lore.kernel.org/r/20211026083214.3375383-1-arnd@kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de> # fix Kconfig typo
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovdpa: add new attribute VDPA_ATTR_DEV_MIN_VQ_SIZE
Wu Zongyong [Fri, 29 Oct 2021 09:14:48 +0000 (17:14 +0800)]
vdpa: add new attribute VDPA_ATTR_DEV_MIN_VQ_SIZE

This attribute advertises the min value of virtqueue size. The value is
1 by default.

Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Link: https://lore.kernel.org/r/2bbc417355c4d22298050b1ba887cecfbde3e85d.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_vdpa: setup correct vq size with callbacks get_vq_num_{max,min}
Wu Zongyong [Fri, 29 Oct 2021 09:14:47 +0000 (17:14 +0800)]
virtio_vdpa: setup correct vq size with callbacks get_vq_num_{max,min}

For the devices which implement the get_vq_num_min callback, the driver
should not negotiate with virtqueue size with the backend vdpa device if
the value returned by get_vq_num_min equals to the value returned by
get_vq_num_max.
This is useful for vdpa devices based on legacy virtio specfication.

Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Link: https://lore.kernel.org/r/bc0551cec6c3f3dd9424b678b7c22d882aebab3a.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovdpa: min vq num of vdpa device cannot be greater than max vq num
Wu Zongyong [Fri, 29 Oct 2021 09:14:46 +0000 (17:14 +0800)]
vdpa: min vq num of vdpa device cannot be greater than max vq num

Just failed to probe the vdpa device if the min virtqueue num returned
by get_vq_num_min is greater than the max virtqueue num returned by
get_vq_num_max.

Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/21199b62cc10b2a9f2cf90eeb63ad080645d881f.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovdpa: add new callback get_vq_num_min in vdpa_config_ops
Wu Zongyong [Fri, 29 Oct 2021 09:14:45 +0000 (17:14 +0800)]
vdpa: add new callback get_vq_num_min in vdpa_config_ops

This callback is optional. For vdpa devices that not support to change
virtqueue size, get_vq_num_min and get_vq_num_max will return the same
value, so that users can choose a correct value for that device.

Suggested-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/f4af5b0abd660d9a29ab6b2f67bd6df10284a230.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovp_vdpa: add vq irq offloading support
Wu Zongyong [Fri, 29 Oct 2021 09:14:44 +0000 (17:14 +0800)]
vp_vdpa: add vq irq offloading support

This patch implements the get_vq_irq() callback for virtio pci devices
to allow irq offloading.

Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/bb091e5505db704dd620f8854a7aebc921d2a752.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovdpa: fix typo
Wu Zongyong [Fri, 29 Oct 2021 09:14:43 +0000 (17:14 +0800)]
vdpa: fix typo

Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/4b5153262e4ba64986bb567d7425ad4829ca7bcc.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-pci: introduce legacy device module
Wu Zongyong [Fri, 29 Oct 2021 09:14:42 +0000 (17:14 +0800)]
virtio-pci: introduce legacy device module

Split common codes from virtio-pci-legacy so vDPA driver can reuse it
later.

Signed-off-by: Wu Zongyong <wuzongyong@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://lore.kernel.org/r/71605acde5e97fcb2760a6973e406279fb1bbd33.1635493219.git.wuzongyong@linux.alibaba.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-blk: add num_request_queues module parameter
Max Gurtovoy [Thu, 2 Sep 2021 20:46:22 +0000 (23:46 +0300)]
virtio-blk: add num_request_queues module parameter

Sometimes a user would like to control the amount of request queues to
be created for a block device. For example, for limiting the memory
footprint of virtio-blk devices.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Link: https://lore.kernel.org/r/20210902204622.54354-1-mgurtovoy@nvidia.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio-blk: avoid preallocating big SGL for data
Max Gurtovoy [Wed, 1 Sep 2021 13:14:34 +0000 (16:14 +0300)]
virtio-blk: avoid preallocating big SGL for data

No need to pre-allocate a big buffer for the IO SGL anymore. If a device
has lots of deep queues, preallocation for the sg list can consume
substantial amounts of memory. For HW virtio-blk device, nr_hw_queues
can be 64 or 128 and each queue's depth might be 128. This means the
resulting preallocation for the data SGLs is big.

Switch to runtime allocation for SGL for lists longer than 2 entries.
This is the approach used by NVMe drivers so it should be reasonable for
virtio block as well. Runtime SGL allocation has always been the case
for the legacy I/O path so this is nothing new.

The preallocated small SGL depends on SG_CHAIN so if the ARCH doesn't
support SG_CHAIN, use only runtime allocation for the SGL.

Re-organize the setup of the IO request to fit the new sg chain
mechanism.

No performance degradation was seen (fio libaio engine with 16 jobs and
128 iodepth):

IO size      IOPs Rand Read (before/after)         IOPs Rand Write (before/after)
--------     ---------------------------------    ----------------------------------
512B          318K/316K                                    329K/325K

4KB           323K/321K                                    353K/349K

16KB          199K/208K                                    250K/275K

128KB         36K/36.1K                                    39.2K/41.7K

Signed-off-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Reviewed-by: Israel Rukshin <israelr@nvidia.com>
Link: https://lore.kernel.org/r/20210901131434.31158-1-mgurtovoy@nvidia.com
Reviewed-by: Feng Li <lifeng1519@gmail.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Tested-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Arnd Bergmann <arnd@arndb.de> # kconfig fixups
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agovirtio_net: clarify tailroom logic
Michael S. Tsirkin [Sat, 9 Oct 2021 14:52:50 +0000 (10:52 -0400)]
virtio_net: clarify tailroom logic

Make tailroom math follow same logic as everything else, subtracing
values in the order in which things are laid out in the buffer.

Tested-by: Corentin Noël <corentin.noel@collabora.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
2 years agoLinux 5.15 v5.15
Linus Torvalds [Sun, 31 Oct 2021 20:53:10 +0000 (13:53 -0700)]
Linux 5.15

2 years agoMerge tag 'perf-tools-fixes-for-v5.15-2021-10-31' of git://git.kernel.org/pub/scm...
Linus Torvalds [Sun, 31 Oct 2021 18:24:06 +0000 (11:24 -0700)]
Merge tag 'perf-tools-fixes-for-v5.15-2021-10-31' of git://git./linux/kernel/git/acme/linux

Pull perf tools fixes from Arnaldo Carvalho de Melo:

 - Fix compilation of callchain related code on powerpc with gcc11+

 - Fix PERF_SAMPLE_WEIGHT_STRUCT support in 'perf script'

 - Check session->header.env.arch before using it, fixing a segmentation
   fault

 - Suppress 'rm dlfilter' build messages

* tag 'perf-tools-fixes-for-v5.15-2021-10-31' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux:
  perf script: Fix PERF_SAMPLE_WEIGHT_STRUCT support
  perf callchain: Fix compilation on powerpc with gcc11+
  perf script: Check session->header.env.arch before using it
  perf build: Suppress 'rm dlfilter' build message

2 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sun, 31 Oct 2021 18:19:02 +0000 (11:19 -0700)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull kvm fixes from Paolo Bonzini:

 - Fixes for s390 interrupt delivery

 - Fixes for Xen emulator bugs showing up as debug kernel WARNs

 - Fix another issue with SEV/ES string I/O VMGEXITs

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: x86: Take srcu lock in post_kvm_run_save()
  KVM: SEV-ES: fix another issue with string I/O VMGEXITs
  KVM: x86/xen: Fix kvm_xen_has_interrupt() sleeping in kvm_vcpu_block()
  KVM: x86: switch pvclock_gtod_sync_lock to a raw spinlock
  KVM: s390: preserve deliverable_mask in __airqs_kick_single_vcpu
  KVM: s390: clear kicked_mask before sleeping again

2 years agoperf script: Fix PERF_SAMPLE_WEIGHT_STRUCT support
Kan Liang [Wed, 29 Sep 2021 15:38:13 +0000 (08:38 -0700)]
perf script: Fix PERF_SAMPLE_WEIGHT_STRUCT support

-F weight in perf script is broken.

  # ./perf mem record
  # ./perf script -F weight
  Samples for 'dummy:HG' event do not have WEIGHT attribute set. Cannot
print 'weight' field.

The sample type, PERF_SAMPLE_WEIGHT_STRUCT, is an alternative of the
PERF_SAMPLE_WEIGHT sample type. They share the same space, weight. The
lower 32 bits are exactly the same for both sample type. The higher 32
bits may be different for different architecture. For a new kernel on
x86, the PERF_SAMPLE_WEIGHT_STRUCT is used. For an old kernel or other
ARCHs, the PERF_SAMPLE_WEIGHT is used.

With -F weight, current perf script will only check the input string
"weight" with the PERF_SAMPLE_WEIGHT sample type. Because the commit
ea8d0ed6eae3 ("perf tools: Support PERF_SAMPLE_WEIGHT_STRUCT") didn't
update the PERF_SAMPLE_WEIGHT_STRUCT sample type for perf script. For a
new kernel on x86, the check fails.

