platform/kernel/linux-starfive.git
9 months agonfc: nci: fix possible NULL pointer dereference in send_acknowledge()
Krzysztof Kozlowski [Fri, 13 Oct 2023 18:41:29 +0000 (20:41 +0200)]
nfc: nci: fix possible NULL pointer dereference in send_acknowledge()

commit 7937609cd387246aed994e81aa4fa951358fba41 upstream.

Handle memory allocation failure from nci_skb_alloc() (calling
alloc_skb()) to avoid possible NULL pointer dereference.

Reported-by: 黄思聪 <huangsicong@iie.ac.cn>
Fixes: 391d8a2da787 ("NFC: Add NCI over SPI receive")
Cc: <stable@vger.kernel.org>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/20231013184129.18738-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agobtrfs: fix stripe length calculation for non-zoned data chunk allocation
Zygo Blaxell [Sat, 7 Oct 2023 05:14:21 +0000 (01:14 -0400)]
btrfs: fix stripe length calculation for non-zoned data chunk allocation

commit 8a540e990d7da36813cb71a4a422712bfba448a4 upstream.

Commit f6fca3917b4d "btrfs: store chunk size in space-info struct"
broke data chunk allocations on non-zoned multi-device filesystems when
using default chunk_size.  Commit 5da431b71d4b "btrfs: fix the max chunk
size and stripe length calculation" partially fixed that, and this patch
completes the fix for that case.

After commit f6fca3917b4d and 5da431b71d4b, the sequence of events for
a data chunk allocation on a non-zoned filesystem is:

        1.  btrfs_create_chunk calls init_alloc_chunk_ctl, which copies
        space_info->chunk_size (default 10 GiB) to ctl->max_stripe_len
        unmodified.  Before f6fca3917b4d, ctl->max_stripe_len value was
        1 GiB for non-zoned data chunks and not configurable.

        2.  btrfs_create_chunk calls gather_device_info which consumes
        and produces more fields of chunk_ctl.

        3.  gather_device_info multiplies ctl->max_stripe_len by
        ctl->dev_stripes (which is 1 in all cases except dup)
        and calls find_free_dev_extent with that number as num_bytes.

        4.  find_free_dev_extent locates the first dev_extent hole on
        a device which is at least as large as num_bytes.  With default
        max_chunk_size from f6fca3917b4d, it finds the first hole which is
        longer than 10 GiB, or the largest hole if that hole is shorter
        than 10 GiB.  This is different from the pre-f6fca3917b4d
        behavior, where num_bytes is 1 GiB, and find_free_dev_extent
        may choose a different hole.

        5.  gather_device_info repeats step 4 with all devices to find
        the first or largest dev_extent hole that can be allocated on
        each device.

        6.  gather_device_info sorts the device list by the hole size
        on each device, using total unallocated space on each device to
        break ties, then returns to btrfs_create_chunk with the list.

        7.  btrfs_create_chunk calls decide_stripe_size_regular.

        8.  decide_stripe_size_regular finds the largest stripe_len that
        fits across the first nr_devs device dev_extent holes that were
        found by gather_device_info (and satisfies other constraints
        on stripe_len that are not relevant here).

        9.  decide_stripe_size_regular caps the length of the stripe it
        computed at 1 GiB.  This cap appeared in 5da431b71d4b to correct
        one of the other regressions introduced in f6fca3917b4d.

        10.  btrfs_create_chunk creates a new chunk with the above
        computed size and number of devices.

At step 4, gather_device_info() has found a location where stripe up to
10 GiB in length could be allocated on several devices, and selected
which devices should have a dev_extent allocated on them, but at step
9, only 1 GiB of the space that was found on each device can be used.
This mismatch causes new suboptimal chunk allocation cases that did not
occur in pre-f6fca3917b4d kernels.

Consider a filesystem using raid1 profile with 3 devices.  After some
balances, device 1 has 10x 1 GiB unallocated space, while devices 2
and 3 have 1x 10 GiB unallocated space, i.e. the same total amount of
space, but distributed across different numbers of dev_extent holes.
For visualization, let's ignore all the chunks that were allocated before
this point, and focus on the remaining holes:

        Device 1:  [_] [_] [_] [_] [_] [_] [_] [_] [_] [_] (10x 1 GiB unallocated)
        Device 2:  [__________] (10 GiB contig unallocated)
        Device 3:  [__________] (10 GiB contig unallocated)

Before f6fca3917b4d, the allocator would fill these optimally by
allocating chunks with dev_extents on devices 1 and 2 ([12]), 1 and 3
([13]), or 2 and 3 ([23]):

        [after 0 chunk allocations]
        Device 1:  [_] [_] [_] [_] [_] [_] [_] [_] [_] [_] (10 GiB)
        Device 2:  [__________] (10 GiB)
        Device 3:  [__________] (10 GiB)

        [after 1 chunk allocation]
        Device 1:  [12] [_] [_] [_] [_] [_] [_] [_] [_] [_]
        Device 2:  [12] [_________] (9 GiB)
        Device 3:  [__________] (10 GiB)

        [after 2 chunk allocations]
        Device 1:  [12] [13] [_] [_] [_] [_] [_] [_] [_] [_] (8 GiB)
        Device 2:  [12] [_________] (9 GiB)
        Device 3:  [13] [_________] (9 GiB)

        [after 3 chunk allocations]
        Device 1:  [12] [13] [12] [_] [_] [_] [_] [_] [_] [_] (7 GiB)
        Device 2:  [12] [12] [________] (8 GiB)
        Device 3:  [13] [_________] (9 GiB)

        [...]

        [after 12 chunk allocations]
        Device 1:  [12] [13] [12] [13] [12] [13] [12] [13] [_] [_] (2 GiB)
        Device 2:  [12] [12] [23] [23] [12] [12] [23] [23] [__] (2 GiB)
        Device 3:  [13] [13] [23] [23] [13] [23] [13] [23] [__] (2 GiB)

        [after 13 chunk allocations]
        Device 1:  [12] [13] [12] [13] [12] [13] [12] [13] [12] [_] (1 GiB)
        Device 2:  [12] [12] [23] [23] [12] [12] [23] [23] [12] [_] (1 GiB)
        Device 3:  [13] [13] [23] [23] [13] [23] [13] [23] [__] (2 GiB)

        [after 14 chunk allocations]
        Device 1:  [12] [13] [12] [13] [12] [13] [12] [13] [12] [13] (full)
        Device 2:  [12] [12] [23] [23] [12] [12] [23] [23] [12] [_] (1 GiB)
        Device 3:  [13] [13] [23] [23] [13] [23] [13] [23] [13] [_] (1 GiB)

        [after 15 chunk allocations]
        Device 1:  [12] [13] [12] [13] [12] [13] [12] [13] [12] [13] (full)
        Device 2:  [12] [12] [23] [23] [12] [12] [23] [23] [12] [23] (full)
        Device 3:  [13] [13] [23] [23] [13] [23] [13] [23] [13] [23] (full)

This allocates all of the space with no waste.  The sorting function used
by gather_device_info considers free space holes above 1 GiB in length
to be equal to 1 GiB, so once find_free_dev_extent locates a sufficiently
long hole on each device, all the holes appear equal in the sort, and the
comparison falls back to sorting devices by total free space.  This keeps
usable space on each device equal so they can all be filled completely.

After f6fca3917b4d, the allocator prefers the devices with larger holes
over the devices with more free space, so it makes bad allocation choices:

        [after 1 chunk allocation]
        Device 1:  [_] [_] [_] [_] [_] [_] [_] [_] [_] [_] (10 GiB)
        Device 2:  [23] [_________] (9 GiB)
        Device 3:  [23] [_________] (9 GiB)

        [after 2 chunk allocations]
        Device 1:  [_] [_] [_] [_] [_] [_] [_] [_] [_] [_] (10 GiB)
        Device 2:  [23] [23] [________] (8 GiB)
        Device 3:  [23] [23] [________] (8 GiB)

        [after 3 chunk allocations]
        Device 1:  [_] [_] [_] [_] [_] [_] [_] [_] [_] [_] (10 GiB)
        Device 2:  [23] [23] [23] [_______] (7 GiB)
        Device 3:  [23] [23] [23] [_______] (7 GiB)

        [...]

        [after 9 chunk allocations]
        Device 1:  [_] [_] [_] [_] [_] [_] [_] [_] [_] [_] (10 GiB)
        Device 2:  [23] [23] [23] [23] [23] [23] [23] [23] [23] [_] (1 GiB)
        Device 3:  [23] [23] [23] [23] [23] [23] [23] [23] [23] [_] (1 GiB)

        [after 10 chunk allocations]
        Device 1:  [12] [_] [_] [_] [_] [_] [_] [_] [_] [_] (9 GiB)
        Device 2:  [23] [23] [23] [23] [23] [23] [23] [23] [12] (full)
        Device 3:  [23] [23] [23] [23] [23] [23] [23] [23] [_] (1 GiB)

        [after 11 chunk allocations]
        Device 1:  [12] [13] [_] [_] [_] [_] [_] [_] [_] [_] (8 GiB)
        Device 2:  [23] [23] [23] [23] [23] [23] [23] [23] [12] (full)
        Device 3:  [23] [23] [23] [23] [23] [23] [23] [23] [13] (full)

No further allocations are possible, with 8 GiB wasted (4 GiB of data
space).  The sort in gather_device_info now considers free space in
holes longer than 1 GiB to be distinct, so it will prefer devices 2 and
3 over device 1 until all but 1 GiB is allocated on devices 2 and 3.
At that point, with only 1 GiB unallocated on every device, the largest
hole length on each device is equal at 1 GiB, so the sort finally moves
to ordering the devices with the most free space, but by this time it
is too late to make use of the free space on device 1.

Note that it's possible to contrive a case where the pre-f6fca3917b4d
allocator fails the same way, but these cases generally have extensive
dev_extent fragmentation as a precondition (e.g. many holes of 768M
in length on one device, and few holes 1 GiB in length on the others).
With the regression in f6fca3917b4d, bad chunk allocation can occur even
under optimal conditions, when all dev_extent holes are exact multiples
of stripe_len in length, as in the example above.

Also note that post-f6fca3917b4d kernels do treat dev_extent holes
larger than 10 GiB as equal, so the bad behavior won't show up on a
freshly formatted filesystem; however, as the filesystem ages and fills
up, and holes ranging from 1 GiB to 10 GiB in size appear, the problem
can show up as a failure to balance after adding or removing devices,
or an unexpected shortfall in available space due to unequal allocation.

To fix the regression and make data chunk allocation work
again, set ctl->max_stripe_len back to the original SZ_1G, or
space_info->chunk_size if that's smaller (the latter can happen if the
user set space_info->chunk_size to less than 1 GiB via sysfs, or it's
a 32 MiB system chunk with a hardcoded chunk_size and stripe_len).

While researching the background of the earlier commits, I found that an
identical fix was already proposed at:

  https://lore.kernel.org/linux-btrfs/de83ac46-a4a3-88d3-85ce-255b7abc5249@gmx.com/

The previous review missed one detail:  ctl->max_stripe_len is used
before decide_stripe_size_regular() is called, when it is too late for
the changes in that function to have any effect.  ctl->max_stripe_len is
not used directly by decide_stripe_size_regular(), but the parameter
does heavily influence the per-device free space data presented to
the function.

Fixes: f6fca3917b4d ("btrfs: store chunk size in space-info struct")
CC: stable@vger.kernel.org # 6.1+
Link: https://lore.kernel.org/linux-btrfs/20231007051421.19657-1-ce3g8jdj@umail.furryterror.org/
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agonet/smc: return the right falback reason when prefix checks fail
Dust Li [Thu, 12 Oct 2023 12:37:29 +0000 (20:37 +0800)]
net/smc: return the right falback reason when prefix checks fail

commit 4abbd2e3c1db671fa1286390f1310aec78386f1d upstream.

In the smc_listen_work(), if smc_listen_prfx_check() failed,
the real reason: SMC_CLC_DECL_DIFFPREFIX was dropped, and
SMC_CLC_DECL_NOSMCDEV was returned.

Althrough this is also kind of SMC_CLC_DECL_NOSMCDEV, but return
the real reason is much friendly for debugging.

Fixes: e49300a6bf62 ("net/smc: add listen processing for SMC-Rv2")
Signed-off-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Alexandra Winter <wintera@linux.ibm.com>
Reviewed-by: Wenjia Zhang <wenjia@linux.ibm.com>
Link: https://lore.kernel.org/r/20231012123729.29307-1-dust.li@linux.alibaba.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoice: reset first in crash dump kernels
Jesse Brandeburg [Wed, 11 Oct 2023 23:33:33 +0000 (16:33 -0700)]
ice: reset first in crash dump kernels

commit 0288c3e709e5fabd51e84715c5c798a02f43061a upstream.

When the system boots into the crash dump kernel after a panic, the ice
networking device may still have pending transactions that can cause errors
or machine checks when the device is re-enabled. This can prevent the crash
dump kernel from loading the driver or collecting the crash data.

To avoid this issue, perform a function level reset (FLR) on the ice device
via PCIe config space before enabling it on the crash kernel. This will
clear any outstanding transactions and stop all queues and interrupts.
Restore the config space after the FLR, otherwise it was found in testing
that the driver wouldn't load successfully.

The following sequence causes the original issue:
- Load the ice driver with modprobe ice
- Enable SR-IOV with 2 VFs: echo 2 > /sys/class/net/eth0/device/sriov_num_vfs
- Trigger a crash with echo c > /proc/sysrq-trigger
- Load the ice driver again (or let it load automatically) with modprobe ice
- The system crashes again during pcim_enable_device()

Fixes: 837f08fdecbe ("ice: Add basic driver framework for Intel(R) E800 Series")
Reported-by: Vishal Agrawal <vagrawal@redhat.com>
Reviewed-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel)
Link: https://lore.kernel.org/r/20231011233334.336092-3-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoice: fix over-shifted variable
Jesse Brandeburg [Tue, 10 Oct 2023 20:30:59 +0000 (13:30 -0700)]
ice: fix over-shifted variable

commit 242e34500a32631f85c2b4eb6cb42a368a39e54f upstream.

Since the introduction of the ice driver the code has been
double-shifting the RSS enabling field, because the define already has
shifts in it and can't have the regular pattern of "a << shiftval &
mask" applied.