Use PERF_SAMPLE_WEIGHT_TYPE, which supports both sample types, to
replace PERF_SAMPLE_WEIGHT

Fixes: ea8d0ed6eae37b01 ("perf tools: Support PERF_SAMPLE_WEIGHT_STRUCT")
Reported-by: Joe Mario <jmario@redhat.com>
Reviewed-by: Kajol Jain <kjain@linux.ibm.com>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Tested-by: Jiri Olsa <jolsa@redhat.com>
Tested-by: Joe Mario <jmario@redhat.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Joe Mario <jmario@redhat.com>
Cc: Andi Kleen <ak@linux.intel.com>
Link: https://lore.kernel.org/r/1632929894-102778-1-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agoperf callchain: Fix compilation on powerpc with gcc11+
Jiri Olsa [Tue, 28 Sep 2021 19:52:53 +0000 (21:52 +0200)]
perf callchain: Fix compilation on powerpc with gcc11+

Got following build fail on powerpc:

    CC      arch/powerpc/util/skip-callchain-idx.o
  In function ‘check_return_reg’,
      inlined from ‘check_return_addr’ at arch/powerpc/util/skip-callchain-idx.c:213:7,
      inlined from ‘arch_skip_callchain_idx’ at arch/powerpc/util/skip-callchain-idx.c:265:7:
  arch/powerpc/util/skip-callchain-idx.c:54:18: error: ‘dwarf_frame_register’ accessing 96 bytes \
  in a region of size 64 [-Werror=stringop-overflow=]
     54 |         result = dwarf_frame_register(frame, ra_regno, ops_mem, &ops, &nops);
        |                  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  arch/powerpc/util/skip-callchain-idx.c: In function ‘arch_skip_callchain_idx’:
  arch/powerpc/util/skip-callchain-idx.c:54:18: note: referencing argument 3 of type ‘Dwarf_Op *’
  In file included from /usr/include/elfutils/libdwfl.h:32,
                   from arch/powerpc/util/skip-callchain-idx.c:10:
  /usr/include/elfutils/libdw.h:1069:12: note: in a call to function ‘dwarf_frame_register’
   1069 | extern int dwarf_frame_register (Dwarf_Frame *frame, int regno,
        |            ^~~~~~~~~~~~~~~~~~~~
  cc1: all warnings being treated as errors

The dwarf_frame_register args changed with [1],
Updating ops_mem accordingly.

[1] https://sourceware.org/git/?p=elfutils.git;a=commit;h=5621fe5443da23112170235dd5cac161e5c75e65

Reviewed-by: Kajol Jain <kjain@linux.ibm.com>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Mark Wieelard <mjw@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Link: https://lore.kernel.org/r/20210928195253.1267023-1-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agoperf script: Check session->header.env.arch before using it
Song Liu [Mon, 4 Oct 2021 05:32:38 +0000 (22:32 -0700)]
perf script: Check session->header.env.arch before using it

When perf.data is not written cleanly, we would like to process existing
data as much as possible (please see f_header.data.size == 0 condition
in perf_session__read_header). However, perf.data with partial data may
crash perf. Specifically, we see crash in 'perf script' for NULL
session->header.env.arch.

Fix this by checking session->header.env.arch before using it to determine
native_arch. Also split the if condition so it is easier to read.

Committer notes:

If it is a pipe, we already assume is a native arch, so no need to check
session->header.env.arch.

Signed-off-by: Song Liu <songliubraving@fb.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: kernel-team@fb.com
Cc: stable@vger.kernel.org
Link: http://lore.kernel.org/lkml/20211004053238.514936-1-songliubraving@fb.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agoperf build: Suppress 'rm dlfilter' build message
Adrian Hunter [Thu, 30 Sep 2021 06:28:49 +0000 (09:28 +0300)]
perf build: Suppress 'rm dlfilter' build message

The following build message:

rm dlfilters/dlfilter-test-api-v0.o

is unwanted.

The object file is being treated as an intermediate file and being
automatically removed. Mark the object file as .SECONDARY to prevent
removal and hence the message.

Requested-by: Arnaldo Carvalho de Melo <acme@kernel.org>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lore.kernel.org/lkml/20210930062849.110416-1-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sat, 30 Oct 2021 22:56:38 +0000 (15:56 -0700)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Three small fixes, all in drivers, and one sizeable update to the UFS
  driver to remove the HPB 2.0 feature that has been objected to by Jens
  and Christoph.

  Although the UFS patch is large and last minute, it's essentially the
  least intrusive way of resolving the objections in time for the 5.15
  release"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: ufs: ufshpb: Remove HPB2.0 flows
  scsi: mpt3sas: Fix reference tag handling for WRITE_INSERT
  scsi: ufs: ufs-exynos: Correct timeout value setting registers
  scsi: ibmvfc: Fix up duplicate response detection

2 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 30 Oct 2021 16:55:46 +0000 (09:55 -0700)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fix from Stephen Boyd:
 "One fix for the composite clk that broke when we changed this clk type
  to use the determine_rate instead of round_rate clk op by default.
  This caused lots of problems on Rockchip SoCs because they heavily use
  the composite clk code to model the clk tree"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: composite: Also consider .determine_rate for rate + mux composites

2 years agoMerge tag 'riscv-for-linus-5.15-rc8' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 30 Oct 2021 16:28:24 +0000 (09:28 -0700)]
Merge tag 'riscv-for-linus-5.15-rc8' of git://git./linux/kernel/git/riscv/linux

Pull RISC-V fixes from Palmer Dabbelt:
 "These are pretty late, but they do fix concrete issues.

   - ensure the trap vector's address is aligned.

   - avoid re-populating the KASAN shadow memory.

   - allow kasan to build without warnings, which have recently become
     errors"

* tag 'riscv-for-linus-5.15-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux:
  riscv: Fix asan-stack clang build
  riscv: Do not re-populate shadow memory with kasan_populate_early_shadow
  riscv: fix misalgned trap vector base address

2 years agoscsi: ufs: ufshpb: Remove HPB2.0 flows
Avri Altman [Sat, 30 Oct 2021 06:23:01 +0000 (09:23 +0300)]
scsi: ufs: ufshpb: Remove HPB2.0 flows

The Host Performance Buffer feature allows UFS read commands to carry the
physical media addresses along with the LBAs, thus allowing less internal
L2P-table switches in the device.  HPB1.0 allowed a single LBA, while
HPB2.0 increases this capacity up to 255 blocks.

Carrying more than a single record, the read operation is no longer purely
of type "read" but a "hybrid" command: Writing the physical address to the
device in one operation and reading back the required payload in another.

The JEDEC HPB spec defines two commands for this operation:
HPB-WRITE-BUFFER (0x2) to write the physical addresses to device, and
HPB-READ to read the payload.

With the current HPB design the UFS driver has no alternative but to divide
the READ request into 2 separate commands: HPB-WRITE-BUFFER and HPB-READ.
This causes a great deal of aggravation to the block layer guys who
demanded that we completely revert the entire HPB driver regardless of the
huge amount of corporate effort already invested in it.

As a compromise, remove only the pieces that implement the 2.0
specification. This is done as a matter of urgency for the final 5.15
release.

Link: https://lore.kernel.org/r/20211030062301.248-1-avri.altman@wdc.com
Tested-by: Avri Altman <avri.altman@wdc.com>
Tested-by: Bean Huo <beanhuo@micron.com>
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Co-developed-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2 years agoMerge tag 'powerpc-5.15-6' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Sat, 30 Oct 2021 00:35:56 +0000 (17:35 -0700)]
Merge tag 'powerpc-5.15-6' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:
 "Three commits fixing some issues introduced with the recent IOMMU
  changes we merged.

  Thanks to Alexey Kardashevskiy"

* tag 'powerpc-5.15-6' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/pseries/iommu: Create huge DMA window if no MMIO32 is present
  powerpc/pseries/iommu: Check if the default window in use before removing it
  powerpc/pseries/iommu: Use correct vfree for it_map

2 years agoMerge tag 'gpio-fixes-for-v5.15' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 30 Oct 2021 00:04:38 +0000 (17:04 -0700)]
Merge tag 'gpio-fixes-for-v5.15' of git://git./linux/kernel/git/brgl/linux

Pull gpio fixes from Bartosz Golaszewski:

 - fix the return value check when parsing the ngpios property in
   gpio-xgs-iproc

 - check the return value of bgpio_init() in gpio-mlxbf2

* tag 'gpio-fixes-for-v5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  gpio: mlxbf2.c: Add check for bgpio_init failure
  gpio: xgs-iproc: fix parsing of ngpios property

2 years agoMerge tag 'block-5.15-2021-10-29' of git://git.kernel.dk/linux-block
Linus Torvalds [Fri, 29 Oct 2021 18:10:29 +0000 (11:10 -0700)]
Merge tag 'block-5.15-2021-10-29' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:

 - NVMe pull request:
      - fix nvmet-tcp header digest verification (Amit Engel)
      - fix a memory leak in nvmet-tcp when releasing a queue (Maurizio
        Lombardi)
      - fix nvme-tcp H2CData PDU send accounting again (Sagi Grimberg)
      - fix digest pointer calculation in nvme-tcp and nvmet-tcp (Varun
        Prakash)
      - fix possible nvme-tcp req->offset corruption (Varun Prakash)

 - Queue drain ordering fix (Ming)

 - Partition check regression for zoned devices (Shin'ichiro)

 - Zone queue restart fix (Naohiro)

* tag 'block-5.15-2021-10-29' of git://git.kernel.dk/linux-block:
  block: Fix partition check for host-aware zoned block devices
  nvmet-tcp: fix header digest verification
  nvmet-tcp: fix data digest pointer calculation
  nvme-tcp: fix data digest pointer calculation
  nvme-tcp: fix possible req->offset corruption
  block: schedule queue restart after BLK_STS_ZONE_RESOURCE
  block: drain queue after disk is removed from sysfs
  nvme-tcp: fix H2CData PDU send accounting (again)
  nvmet-tcp: fix a memory leak when releasing a queue

2 years agoscsi: mpt3sas: Fix reference tag handling for WRITE_INSERT
Martin K. Petersen [Thu, 28 Oct 2021 03:42:02 +0000 (23:42 -0400)]
scsi: mpt3sas: Fix reference tag handling for WRITE_INSERT

Testing revealed a problem with how the reference tag was handled for
a WRITE_INSERT operation. The SCSI_PROT_REF_CHECK flag is not set when
the controller is asked to generate the protection information
(i.e. not DIX). And as a result the initial reference tag would not be
set in the WRITE_INSERT case.