Most places in the code got it right, but one line was still wrong. Fix
this one location for easy backports to stable. An in-progress patch
fixes the defines to "standard" and will be applied as part of the
regular -next process sometime after this one.

Fixes: d76a60ba7afb ("ice: Add support for VLANs and offloads")
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
CC: stable@vger.kernel.org
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel)
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://lore.kernel.org/r/20231010203101.406248-1-jacob.e.keller@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoBluetooth: avoid memcmp() out of bounds warning
Arnd Bergmann [Mon, 9 Oct 2023 20:31:31 +0000 (22:31 +0200)]
Bluetooth: avoid memcmp() out of bounds warning

commit 9d1a3c74746428102d55371fbf74b484733937d9 upstream.

bacmp() is a wrapper around memcpy(), which contain compile-time
checks for buffer overflow. Since the hci_conn_request_evt() also calls
bt_dev_dbg() with an implicit NULL pointer check, the compiler is now
aware of a case where 'hdev' is NULL and treats this as meaning that
zero bytes are available:

In file included from net/bluetooth/hci_event.c:32:
In function 'bacmp',
    inlined from 'hci_conn_request_evt' at net/bluetooth/hci_event.c:3276:7:
include/net/bluetooth/bluetooth.h:364:16: error: 'memcmp' specified bound 6 exceeds source size 0 [-Werror=stringop-overread]
  364 |         return memcmp(ba1, ba2, sizeof(bdaddr_t));
      |                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Add another NULL pointer check before the bacmp() to ensure the compiler
understands the code flow enough to not warn about it.  Since the patch
that introduced the warning is marked for stable backports, this one
should also go that way to avoid introducing build regressions.

Fixes: 1ffc6f8cc332 ("Bluetooth: Reject connection with the device which has same BD_ADDR")
Cc: Kees Cook <keescook@chromium.org>
Cc: "Lee, Chun-Yi" <jlee@suse.com>
Cc: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: stable@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoBluetooth: hci_event: Fix coding style
Luiz Augusto von Dentz [Thu, 5 Oct 2023 21:12:19 +0000 (14:12 -0700)]
Bluetooth: hci_event: Fix coding style

commit 35d91d95a0cd61ebb90e0246dc917fd25e519b8c upstream.

This fixes the following code style problem:

ERROR: that open brace { should be on the previous line
+ if (!bacmp(&hdev->bdaddr, &ev->bdaddr))
+ {

Fixes: 1ffc6f8cc332 ("Bluetooth: Reject connection with the device which has same BD_ADDR")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoBluetooth: vhci: Fix race when opening vhci device
Arkadiusz Bokowy [Wed, 20 Sep 2023 15:30:07 +0000 (17:30 +0200)]
Bluetooth: vhci: Fix race when opening vhci device

commit 92d4abd66f7080075793970fc8f241239e58a9e7 upstream.

When the vhci device is opened in the two-step way, i.e.: open device
then write a vendor packet with requested controller type, the device
shall respond with a vendor packet which includes HCI index of created
interface.

When the virtual HCI is created, the host sends a reset request to the
controller. This request is processed by the vhci_send_frame() function.
However, this request is send by a different thread, so it might happen
that this HCI request will be received before the vendor response is
queued in the read queue. This results in the HCI vendor response and
HCI reset request inversion in the read queue which leads to improper
behavior of btvirt:

> dmesg
[1754256.640122] Bluetooth: MGMT ver 1.22
[1754263.023806] Bluetooth: MGMT ver 1.22
[1754265.043775] Bluetooth: hci1: Opcode 0x c03 failed: -110

In order to synchronize vhci two-step open/setup process with virtual
HCI initialization, this patch adds internal lock when queuing data in
the vhci_send_frame() function.

Signed-off-by: Arkadiusz Bokowy <arkadiusz.bokowy@gmail.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoBluetooth: Fix a refcnt underflow problem for hci_conn
Ziyang Xuan [Wed, 4 Oct 2023 12:42:24 +0000 (20:42 +0800)]
Bluetooth: Fix a refcnt underflow problem for hci_conn

commit c7f59461f5a78994613afc112cdd73688aef9076 upstream.

Syzbot reports a warning as follows:

WARNING: CPU: 1 PID: 26946 at net/bluetooth/hci_conn.c:619
hci_conn_timeout+0x122/0x210 net/bluetooth/hci_conn.c:619
...
Call Trace:
 <TASK>
 process_one_work+0x884/0x15c0 kernel/workqueue.c:2630
 process_scheduled_works kernel/workqueue.c:2703 [inline]
 worker_thread+0x8b9/0x1290 kernel/workqueue.c:2784
 kthread+0x33c/0x440 kernel/kthread.c:388
 ret_from_fork+0x45/0x80 arch/x86/kernel/process.c:147
 ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:304
 </TASK>

It is because the HCI_EV_SIMPLE_PAIR_COMPLETE event handler drops
hci_conn directly without check Simple Pairing whether be enabled. But
the Simple Pairing process can only be used if both sides have the
support enabled in the host stack.

Add hci_conn_ssp_enabled() for hci_conn in HCI_EV_IO_CAPA_REQUEST and
HCI_EV_SIMPLE_PAIR_COMPLETE event handlers to fix the problem.

Fixes: 0493684ed239 ("[Bluetooth] Disable disconnect timer during Simple Pairing")
Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoBluetooth: Reject connection with the device which has same BD_ADDR
Lee, Chun-Yi [Sun, 1 Oct 2023 08:59:58 +0000 (16:59 +0800)]
Bluetooth: Reject connection with the device which has same BD_ADDR

commit 1ffc6f8cc33268731fcf9629fc4438f6db1191fc upstream.

This change is used to relieve CVE-2020-26555. The description of
the CVE:

Bluetooth legacy BR/EDR PIN code pairing in Bluetooth Core Specification
1.0B through 5.2 may permit an unauthenticated nearby device to spoof
the BD_ADDR of the peer device to complete pairing without knowledge
of the PIN. [1]

The detail of this attack is in IEEE paper:
BlueMirror: Reflections on Bluetooth Pairing and Provisioning Protocols
[2]

It's a reflection attack. The paper mentioned that attacker can induce
the attacked target to generate null link key (zero key) without PIN
code. In BR/EDR, the key generation is actually handled in the controller
which is below HCI.

A condition of this attack is that attacker should change the
BR_ADDR of his hacking device (Host B) to equal to the BR_ADDR with
the target device being attacked (Host A).

Thus, we reject the connection with device which has same BD_ADDR
both on HCI_Create_Connection and HCI_Connection_Request to prevent
the attack. A similar implementation also shows in btstack project.
[3][4]

Cc: stable@vger.kernel.org
Link: https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2020-26555
Link: https://ieeexplore.ieee.org/abstract/document/9474325/authors#authors
Link: https://github.com/bluekitchen/btstack/blob/master/src/hci.c#L3523
Link: https://github.com/bluekitchen/btstack/blob/master/src/hci.c#L7297
Signed-off-by: Lee, Chun-Yi <jlee@suse.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoBluetooth: hci_event: Ignore NULL link key
Lee, Chun-Yi [Sun, 1 Oct 2023 08:59:31 +0000 (16:59 +0800)]
Bluetooth: hci_event: Ignore NULL link key

commit 33155c4aae5260475def6f7438e4e35564f4f3ba upstream.

This change is used to relieve CVE-2020-26555. The description of the
CVE:

Bluetooth legacy BR/EDR PIN code pairing in Bluetooth Core Specification
1.0B through 5.2 may permit an unauthenticated nearby device to spoof
the BD_ADDR of the peer device to complete pairing without knowledge
of the PIN. [1]

The detail of this attack is in IEEE paper:
BlueMirror: Reflections on Bluetooth Pairing and Provisioning Protocols
[2]

It's a reflection attack. The paper mentioned that attacker can induce
the attacked target to generate null link key (zero key) without PIN
code. In BR/EDR, the key generation is actually handled in the controller
which is below HCI.

Thus, we can ignore null link key in the handler of "Link Key Notification
event" to relieve the attack. A similar implementation also shows in
btstack project. [3]

v3: Drop the connection when null link key be detected.

v2:
- Used Link: tag instead of Closes:
- Used bt_dev_dbg instead of BT_DBG
- Added Fixes: tag

Cc: stable@vger.kernel.org
Fixes: 55ed8ca10f35 ("Bluetooth: Implement link key handling for the management interface")
Link: https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2020-26555
Link: https://ieeexplore.ieee.org/abstract/document/9474325/authors#authors
Link: https://github.com/bluekitchen/btstack/blob/master/src/hci.c#L3722
Signed-off-by: Lee, Chun-Yi <jlee@suse.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoigc: Fix race condition in PTP tx code
Vinicius Costa Gomes [Wed, 7 Jun 2023 21:32:29 +0000 (14:32 -0700)]
igc: Fix race condition in PTP tx code

commit 9c50e2b150c8ee0eee5f8154e2ad168cdd748877 upstream.

Currently, the igc driver supports timestamping only one tx packet at a
time. During the transmission flow, the skb that requires hardware
timestamping is saved in adapter->ptp_tx_skb. Once hardware has the
timestamp, an interrupt is delivered, and adapter->ptp_tx_work is
scheduled. In igc_ptp_tx_work(), we read the timestamp register, update
adapter->ptp_tx_skb, and notify the network stack.

While the thread executing the transmission flow (the user process
running in kernel mode) and the thread executing ptp_tx_work don't
access adapter->ptp_tx_skb concurrently, there are two other places
where adapter->ptp_tx_skb is accessed: igc_ptp_tx_hang() and
igc_ptp_suspend().

igc_ptp_tx_hang() is executed by the adapter->watchdog_task worker
thread which runs periodically so it is possible we have two threads
accessing ptp_tx_skb at the same time. Consider the following scenario:
right after __IGC_PTP_TX_IN_PROGRESS is set in igc_xmit_frame_ring(),
igc_ptp_tx_hang() is executed. Since adapter->ptp_tx_start hasn't been
written yet, this is considered a timeout and adapter->ptp_tx_skb is
cleaned up.

This patch fixes the issue described above by adding the ptp_tx_lock to
protect access to ptp_tx_skb and ptp_tx_start fields from igc_adapter.
Since igc_xmit_frame_ring() called in atomic context by the networking
stack, ptp_tx_lock is defined as a spinlock, and the irq safe variants
of lock/unlock are used.

With the introduction of the ptp_tx_lock, the __IGC_PTP_TX_IN_PROGRESS
flag doesn't provide much of a use anymore so this patch gets rid of it.

Fixes: 2c344ae24501 ("igc: Add support for TX timestamping")
Signed-off-by: Andre Guedes <andre.guedes@intel.com>
Signed-off-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Reviewed-by: Kurt Kanzenbach <kurt@linutronix.de>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoigc: Add condition for qbv_config_change_errors counter
Muhammad Husaini Zulkifli [Mon, 15 May 2023 06:03:36 +0000 (14:03 +0800)]
igc: Add condition for qbv_config_change_errors counter

commit ed89b74d2dc920cb61d3094e0e97ec8775b13086 upstream.

Add condition to increase the qbv counter during taprio qbv
configuration only.

There might be a case when TC already been setup then user configure
the ETF/CBS qdisc and this counter will increase if no condition above.

Fixes: ae4fe4698300 ("igc: Add qbv_config_change_errors counter")
Signed-off-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoigc: Add qbv_config_change_errors counter
Muhammad Husaini Zulkifli [Thu, 16 Feb 2023 02:07:31 +0000 (10:07 +0800)]
igc: Add qbv_config_change_errors counter

commit ae4fe46983007bc46d87dcb284a5e5851c3e1c84 upstream.

Add ConfigChangeError(qbv_config_change_errors) when user try to set the
AdminBaseTime to past value while the current GCL is still running.

The ConfigChangeError counter should not be increased when a gate control
list is scheduled into the future.

User can use "ethtool -S <interface> | grep qbv_config_change_errors"
command to check the counter values.

Signed-off-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoigc: Remove reset adapter task for i226 during disable tsn config
Muhammad Husaini Zulkifli [Wed, 14 Dec 2022 16:29:09 +0000 (00:29 +0800)]
igc: Remove reset adapter task for i226 during disable tsn config

commit 1d1b4c63ba739c6ca695cb2ea13fefa9dfbff60d upstream.

I225 have limitation when programming the BaseTime register which required
a power cycle of the controller. This limitation already lifted in I226.
This patch removes the restriction so that when user configure/remove any
TSN mode, it would not go into power cycle reset adapter.

How to test:

Schedule any gate control list configuration or delete it.

Example:

1)

BASE_TIME=$(date +%s%N)
tc qdisc replace dev $interface_name parent root handle 100 taprio \
    num_tc 4 \
    map 3 1 0 2 3 3 3 3 3 3 3 3 3 3 3 3 \
    queues 1@0 1@1 1@2 1@3 \
    base-time $BASE_TIME \
    sched-entry S 0F 1000000 \
    flags 0x2

2) tc qdisc del dev $intername_name root

Signed-off-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoigc: enable Qbv configuration for 2nd GCL
Tan Tee Min [Wed, 14 Dec 2022 16:29:08 +0000 (00:29 +0800)]
igc: enable Qbv configuration for 2nd GCL

commit 5ac1231ac14d1b8a1098048e51cad45f11b85c0a upstream.

Make reset task only executes for i225 and Qbv disabling to allow
i226 configure for 2nd GCL without resetting the adapter.

In i226, Tx won't hang if there is a GCL is already running, so in
this case we don't need to set FutScdDis bit.

Signed-off-by: Tan Tee Min <tee.min.tan@linux.intel.com>
Signed-off-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoigc: remove I226 Qbv BaseTime restriction
Muhammad Husaini Zulkifli [Wed, 14 Dec 2022 16:29:07 +0000 (00:29 +0800)]
igc: remove I226 Qbv BaseTime restriction

commit b8897dc54e3bc9d25281bbb42a7d730782ff4588 upstream.

Remove the Qbv BaseTime restriction for I226 so that the BaseTime can be
scheduled to the future time. A new register bit of Tx Qav Control
(Bit-7: FutScdDis) was introduced to allow I226 scheduling future time as
Qbv BaseTime and not having the Tx hang timeout issue.