Separate handling of the REF_CHECK and REF_INCREMENT flags to align
with both the DIX spec and the MPI implementation.

Link: https://lore.kernel.org/r/20211028034202.24225-1-martin.petersen@oracle.com
Fixes: b3e2c72af1d5 ("scsi: mpt3sas: Use the proper SCSI midlayer interfaces for PI")
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
2 years agoMerge tag 'mmc-v5.15-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc
Linus Torvalds [Fri, 29 Oct 2021 17:54:44 +0000 (10:54 -0700)]
Merge tag 'mmc-v5.15-rc5' of git://git./linux/kernel/git/ulfh/mmc

Pull MMC fixes from Ulf Hansson:

 - tmio: Re-enable card irqs after a reset

 - mtk-sd: Fixup probing of cqhci for crypto

 - cqhci: Fix support for suspend/resume

 - vub300: Fix control-message timeouts

 - dw_mmc-exynos: Fix support for tuning

 - winbond: Silences build errors on M68K

 - sdhci-esdhc-imx: Fix support for tuning

 - sdhci-pci: Read card detect from ACPI for Intel Merrifield

 - sdhci: Fix eMMC support for Thundercomm TurboX CM2290

* tag 'mmc-v5.15-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc:
  mmc: tmio: reenable card irqs after the reset callback
  mmc: mediatek: Move cqhci init behind ungate clock
  mmc: cqhci: clear HALT state after CQE enable
  mmc: vub300: fix control-message timeouts
  mmc: dw_mmc: exynos: fix the finding clock sample value
  mmc: winbond: don't build on M68K
  mmc: sdhci-esdhc-imx: clear the buffer_read_ready to reset standard tuning circuit
  mmc: sdhci-pci: Read card detect from ACPI for Intel Merrifield
  mmc: sdhci: Map more voltage level to SDHCI_POWER_330

2 years agoMerge tag 'for-5.15-rc7-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave...
Linus Torvalds [Fri, 29 Oct 2021 17:46:59 +0000 (10:46 -0700)]
Merge tag 'for-5.15-rc7-tag' of git://git./linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:
 "Last minute fixes for crash on 32bit architectures when compression is
  in use. It's a regression introduced in 5.15-rc and I'd really like
  not let this into the final release, fixes via stable trees would add
  unnecessary delay.

  The problem is on 32bit architectures with highmem enabled, the pages
  for compression may need to be kmapped, while the patches removed that
  as we don't use GFP_HIGHMEM allocations anymore. The pages that don't
  come from local allocation still may be from highmem. Despite being on
  32bit there's enough such ARM machines in use so it's not a marginal
  issue.

  I did full reverts of the patches one by one instead of a huge one.
  There's one exception for the "lzo" revert as there was an
  intermediate patch touching the same code to make it compatible with
  subpage. I can't revert that one too, so the revert in lzo.c is
  manual. Qu Wenruo has worked on that with me and verified the changes"

* tag 'for-5.15-rc7-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  Revert "btrfs: compression: drop kmap/kunmap from lzo"
  Revert "btrfs: compression: drop kmap/kunmap from zlib"
  Revert "btrfs: compression: drop kmap/kunmap from zstd"
  Revert "btrfs: compression: drop kmap/kunmap from generic helpers"

2 years agoMerge tag 'trace-v5.15-rc6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/rosted...
Linus Torvalds [Fri, 29 Oct 2021 17:41:07 +0000 (10:41 -0700)]
Merge tag 'trace-v5.15-rc6-3' of git://git./linux/kernel/git/rostedt/linux-trace

Pull tracing comment fixes from Steven Rostedt:

 - Some bots have informed me that some of the ftrace functions
   kernel-doc has formatting issues.

 - Also, fix my snake instinct.

* tag 'trace-v5.15-rc6-3' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
  tracing: Fix misspelling of "missing"
  ftrace: Fix kernel-doc formatting issues

2 years agoMerge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Fri, 29 Oct 2021 17:17:08 +0000 (10:17 -0700)]
Merge branch 'linus' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fix from Herbert Xu:
 "Fix a build-time warning in x86/sm4"

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: x86/sm4 - Fix invalid section entry size

2 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Fri, 29 Oct 2021 17:03:07 +0000 (10:03 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge misc fixes from Andrew Morton:
 "11 patches.

  Subsystems affected by this patch series: mm (memcg, memory-failure,
  oom-kill, secretmem, vmalloc, hugetlb, damon, and tools), and ocfs2"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  tools/testing/selftests/vm/split_huge_page_test.c: fix application of sizeof to pointer
  mm/damon/core-test: fix wrong expectations for 'damon_split_regions_of()'
  mm: khugepaged: skip huge page collapse for special files
  mm, thp: bail out early in collapse_file for writeback page
  mm/vmalloc: fix numa spreading for large hash tables
  mm/secretmem: avoid letting secretmem_users drop to zero
  ocfs2: fix race between searching chunks and release journal_head from buffer_head
  mm/oom_kill.c: prevent a race between process_mrelease and exit_mmap
  mm: filemap: check if THP has hwpoisoned subpage for PMD page fault
  mm: hwpoison: remove the unnecessary THP check
  memcg: page_alloc: skip bulk allocator for __GFP_ACCOUNT

2 years agoriscv: Fix asan-stack clang build
Alexandre Ghiti [Fri, 29 Oct 2021 04:59:27 +0000 (06:59 +0200)]
riscv: Fix asan-stack clang build

Nathan reported that because KASAN_SHADOW_OFFSET was not defined in
Kconfig, it prevents asan-stack from getting disabled with clang even
when CONFIG_KASAN_STACK is disabled: fix this by defining the
corresponding config.

Reported-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alexandre Ghiti <alexandre.ghiti@canonical.com>
Fixes: 8ad8b72721d0 ("riscv: Add KASAN support")
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
2 years agoriscv: Do not re-populate shadow memory with kasan_populate_early_shadow
Alexandre Ghiti [Fri, 29 Oct 2021 04:59:26 +0000 (06:59 +0200)]
riscv: Do not re-populate shadow memory with kasan_populate_early_shadow

When calling this function, all the shadow memory is already populated
with kasan_early_shadow_pte which has PAGE_KERNEL protection.
kasan_populate_early_shadow write-protects the mapping of the range
of addresses passed in argument in zero_pte_populate, which actually
write-protects all the shadow memory mapping since kasan_early_shadow_pte
is used for all the shadow memory at this point. And then when using
memblock API to populate the shadow memory, the first write access to the
kernel stack triggers a trap. This becomes visible with the next commit
that contains a fix for asan-stack.

We already manually populate all the shadow memory in kasan_early_init
and we write-protect kasan_early_shadow_pte at the end of kasan_init
which makes the calls to kasan_populate_early_shadow superfluous so
we can remove them.

Signed-off-by: Alexandre Ghiti <alexandre.ghiti@canonical.com>
Fixes: e178d670f251 ("riscv/kasan: add KASAN_VMALLOC support")
Fixes: 8ad8b72721d0 ("riscv: Add KASAN support")
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
2 years agotracing: Fix misspelling of "missing"
Steven Rostedt (VMware) [Fri, 29 Oct 2021 13:54:14 +0000 (09:54 -0400)]
tracing: Fix misspelling of "missing"

My snake instinct was on and I wrote "misssing" instead of "missing".

Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2 years agoftrace: Fix kernel-doc formatting issues
Steven Rostedt (VMware) [Fri, 29 Oct 2021 13:52:23 +0000 (09:52 -0400)]
ftrace: Fix kernel-doc formatting issues

Some functions had kernel-doc that used a comma instead of a hash to
separate the function name from the one line description.

Also, the "ftrace_is_dead()" had an incomplete description.

Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
2 years agoRevert "btrfs: compression: drop kmap/kunmap from lzo"
David Sterba [Wed, 27 Oct 2021 08:44:21 +0000 (10:44 +0200)]
Revert "btrfs: compression: drop kmap/kunmap from lzo"

This reverts commit 8c945d32e60427cbc0859cf7045bbe6196bb03d8.

The kmaps in compression code are still needed and cause crashes on
32bit machines (ARM, x86). Reproducible eg. by running fstest btrfs/004
with enabled LZO or ZSTD compression.

The revert does not apply cleanly due to changes in a6e66e6f8c1b
("btrfs: rework lzo_decompress_bio() to make it subpage compatible")
that reworked the page iteration so the revert is done to be equivalent
to the original code.