Besides, according to datasheet section 7.5.2.9.3.3, FutScdDis bit has to
be configured first before the cycle time and base time.

Indeed the FutScdDis bit is only active on re-configuration, thus we have
to set the BASET_L to zero and then only set it to the desired value.

Please also note that the Qbv configuration flow is moved around based on
the Qbv programming guideline that is documented in the latest datasheet.

Co-developed-by: Tan Tee Min <tee.min.tan@linux.intel.com>
Signed-off-by: Tan Tee Min <tee.min.tan@linux.intel.com>
Signed-off-by: Muhammad Husaini Zulkifli <muhammad.husaini.zulkifli@intel.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agolib/Kconfig.debug: do not enable DEBUG_PREEMPT by default
Hyeonggon Yoo [Sat, 21 Jan 2023 03:39:42 +0000 (12:39 +0900)]
lib/Kconfig.debug: do not enable DEBUG_PREEMPT by default

commit cc6003916ed46d7a67d91ee32de0f9138047d55f upstream.

In workloads where this_cpu operations are frequently performed,
enabling DEBUG_PREEMPT may result in significant increase in
runtime overhead due to frequent invocation of
__this_cpu_preempt_check() function.

This can be demonstrated through benchmarks such as hackbench where this
configuration results in a 10% reduction in performance, primarily due to
the added overhead within memcg charging path.

Therefore, do not to enable DEBUG_PREEMPT by default and make users aware
of its potential impact on performance in some workloads.

hackbench-process-sockets
      debug_preempt  no_debug_preempt
Amean     1       0.4743 (   0.00%)      0.4295 *   9.45%*
Amean     4       1.4191 (   0.00%)      1.2650 *  10.86%*
Amean     7       2.2677 (   0.00%)      2.0094 *  11.39%*
Amean     12      3.6821 (   0.00%)      3.2115 *  12.78%*
Amean     21      6.6752 (   0.00%)      5.7956 *  13.18%*
Amean     30      9.6646 (   0.00%)      8.5197 *  11.85%*
Amean     48     15.3363 (   0.00%)     13.5559 *  11.61%*
Amean     79     24.8603 (   0.00%)     22.0597 *  11.27%*
Amean     96     30.1240 (   0.00%)     26.8073 *  11.01%*

Link: https://lkml.kernel.org/r/20230121033942.350387-1-42.hyeyoo@gmail.com
Signed-off-by: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Davidlohr Bueso <dave@stgolabs.net>
Cc: Ben Segall <bsegall@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Dennis Zhou <dennis@kernel.org>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Steven Rostedt (Google) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Luiz Capitulino <luizcap@amazon.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoLinux 6.1.59
Greg Kroah-Hartman [Thu, 19 Oct 2023 21:08:58 +0000 (23:08 +0200)]
Linux 6.1.59

Link: https://lore.kernel.org/r/20231016084000.050926073@linuxfoundation.org
Tested-by: Ricardo B. Marliere <ricardo@marliere.net>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Tested-by: Florian Fainelli <florian.fainelli@broadcom.com>
Tested-by: SeongJae Park <sj@kernel.org>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Bagas Sanjaya <bagasdotme@gmail.com>
Tested-by: Ron Economos <re@w6rz.net>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Pavel Machek (CIP) <pavel@denx.de>
Tested-by: Takeshi Ogasawara <takeshi.ogasawara@futuring-girl.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoALSA: hda/realtek - Fixed two speaker platform
Kailang Yang [Tue, 12 Sep 2023 07:31:49 +0000 (15:31 +0800)]
ALSA: hda/realtek - Fixed two speaker platform

commit fb6254df09bba303db2a1002085f6c0b90a456ed upstream.

If system has two speakers and one connect to 0x14 pin, use this
function will disable it.

Fixes: e43252db7e20 ("ALSA: hda/realtek - ALC287 I2S speaker platform support")
Signed-off-by: Kailang Yang <kailang@realtek.com>
Link: https://lore.kernel.org/r/e3f2aac3fe6a47079d728a6443358cc2@realtek.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agopowerpc/64e: Fix wrong test in __ptep_test_and_clear_young()
Christophe Leroy [Mon, 25 Sep 2023 18:31:16 +0000 (20:31 +0200)]
powerpc/64e: Fix wrong test in __ptep_test_and_clear_young()

[ Upstream commit 5ea0bbaa32e8f54e9a57cfee4a3b8769b80be0d2 ]

Commit 45201c879469 ("powerpc/nohash: Remove hash related code from
nohash headers.") replaced:

  if ((pte_val(*ptep) & (_PAGE_ACCESSED | _PAGE_HASHPTE)) == 0)
return 0;

By:

  if (pte_young(*ptep))
return 0;

But it should be:

  if (!pte_young(*ptep))
return 0;

Fix it.

Fixes: 45201c879469 ("powerpc/nohash: Remove hash related code from nohash headers.")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/8bb7f06494e21adada724ede47a4c3d97e879d40.1695659959.git.christophe.leroy@csgroup.eu
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agopowerpc/8xx: Fix pte_access_permitted() for PAGE_NONE
Christophe Leroy [Mon, 25 Sep 2023 18:31:15 +0000 (20:31 +0200)]
powerpc/8xx: Fix pte_access_permitted() for PAGE_NONE

[ Upstream commit 5d9cea8a552ee122e21fbd5a3c5d4eb85f648e06 ]

On 8xx, PAGE_NONE is handled by setting _PAGE_NA instead of clearing
_PAGE_USER.

But then pte_user() returns 1 also for PAGE_NONE.

As _PAGE_NA prevent reads, add a specific version of pte_read()
that returns 0 when _PAGE_NA is set instead of always returning 1.

Fixes: 351750331fc1 ("powerpc/mm: Introduce _PAGE_NA")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/57bcfbe578e43123f9ed73e040229b80f1ad56ec.1695659959.git.christophe.leroy@csgroup.eu
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agodmaengine: mediatek: Fix deadlock caused by synchronize_irq()
Duoming Zhou [Sun, 6 Aug 2023 03:25:11 +0000 (11:25 +0800)]
dmaengine: mediatek: Fix deadlock caused by synchronize_irq()

[ Upstream commit 01f1ae2733e2bb4de92fefcea5fda847d92aede1 ]

The synchronize_irq(c->irq) will not return until the IRQ handler
mtk_uart_apdma_irq_handler() is completed. If the synchronize_irq()
holds a spin_lock and waits the IRQ handler to complete, but the
IRQ handler also needs the same spin_lock. The deadlock will happen.
The process is shown below:

          cpu0                        cpu1
mtk_uart_apdma_device_pause() | mtk_uart_apdma_irq_handler()
  spin_lock_irqsave()         |
                              |   spin_lock_irqsave()
  //hold the lock to wait     |
  synchronize_irq()           |

This patch reorders the synchronize_irq(c->irq) outside the spin_lock
in order to mitigate the bug.

Fixes: 9135408c3ace ("dmaengine: mediatek: Add MediaTek UART APDMA support")
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Reviewed-by: Eugen Hristev <eugen.hristev@collabora.com>
Link: https://lore.kernel.org/r/20230806032511.45263-1-duoming@zju.edu.cn
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agodmaengine: idxd: use spin_lock_irqsave before wait_event_lock_irq
Rex Zhang [Sat, 16 Sep 2023 06:06:19 +0000 (14:06 +0800)]
dmaengine: idxd: use spin_lock_irqsave before wait_event_lock_irq

[ Upstream commit c0409dd3d151f661e7e57b901a81a02565df163c ]

In idxd_cmd_exec(), wait_event_lock_irq() explicitly calls
spin_unlock_irq()/spin_lock_irq(). If the interrupt is on before entering
wait_event_lock_irq(), it will become off status after
wait_event_lock_irq() is called. Later, wait_for_completion() may go to
sleep but irq is disabled. The scenario is warned in might_sleep().

Fix it by using spin_lock_irqsave() instead of the primitive spin_lock()
to save the irq status before entering wait_event_lock_irq() and using
spin_unlock_irqrestore() instead of the primitive spin_unlock() to restore
the irq status before entering wait_for_completion().

Before the change:
idxd_cmd_exec() {
interrupt is on
spin_lock()                        // interrupt is on
wait_event_lock_irq()
spin_unlock_irq()  // interrupt is enabled
...
spin_lock_irq()    // interrupt is disabled
spin_unlock()                      // interrupt is still disabled
wait_for_completion()              // report "BUG: sleeping function
   // called from invalid context...
   // in_atomic() irqs_disabled()"
}

After applying spin_lock_irqsave():
idxd_cmd_exec() {
interrupt is on
spin_lock_irqsave()                // save the on state
   // interrupt is disabled
wait_event_lock_irq()
spin_unlock_irq()  // interrupt is enabled
...
spin_lock_irq()    // interrupt is disabled
spin_unlock_irqrestore()           // interrupt is restored to on
wait_for_completion()              // No Call trace
}

Fixes: f9f4082dbc56 ("dmaengine: idxd: remove interrupt disable for cmd_lock")
Signed-off-by: Rex Zhang <rex.zhang@intel.com>
Signed-off-by: Lijun Pan <lijun.pan@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Fenghua Yu <fenghua.yu@intel.com>
Link: https://lore.kernel.org/r/20230916060619.3744220-1-rex.zhang@intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agox86/alternatives: Disable KASAN in apply_alternatives()
Kirill A. Shutemov [Thu, 12 Oct 2023 10:04:24 +0000 (13:04 +0300)]
x86/alternatives: Disable KASAN in apply_alternatives()

commit d35652a5fc9944784f6f50a5c979518ff8dacf61 upstream.

Fei has reported that KASAN triggers during apply_alternatives() on
a 5-level paging machine:

BUG: KASAN: out-of-bounds in rcu_is_watching()
Read of size 4 at addr ff110003ee6419a0 by task swapper/0/0
...
__asan_load4()
rcu_is_watching()
trace_hardirqs_on()
text_poke_early()
apply_alternatives()
...

On machines with 5-level paging, cpu_feature_enabled(X86_FEATURE_LA57)
gets patched. It includes KASAN code, where KASAN_SHADOW_START depends on
__VIRTUAL_MASK_SHIFT, which is defined with cpu_feature_enabled().

KASAN gets confused when apply_alternatives() patches the
KASAN_SHADOW_START users. A test patch that makes KASAN_SHADOW_START
static, by replacing __VIRTUAL_MASK_SHIFT with 56, works around the issue.

Fix it for real by disabling KASAN while the kernel is patching alternatives.

[ mingo: updated the changelog ]

Fixes: 6657fca06e3f ("x86/mm: Allow to boot without LA57 if CONFIG_X86_5LEVEL=y")
Reported-by: Fei Yang <fei.yang@intel.com>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20231012100424.1456-1-kirill.shutemov@linux.intel.com
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: cdnsp: Fixes issue with dequeuing not queued requests
Pawel Laszczak [Thu, 13 Jul 2023 08:14:29 +0000 (04:14 -0400)]
usb: cdnsp: Fixes issue with dequeuing not queued requests

commit 34f08eb0ba6e4869bbfb682bf3d7d0494ffd2f87 upstream.

Gadget ACM while unloading module try to dequeue not queued usb
request which causes the kernel to crash.
Patch adds extra condition to check whether usb request is processed
by CDNSP driver.

cc: stable@vger.kernel.org
Fixes: 3d82904559f4 ("usb: cdnsp: cdns3 Add main part of Cadence USBSSP DRD Driver")
Signed-off-by: Pawel Laszczak <pawell@cadence.com>
Acked-by: Peter Chen <peter.chen@kernel.org>
Link: https://lore.kernel.org/r/20230713081429.326660-1-pawell@cadence.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: gadget: ncm: Handle decoding of multiple NTB's in unwrap call
Krishna Kurapati [Wed, 27 Sep 2023 10:58:58 +0000 (16:28 +0530)]
usb: gadget: ncm: Handle decoding of multiple NTB's in unwrap call

commit 427694cfaafa565a3db5c5ea71df6bc095dca92f upstream.

When NCM is used with hosts like Windows PC, it is observed that there are
multiple NTB's contained in one usb request giveback. Since the driver
unwraps the obtained request data assuming only one NTB is present, we
loose the subsequent NTB's present resulting in data loss.

Fix this by checking the parsed block length with the obtained data
length in usb request and continue parsing after the last byte of current
NTB.

Cc: stable@vger.kernel.org
Fixes: 9f6ce4240a2b ("usb: gadget: f_ncm.c added")
Signed-off-by: Krishna Kurapati <quic_kriskura@quicinc.com>
Reviewed-by: Maciej Żenczykowski <maze@google.com>
Link: https://lore.kernel.org/r/20230927105858.12950-1-quic_kriskura@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: gadget: udc-xilinx: replace memcpy with memcpy_toio
Piyush Mehta [Fri, 29 Sep 2023 12:15:14 +0000 (17:45 +0530)]
usb: gadget: udc-xilinx: replace memcpy with memcpy_toio

commit 3061b6491f491197a35e14e49f805d661b02acd4 upstream.

For ARM processor, unaligned access to device memory is not allowed.
Method memcpy does not take care of alignment.

USB detection failure with the unalingned address of memory, with
below kernel crash. To fix the unalingned address kernel panic,
replace memcpy with memcpy_toio method.

Kernel crash:
Unable to handle kernel paging request at virtual address ffff80000c05008a
Mem abort info:
  ESR = 0x96000061
  EC = 0x25: DABT (current EL), IL = 32 bits
  SET = 0, FnV = 0
  EA = 0, S1PTW = 0
  FSC = 0x21: alignment fault
Data abort info:
  ISV = 0, ISS = 0x00000061
  CM = 0, WnR = 1
swapper pgtable: 4k pages, 48-bit VAs, pgdp=000000000143b000
[ffff80000c05008a] pgd=100000087ffff003, p4d=100000087ffff003,
pud=100000087fffe003, pmd=1000000800bcc003, pte=00680000a0010713
Internal error: Oops: 96000061 [#1] SMP
Modules linked in:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.15.19-xilinx-v2022.1 #1
Hardware name: ZynqMP ZCU102 Rev1.0 (DT)
pstate: 200000c5 (nzCv daIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
pc : __memcpy+0x30/0x260
lr : __xudc_ep0_queue+0xf0/0x110
sp : ffff800008003d00
x29: ffff800008003d00 x28: ffff800009474e80 x27: 00000000000000a0
x26: 0000000000000100 x25: 0000000000000012 x24: ffff000800bc8080
x23: 0000000000000001 x22: 0000000000000012 x21: ffff000800bc8080
x20: 0000000000000012 x19: ffff000800bc8080 x18: 0000000000000000
x17: ffff800876482000 x16: ffff800008004000 x15: 0000000000004000
x14: 00001f09785d0400 x13: 0103020101005567 x12: 0781400000000200
x11: 00000000c5672a10 x10: 00000000000008d0 x9 : ffff800009463cf0
x8 : ffff8000094757b0 x7 : 0201010055670781 x6 : 4000000002000112
x5 : ffff80000c05009a x4 : ffff000800a15012 x3 : ffff00080362ad80
x2 : 0000000000000012 x1 : ffff000800a15000 x0 : ffff80000c050088
Call trace:
 __memcpy+0x30/0x260
 xudc_ep0_queue+0x3c/0x60
 usb_ep_queue+0x38/0x44
 composite_ep0_queue.constprop.0+0x2c/0xc0
 composite_setup+0x8d0/0x185c
 configfs_composite_setup+0x74/0xb0
 xudc_irq+0x570/0xa40
 __handle_irq_event_percpu+0x58/0x170
 handle_irq_event+0x60/0x120
 handle_fasteoi_irq+0xc0/0x220
 handle_domain_irq+0x60/0x90
 gic_handle_irq+0x74/0xa0
 call_on_irq_stack+0x2c/0x60
 do_interrupt_handler+0x54/0x60
 el1_interrupt+0x30/0x50
 el1h_64_irq_handler+0x18/0x24
 el1h_64_irq+0x78/0x7c
 arch_cpu_idle+0x18/0x2c
 do_idle+0xdc/0x15c
 cpu_startup_entry+0x28/0x60
 rest_init+0xc8/0xe0
 arch_call_rest_init+0x10/0x1c
 start_kernel+0x694/0x6d4
 __primary_switched+0xa4/0xac

Fixes: 1f7c51660034 ("usb: gadget: Add xilinx usb2 device support")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/all/202209020044.CX2PfZzM-lkp@intel.com/
Cc: stable@vger.kernel.org
Signed-off-by: Piyush Mehta <piyush.mehta@amd.com>
Link: https://lore.kernel.org/r/20230929121514.13475-1-piyush.mehta@amd.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: typec: ucsi: Clear EVENT_PENDING bit if ucsi_send_command fails
Prashanth K [Mon, 11 Sep 2023 09:04:15 +0000 (14:34 +0530)]
usb: typec: ucsi: Clear EVENT_PENDING bit if ucsi_send_command fails

commit a00e197daec52bcd955e118f5f57d706da5bfe50 upstream.

Currently if ucsi_send_command() fails, then we bail out without
clearing EVENT_PENDING flag. So when the next connector change
event comes, ucsi_connector_change() won't queue the con->work,
because of which none of the new events will be processed.

Fix this by clearing EVENT_PENDING flag if ucsi_send_command()
fails.

Cc: stable@vger.kernel.org # 5.16
Fixes: 512df95b9432 ("usb: typec: ucsi: Better fix for missing unplug events issue")
Signed-off-by: Prashanth K <quic_prashk@quicinc.com>
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/1694423055-8440-1-git-send-email-quic_prashk@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: typec: altmodes/displayport: Signal hpd low when exiting mode
RD Babiera [Mon, 9 Oct 2023 21:00:58 +0000 (21:00 +0000)]
usb: typec: altmodes/displayport: Signal hpd low when exiting mode

commit 89434b069e460967624903b049e5cf5c9e6b99b9 upstream.

Upon receiving an ACK for a sent EXIT_MODE message, the DisplayPort
driver currently resets the status and configuration of the port partner.
The hpd signal is not updated despite being part of the status, so the
Display stack can still transmit video despite typec_altmode_exit placing
the lanes in a Safe State.

Set hpd to low when a sent EXIT_MODE message is ACK'ed.

Fixes: 0e3bb7d6894d ("usb: typec: Add driver for DisplayPort alternate mode")
Cc: stable@vger.kernel.org
Signed-off-by: RD Babiera <rdbabiera@google.com>
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20231009210057.3773877-2-rdbabiera@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agocounter: microchip-tcb-capture: Fix the use of internal GCLK logic
Dharma Balasubiramani [Tue, 5 Sep 2023 10:08:35 +0000 (15:38 +0530)]
counter: microchip-tcb-capture: Fix the use of internal GCLK logic

commit df8fdd01c98b99d04915c04f3a5ce73f55456b7c upstream.

As per the datasheet, the clock selection Bits 2:0 – TCCLKS[2:0] should
be set to 0 while using the internal GCLK (TIMER_CLOCK1).

Fixes: 106b104137fd ("counter: Add microchip TCB capture counter")
Signed-off-by: Dharma Balasubiramani <dharma.b@microchip.com>
Link: https://lore.kernel.org/r/20230905100835.315024-1-dharma.b@microchip.com
Signed-off-by: William Breathitt Gray <william.gray@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agocounter: chrdev: fix getting array extensions
Fabrice Gasnier [Tue, 29 Aug 2023 13:40:22 +0000 (15:40 +0200)]
counter: chrdev: fix getting array extensions

commit 3170256d7bc1ef81587caf4b83573eb1f5bb4fb6 upstream.

When trying to watch a component array extension, and the array isn't the
first extended element, it fails as the type comparison is always done on
the 1st element. Fix it by indexing the 'ext' array.

Example on a dummy struct counter_comp:
static struct counter_comp dummy[] = {
COUNTER_COMP_DIRECTION(..),
...,
COUNTER_COMP_ARRAY_CAPTURE(...),
};
static struct counter_count dummy_cnt = {
...
.ext = dummy,
.num_ext = ARRAY_SIZE(dummy),
}

Currently, counter_get_ext() returns -EINVAL when trying to add a watch
event on one of the capture array element in such example.

Fixes: d2011be1e22f ("counter: Introduce the COUNTER_COMP_ARRAY component type")
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@foss.st.com>
Link: https://lore.kernel.org/r/20230829134029.2402868-2-fabrice.gasnier@foss.st.com
Signed-off-by: William Breathitt Gray <william.gray@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoscsi: ufs: core: Correct clear TM error log
Peter Wang [Tue, 3 Oct 2023 02:20:02 +0000 (10:20 +0800)]
scsi: ufs: core: Correct clear TM error log

commit a20c4350c6a12405b7f732b3ee6801ffe2cc45ce upstream.

The clear TM function error log status was inverted.

Fixes: 4693fad7d6d4 ("scsi: ufs: core: Log error handler activity")
Signed-off-by: Peter Wang <peter.wang@mediatek.com>
Link: https://lore.kernel.org/r/20231003022002.25578-1-peter.wang@mediatek.com
Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Stanley Chu <stanley.chu@mediatek.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agopinctrl: avoid unsafe code pattern in find_pinctrl()
Dmitry Torokhov [Wed, 20 Sep 2023 18:09:10 +0000 (11:09 -0700)]
pinctrl: avoid unsafe code pattern in find_pinctrl()

commit c153a4edff6ab01370fcac8e46f9c89cca1060c2 upstream.

The code in find_pinctrl() takes a mutex and traverses a list of pinctrl
structures. Later the caller bumps up reference count on the found
structure. Such pattern is not safe as pinctrl that was found may get
deleted before the caller gets around to increasing the reference count.

Fix this by taking the reference count in find_pinctrl(), while it still
holds the mutex.

Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Link: https://lore.kernel.org/r/ZQs1RgTKg6VJqmPs@google.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodma-buf: add dma_fence_timestamp helper
Christian König [Fri, 8 Sep 2023 08:27:23 +0000 (10:27 +0200)]
dma-buf: add dma_fence_timestamp helper

commit b83ce9cb4a465b8f9a3fa45561b721a9551f60e3 upstream.

When a fence signals there is a very small race window where the timestamp
isn't updated yet. sync_file solves this by busy waiting for the
timestamp to appear, but on other ocassions didn't handled this
correctly.

Provide a dma_fence_timestamp() helper function for this and use it in
all appropriate cases.

Another alternative would be to grab the spinlock when that happens.

v2 by teddy: add a wait parameter to wait for the timestamp to show up, in case
   the accurate timestamp is needed and/or the timestamp is not based on
   ktime (e.g. hw timestamp)
v3 chk: drop the parameter again for unified handling

Signed-off-by: Yunxiang Li <Yunxiang.Li@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Fixes: 1774baa64f93 ("drm/scheduler: Change scheduled fence track v2")
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
CC: stable@vger.kernel.org
Link: https://patchwork.freedesktop.org/patch/msgid/20230929104725.2358-1-christian.koenig@amd.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agocgroup: Remove duplicates in cgroup v1 tasks file
Michal Koutný [Mon, 9 Oct 2023 13:58:11 +0000 (15:58 +0200)]
cgroup: Remove duplicates in cgroup v1 tasks file

commit 1ca0b605150501b7dc59f3016271da4eb3e96fce upstream.

One PID may appear multiple times in a preloaded pidlist.
(Possibly due to PID recycling but we have reports of the same
task_struct appearing with different PIDs, thus possibly involving
transfer of PID via de_thread().)

Because v1 seq_file iterator uses PIDs as position, it leads to
a message:
> seq_file: buggy .next function kernfs_seq_next did not update position index

Conservative and quick fix consists of removing duplicates from `tasks`
file (as opposed to removing pidlists altogether). It doesn't affect
correctness (it's sufficient to show a PID once), performance impact
would be hidden by unconditional sorting of the pidlist already in place
(asymptotically).

Link: https://lore.kernel.org/r/20230823174804.23632-1-mkoutny@suse.com/
Suggested-by: Firo Yang <firo.yang@suse.com>
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: typec: ucsi: Use GET_CAPABILITY attributes data to set power supply scope
Mario Limonciello [Mon, 9 Oct 2023 18:46:43 +0000 (13:46 -0500)]
usb: typec: ucsi: Use GET_CAPABILITY attributes data to set power supply scope

commit c9ca8de2eb15f9da24113e652980c61f95a47530 upstream.

On some OEM systems, adding a W7900 dGPU triggers RAS errors and hangs
at a black screen on startup.  This issue occurs only if `ucsi_acpi` has
loaded before `amdgpu` has loaded.  The reason for this failure is that
`amdgpu` uses power_supply_is_system_supplied() to determine if running
on AC or DC power at startup. If this value is reported incorrectly the
dGPU will also be programmed incorrectly and trigger errors.

power_supply_is_system_supplied() reports the wrong value because UCSI
power supplies provided as part of the system don't properly report the
scope as "DEVICE" scope (not powering the system).

In order to fix this issue check the capabilities reported from the UCSI
power supply to ensure that it supports charging a battery and that it can
be powered by AC.  Mark the scope accordingly.

Cc: stable@vger.kernel.org
Fixes: a7fbfd44c020 ("usb: typec: ucsi: Mark dGPUs as DEVICE scope")
Link: https://www.intel.com/content/www/us/en/products/docs/io/universal-serial-bus/usb-type-c-ucsi-spec.html
Reviewed-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20231009184643.129986-1-mario.limonciello@amd.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agonfp: flower: avoid rmmod nfp crash issues
Yanguo Li [Mon, 9 Oct 2023 11:21:55 +0000 (13:21 +0200)]
nfp: flower: avoid rmmod nfp crash issues

commit 14690995c14109852c7ba6e316045c02e4254272 upstream.

When there are CT table entries, and you rmmod nfp, the following
events can happen:

task1:
    nfp_net_pci_remove
          ↓
    nfp_flower_stop->(asynchronous)tcf_ct_flow_table_cleanup_work(3)
          ↓
    nfp_zone_table_entry_destroy(1)

task2:
    nfp_fl_ct_handle_nft_flow(2)

When the execution order is (1)->(2)->(3), it will crash. Therefore, in
the function nfp_fl_ct_del_flow, nf_flow_table_offload_del_cb needs to
be executed synchronously.

At the same time, in order to solve the deadlock problem and the problem
of rtnl_lock sometimes failing, replace rtnl_lock with the private
nfp_fl_lock.

Fixes: 7cc93d888df7 ("nfp: flower-ct: remove callback delete deadlock")
Cc: stable@vger.kernel.org
Signed-off-by: Yanguo Li <yanguo.li@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agomctp: perform route lookups under a RCU read-side lock
Jeremy Kerr [Mon, 9 Oct 2023 07:56:45 +0000 (15:56 +0800)]
mctp: perform route lookups under a RCU read-side lock

commit 5093bbfc10ab6636b32728e35813cbd79feb063c upstream.

Our current route lookups (mctp_route_lookup and mctp_route_lookup_null)
traverse the net's route list without the RCU read lock held. This means
the route lookup is subject to preemption, resulting in an potential
grace period expiry, and so an eventual kfree() while we still have the
route pointer.

Add the proper read-side critical section locks around the route
lookups, preventing premption and a possible parallel kfree.

The remaining net->mctp.routes accesses are already under a
rcu_read_lock, or protected by the RTNL for updates.

Based on an analysis from Sili Luo <rootlab@huawei.com>, where
introducing a delay in the route lookup could cause a UAF on
simultaneous sendmsg() and route deletion.

Reported-by: Sili Luo <rootlab@huawei.com>
Fixes: 889b7da23abf ("mctp: Add initial routing framework")
Cc: stable@vger.kernel.org
Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/29c4b0e67dc1bf3571df3982de87df90cae9b631.1696837310.git.jk@codeconstruct.com.au
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agotee: amdtee: fix use-after-free vulnerability in amdtee_close_session
Rijo Thomas [Fri, 29 Sep 2023 07:00:24 +0000 (12:30 +0530)]
tee: amdtee: fix use-after-free vulnerability in amdtee_close_session

commit f4384b3e54ea813868bb81a861bf5b2406e15d8f upstream.