Link: https://lore.kernel.org/all/CAJCQCtT+OuemovPO7GZk8Y8=qtOObr0XTDp8jh4OHD6y84AFxw@mail.gmail.com/
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214839
Tested-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2 years agoRevert "btrfs: compression: drop kmap/kunmap from zlib"
David Sterba [Wed, 27 Oct 2021 08:42:43 +0000 (10:42 +0200)]
Revert "btrfs: compression: drop kmap/kunmap from zlib"

This reverts commit 696ab562e6df9fbafd6052d8ce4aafcb2ed16069.

The kmaps in compression code are still needed and cause crashes on
32bit machines (ARM, x86). Reproducible eg. by running fstest btrfs/004
with enabled LZO or ZSTD compression.

Link: https://lore.kernel.org/all/CAJCQCtT+OuemovPO7GZk8Y8=qtOObr0XTDp8jh4OHD6y84AFxw@mail.gmail.com/
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214839
Signed-off-by: David Sterba <dsterba@suse.com>
2 years agoRevert "btrfs: compression: drop kmap/kunmap from zstd"
David Sterba [Wed, 27 Oct 2021 08:42:27 +0000 (10:42 +0200)]
Revert "btrfs: compression: drop kmap/kunmap from zstd"

This reverts commit bbaf9715f3f5b5ff0de71da91fcc34ee9c198ed8.

The kmaps in compression code are still needed and cause crashes on
32bit machines (ARM, x86). Reproducible eg. by running fstest btrfs/004
with enabled LZO or ZSTD compression.

Example stacktrace with ZSTD on a 32bit ARM machine:

  Unable to handle kernel NULL pointer dereference at virtual address 00000000
  pgd = c4159ed3
  [00000000] *pgd=00000000
  Internal error: Oops: 5 [#1] PREEMPT SMP ARM
  Modules linked in:
  CPU: 0 PID: 210 Comm: kworker/u2:3 Not tainted 5.14.0-rc79+ #12
  Hardware name: Allwinner sun4i/sun5i Families
  Workqueue: btrfs-delalloc btrfs_work_helper
  PC is at mmiocpy+0x48/0x330
  LR is at ZSTD_compressStream_generic+0x15c/0x28c

  (mmiocpy) from [<c0629648>] (ZSTD_compressStream_generic+0x15c/0x28c)
  (ZSTD_compressStream_generic) from [<c06297dc>] (ZSTD_compressStream+0x64/0xa0)
  (ZSTD_compressStream) from [<c049444c>] (zstd_compress_pages+0x170/0x488)
  (zstd_compress_pages) from [<c0496798>] (btrfs_compress_pages+0x124/0x12c)
  (btrfs_compress_pages) from [<c043c068>] (compress_file_range+0x3c0/0x834)
  (compress_file_range) from [<c043c4ec>] (async_cow_start+0x10/0x28)
  (async_cow_start) from [<c0475c3c>] (btrfs_work_helper+0x100/0x230)
  (btrfs_work_helper) from [<c014ef68>] (process_one_work+0x1b4/0x418)
  (process_one_work) from [<c014f210>] (worker_thread+0x44/0x524)
  (worker_thread) from [<c0156aa4>] (kthread+0x180/0x1b0)
  (kthread) from [<c0100150>]

Link: https://lore.kernel.org/all/CAJCQCtT+OuemovPO7GZk8Y8=qtOObr0XTDp8jh4OHD6y84AFxw@mail.gmail.com/
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214839
Signed-off-by: David Sterba <dsterba@suse.com>
2 years agotools/testing/selftests/vm/split_huge_page_test.c: fix application of sizeof to pointer
David Yang [Thu, 28 Oct 2021 21:36:36 +0000 (14:36 -0700)]
tools/testing/selftests/vm/split_huge_page_test.c: fix application of sizeof to pointer

The coccinelle check report:

  ./tools/testing/selftests/vm/split_huge_page_test.c:344:36-42:
  ERROR: application of sizeof to pointer

Use "strlen" to fix it.

Link: https://lkml.kernel.org/r/20211012030116.184027-1-davidcomponentone@gmail.com
Signed-off-by: David Yang <davidcomponentone@gmail.com>
Reported-by: Zeal Robot <zealci@zte.com.cn>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomm/damon/core-test: fix wrong expectations for 'damon_split_regions_of()'
SeongJae Park [Thu, 28 Oct 2021 21:36:33 +0000 (14:36 -0700)]
mm/damon/core-test: fix wrong expectations for 'damon_split_regions_of()'

Kunit test cases for 'damon_split_regions_of()' expects the number of
regions after calling the function will be same to their request
('nr_sub').  However, the requested number is just an upper-limit,
because the function randomly decides the size of each sub-region.

This fixes the wrong expectation.

Link: https://lkml.kernel.org/r/20211028090628.14948-1-sj@kernel.org
Fixes: 17ccae8bb5c9 ("mm/damon: add kunit tests")
Signed-off-by: SeongJae Park <sj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomm: khugepaged: skip huge page collapse for special files
Yang Shi [Thu, 28 Oct 2021 21:36:30 +0000 (14:36 -0700)]
mm: khugepaged: skip huge page collapse for special files

The read-only THP for filesystems will collapse THP for files opened
readonly and mapped with VM_EXEC.  The intended usecase is to avoid TLB
misses for large text segments.  But it doesn't restrict the file types
so a THP could be collapsed for a non-regular file, for example, block
device, if it is opened readonly and mapped with EXEC permission.  This
may cause bugs, like [1] and [2].

This is definitely not the intended usecase, so just collapse THP for
regular files in order to close the attack surface.

[shy828301@gmail.com: fix vm_file check [3]]

Link: https://lore.kernel.org/lkml/CACkBjsYwLYLRmX8GpsDpMthagWOjWWrNxqY6ZLNQVr6yx+f5vA@mail.gmail.com/
Link: https://lore.kernel.org/linux-mm/000000000000c6a82505ce284e4c@google.com/
Link: https://lkml.kernel.org/r/CAHbLzkqTW9U3VvTu1Ki5v_cLRC9gHW+znBukg_ycergE0JWj-A@mail.gmail.com
Link: https://lkml.kernel.org/r/20211027195221.3825-1-shy828301@gmail.com
Fixes: 99cb0dbd47a1 ("mm,thp: add read-only THP support for (non-shmem) FS")
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reported-by: Hao Sun <sunhao.th@gmail.com>
Reported-by: syzbot+aae069be1de40fb11825@syzkaller.appspotmail.com
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: Andrea Righi <andrea.righi@canonical.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomm, thp: bail out early in collapse_file for writeback page
Rongwei Wang [Thu, 28 Oct 2021 21:36:27 +0000 (14:36 -0700)]
mm, thp: bail out early in collapse_file for writeback page

Currently collapse_file does not explicitly check PG_writeback, instead,
page_has_private and try_to_release_page are used to filter writeback
pages.  This does not work for xfs with blocksize equal to or larger
than pagesize, because in such case xfs has no page->private.

This makes collapse_file bail out early for writeback page.  Otherwise,
xfs end_page_writeback will panic as follows.

  page:fffffe00201bcc80 refcount:0 mapcount:0 mapping:ffff0003f88c86a8 index:0x0 pfn:0x84ef32
  aops:xfs_address_space_operations [xfs] ino:30000b7 dentry name:"libtest.so"
  flags: 0x57fffe0000008027(locked|referenced|uptodate|active|writeback)
  raw: 57fffe0000008027 ffff80001b48bc28 ffff80001b48bc28 ffff0003f88c86a8
  raw: 0000000000000000 0000000000000000 00000000ffffffff ffff0000c3e9a000
  page dumped because: VM_BUG_ON_PAGE(((unsigned int) page_ref_count(page) + 127u <= 127u))
  page->mem_cgroup:ffff0000c3e9a000
  ------------[ cut here ]------------
  kernel BUG at include/linux/mm.h:1212!
  Internal error: Oops - BUG: 0 [#1] SMP
  Modules linked in:
  BUG: Bad page state in process khugepaged  pfn:84ef32
   xfs(E)
  page:fffffe00201bcc80 refcount:0 mapcount:0 mapping:0 index:0x0 pfn:0x84ef32
   libcrc32c(E) rfkill(E) aes_ce_blk(E) crypto_simd(E) ...
  CPU: 25 PID: 0 Comm: swapper/25 Kdump: loaded Tainted: ...
  pstate: 60400005 (nZCv daif +PAN -UAO -TCO BTYPE=--)
  Call trace:
    end_page_writeback+0x1c0/0x214
    iomap_finish_page_writeback+0x13c/0x204
    iomap_finish_ioend+0xe8/0x19c
    iomap_writepage_end_bio+0x38/0x50
    bio_endio+0x168/0x1ec
    blk_update_request+0x278/0x3f0
    blk_mq_end_request+0x34/0x15c
    virtblk_request_done+0x38/0x74 [virtio_blk]
    blk_done_softirq+0xc4/0x110
    __do_softirq+0x128/0x38c
    __irq_exit_rcu+0x118/0x150
    irq_exit+0x1c/0x30
    __handle_domain_irq+0x8c/0xf0
    gic_handle_irq+0x84/0x108
    el1_irq+0xcc/0x180
    arch_cpu_idle+0x18/0x40
    default_idle_call+0x4c/0x1a0
    cpuidle_idle_call+0x168/0x1e0
    do_idle+0xb4/0x104
    cpu_startup_entry+0x30/0x9c
    secondary_start_kernel+0x104/0x180
  Code: d4210000 b0006161 910c8021 94013f4d (d4210000)
  ---[ end trace 4a88c6a074082f8c ]---
  Kernel panic - not syncing: Oops - BUG: Fatal exception in interrupt

Link: https://lkml.kernel.org/r/20211022023052.33114-1-rongwei.wang@linux.alibaba.com
Fixes: 99cb0dbd47a1 ("mm,thp: add read-only THP support for (non-shmem) FS")
Signed-off-by: Rongwei Wang <rongwei.wang@linux.alibaba.com>
Signed-off-by: Xu Yu <xuyu@linux.alibaba.com>
Suggested-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Yang Shi <shy828301@gmail.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Song Liu <song@kernel.org>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomm/vmalloc: fix numa spreading for large hash tables
Chen Wandun [Thu, 28 Oct 2021 21:36:24 +0000 (14:36 -0700)]
mm/vmalloc: fix numa spreading for large hash tables

Eric Dumazet reported a strange numa spreading info in [1], and found
commit 121e6f3258fe ("mm/vmalloc: hugepage vmalloc mappings") introduced
this issue [2].