There is a potential race condition in amdtee_close_session that may
cause use-after-free in amdtee_open_session. For instance, if a session
has refcount == 1, and one thread tries to free this session via:

    kref_put(&sess->refcount, destroy_session);

the reference count will get decremented, and the next step would be to
call destroy_session(). However, if in another thread,
amdtee_open_session() is called before destroy_session() has completed
execution, alloc_session() may return 'sess' that will be freed up
later in destroy_session() leading to use-after-free in
amdtee_open_session.

To fix this issue, treat decrement of sess->refcount and removal of
'sess' from session list in destroy_session() as a critical section, so
that it is executed atomically.

Fixes: 757cc3e9ff1d ("tee: add AMD-TEE driver")
Cc: stable@vger.kernel.org
Signed-off-by: Rijo Thomas <Rijo-john.Thomas@amd.com>
Reviewed-by: Sumit Garg <sumit.garg@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoInput: goodix - ensure int GPIO is in input for gpio_count == 1 && gpio_int_idx ...
Hans de Goede [Wed, 4 Oct 2023 14:18:31 +0000 (07:18 -0700)]
Input: goodix - ensure int GPIO is in input for gpio_count == 1 && gpio_int_idx == 0 case

commit 423622a90abb243944d1517b9f57db53729e45c4 upstream.

Add a special case for gpio_count == 1 && gpio_int_idx == 0 to
goodix_add_acpi_gpio_mappings().

It seems that on newer x86/ACPI devices the reset and irq GPIOs are no
longer listed as GPIO resources instead there is only 1 GpioInt resource
and _PS0 does the whole reset sequence for us.

This means that we must call acpi_device_fix_up_power() on these devices
to ensure that the chip is reset before we try to use it.

This part was already fixed in commit 3de93e6ed2df ("Input: goodix - call
acpi_device_fix_up_power() in some cases") by adding a call to
acpi_device_fix_up_power() to the generic "Unexpected ACPI resources"
catch all.

But it turns out that this case on some hw needs some more special
handling. Specifically the firmware may bootup with the IRQ pin in
output mode. The reset sequence from ACPI _PS0 (executed by
acpi_device_fix_up_power()) should put the pin in input mode,
but the GPIO subsystem has cached the direction at bootup, causing
request_irq() to fail due to gpiochip_lock_as_irq() failure:

[    9.119864] Goodix-TS i2c-GDIX1002:00: Unexpected ACPI resources: gpio_count 1, gpio_int_idx 0
[    9.317443] Goodix-TS i2c-GDIX1002:00: ID 911, version: 1060
[    9.321902] input: Goodix Capacitive TouchScreen as /devices/pci0000:00/0000:00:17.0/i2c_designware.4/i2c-5/i2c-GDIX1002:00/input/input8
[    9.327840] gpio gpiochip0: (INT3453:00): gpiochip_lock_as_irq: tried to flag a GPIO set as output for IRQ
[    9.327856] gpio gpiochip0: (INT3453:00): unable to lock HW IRQ 26 for IRQ
[    9.327861] genirq: Failed to request resources for GDIX1002:00 (irq 131) on irqchip intel-gpio
[    9.327912] Goodix-TS i2c-GDIX1002:00: request IRQ failed: -5

Fix this by adding a special case for gpio_count == 1 && gpio_int_idx == 0
which adds an ACPI GPIO lookup table for the int GPIO even though we cannot
use it for reset purposes (as there is no reset GPIO).

Adding the lookup will make the gpiod_int = gpiod_get(..., GPIOD_IN) call
succeed, which will explicitly set the direction to input fixing the issue.

Note this re-uses the acpi_goodix_int_first_gpios[] lookup table, since
there is only 1 GPIO in the ACPI resources the reset entry in that
lookup table will amount to a no-op.

Reported-and-tested-by: Michael Smith <1973.mjsmith@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20231003215144.69527-1-hdegoede@redhat.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoInput: i8042 - add Fujitsu Lifebook E5411 to i8042 quirk table
Szilard Fabian [Wed, 4 Oct 2023 12:47:01 +0000 (05:47 -0700)]
Input: i8042 - add Fujitsu Lifebook E5411 to i8042 quirk table

commit 80f39e1c27ba9e5a1ea7e68e21c569c9d8e46062 upstream.

In the initial boot stage the integrated keyboard of Fujitsu Lifebook E5411
refuses to work and it's not possible to type for example a dm-crypt
passphrase without the help of an external keyboard.

i8042.nomux kernel parameter resolves this issue but using that a PS/2
mouse is detected. This input device is unused even when the i2c-hid-acpi
kernel module is blacklisted making the integrated ELAN touchpad
(04F3:308A) not working at all.

Since the integrated touchpad is managed by the i2c_designware input
driver in the Linux kernel and you can't find a PS/2 mouse port on the
computer I think it's safe to not use the PS/2 mouse port at all.

Signed-off-by: Szilard Fabian <szfabian@bluemarch.art>
Link: https://lore.kernel.org/r/20231004011749.101789-1-szfabian@bluemarch.art
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoInput: xpad - add PXN V900 support
Matthias Berndt [Fri, 13 Oct 2023 22:04:36 +0000 (15:04 -0700)]
Input: xpad - add PXN V900 support

commit a65cd7ef5a864bdbbe037267c327786b7759d4c6 upstream.

Add VID and PID to the xpad_device table to allow driver to use the PXN
V900 steering wheel, which is XTYPE_XBOX360 compatible in xinput mode.

Signed-off-by: Matthias Berndt <matthias_berndt@gmx.de>
Link: https://lore.kernel.org/r/4932699.31r3eYUQgx@fedora
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoInput: psmouse - fix fast_reconnect function for PS/2 mode
Jeffery Miller [Fri, 13 Oct 2023 22:23:49 +0000 (15:23 -0700)]
Input: psmouse - fix fast_reconnect function for PS/2 mode

commit e2cb5cc822b6c9ee72c56ce1d81671b22c05406a upstream.

When the SMBus connection is attempted psmouse_smbus_init() sets
the fast_reconnect pointer to psmouse_smbus_reconnecti(). If SMBus
initialization fails, elantech_setup_ps2() and synaptics_init_ps2() will
fallback to PS/2 mode, replacing the psmouse private data. This can cause
issues on resume, since psmouse_smbus_reconnect() expects to find an
instance of struct psmouse_smbus_dev in psmouse->private.

The issue was uncovered when in 92e24e0e57f7 ("Input: psmouse - add
delay when deactivating for SMBus mode") psmouse_smbus_reconnect()
started attempting to use more of the data structure. The commit was
since reverted, not because it was at fault, but because there was found
a better way of doing what it was attempting to do.

Fix the problem by resetting the fast_reconnect pointer in psmouse
structure in elantech_setup_ps2() and synaptics_init_ps2() when the PS/2
mode is used.

Reported-by: Thorsten Leemhuis <linux@leemhuis.info>
Tested-by: Thorsten Leemhuis <linux@leemhuis.info>
Signed-off-by: Jeffery Miller <jefferymiller@google.com>
Fixes: bf232e460a35 ("Input: psmouse-smbus - allow to control psmouse_deactivate")
Link: https://lore.kernel.org/r/20231005002249.554877-1-jefferymiller@google.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoInput: powermate - fix use-after-free in powermate_config_complete
Javier Carrasco [Sat, 14 Oct 2023 03:11:33 +0000 (20:11 -0700)]
Input: powermate - fix use-after-free in powermate_config_complete

commit 5c15c60e7be615f05a45cd905093a54b11f461bc upstream.

syzbot has found a use-after-free bug [1] in the powermate driver. This
happens when the device is disconnected, which leads to a memory free from
the powermate_device struct.  When an asynchronous control message
completes after the kfree and its callback is invoked, the lock does not
exist anymore and hence the bug.

Use usb_kill_urb() on pm->config to cancel any in-progress requests upon
device disconnection.

[1] https://syzkaller.appspot.com/bug?extid=0434ac83f907a1dbdd1e

Signed-off-by: Javier Carrasco <javier.carrasco.cruz@gmail.com>
Reported-by: syzbot+0434ac83f907a1dbdd1e@syzkaller.appspotmail.com
Link: https://lore.kernel.org/r/20230916-topic-powermate_use_after_free-v3-1-64412b81a7a2@gmail.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoceph: fix type promotion bug on 32bit systems
Dan Carpenter [Sat, 7 Oct 2023 08:52:39 +0000 (11:52 +0300)]
ceph: fix type promotion bug on 32bit systems

commit 07bb00ef00ace88dd6f695fadbba76565756e55c upstream.

In this code "ret" is type long and "src_objlen" is unsigned int.  The
problem is that on 32bit systems, when we do the comparison signed longs
are type promoted to unsigned int.  So negative error codes from
do_splice_direct() are treated as success instead of failure.

Cc: stable@vger.kernel.org
Fixes: 1b0c3b9f91f0 ("ceph: re-org copy_file_range and fix some error paths")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoceph: fix incorrect revoked caps assert in ceph_fill_file_size()
Xiubo Li [Wed, 6 Sep 2023 06:22:07 +0000 (14:22 +0800)]
ceph: fix incorrect revoked caps assert in ceph_fill_file_size()

commit 15c0a870dc44ed14e01efbdd319d232234ee639f upstream.

When truncating the inode the MDS will acquire the xlock for the
ifile Locker, which will revoke the 'Frwsxl' caps from the clients.
But when the client just releases and flushes the 'Fw' caps to MDS,
for exmaple, and once the MDS receives the caps flushing msg it
just thought the revocation has finished. Then the MDS will continue
truncating the inode and then issued the truncate notification to
all the clients. While just before the clients receives the cap
flushing ack they receive the truncation notification, the clients
will detecte that the 'issued | dirty' is still holding the 'Fw'
caps.

Cc: stable@vger.kernel.org
Link: https://tracker.ceph.com/issues/56693
Fixes: b0d7c2231015 ("ceph: introduce i_truncate_mutex")
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Milind Changire <mchangir@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agolibceph: use kernel_connect()
Jordan Rife [Wed, 4 Oct 2023 23:38:27 +0000 (18:38 -0500)]
libceph: use kernel_connect()

commit 7563cf17dce0a875ba3d872acdc63a78ea344019 upstream.

Direct calls to ops->connect() can overwrite the address parameter when
used in conjunction with BPF SOCK_ADDR hooks. Recent changes to
kernel_connect() ensure that callers are insulated from such side
effects. This patch wraps the direct call to ops->connect() with
kernel_connect() to prevent unexpected changes to the address passed to
ceph_tcp_connect().

This change was originally part of a larger patch targeting the net tree
addressing all instances of unprotected calls to ops->connect()
throughout the kernel, but this change was split up into several patches
targeting various trees.

Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/netdev/20230821100007.559638-1-jrife@google.com/
Link: https://lore.kernel.org/netdev/9944248dba1bce861375fcce9de663934d933ba9.camel@redhat.com/
Fixes: d74bad4e74ee ("bpf: Hooks for sys_connect")
Signed-off-by: Jordan Rife <jrife@google.com>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agopowerpc/47x: Fix 47x syscall return crash
Michael Ellerman [Tue, 10 Oct 2023 11:47:50 +0000 (22:47 +1100)]
powerpc/47x: Fix 47x syscall return crash

commit f0eee815babed70a749d2496a7678be5b45b4c14 upstream.

Eddie reported that newer kernels were crashing during boot on his 476
FSP2 system:

  kernel tried to execute user page (b7ee2000) - exploit attempt? (uid: 0)
  BUG: Unable to handle kernel instruction fetch
  Faulting instruction address: 0xb7ee2000
  Oops: Kernel access of bad area, sig: 11 [#1]
  BE PAGE_SIZE=4K FSP-2
  Modules linked in:
  CPU: 0 PID: 61 Comm: mount Not tainted 6.1.55-d23900f.ppcnf-fsp2 #1
  Hardware name: ibm,fsp2 476fpe 0x7ff520c0 FSP-2
  NIP:  b7ee2000 LR: 8c008000 CTR: 00000000
  REGS: bffebd83 TRAP: 0400   Not tainted (6.1.55-d23900f.ppcnf-fs p2)
  MSR:  00000030 <IR,DR>  CR: 00001000  XER: 20000000
  GPR00: c00110ac bffebe63 bffebe7e bffebe88 8c008000 00001000 00000d12 b7ee2000
  GPR08: 00000033 00000000 00000000 c139df10 48224824 1016c314 10160000 00000000
  GPR16: 10160000 10160000 00000008 00000000 10160000 00000000 10160000 1017f5b0
  GPR24: 1017fa50 1017f4f0 1017fa50 1017f740 1017f630 00000000 00000000 1017f4f0
  NIP [b7ee2000] 0xb7ee2000
  LR [8c008000] 0x8c008000
  Call Trace:
  Instruction dump:
  XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX
  XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX
  ---[ end trace 0000000000000000 ]---

The problem is in ret_from_syscall where the check for
icache_44x_need_flush is done. When the flush is needed the code jumps
out-of-line to do the flush, and then intends to jump back to continue
the syscall return.

However the branch back to label 1b doesn't return to the correct
location, instead branching back just prior to the return to userspace,
causing bogus register values to be used by the rfi.

The breakage was introduced by commit 6f76a01173cc
("powerpc/syscall: implement system call entry/exit logic in C for PPC32") which
inadvertently removed the "1" label and reused it elsewhere.

Fix it by adding named local labels in the correct locations. Note that
the return label needs to be outside the ifdef so that CONFIG_PPC_47x=n
compiles.

Fixes: 6f76a01173cc ("powerpc/syscall: implement system call entry/exit logic in C for PPC32")
Cc: stable@vger.kernel.org # v5.12+
Reported-by: Eddie James <eajames@linux.ibm.com>
Tested-by: Eddie James <eajames@linux.ibm.com>
Link: https://lore.kernel.org/linuxppc-dev/fdaadc46-7476-9237-e104-1d2168526e72@linux.ibm.com/
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Link: https://msgid.link/20231010114750.847794-1-mpe@ellerman.id.au
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agothunderbolt: Restart XDomain discovery handshake after failure
Mika Westerberg [Thu, 7 Sep 2023 13:02:30 +0000 (16:02 +0300)]
thunderbolt: Restart XDomain discovery handshake after failure

commit 308092d080852f8997126e5b3507536162416f4a upstream.

Alex reported that after rebooting the other host the peer-to-peer link
does not come up anymore. The reason for this is that the host that was
not rebooted tries to send the UUID request only 10 times according to
the USB4 Inter-Domain spec and gives up if it does not get reply. Then
when the other side is actually ready it cannot get the link established
anymore. The USB4 Inter-Domain spec requires that the discovery protocol
is restarted in that case so implement this now.

Reported-by: Alex Balcanquall <alex@alexbal.com>
Fixes: 8e1de7042596 ("thunderbolt: Add support for XDomain lane bonding")
Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agothunderbolt: Check that lane 1 is in CL0 before enabling lane bonding
Mika Westerberg [Tue, 22 Aug 2023 13:36:18 +0000 (16:36 +0300)]
thunderbolt: Check that lane 1 is in CL0 before enabling lane bonding

commit a9fdf5f933a6f2b358fad0194b1287b67f6704b1 upstream.

Marek reported that when BlackMagic UltraStudio device is connected the
kernel repeatedly tries to enable lane bonding without success making
the device non-functional. It looks like the device does not have lane 1
connected at all so even though it is enabled we should not try to bond
the lanes. For this reason check that lane 1 is in fact CL0 (connected,
active) before attempting to bond the lanes.

Reported-by: Marek Šanta <teslan223@gmail.com>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217737
Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agothunderbolt: Workaround an IOMMU fault on certain systems with Intel Maple Ridge
Mika Westerberg [Fri, 18 Aug 2023 12:27:46 +0000 (15:27 +0300)]
thunderbolt: Workaround an IOMMU fault on certain systems with Intel Maple Ridge

commit 582620d9f6b352552bc9a3316fe2b1c3acd8742d upstream.

On some systems the IOMMU blocks the first couple of driver ready
messages to the connection manager firmware as can be seen in below
excerpts:

  thunderbolt 0000:06:00.0: AMD-Vi: Event logged [IO_PAGE_FAULT domain=0x0010 address=0xbb0e3400 flags=0x0020]

or

  DMAR: DRHD: handling fault status reg 2
  DMAR: [DMA Write] Request device [04:00.0] PASID ffffffff fault addr 69974000 [fault reason 05] PTE Write access is not set

The reason is unknown and hard to debug because we were not able to
reproduce this locally. This only happens on certain systems with Intel
Maple Ridge Thunderbolt controller. If there is a device connected when
the driver is loaded the issue does not happen either. Only when there
is nothing connected (so typically when the system is booted up).

We can work this around by sending the driver ready several times. After
a couple of retries the message goes through and the controller works
just fine. For this reason make the number of retries a parameter for
icm_request() and then for Maple Ridge (and Titan Ridge as they us the
same function but this should not matter) increase number of retries
while shortening the timeout accordingly.

Reported-by: Werner Sembach <wse@tuxedocomputers.com>
Reported-by: Konrad J Hambrick <kjhambrick@gmail.com>
Reported-by: Calvin Walton <calvin.walton@kepstin.ca>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=214259
Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agomcb: remove is_added flag from mcb_device struct
Jorge Sanjuan Garcia [Wed, 6 Sep 2023 11:49:26 +0000 (11:49 +0000)]
mcb: remove is_added flag from mcb_device struct

commit 0f28ada1fbf0054557cddcdb93ad17f767105208 upstream.

When calling mcb_bus_add_devices(), both mcb devices and the mcb
bus will attempt to attach a device to a driver because they share
the same bus_type. This causes an issue when trying to cast the
container of the device to mcb_device struct using to_mcb_device(),
leading to a wrong cast when the mcb_bus is added. A crash occurs
when freing the ida resources as the bus numbering of mcb_bus gets
confused with the is_added flag on the mcb_device struct.

The only reason for this cast was to keep an is_added flag on the
mcb_device struct that does not seem necessary. The function
device_attach() handles already bound devices and the mcb subsystem
does nothing special with this is_added flag so remove it completely.

Fixes: 18d288198099 ("mcb: Correctly initialize the bus's device")
Cc: stable <stable@kernel.org>
Signed-off-by: Jorge Sanjuan Garcia <jorge.sanjuangarcia@duagon.com>
Co-developed-by: Jose Javier Rodriguez Barbarin <JoseJavier.Rodriguez@duagon.com>
Signed-off-by: Jose Javier Rodriguez Barbarin <JoseJavier.Rodriguez@duagon.com>
Link: https://lore.kernel.org/r/20230906114901.63174-2-JoseJavier.Rodriguez@duagon.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agox86/cpu: Fix AMD erratum #1485 on Zen4-based CPUs
Borislav Petkov (AMD) [Sat, 7 Oct 2023 10:57:02 +0000 (12:57 +0200)]
x86/cpu: Fix AMD erratum #1485 on Zen4-based CPUs

commit f454b18e07f518bcd0c05af17a2239138bff52de upstream.

Fix erratum #1485 on Zen4 parts where running with STIBP disabled can
cause an #UD exception. The performance impact of the fix is negligible.

Reported-by: René Rebe <rene@exactcode.de>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Tested-by: René Rebe <rene@exactcode.de>
Cc: <stable@kernel.org>
Link: https://lore.kernel.org/r/D99589F4-BC5D-430B-87B2-72C20370CF57@exactcode.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoperf/x86/lbr: Filter vsyscall addresses
JP Kobryn [Fri, 6 Oct 2023 18:57:26 +0000 (11:57 -0700)]
perf/x86/lbr: Filter vsyscall addresses

commit e53899771a02f798d436655efbd9d4b46c0f9265 upstream.

We found that a panic can occur when a vsyscall is made while LBR sampling
is active. If the vsyscall is interrupted (NMI) for perf sampling, this
call sequence can occur (most recent at top):

    __insn_get_emulate_prefix()
    insn_get_emulate_prefix()
    insn_get_prefixes()
    insn_get_opcode()
    decode_branch_type()
    get_branch_type()
    intel_pmu_lbr_filter()
    intel_pmu_handle_irq()
    perf_event_nmi_handler()

Within __insn_get_emulate_prefix() at frame 0, a macro is called:

    peek_nbyte_next(insn_byte_t, insn, i)

Within this macro, this dereference occurs:

    (insn)->next_byte

Inspecting registers at this point, the value of the next_byte field is the
address of the vsyscall made, for example the location of the vsyscall
version of gettimeofday() at 0xffffffffff600000. The access to an address
in the vsyscall region will trigger an oops due to an unhandled page fault.

To fix the bug, filtering for vsyscalls can be done when
determining the branch type. This patch will return
a "none" branch if a kernel address if found to lie in the
vsyscall region.

Suggested-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: JP Kobryn <inwardvessel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoksmbd: not allow to open file if delelete on close bit is set
Namjae Jeon [Fri, 6 Oct 2023 01:41:36 +0000 (10:41 +0900)]
ksmbd: not allow to open file if delelete on close bit is set

commit f43328357defc0dc9d28dbd06dc3361fd2b22e28 upstream.

Cthon test fail with the following error.

check for proper open/unlink operation
nfsjunk files before unlink:
  -rwxr-xr-x 1 root root 0  9월 25 11:03 ./nfs2y8Jm9
./nfs2y8Jm9 open; unlink ret = 0
nfsjunk files after unlink:
  -rwxr-xr-x 1 root root 0  9월 25 11:03 ./nfs2y8Jm9
data compare ok
nfsjunk files after close:
  ls: cannot access './nfs2y8Jm9': No such file or directory
special tests failed

Cthon expect to second unlink failure when file is already unlinked.
ksmbd can not allow to open file if flags of ksmbd inode is set with
S_DEL_ON_CLS flags.

Cc: stable@vger.kernel.org
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoACPI: EC: Add quirk for the HP Pavilion Gaming 15-dk1xxx
Hans de Goede [Wed, 20 Sep 2023 13:05:06 +0000 (15:05 +0200)]
ACPI: EC: Add quirk for the HP Pavilion Gaming 15-dk1xxx

commit cd4aece493f99f95d41edcce32927d70a5dde923 upstream.

Added GPE quirk entry for the HP Pavilion Gaming 15-dk1xxx.
There is a quirk entry for 2 15-c..... laptops, this is
for a new version which has 15-dk1xxx as identifier.

This fixes the LID switch and rfkill and brightness hotkeys
not working.

Closes: https://github.com/systemd/systemd/issues/28942
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoACPI: resource: Skip IRQ override on ASUS ExpertBook B1402CBA
Hans de Goede [Tue, 12 Sep 2023 10:08:27 +0000 (12:08 +0200)]
ACPI: resource: Skip IRQ override on ASUS ExpertBook B1402CBA

commit c1ed72171ed580fbf159e703b77685aa4b0d0df5 upstream.

Like various other ASUS ExpertBook-s, the ASUS ExpertBook B1402CBA
has an ACPI DSDT table that describes IRQ 1 as ActiveLow while
the kernel overrides it to EdgeHigh.

This prevents the keyboard from working. To fix this issue, add this laptop
to the skip_override_table so that the kernel does not override IRQ 1.

Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217901
Cc: stable@vger.kernel.org
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodrm/amd/display: Don't set dpms_off for seamless boot
Daniel Miess [Fri, 29 Sep 2023 17:04:33 +0000 (13:04 -0400)]
drm/amd/display: Don't set dpms_off for seamless boot

commit 23645bca98304a2772f0de96f97370dd567d0ae6 upstream.

[Why]
eDPs fail to light up with seamless boot enabled

[How]
When seamless boot is enabled don't configure dpms_off
in disable_vbios_mode_if_required.

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Cc: Mario Limonciello <mario.limonciello@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Daniel Miess <daniel.miess@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodrm/amdgpu: add missing NULL check
Christian König [Fri, 6 Oct 2023 12:04:04 +0000 (14:04 +0200)]
drm/amdgpu: add missing NULL check

commit ff89f064dca38e2203790bf876cc7756b8ab2961 upstream.

bo->tbo.resource can easily be NULL here.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2902
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
CC: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodrm/atomic-helper: relax unregistered connector check
Simon Ser [Thu, 5 Oct 2023 13:16:32 +0000 (13:16 +0000)]
drm/atomic-helper: relax unregistered connector check

commit 2b7947bd32e243c52870d54141d3b4ea6775e63d upstream.

The driver might pull connectors which weren't submitted by
user-space into the atomic state. For instance,
intel_dp_mst_atomic_master_trans_check() pulls in connectors
sharing the same DP-MST stream. However, if the connector is
unregistered, this later fails with:

    [  559.425658] i915 0000:00:02.0: [drm:drm_atomic_helper_check_modeset] [CONNECTOR:378:DP-7] is not registered

Skip the unregistered connector check to allow user-space to turn
off connectors one-by-one.

See this wlroots issue:
https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3407

Previous discussion:
https://lore.kernel.org/intel-gfx/Y6GX7z17WmDSKwta@ideak-desk.fi.intel.com/

Signed-off-by: Simon Ser <contact@emersion.fr>
Cc: stable@vger.kernel.org
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231005131623.114379-1-contact@emersion.fr
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoarm64: dts: mediatek: mt8195-demo: update and reorder reserved memory regions
Macpaul Lin [Tue, 3 Oct 2023 11:13:45 +0000 (13:13 +0200)]
arm64: dts: mediatek: mt8195-demo: update and reorder reserved memory regions

commit 6cd2a30b96a4b2d270bc1ef1611429dc3fa63327 upstream.

The dts file of the MediaTek MT8195 demo board has been updated to include
new reserved memory regions.
These reserved memory regions are:
 - SCP
 - VPU,
 - Sound DMA
 - APU.

These regions are defined with the "shared-dma-pool" compatible property.
In addition, the existing reserved memory regions have been reordered by
their addresses to improve readability and maintainability of the DTS
file.

Cc: stable@vger.kernel.org # 6.1, 6.4, 6.5
Fixes: e4a417520101 ("arm64: dts: mediatek: mt8195-demo: fix the memory size of node secmon")
Signed-off-by: Macpaul Lin <macpaul.lin@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20230905034511.11232-2-macpaul.lin@mediatek.com
Link: https://lore.kernel.org/r/20231003-mediatek-fixes-v6-7-v1-3-dad7cd62a8ff@collabora.com
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoarm64: dts: mediatek: mt8195-demo: fix the memory size to 8GB
Macpaul Lin [Tue, 3 Oct 2023 11:13:44 +0000 (13:13 +0200)]
arm64: dts: mediatek: mt8195-demo: fix the memory size to 8GB

commit 25389c03c21c9587dd21c768d1cbfa514a3ca211 upstream.

The onboard dram of mt8195-demo board is 8GB.

Cc: stable@vger.kernel.org # 6.1, 6.4, 6.5
Fixes: 6147314aeedc ("arm64: dts: mediatek: Add device-tree for MT8195 Demo board")
Signed-off-by: Macpaul Lin <macpaul.lin@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20230905034511.11232-1-macpaul.lin@mediatek.com
Link: https://lore.kernel.org/r/20231003-mediatek-fixes-v6-7-v1-2-dad7cd62a8ff@collabora.com
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: addac: Kconfig: update ad74413r selections
Antoniu Miclaus [Tue, 12 Sep 2023 08:54:21 +0000 (11:54 +0300)]
iio: addac: Kconfig: update ad74413r selections

commit b120dd3a15582fb7a959cecb05e4d9814fcba386 upstream.

Building ad74413r without selecting IIO_BUFFER and
IIO_TRIGGERED_BUFFER generates error with respect to the iio trigger
functions that are used within the driver.
Update the Kconfig accordingly.

Fixes: fea251b6a5db ("iio: addac: add AD74413R driver")
Signed-off-by: Antoniu Miclaus <antoniu.miclaus@analog.com>
Link: https://lore.kernel.org/r/20230912085421.51102-1-antoniu.miclaus@analog.com
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: pressure: ms5611: ms5611_prom_is_valid false negative bug
Alexander Zangerl [Wed, 20 Sep 2023 00:01:10 +0000 (10:01 +1000)]
iio: pressure: ms5611: ms5611_prom_is_valid false negative bug

commit fd39d9668f2ce9f4b05ad55e8c8d80c098073e0b upstream.

The ms5611 driver falsely rejects lots of MS5607-02BA03-50 chips
with "PROM integrity check failed" because it doesn't accept a prom crc
value of zero as legitimate.