Dig into the difference before and after this patch, page allocation has
some difference:

before:
  alloc_large_system_hash
    __vmalloc
      __vmalloc_node(..., NUMA_NO_NODE, ...)
        __vmalloc_node_range
          __vmalloc_area_node
            alloc_page /* because NUMA_NO_NODE, so choose alloc_page branch */
              alloc_pages_current
                alloc_page_interleave /* can be proved by print policy mode */

after:
  alloc_large_system_hash
    __vmalloc
      __vmalloc_node(..., NUMA_NO_NODE, ...)
        __vmalloc_node_range
          __vmalloc_area_node
            alloc_pages_node /* choose nid by nuam_mem_id() */
              __alloc_pages_node(nid, ....)

So after commit 121e6f3258fe ("mm/vmalloc: hugepage vmalloc mappings"),
it will allocate memory in current node instead of interleaving allocate
memory.

Link: https://lore.kernel.org/linux-mm/CANn89iL6AAyWhfxdHO+jaT075iOa3XcYn9k6JJc7JR2XYn6k_Q@mail.gmail.com/
Link: https://lore.kernel.org/linux-mm/CANn89iLofTR=AK-QOZY87RdUZENCZUT4O6a0hvhu3_EwRMerOg@mail.gmail.com/
Link: https://lkml.kernel.org/r/20211021080744.874701-2-chenwandun@huawei.com
Fixes: 121e6f3258fe ("mm/vmalloc: hugepage vmalloc mappings")
Signed-off-by: Chen Wandun <chenwandun@huawei.com>
Reported-by: Eric Dumazet <edumazet@google.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Hanjun Guo <guohanjun@huawei.com>
Cc: Uladzislau Rezki <urezki@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomm/secretmem: avoid letting secretmem_users drop to zero
Kees Cook [Thu, 28 Oct 2021 21:36:21 +0000 (14:36 -0700)]
mm/secretmem: avoid letting secretmem_users drop to zero

Quoting Dmitry:
 "refcount_inc() needs to be done before fd_install(). After
  fd_install() finishes, the fd can be used by userspace and
  we can have secret data in memory before the refcount_inc().

  A straightforward misuse where a user will predict the returned
  fd in another thread before the syscall returns and will use it
  to store secret data is somewhat dubious because such a user just
  shoots themself in the foot.

  But a more interesting misuse would be to close the predicted fd
  and decrement the refcount before the corresponding refcount_inc,
  this way one can briefly drop the refcount to zero while there are
  other users of secretmem."

Move fd_install() after refcount_inc().

Link: https://lkml.kernel.org/r/20211021154046.880251-1-keescook@chromium.org
Link: https://lore.kernel.org/lkml/CACT4Y+b1sW6-Hkn8HQYw_SsT7X3tp-CJNh2ci0wG3ZnQz9jjig@mail.gmail.com
Fixes: 9a436f8ff631 ("PM: hibernate: disable when there are active secretmem users")
Signed-off-by: Kees Cook <keescook@chromium.org>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Jordy Zomer <jordy@pwning.systems>
Cc: Mike Rapoport <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agoocfs2: fix race between searching chunks and release journal_head from buffer_head
Gautham Ananthakrishna [Thu, 28 Oct 2021 21:36:17 +0000 (14:36 -0700)]
ocfs2: fix race between searching chunks and release journal_head from buffer_head

Encountered a race between ocfs2_test_bg_bit_allocatable() and
jbd2_journal_put_journal_head() resulting in the below vmcore.

  PID: 106879  TASK: ffff880244ba9c00  CPU: 2   COMMAND: "loop3"
  Call trace:
    panic
    oops_end
    no_context
    __bad_area_nosemaphore
    bad_area_nosemaphore
    __do_page_fault
    do_page_fault
    page_fault
      [exception RIP: ocfs2_block_group_find_clear_bits+316]
    ocfs2_block_group_find_clear_bits [ocfs2]
    ocfs2_cluster_group_search [ocfs2]
    ocfs2_search_chain [ocfs2]
    ocfs2_claim_suballoc_bits [ocfs2]
    __ocfs2_claim_clusters [ocfs2]
    ocfs2_claim_clusters [ocfs2]
    ocfs2_local_alloc_slide_window [ocfs2]
    ocfs2_reserve_local_alloc_bits [ocfs2]
    ocfs2_reserve_clusters_with_limit [ocfs2]
    ocfs2_reserve_clusters [ocfs2]
    ocfs2_lock_refcount_allocators [ocfs2]
    ocfs2_make_clusters_writable [ocfs2]
    ocfs2_replace_cow [ocfs2]
    ocfs2_refcount_cow [ocfs2]
    ocfs2_file_write_iter [ocfs2]
    lo_rw_aio
    loop_queue_work
    kthread_worker_fn
    kthread
    ret_from_fork

When ocfs2_test_bg_bit_allocatable() called bh2jh(bg_bh), the
bg_bh->b_private NULL as jbd2_journal_put_journal_head() raced and
released the jounal head from the buffer head.  Needed to take bit lock
for the bit 'BH_JournalHead' to fix this race.

Link: https://lkml.kernel.org/r/1634820718-6043-1-git-send-email-gautham.ananthakrishna@oracle.com
Signed-off-by: Gautham Ananthakrishna <gautham.ananthakrishna@oracle.com>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: <rajesh.sivaramasubramaniom@oracle.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Jun Piao <piaojun@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>
2 years agomm/oom_kill.c: prevent a race between process_mrelease and exit_mmap
Suren Baghdasaryan [Thu, 28 Oct 2021 21:36:14 +0000 (14:36 -0700)]
mm/oom_kill.c: prevent a race between process_mrelease and exit_mmap

Race between process_mrelease and exit_mmap, where free_pgtables is
called while __oom_reap_task_mm is in progress, leads to kernel crash
during pte_offset_map_lock call.  oom-reaper avoids this race by setting
MMF_OOM_VICTIM flag and causing exit_mmap to take and release
mmap_write_lock, blocking it until oom-reaper releases mmap_read_lock.

Reusing MMF_OOM_VICTIM for process_mrelease would be the simplest way to
fix this race, however that would be considered a hack.  Fix this race
by elevating mm->mm_users and preventing exit_mmap from executing until
process_mrelease is finished.  Patch slightly refactors the code to
adapt for a possible mmget_not_zero failure.

This fix has considerable negative impact on process_mrelease
performance and will likely need later optimization.

Link: https://lkml.kernel.org/r/20211022014658.263508-1-surenb@google.com
Fixes: 884a7e5964e0 ("mm: introduce process_mrelease system call")
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Roman Gushchin <guro@fb.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Christian Brauner <christian@brauner.io>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jann Horn <jannh@google.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Christian Brauner <christian.brauner@ubuntu.com>
Cc: Florian Weimer <fweimer@redhat.com>
Cc: Jan Engelhardt <jengelh@inai.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomm: filemap: check if THP has hwpoisoned subpage for PMD page fault
Yang Shi [Thu, 28 Oct 2021 21:36:11 +0000 (14:36 -0700)]
mm: filemap: check if THP has hwpoisoned subpage for PMD page fault

When handling shmem page fault the THP with corrupted subpage could be
PMD mapped if certain conditions are satisfied.  But kernel is supposed
to send SIGBUS when trying to map hwpoisoned page.

There are two paths which may do PMD map: fault around and regular
fault.

Before commit f9ce0be71d1f ("mm: Cleanup faultaround and finish_fault()
codepaths") the thing was even worse in fault around path.  The THP
could be PMD mapped as long as the VMA fits regardless what subpage is
accessed and corrupted.  After this commit as long as head page is not
corrupted the THP could be PMD mapped.

In the regular fault path the THP could be PMD mapped as long as the
corrupted page is not accessed and the VMA fits.

This loophole could be fixed by iterating every subpage to check if any
of them is hwpoisoned or not, but it is somewhat costly in page fault
path.

So introduce a new page flag called HasHWPoisoned on the first tail
page.  It indicates the THP has hwpoisoned subpage(s).  It is set if any
subpage of THP is found hwpoisoned by memory failure and after the
refcount is bumped successfully, then cleared when the THP is freed or
split.

The soft offline path doesn't need this since soft offline handler just
marks a subpage hwpoisoned when the subpage is migrated successfully.
But shmem THP didn't get split then migrated at all.

Link: https://lkml.kernel.org/r/20211020210755.23964-3-shy828301@gmail.com
Fixes: 800d8c63b2e9 ("shmem: add huge pages support")
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Suggested-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Peter Xu <peterx@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomm: hwpoison: remove the unnecessary THP check
Yang Shi [Thu, 28 Oct 2021 21:36:07 +0000 (14:36 -0700)]
mm: hwpoison: remove the unnecessary THP check

When handling THP hwpoison checked if the THP is in allocation or free
stage since hwpoison may mistreat it as hugetlb page.  After commit
415c64c1453a ("mm/memory-failure: split thp earlier in memory error
handling") the problem has been fixed, so this check is no longer
needed.  Remove it.  The side effect of the removal is hwpoison may
report unsplit THP instead of unknown error for shmem THP.  It seems not
like a big deal.

The following patch "mm: filemap: check if THP has hwpoisoned subpage
for PMD page fault" depends on this, which fixes shmem THP with
hwpoisoned subpage(s) are mapped PMD wrongly.  So this patch needs to be
backported to -stable as well.

Link: https://lkml.kernel.org/r/20211020210755.23964-2-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Suggested-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Peter Xu <peterx@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agomemcg: page_alloc: skip bulk allocator for __GFP_ACCOUNT
Shakeel Butt [Thu, 28 Oct 2021 21:36:04 +0000 (14:36 -0700)]
memcg: page_alloc: skip bulk allocator for __GFP_ACCOUNT

Commit 5c1f4e690eec ("mm/vmalloc: switch to bulk allocator in
__vmalloc_area_node()") switched to bulk page allocator for order 0
allocation backing vmalloc.  However bulk page allocator does not
support __GFP_ACCOUNT allocations and there are several users of
kvmalloc(__GFP_ACCOUNT).

For now make __GFP_ACCOUNT allocations bypass bulk page allocator.  In
future if there is workload that can be significantly improved with the
bulk page allocator with __GFP_ACCCOUNT support, we can revisit the
decision.

Link: https://lkml.kernel.org/r/20211014151607.2171970-1-shakeelb@google.com
Fixes: 5c1f4e690eec ("mm/vmalloc: switch to bulk allocator in __vmalloc_area_node()")
Signed-off-by: Shakeel Butt <shakeelb@google.com>
Reported-by: Vasily Averin <vvs@virtuozzo.com>
Tested-by: Vasily Averin <vvs@virtuozzo.com>
Acked-by: David Hildenbrand <david@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Roman Gushchin <guro@fb.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agoMerge tag 'libnvdimm-fixes-5.15-rc8' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 28 Oct 2021 23:50:25 +0000 (16:50 -0700)]
Merge tag 'libnvdimm-fixes-5.15-rc8' of git://git./linux/kernel/git/nvdimm/nvdimm

Pull libnvdimm fix from Dan Williams:

 - Fix a regression introduced in v5.15-rc6 that caused nvdimm namespace
   shutdown to hang due to reworks in the block layer q_usage_count.

* tag 'libnvdimm-fixes-5.15-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
  nvdimm/pmem: stop using q_usage_count as external pgmap refcount

2 years agommc: tmio: reenable card irqs after the reset callback
Wolfram Sang [Thu, 28 Oct 2021 19:51:49 +0000 (21:51 +0200)]
mmc: tmio: reenable card irqs after the reset callback

The reset callback may clear the internal card detect interrupts, so
make sure to reenable them if needed.

Fixes: b4d86f37eacb ("mmc: renesas_sdhi: do hard reset if possible")
Reported-by: Biju Das <biju.das.jz@bp.renesas.com>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20211028195149.8003-1-wsa+renesas@sang-engineering.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
2 years agoMerge tag 'drm-fixes-2021-10-29' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Thu, 28 Oct 2021 19:17:01 +0000 (12:17 -0700)]
Merge tag 'drm-fixes-2021-10-29' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Quiet but not too quiet, I blame Halloween.

  The first set of amdgpu fixes missed last week, hence why this has a
  few more of them, it's mostly display fixes for new GPUs and some
  debugfs OOB stuff.

  The i915 patches have one to remove a tracepoint possible issue before
  it's a real problem, the others around cflush and display are cc'ed to
  stable as well.

  Otherwise it's just a few misc fixes.

  Summary:

  MAINTAINERS:
   - Fix the path pattern

  ttm:
   - Fix fence leak in ttm_transfered_destroy.

  core:
   - Add GPD Win3 rotation quirk

  i915:
   - Remove unconditional clflushes
   - Fix oops on boot due to sync state on disabled DP encoders
   - Revert backend specific data added to tracepoints
   - Remove useless and incorrect memory frequence calculation

  panel:
   - Add quirk for Aya Neo 2021

  seltest:
   - Reset property count for each drm damage selftest so full run will
     work correctly.

  amdgpu:
   - Fix two potential out of bounds writes in debugfs
   - Fix revision handling for Yellow Carp
   - Display fixes for Yellow Carp
   - Display fixes for DCN 3.1"

* tag 'drm-fixes-2021-10-29' of git://anongit.freedesktop.org/drm/drm: (21 commits)
  MAINTAINERS: dri-devel is for all of drivers/gpu
  drm/i915: Revert 'guc_id' from i915_request tracepoint
  drm/amd/display: Fix deadlock when falling back to v2 from v3
  drm/amd/display: Fallback to clocks which meet requested voltage on DCN31
  drm/amdgpu: Fix even more out of bound writes from debugfs
  drm: panel-orientation-quirks: Add quirk for GPD Win3
  drm/i915/dp: Skip the HW readout of DPCD on disabled encoders
  drm/i915: Catch yet another unconditioal clflush
  drm/i915: Convert unconditional clflush to drm_clflush_virt_range()
  drm/i915/selftests: Properly reset mock object propers for each test
  drm: panel-orientation-quirks: Add quirk for Aya Neo 2021
  drm/ttm: fix memleak in ttm_transfered_destroy
  drm/amdgpu: support B0&B1 external revision id for yellow carp
  drm/amd/display: Moved dccg init to after bios golden init
  drm/amd/display: Increase watermark latencies for DCN3.1
  drm/amd/display: increase Z9 latency to workaround underflow in Z9
  drm/amd/display: Require immediate flip support for DCN3.1 planes
  drm/amd/display: Fix prefetch bandwidth calculation for DCN3.1
  drm/amd/display: Limit display scaling to up to true 4k for DCN 3.1
  drm/amdgpu: fix out of bounds write
  ...

2 years agoMAINTAINERS: dri-devel is for all of drivers/gpu
Daniel Vetter [Thu, 28 Oct 2021 17:08:57 +0000 (19:08 +0200)]
MAINTAINERS: dri-devel is for all of drivers/gpu

Somehow we only have a list of subdirectories, which apparently made
it harder for folks to find the gpu maintainers. Fix that.

References: https://lore.kernel.org/dri-devel/YXrAAZlxxStNFG%2FK@phenom.ffwll.local/
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Steven Rostedt <rostedt@goodmis.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211028170857.4029606-1-daniel.vetter@ffwll.ch
2 years agoMerge tag 'drm-intel-fixes-2021-10-28' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Thu, 28 Oct 2021 18:46:14 +0000 (04:46 +1000)]
Merge tag 'drm-intel-fixes-2021-10-28' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

drm/i915 fixes for v5.15 final:
- Remove unconditional clflushes
- Fix oops on boot due to sync state on disabled DP encoders
- Revert backend specific data added to tracepoints
- Remove useless and incorrect memory frequence calculation

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/8735olh27y.fsf@intel.com
2 years agoMerge tag 'net-5.15-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 28 Oct 2021 17:17:31 +0000 (10:17 -0700)]
Merge tag 'net-5.15-rc8' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Including fixes from WiFi (mac80211), and BPF.

  Current release - regressions:

   - skb_expand_head: adjust skb->truesize to fix socket memory
     accounting

   - mptcp: fix corrupt receiver key in MPC + data + checksum

  Previous releases - regressions:

   - multicast: calculate csum of looped-back and forwarded packets

   - cgroup: fix memory leak caused by missing cgroup_bpf_offline

   - cfg80211: fix management registrations locking, prevent list
     corruption

   - cfg80211: correct false positive in bridge/4addr mode check

   - tcp_bpf: fix race in the tcp_bpf_send_verdict resulting in reusing
     previous verdict

  Previous releases - always broken:

   - sctp: enhancements for the verification tag, prevent attackers from
     killing SCTP sessions

   - tipc: fix size validations for the MSG_CRYPTO type

   - mac80211: mesh: fix HE operation element length check, prevent out
     of bound access

   - tls: fix sign of socket errors, prevent positive error codes being
     reported from read()/write()

   - cfg80211: scan: extend RCU protection in
     cfg80211_add_nontrans_list()

   - implement ->sock_is_readable() for UDP and AF_UNIX, fix poll() for
     sockets in a BPF sockmap

   - bpf: fix potential race in tail call compatibility check resulting
     in two operations which would make the map incompatible succeeding

   - bpf: prevent increasing bpf_jit_limit above max

   - bpf: fix error usage of map_fd and fdget() in generic batch update

   - phy: ethtool: lock the phy for consistency of results

   - prevent infinite while loop in skb_tx_hash() when Tx races with
     driver reconfiguring the queue <> traffic class mapping