According to the datasheet for this chip (and the manufacturer's
application note about the PROM CRC), none of the possible values for the
CRC are excluded - but the current code in ms5611_prom_is_valid() ends with

return crc_orig != 0x0000 && crc == crc_orig

Discussed with the driver author (Tomasz Duszynski) and he indicated that
at that time (2015) he was dealing with some faulty chip samples which
returned blank data under some circumstances and/or followed example code
which indicated CRC zero being bad.

As far as I can tell this exception should not be applied anymore; We've
got a few hundred custom boards here with this chip where large numbers
of the prom have a legitimate CRC value 0, and do work fine, but which the
current driver code wrongly rejects.

Signed-off-by: Alexander Zangerl <az@breathe-safe.com>
Fixes: c0644160a8b5 ("iio: pressure: add support for MS5611 pressure and temperature sensor")
Link: https://lore.kernel.org/r/2535-1695168070.831792@Ze3y.dhYT.s3fx
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: pressure: dps310: Adjust Timeout Settings
Lakshmi Yadlapati [Tue, 29 Aug 2023 18:02:22 +0000 (13:02 -0500)]
iio: pressure: dps310: Adjust Timeout Settings

commit 901a293fd96fb9bab843ba4cc7be3094a5aa7c94 upstream.

The DPS310 sensor chip has been encountering intermittent errors while
reading the sensor device across various system designs. This issue causes
the chip to become "stuck," preventing the indication of "ready" status
for pressure and temperature measurements in the MEAS_CFG register.

To address this issue, this commit fixes the timeout settings to improve
sensor stability:
- After sending a reset command to the chip, the timeout has been extended
  from 2.5 ms to 15 ms, aligning with the DPS310 specification.
- The read timeout value of the MEAS_CFG register has been adjusted from
  20ms to 30ms to match the specification.

Signed-off-by: Lakshmi Yadlapati <lakshmiy@us.ibm.com>
Fixes: 7b4ab4abcea4 ("iio: pressure: dps310: Reset chip after timeout")
Link: https://lore.kernel.org/r/20230829180222.3431926-2-lakshmiy@us.ibm.com
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: admv1013: add mixer_vgate corner cases
Antoniu Miclaus [Mon, 7 Aug 2023 14:38:05 +0000 (17:38 +0300)]
iio: admv1013: add mixer_vgate corner cases

commit 287d998af24326b009ae0956820a3188501b34a0 upstream.

Include the corner cases in the computation of the MIXER_VGATE register
value.

According to the datasheet: The MIXER_VGATE values follows the VCM such
as, that for a 0V to 1.8V VCM, MIXER_VGATE = 23.89 VCM + 81, and for a >
1.8V to 2.6V VCM, MIXER_VGATE = 23.75 VCM + 1.25.

Fixes: da35a7b526d9 ("iio: frequency: admv1013: add support for ADMV1013")
Signed-off-by: Antoniu Miclaus <antoniu.miclaus@analog.com>
Reviewed-by: Nuno Sa <nuno.sa@analog.com>
Link: https://lore.kernel.org/r/20230807143806.6954-1-antoniu.miclaus@analog.com
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: dac: ad3552r: Correct device IDs
Marcelo Schmitt [Thu, 3 Aug 2023 19:56:23 +0000 (16:56 -0300)]
iio: dac: ad3552r: Correct device IDs

commit 9a85653ed3b9a9b7b31d95a34b64b990c3d33ca1 upstream.

Device IDs for AD3542R and AD3552R were swapped leading to unintended
collection of DAC output ranges being used for each design.
Change device ID values so they are correct for each DAC chip.

Fixes: 8f2b54824b28 ("drivers:iio:dac: Add AD3552R driver support")
Signed-off-by: Marcelo Schmitt <marcelo.schmitt1@gmail.com>
Reported-by: Chandrakant Minajigi <Chandrakant.Minajigi@analog.com>
Link: https://lore.kernel.org/r/011f480220799fbfabdd53896f8a2f251ad995ad.1691091324.git.marcelo.schmitt1@gmail.com
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: adc: imx8qxp: Fix address for command buffer registers
Philipp Rossak [Mon, 4 Sep 2023 22:02:04 +0000 (00:02 +0200)]
iio: adc: imx8qxp: Fix address for command buffer registers

commit 850101b3598277794f92a9e363a60a66e0d42890 upstream.

The ADC Command Buffer Register high and low are currently pointing to
the wrong address and makes it impossible to perform correct
ADC measurements over all channels.

According to the datasheet of the imx8qxp the ADC_CMDL register starts
at address 0x100 and the ADC_CMDH register starts at address 0x104.

This bug seems to be in the kernel since the introduction of this
driver.

This can be observed by checking all raw voltages of the adc and they
are all nearly identical:

cat /sys/bus/iio/devices/iio\:device0/in_voltage*_raw
3498
3494
3491
3491
3489
3490
3490
3490

Fixes: 1e23dcaa1a9fa ("iio: imx8qxp-adc: Add driver support for NXP IMX8QXP ADC")
Signed-off-by: Philipp Rossak <embed3d@gmail.com>
Acked-by: Haibo Chen <haibo.chen@nxp.com>
Link: https://lore.kernel.org/r/20230904220204.23841-1-embed3d@gmail.com
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: imu: bno055: Fix missing Kconfig dependencies
Jonathan Cameron [Sun, 3 Sep 2023 11:30:52 +0000 (12:30 +0100)]
iio: imu: bno055: Fix missing Kconfig dependencies

commit c9b9cfe7d342683f624a89c3b617be18aff879e8 upstream.

This driver uses IIO triggered buffers so it needs to select them in
Kconfig.

on riscv-32bit:

/opt/crosstool/gcc-13.2.0-nolibc/riscv32-linux/bin/riscv32-linux-ld: drivers/iio/imu/bno055/bno055.o: in function `.L367':
bno055.c:(.text+0x2c96): undefined reference to `devm_iio_triggered_buffer_setup_ext'

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Closes: https://lore.kernel.org/linux-next/40566b4b-3950-81fe-ff14-871d8c447627@infradead.org/
Fixes: 4aefe1c2bd0c ("iio: imu: add Bosch Sensortec BNO055 core driver")
Cc: Andrea Merello <andrea.merello@iit.it>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Tested-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230903113052.846298-1-jic23@kernel.org
Cc: <Stable@vger.kernel.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agoiio: pressure: bmp280: Fix NULL pointer exception
Phil Elwell [Fri, 11 Aug 2023 15:58:29 +0000 (16:58 +0100)]
iio: pressure: bmp280: Fix NULL pointer exception

commit 85dfb43bf69281adb1f345dfd9a39faf2e5a718d upstream.

The bmp085 EOC IRQ support is optional, but the driver's common probe
function queries the IRQ properties whether or not it exists, which
can trigger a NULL pointer exception. Avoid any exception by making
the query conditional on the possession of a valid IRQ.

Fixes: aae953949651 ("iio: pressure: bmp280: add support for BMP085 EOC interrupt")
Signed-off-by: Phil Elwell <phil@raspberrypi.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20230811155829.51208-1-phil@raspberrypi.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: musb: Modify the "HWVers" register address
Xingxing Luo [Fri, 22 Sep 2023 07:59:29 +0000 (15:59 +0800)]
usb: musb: Modify the "HWVers" register address

commit 6658a62e1ddf726483cb2d8bf45ea3f9bd533074 upstream.

musb HWVers rgister address is not 0x69, if we operate the
wrong address 0x69, it will cause a kernel crash, because
there is no register corresponding to this address in the
additional control register of musb. In fact, HWVers has
been defined in musb_register.h, and the name is
"MUSB_HWVERS", so We need to use this macro instead of 0x69.

Fixes: c2365ce5d5a0 ("usb: musb: replace hard coded registers with defines")
Cc: stable@vger.kernel.org
Signed-off-by: Xingxing Luo <xingxing.luo@unisoc.com>
Link: https://lore.kernel.org/r/20230922075929.31074-1-xingxing.luo@unisoc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: musb: Get the musb_qh poniter after musb_giveback
Xingxing Luo [Tue, 19 Sep 2023 03:30:55 +0000 (11:30 +0800)]
usb: musb: Get the musb_qh poniter after musb_giveback

commit 33d7e37232155aadebe4145dcc592f00dabd7a2b upstream.

When multiple threads are performing USB transmission, musb->lock will be
unlocked when musb_giveback is executed. At this time, qh may be released
in the dequeue process in other threads, resulting in a wild pointer, so
it needs to be here get qh again, and judge whether qh is NULL, and when
dequeue, you need to set qh to NULL.

Fixes: dbac5d07d13e ("usb: musb: host: don't start next rx urb if current one failed")
Cc: stable@vger.kernel.org
Signed-off-by: Xingxing Luo <xingxing.luo@unisoc.com>
Link: https://lore.kernel.org/r/20230919033055.14085-1-xingxing.luo@unisoc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: hub: Guard against accesses to uninitialized BOS descriptors
Ricardo Cañuelo [Wed, 30 Aug 2023 10:04:18 +0000 (12:04 +0200)]
usb: hub: Guard against accesses to uninitialized BOS descriptors

commit f74a7afc224acd5e922c7a2e52244d891bbe44ee upstream.

Many functions in drivers/usb/core/hub.c and drivers/usb/core/hub.h
access fields inside udev->bos without checking if it was allocated and
initialized. If usb_get_bos_descriptor() fails for whatever
reason, udev->bos will be NULL and those accesses will result in a
crash:

BUG: kernel NULL pointer dereference, address: 0000000000000018
PGD 0 P4D 0
Oops: 0000 [#1] PREEMPT SMP NOPTI
CPU: 5 PID: 17818 Comm: kworker/5:1 Tainted: G W 5.15.108-18910-gab0e1cb584e1 #1 <HASH:1f9e 1>
Hardware name: Google Kindred/Kindred, BIOS Google_Kindred.12672.413.0 02/03/2021
Workqueue: usb_hub_wq hub_event
RIP: 0010:hub_port_reset+0x193/0x788
Code: 89 f7 e8 20 f7 15 00 48 8b 43 08 80 b8 96 03 00 00 03 75 36 0f b7 88 92 03 00 00 81 f9 10 03 00 00 72 27 48 8b 80 a8 03 00 00 <48> 83 78 18 00 74 19 48 89 df 48 8b 75 b0 ba 02 00 00 00 4c 89 e9
RSP: 0018:ffffab740c53fcf8 EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffffa1bc5f678000 RCX: 0000000000000310
RDX: fffffffffffffdff RSI: 0000000000000286 RDI: ffffa1be9655b840
RBP: ffffab740c53fd70 R08: 00001b7d5edaa20c R09: ffffffffb005e060
R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000000
R13: ffffab740c53fd3e R14: 0000000000000032 R15: 0000000000000000
FS: 0000000000000000(0000) GS:ffffa1be96540000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000018 CR3: 000000022e80c005 CR4: 00000000003706e0
Call Trace:
hub_event+0x73f/0x156e
? hub_activate+0x5b7/0x68f
process_one_work+0x1a2/0x487
worker_thread+0x11a/0x288
kthread+0x13a/0x152
? process_one_work+0x487/0x487
? kthread_associate_blkcg+0x70/0x70
ret_from_fork+0x1f/0x30

Fall back to a default behavior if the BOS descriptor isn't accessible
and skip all the functionalities that depend on it: LPM support checks,
Super Speed capabilitiy checks, U1/U2 states setup.

Signed-off-by: Ricardo Cañuelo <ricardo.canuelo@collabora.com>
Cc: stable <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20230830100418.1952143-1-ricardo.canuelo@collabora.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: cdns3: Modify the return value of cdns_set_active () to void when CONFIG_PM_SLEE...
Xiaolei Wang [Tue, 26 Sep 2023 07:53:33 +0000 (15:53 +0800)]
usb: cdns3: Modify the return value of cdns_set_active () to void when CONFIG_PM_SLEEP is disabled

commit 9f35d612da5592f1bf1cae44ec1e023df37bea12 upstream.

The return type of cdns_set_active () is inconsistent
depending on whether CONFIG_PM_SLEEP is enabled, so the
return value is modified to void type.

Reported-by: Pavel Machek <pavel@denx.de>
Closes: https://lore.kernel.org/all/ZP7lIKUzD68XA91j@duo.ucw.cz/
Fixes: 2319b9c87fe2 ("usb: cdns3: Put the cdns set active part outside the spin lock")
Cc: stable@vger.kernel.org
Signed-off-by: Xiaolei Wang <xiaolei.wang@windriver.com>
Reviewed-by: Pavel Machek <pavel@denx.de>
Reviewed-by: Roger Quadros <rogerq@kernel.org>
Acked-by: Peter Chen <peter.chen@kernel.org>
Link: https://lore.kernel.org/r/20230926075333.1791011-1-xiaolei.wang@windriver.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: dwc3: Soft reset phy on probe for host
Thinh Nguyen [Wed, 13 Sep 2023 00:52:15 +0000 (00:52 +0000)]
usb: dwc3: Soft reset phy on probe for host

commit 8bea147dfdf823eaa8d3baeccc7aeb041b41944b upstream.

When there's phy initialization, we need to initiate a soft-reset
sequence. That's done through USBCMD.HCRST in the xHCI driver and its
initialization, However, the dwc3 driver may modify core configs before
the soft-reset. This may result in some connection instability. So,
ensure the phy is ready before the controller updates the GCTL.PRTCAPDIR
or other settings by issuing phy soft-reset.

Note that some host-mode configurations may not expose device registers
to initiate the controller soft-reset (via DCTL.CoreSftRst). So we reset
through GUSB3PIPECTL and GUSB2PHYCFG instead.

Cc: stable@vger.kernel.org
Fixes: e835c0a4e23c ("usb: dwc3: don't reset device side if dwc3 was configured as host-only")
Reported-by: Kenta Sato <tosainu.maple@gmail.com>
Closes: https://lore.kernel.org/linux-usb/ZPUciRLUcjDywMVS@debian.me/
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Tested-by: Kenta Sato <tosainu.maple@gmail.com>
Link: https://lore.kernel.org/r/70aea513215d273669152696cc02b20ddcdb6f1a.1694564261.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agonet: usb: dm9601: fix uninitialized variable use in dm9601_mdio_read
Javier Carrasco [Mon, 9 Oct 2023 22:26:14 +0000 (00:26 +0200)]
net: usb: dm9601: fix uninitialized variable use in dm9601_mdio_read

commit 8f8abb863fa5a4cc18955c6a0e17af0ded3e4a76 upstream.

syzbot has found an uninit-value bug triggered by the dm9601 driver [1].