   - usbnet: fixes for bad HW conjured by syzbot

   - xen: stop tx queues during live migration, prevent UAF

   - net-sysfs: initialize uid and gid before calling
     net_ns_get_ownership

   - mlxsw: prevent Rx stalls under memory pressure"

* tag 'net-5.15-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (67 commits)
  Revert "net: hns3: fix pause config problem after autoneg disabled"
  mptcp: fix corrupt receiver key in MPC + data + checksum
  riscv, bpf: Fix potential NULL dereference
  octeontx2-af: Fix possible null pointer dereference.
  octeontx2-af: Display all enabled PF VF rsrc_alloc entries.
  octeontx2-af: Check whether ipolicers exists
  net: ethernet: microchip: lan743x: Fix skb allocation failure
  net/tls: Fix flipped sign in async_wait.err assignment
  net/tls: Fix flipped sign in tls_err_abort() calls
  net/smc: Correct spelling mistake to TCPF_SYN_RECV
  net/smc: Fix smc_link->llc_testlink_time overflow
  nfp: bpf: relax prog rejection for mtu check through max_pkt_offset
  vmxnet3: do not stop tx queues after netif_device_detach()
  r8169: Add device 10ec:8162 to driver r8169
  ptp: Document the PTP_CLK_MAGIC ioctl number
  usbnet: fix error return code in usbnet_probe()
  net: hns3: adjust string spaces of some parameters of tx bd info in debugfs
  net: hns3: expand buffer len for some debugfs command
  net: hns3: add more string spaces for dumping packets number of queue info in debugfs
  net: hns3: fix data endian problem of some functions of debugfs
  ...

2 years agoMerge tag 'spi-fix-v5.15-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/brooni...
Linus Torvalds [Thu, 28 Oct 2021 17:04:39 +0000 (10:04 -0700)]
Merge tag 'spi-fix-v5.15-rc7' of git://git./linux/kernel/git/broonie/spi

Pull spi fixes from Mark Brown:
 "A couple of final driver specific fixes for v5.15, one fixing
  potential ID collisions between two instances of the Altera driver and
  one making Microwire full duplex mode actually work on pl022"

* tag 'spi-fix-v5.15-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: spl022: fix Microwire full duplex mode
  spi: altera: Change to dynamic allocation of spi id

2 years agoMerge tag 'regmap-fix-v5.15-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 28 Oct 2021 17:00:58 +0000 (10:00 -0700)]
Merge tag 'regmap-fix-v5.15-rc7' of git://git./linux/kernel/git/broonie/regmap

Pull regmap fix from Mark Brown:
 "This fixes a potential double free when handling an out of memory
  error inserting a node into an rbtree regcache"

* tag 'regmap-fix-v5.15-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
  regmap: Fix possible double-free in regcache_rbtree_exit()

2 years agoMerge tag 'linux-watchdog-5.15-rc7' of git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Thu, 28 Oct 2021 16:55:25 +0000 (09:55 -0700)]
Merge tag 'linux-watchdog-5.15-rc7' of git://linux-watchdog.org/linux-watchdog

Pull watchdog fixes from Wim Van Sebroeck:
 "I overlooked Guenters request to sent this upstream earlier, so it's a
  bit late in the release cycle.

  This contains:

   - Revert "watchdog: iTCO_wdt: Account for rebooting on second
     timeout"

   - sbsa: only use 32-bit accessors

   - sbsa: drop unneeded MODULE_ALIAS

   - ixp4xx_wdt: Fix address space warning

   - Fix OMAP watchdog early handling"

* tag 'linux-watchdog-5.15-rc7' of git://www.linux-watchdog.org/linux-watchdog:
  watchdog: Fix OMAP watchdog early handling
  watchdog: ixp4xx_wdt: Fix address space warning
  watchdog: sbsa: drop unneeded MODULE_ALIAS
  watchdog: sbsa: only use 32-bit accessors
  Revert "watchdog: iTCO_wdt: Account for rebooting on second timeout"

2 years agoMerge tag 'trace-v5.15-rc6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rosted...
Linus Torvalds [Thu, 28 Oct 2021 16:50:56 +0000 (09:50 -0700)]
Merge tag 'trace-v5.15-rc6-2' of git://git./linux/kernel/git/rostedt/linux-trace

Pull tracing fix from Steven Rostedt:
 "Do not WARN when attaching event probe to non-existent event

  If the user tries to attach an event probe (eprobe) to an event that
  does not exist, it will trigger a warning. There's an error check that
  only expects memory issues otherwise it is considered a bug. But
  changes in the code to move around the locking made it that it can
  error out if the user attempts to attach to an event that does not
  exist, returning an -ENODEV. As this path can be caused by user space
  putting in a bad value, do not trigger a WARN"

* tag 'trace-v5.15-rc6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
  tracing: Do not warn when connecting eprobe to non existing event

2 years agoRevert "net: hns3: fix pause config problem after autoneg disabled"
Guangbin Huang [Thu, 28 Oct 2021 14:06:24 +0000 (22:06 +0800)]
Revert "net: hns3: fix pause config problem after autoneg disabled"

This reverts commit 3bda2e5df476417b6d08967e2d84234a59d57b1c.

According to discussion with Andrew as follow:
https://lore.kernel.org/netdev/09eda9fe-196b-006b-6f01-f54e75715961@huawei.com/

HNS3 driver needs to separate pause autoneg from general autoneg, so revert
this incorrect patch.

Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Link: https://lore.kernel.org/r/20211028140624.53149-1-huangguangbin2@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agomptcp: fix corrupt receiver key in MPC + data + checksum
Davide Caratti [Wed, 27 Oct 2021 20:38:55 +0000 (13:38 -0700)]
mptcp: fix corrupt receiver key in MPC + data + checksum

using packetdrill it's possible to observe that the receiver key contains
random values when clients transmit MP_CAPABLE with data and checksum (as
specified in RFC8684 §3.1). Fix the layout of mptcp_out_options, to avoid
using the skb extension copy when writing the MP_CAPABLE sub-option.

Fixes: d7b269083786 ("mptcp: shrink mptcp_out_options struct")
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/233
Reported-by: Poorva Sonparote <psonparo@redhat.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Link: https://lore.kernel.org/r/20211027203855.264600-1-mathew.j.martineau@linux.intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoriscv, bpf: Fix potential NULL dereference
Björn Töpel [Thu, 28 Oct 2021 12:51:15 +0000 (14:51 +0200)]
riscv, bpf: Fix potential NULL dereference

The bpf_jit_binary_free() function requires a non-NULL argument. When
the RISC-V BPF JIT fails to converge in NR_JIT_ITERATIONS steps,
jit_data->header will be NULL, which triggers a NULL
dereference. Avoid this by checking the argument, prior calling the
function.

Fixes: ca6cb5447cec ("riscv, bpf: Factor common RISC-V JIT code")
Signed-off-by: Björn Töpel <bjorn@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/r/20211028125115.514587-1-bjorn@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoKVM: x86: Take srcu lock in post_kvm_run_save()
David Woodhouse [Tue, 26 Oct 2021 03:12:38 +0000 (04:12 +0100)]
KVM: x86: Take srcu lock in post_kvm_run_save()

The Xen interrupt injection for event channels relies on accessing the
guest's vcpu_info structure in __kvm_xen_has_interrupt(), through a
gfn_to_hva_cache.

This requires the srcu lock to be held, which is mostly the case except
for this code path:

[   11.822877] WARNING: suspicious RCU usage
[   11.822965] -----------------------------
[   11.823013] include/linux/kvm_host.h:664 suspicious rcu_dereference_check() usage!
[   11.823131]
[   11.823131] other info that might help us debug this:
[   11.823131]
[   11.823196]
[   11.823196] rcu_scheduler_active = 2, debug_locks = 1
[   11.823253] 1 lock held by dom:0/90:
[   11.823292]  #0: ffff998956ec8118 (&vcpu->mutex){+.+.}, at: kvm_vcpu_ioctl+0x85/0x680
[   11.823379]
[   11.823379] stack backtrace:
[   11.823428] CPU: 2 PID: 90 Comm: dom:0 Kdump: loaded Not tainted 5.4.34+ #5
[   11.823496] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.12.1-0-ga5cab58e9a3f-prebuilt.qemu.org 04/01/2014
[   11.823612] Call Trace:
[   11.823645]  dump_stack+0x7a/0xa5
[   11.823681]  lockdep_rcu_suspicious+0xc5/0x100
[   11.823726]  __kvm_xen_has_interrupt+0x179/0x190
[   11.823773]  kvm_cpu_has_extint+0x6d/0x90
[   11.823813]  kvm_cpu_accept_dm_intr+0xd/0x40
[   11.823853]  kvm_vcpu_ready_for_interrupt_injection+0x20/0x30
              < post_kvm_run_save() inlined here >
[   11.823906]  kvm_arch_vcpu_ioctl_run+0x135/0x6a0
[   11.823947]  kvm_vcpu_ioctl+0x263/0x680

Fixes: 40da8ccd724f ("KVM: x86/xen: Add event channel interrupt vector upcall")
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Cc: stable@vger.kernel.org
Message-Id: <606aaaf29fca3850a63aa4499826104e77a72346.camel@infradead.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2 years agoMerge tag 'nvme-5.15-2021-10-28' of git://git.infradead.org/nvme into block-5.15
Jens Axboe [Thu, 28 Oct 2021 14:34:01 +0000 (08:34 -0600)]
Merge tag 'nvme-5.15-2021-10-28' of git://git.infradead.org/nvme into block-5.15

Pull NVMe fixes from Christoph:

"nvme fixe for Linux 5.15

 - fix nvmet-tcp header digest verification (Amit Engel)
 - fix a memory leak in nvmet-tcp when releasing a queue
   (Maurizio Lombardi)
 - fix nvme-tcp H2CData PDU send accounting again (Sagi Grimberg)
 - fix digest pointer calculation in nvme-tcp and nvmet-tcp
   (Varun Prakash)
 - fix possible nvme-tcp req->offset corruption (Varun Prakash)"

* tag 'nvme-5.15-2021-10-28' of git://git.infradead.org/nvme:
  nvmet-tcp: fix header digest verification
  nvmet-tcp: fix data digest pointer calculation
  nvme-tcp: fix data digest pointer calculation
  nvme-tcp: fix possible req->offset corruption
  nvme-tcp: fix H2CData PDU send accounting (again)
  nvmet-tcp: fix a memory leak when releasing a queue

2 years agoMerge branch 'octeontx2-debugfs-fixes'
David S. Miller [Thu, 28 Oct 2021 13:47:37 +0000 (14:47 +0100)]
Merge branch 'octeontx2-debugfs-fixes'

Rakesh Babu Saladi says:

====================
RVU Debugfs fix updates.

The following patch series consists of the patch fixes done over
rvu_debugfs.c and rvu_nix.c files.

Patch 1: Check and return if ipolicers do not exists.
Patch 2: Fix rsrc_alloc to print all enabled PF/VF entries with list of LFs
allocated for each functional block.
Patch 3: Fix possible null pointer dereference.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoocteontx2-af: Fix possible null pointer dereference.
Rakesh Babu Saladi [Wed, 27 Oct 2021 17:32:34 +0000 (23:02 +0530)]
octeontx2-af: Fix possible null pointer dereference.

This patch fixes possible null pointer dereference in files
"rvu_debugfs.c" and "rvu_nix.c"

Fixes: 8756828a8148 ("octeontx2-af: Add NPA aura and pool contexts to debugfs")
Fixes: 9a946def264d ("octeontx2-af: Modify nix_vtag_cfg mailbox to support TX VTAG entries")
Signed-off-by: Rakesh Babu Saladi <rsaladi2@marvell.com>
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoocteontx2-af: Display all enabled PF VF rsrc_alloc entries.
Rakesh Babu [Wed, 27 Oct 2021 17:32:33 +0000 (23:02 +0530)]
octeontx2-af: Display all enabled PF VF rsrc_alloc entries.

Currently, we are using a fixed buffer size of length 2048 to display
rsrc_alloc output. As a result a maximum of 2048 characters of
rsrc_alloc output is displayed, which may lead sometimes to display only
partial output. This patch fixes this dependency on max limit of buffer
size and displays all PF VF entries.

Each column of the debugfs entry "rsrc_alloc" uses a fixed width of 12
characters to print the list of LFs of each block for a PF/VF. If the
length of list of LFs of a block exceeds this fixed width then the list
gets truncated and displays only a part of the list. This patch fixes
this by using the maximum possible length of list of LFs among all
blocks of all PFs and VFs entries as the width size.

Fixes: f7884097141b ("octeontx2-af: Formatting debugfs entry rsrc_alloc.")
Fixes: 23205e6d06d4 ("octeontx2-af: Dump current resource provisioning status")
Signed-off-by: Rakesh Babu <rsaladi2@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <Sunil.Goutham@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoocteontx2-af: Check whether ipolicers exists
Subbaraya Sundeep [Wed, 27 Oct 2021 17:32:32 +0000 (23:02 +0530)]
octeontx2-af: Check whether ipolicers exists

While displaying ingress policers information in
debugfs check whether ingress policers exist in
the hardware or not because some platforms(CN9XXX)
do not have this feature.

Fixes: e7d8971763f3 ("octeontx2-af: cn10k: Debugfs support for bandwidth")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Rakesh Babu <rsaladi2@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: ethernet: microchip: lan743x: Fix skb allocation failure
Yuiko Oshino [Wed, 27 Oct 2021 18:23:02 +0000 (14:23 -0400)]
net: ethernet: microchip: lan743x: Fix skb allocation failure

The driver allocates skb during ndo_open with GFP_ATOMIC which has high chance of failure when there are multiple instances.
GFP_KERNEL is enough while open and use GFP_ATOMIC only from interrupt context.

Fixes: 23f0703c125b ("lan743x: Add main source files for new lan743x driver")
Signed-off-by: Yuiko Oshino <yuiko.oshino@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/tls: Fix flipped sign in async_wait.err assignment
Daniel Jordan [Wed, 27 Oct 2021 21:59:21 +0000 (17:59 -0400)]
net/tls: Fix flipped sign in async_wait.err assignment

sk->sk_err contains a positive number, yet async_wait.err wants the
opposite.  Fix the missed sign flip, which Jakub caught by inspection.

Fixes: a42055e8d2c3 ("net/tls: Add support for async encryption of records for performance")
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/tls: Fix flipped sign in tls_err_abort() calls
Daniel Jordan [Wed, 27 Oct 2021 21:59:20 +0000 (17:59 -0400)]
net/tls: Fix flipped sign in tls_err_abort() calls

sk->sk_err appears to expect a positive value, a convention that ktls
doesn't always follow and that leads to memory corruption in other code.
For instance,

    [kworker]
    tls_encrypt_done(..., err=<negative error from crypto request>)
      tls_err_abort(.., err)
        sk->sk_err = err;

    [task]
    splice_from_pipe_feed
      ...
        tls_sw_do_sendpage
          if (sk->sk_err) {
            ret = -sk->sk_err;  // ret is positive

    splice_from_pipe_feed (continued)
      ret = actor(...)  // ret is still positive and interpreted as bytes
                        // written, resulting in underflow of buf->len and
                        // sd->len, leading to huge buf->offset and bogus
                        // addresses computed in later calls to actor()

Fix all tls_err_abort() callers to pass a negative error code
consistently and centralize the error-prone sign flip there, throwing in
a warning to catch future misuse and uninlining the function so it
really does only warn once.

Cc: stable@vger.kernel.org
Fixes: c46234ebb4d1e ("tls: RX path for ktls")
Reported-by: syzbot+b187b77c8474f9648fae@syzkaller.appspotmail.com
Signed-off-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'SMC-fixes'
David S. Miller [Thu, 28 Oct 2021 12:04:29 +0000 (13:04 +0100)]
Merge branch 'SMC-fixes'

Tony Lu says:

====================
Fixes for SMC

There are some fixes for SMC.

v1->v2:
- fix wrong email address.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Correct spelling mistake to TCPF_SYN_RECV
Wen Gu [Thu, 28 Oct 2021 07:13:47 +0000 (15:13 +0800)]
net/smc: Correct spelling mistake to TCPF_SYN_RECV

There should use TCPF_SYN_RECV instead of TCP_SYN_RECV.

Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Reviewed-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Fix smc_link->llc_testlink_time overflow
Tony Lu [Thu, 28 Oct 2021 07:13:45 +0000 (15:13 +0800)]
net/smc: Fix smc_link->llc_testlink_time overflow

The value of llc_testlink_time is set to the value stored in
net->ipv4.sysctl_tcp_keepalive_time when linkgroup init. The value of
sysctl_tcp_keepalive_time is already jiffies, so we don't need to
multiply by HZ, which would cause smc_link->llc_testlink_time overflow,
and test_link send flood.

Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
Reviewed-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Reviewed-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonfp: bpf: relax prog rejection for mtu check through max_pkt_offset
Yu Xiao [Thu, 28 Oct 2021 10:00:36 +0000 (12:00 +0200)]
nfp: bpf: relax prog rejection for mtu check through max_pkt_offset

MTU change is refused whenever the value of new MTU is bigger than
the max packet bytes that fits in NFP Cluster Target Memory (CTM).
However, an eBPF program doesn't always need to access the whole
packet data.

The maximum direct packet access (DPA) offset has always been
caculated by verifier and stored in the max_pkt_offset field of prog
aux data.

Signed-off-by: Yu Xiao <yu.xiao@corigine.com>
Reviewed-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Reviewed-by: Niklas Soderlund <niklas.soderlund@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agovmxnet3: do not stop tx queues after netif_device_detach()
Dongli Zhang [Tue, 26 Oct 2021 21:50:31 +0000 (14:50 -0700)]
vmxnet3: do not stop tx queues after netif_device_detach()

The netif_device_detach() conditionally stops all tx queues if the queues
are running. There is no need to call netif_tx_stop_all_queues() again.

Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agommc: mediatek: Move cqhci init behind ungate clock
Wenbin Mei [Thu, 28 Oct 2021 02:20:49 +0000 (10:20 +0800)]
mmc: mediatek: Move cqhci init behind ungate clock

We must enable clock before cqhci init, because crypto needs read
information from CQHCI registers, otherwise, it will hang in MediaTek mmc
host controller.

Signed-off-by: Wenbin Mei <wenbin.mei@mediatek.com>
Fixes: 88bd652b3c74 ("mmc: mediatek: command queue support")
Cc: stable@vger.kernel.org
Acked-by: Chaotian Jing <chaotian.jing@mediatek.com>
Link: https://lore.kernel.org/r/20211028022049.22129-1-wenbin.mei@mediatek.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>