This error happens because the variable res is not updated if the call
to dm_read_shared_word returns an error. In this particular case -EPROTO
was returned and res stayed uninitialized.

This can be avoided by checking the return value of dm_read_shared_word
and propagating the error if the read operation failed.

[1] https://syzkaller.appspot.com/bug?extid=1f53a30781af65d2c955

Cc: stable@vger.kernel.org
Signed-off-by: Javier Carrasco <javier.carrasco.cruz@gmail.com>
Reported-and-tested-by: syzbot+1f53a30781af65d2c955@syzkaller.appspotmail.com
Acked-by: Peter Korsgaard <peter@korsgaard.com>
Fixes: d0374f4f9c35cdfbee0 ("USB: Davicom DM9601 usbnet driver")
Link: https://lore.kernel.org/r/20231009-topic-dm9601_uninit_mdio_read-v2-1-f2fe39739b6c@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agousb: xhci: xhci-ring: Use sysdev for mapping bounce buffer
Wesley Cheng [Fri, 15 Sep 2023 14:31:05 +0000 (17:31 +0300)]
usb: xhci: xhci-ring: Use sysdev for mapping bounce buffer

commit 41a43013d2366db5b88b42bbcd8e8f040b6ccf21 upstream.

As mentioned in:
  commit 474ed23a6257 ("xhci: align the last trb before link if it is
easily splittable.")

A bounce buffer is utilized for ensuring that transfers that span across
ring segments are aligned to the EP's max packet size.  However, the device
that is used to map the DMA buffer to is currently using the XHCI HCD,
which does not carry any DMA operations in certain configrations.
Migration to using the sysdev entry was introduced for DWC3 based
implementations where the IOMMU operations are present.

Replace the reference to the controller device to sysdev instead.  This
allows the bounce buffer to be properly mapped to any implementations that
have an IOMMU involved.

cc: stable@vger.kernel.org
Fixes: 4c39d4b949d3 ("usb: xhci: use bus->sysdev for DMA configuration")
Signed-off-by: Wesley Cheng <quic_wcheng@quicinc.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20230915143108.1532163-2-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodmaengine: stm32-mdma: set in_flight_bytes in case CRQA flag is set
Amelie Delaunay [Wed, 4 Oct 2023 16:35:30 +0000 (18:35 +0200)]
dmaengine: stm32-mdma: set in_flight_bytes in case CRQA flag is set

commit 584970421725b7805db84714b857851fdf7203a9 upstream.

CRQA flag is set by hardware when the channel request become active and
the channel is enabled. It is cleared by hardware, when the channel request
is completed.
So when it is set, it means MDMA is transferring bytes.
This information is useful in case of STM32 DMA and MDMA chaining,
especially when the user pauses DMA before stopping it, to trig one last
MDMA transfer to get the latest bytes of the SRAM buffer to the
destination buffer.
STM32 DCMI driver can then use this to know if the last MDMA transfer in
case of chaining is done.

Fixes: 696874322771 ("dmaengine: stm32-mdma: add support to be triggered by STM32 DMA")
Signed-off-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20231004163531.2864160-3-amelie.delaunay@foss.st.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodmaengine: stm32-mdma: use Link Address Register to compute residue
Amelie Delaunay [Wed, 4 Oct 2023 16:35:29 +0000 (18:35 +0200)]
dmaengine: stm32-mdma: use Link Address Register to compute residue

commit a4b306eb83579c07b63dc65cd5bae53b7b4019d0 upstream.

Current implementation relies on curr_hwdesc index. But to keep this index
up to date, Block Transfer interrupt (BTIE) has to be enabled.
If it is not, curr_hwdesc is not updated, and then residue is not reliable.
Rely on Link Address Register instead. And disable BTIE interrupt
in stm32_mdma_setup_xfer() because it is no more needed in case of
_prep_slave_sg() to maintain curr_hwdesc up to date.
It avoids extra interrupts and also ensures a reliable residue. These
improvements are required for STM32 DCMI camera capture use case, which
need STM32 DMA and MDMA chaining for good performance.

Fixes: 696874322771 ("dmaengine: stm32-mdma: add support to be triggered by STM32 DMA")
Signed-off-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20231004163531.2864160-2-amelie.delaunay@foss.st.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodmaengine: stm32-dma: fix residue in case of MDMA chaining
Amelie Delaunay [Wed, 4 Oct 2023 15:50:24 +0000 (17:50 +0200)]
dmaengine: stm32-dma: fix residue in case of MDMA chaining

commit 67e13e89742c3b21ce177f612bf9ef32caae6047 upstream.

In case of MDMA chaining, DMA is configured in Double-Buffer Mode (DBM)
with two periods, but if transfer has been prepared with _prep_slave_sg(),
the transfer is not marked cyclic (=!chan->desc->cyclic). However, as DBM
is activated for MDMA chaining, residue computation must take into account
cyclic constraints.

With only two periods in MDMA chaining, and no update due to Transfer
Complete interrupt masked, n_sg is always 0. If DMA current memory address
(depending on SxCR.CT and SxM0AR/SxM1AR) does not correspond, it means n_sg
should be increased.
Then, the residue of the current period is the one read from SxNDTR and
should not be overwritten with the full period length.

Fixes: 723795173ce1 ("dmaengine: stm32-dma: add support to trigger STM32 MDMA")
Signed-off-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20231004155024.2609531-2-amelie.delaunay@foss.st.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodmaengine: stm32-dma: fix stm32_dma_prep_slave_sg in case of MDMA chaining
Amelie Delaunay [Wed, 4 Oct 2023 15:50:23 +0000 (17:50 +0200)]
dmaengine: stm32-dma: fix stm32_dma_prep_slave_sg in case of MDMA chaining

commit 2df467e908ce463cff1431ca1b00f650f7a514b4 upstream.

Current Target (CT) have to be reset when starting an MDMA chaining use
case, as Double Buffer mode is activated. It ensures the DMA will start
processing the first memory target (pointed with SxM0AR).

Fixes: 723795173ce1 ("dmaengine: stm32-dma: add support to trigger STM32 MDMA")
Signed-off-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20231004155024.2609531-1-amelie.delaunay@foss.st.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agodmaengine: stm32-mdma: abort resume if no ongoing transfer
Amelie Delaunay [Wed, 4 Oct 2023 16:35:28 +0000 (18:35 +0200)]
dmaengine: stm32-mdma: abort resume if no ongoing transfer

commit 81337b9a72dc58a5fa0ae8a042e8cb59f9bdec4a upstream.

chan->desc can be null, if transfer is terminated when resume is called,
leading to a NULL pointer when retrieving the hwdesc.
To avoid this case, check that chan->desc is not null and channel is
disabled (transfer previously paused or terminated).

Fixes: a4ffb13c8946 ("dmaengine: Add STM32 MDMA driver")
Signed-off-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20231004163531.2864160-1-amelie.delaunay@foss.st.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
9 months agotcp: enforce receive buffer memory limits by allowing the tcp window to shrink
mfreemon@cloudflare.com [Mon, 12 Jun 2023 03:05:24 +0000 (22:05 -0500)]
tcp: enforce receive buffer memory limits by allowing the tcp window to shrink

[ Upstream commit b650d953cd391595e536153ce30b4aab385643ac ]

Under certain circumstances, the tcp receive buffer memory limit
set by autotuning (sk_rcvbuf) is increased due to incoming data
packets as a result of the window not closing when it should be.
This can result in the receive buffer growing all the way up to
tcp_rmem[2], even for tcp sessions with a low BDP.

To reproduce:  Connect a TCP session with the receiver doing
nothing and the sender sending small packets (an infinite loop
of socket send() with 4 bytes of payload with a sleep of 1 ms
in between each send()).  This will cause the tcp receive buffer
to grow all the way up to tcp_rmem[2].

As a result, a host can have individual tcp sessions with receive
buffers of size tcp_rmem[2], and the host itself can reach tcp_mem
limits, causing the host to go into tcp memory pressure mode.

The fundamental issue is the relationship between the granularity
of the window scaling factor and the number of byte ACKed back
to the sender.  This problem has previously been identified in
RFC 7323, appendix F [1].

The Linux kernel currently adheres to never shrinking the window.

In addition to the overallocation of memory mentioned above, the
current behavior is functionally incorrect, because once tcp_rmem[2]
is reached when no remediations remain (i.e. tcp collapse fails to
free up any more memory and there are no packets to prune from the
out-of-order queue), the receiver will drop in-window packets
resulting in retransmissions and an eventual timeout of the tcp
session.  A receive buffer full condition should instead result
in a zero window and an indefinite wait.

In practice, this problem is largely hidden for most flows.  It
is not applicable to mice flows.  Elephant flows can send data
fast enough to "overrun" the sk_rcvbuf limit (in a single ACK),
triggering a zero window.

But this problem does show up for other types of flows.  Examples
are websockets and other type of flows that send small amounts of
data spaced apart slightly in time.  In these cases, we directly
encounter the problem described in [1].

RFC 7323, section 2.4 [2], says there are instances when a retracted
window can be offered, and that TCP implementations MUST ensure
that they handle a shrinking window, as specified in RFC 1122,
section 4.2.2.16 [3].  All prior RFCs on the topic of tcp window
management have made clear that sender must accept a shrunk window
from the receiver, including RFC 793 [4] and RFC 1323 [5].

This patch implements the functionality to shrink the tcp window
when necessary to keep the right edge within the memory limit by
autotuning (sk_rcvbuf).  This new functionality is enabled with
the new sysctl: net.ipv4.tcp_shrink_window

Additional information can be found at:
https://blog.cloudflare.com/unbounded-memory-usage-by-tcp-for-receive-buffers-and-how-we-fixed-it/

[1] https://www.rfc-editor.org/rfc/rfc7323#appendix-F
[2] https://www.rfc-editor.org/rfc/rfc7323#section-2.4
[3] https://www.rfc-editor.org/rfc/rfc1122#page-91
[4] https://www.rfc-editor.org/rfc/rfc793
[5] https://www.rfc-editor.org/rfc/rfc1323

Signed-off-by: Mike Freemon <mfreemon@cloudflare.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agoworkqueue: Override implicit ordered attribute in workqueue_apply_unbound_cpumask()
Waiman Long [Wed, 11 Oct 2023 02:48:42 +0000 (22:48 -0400)]
workqueue: Override implicit ordered attribute in workqueue_apply_unbound_cpumask()

[ Upstream commit ca10d851b9ad0338c19e8e3089e24d565ebfffd7 ]

Commit 5c0338c68706 ("workqueue: restore WQ_UNBOUND/max_active==1
to be ordered") enabled implicit ordered attribute to be added to
WQ_UNBOUND workqueues with max_active of 1. This prevented the changing
of attributes to these workqueues leading to fix commit 0a94efb5acbb
("workqueue: implicit ordered attribute should be overridable").

However, workqueue_apply_unbound_cpumask() was not updated at that time.
So sysfs changes to wq_unbound_cpumask has no effect on WQ_UNBOUND
workqueues with implicit ordered attribute. Since not all WQ_UNBOUND
workqueues are visible on sysfs, we are not able to make all the
necessary cpumask changes even if we iterates all the workqueue cpumasks
in sysfs and changing them one by one.

Fix this problem by applying the corresponding change made
to apply_workqueue_attrs_locked() in the fix commit to
workqueue_apply_unbound_cpumask().

Fixes: 5c0338c68706 ("workqueue: restore WQ_UNBOUND/max_active==1 to be ordered")
Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonfc: nci: assert requested protocol is valid
Jeremy Cline [Mon, 9 Oct 2023 20:00:54 +0000 (16:00 -0400)]
nfc: nci: assert requested protocol is valid

[ Upstream commit 354a6e707e29cb0c007176ee5b8db8be7bd2dee0 ]

The protocol is used in a bit mask to determine if the protocol is
supported. Assert the provided protocol is less than the maximum
defined so it doesn't potentially perform a shift-out-of-bounds and
provide a clearer error for undefined protocols vs unsupported ones.

Fixes: 6a2968aaf50c ("NFC: basic NCI protocol implementation")
Reported-and-tested-by: syzbot+0839b78e119aae1fec78@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=0839b78e119aae1fec78
Signed-off-by: Jeremy Cline <jeremy@jcline.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/20231009200054.82557-1-jeremy@jcline.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agopinctrl: renesas: rzn1: Enable missing PINMUX
Ralph Siemsen [Wed, 4 Oct 2023 20:00:08 +0000 (16:00 -0400)]
pinctrl: renesas: rzn1: Enable missing PINMUX

[ Upstream commit f055ff23c331f28aa4ace4b72dc56f63b9a726c8 ]

Enable pin muxing (eg. programmable function), so that the RZ/N1 GPIO
pins will be configured as specified by the pinmux in the DTS.

This used to be enabled implicitly via CONFIG_GENERIC_PINMUX_FUNCTIONS,
however that was removed, since the RZ/N1 driver does not call any of
the generic pinmux functions.

Fixes: 1308fb4e4eae14e6 ("pinctrl: rzn1: Do not select GENERIC_PIN{CTRL_GROUPS,MUX_FUNCTIONS}")
Signed-off-by: Ralph Siemsen <ralph.siemsen@linaro.org>
Reviewed-by: Miquel Raynal <miquel.raynal@bootlin.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20231004200008.1306798-1-ralph.siemsen@linaro.org
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonet/smc: Fix pos miscalculation in statistics
Nils Hoppmann [Mon, 9 Oct 2023 14:40:48 +0000 (16:40 +0200)]
net/smc: Fix pos miscalculation in statistics

[ Upstream commit a950a5921db450c74212327f69950ff03419483a ]

SMC_STAT_PAYLOAD_SUB(_smc_stats, _tech, key, _len, _rc) will calculate
wrong bucket positions for payloads of exactly 4096 bytes and
(1 << (m + 12)) bytes, with m == SMC_BUF_MAX - 1.

Intended bucket distribution:
Assume l == size of payload, m == SMC_BUF_MAX - 1.

Bucket 0                : 0 < l <= 2^13
Bucket n, 1 <= n <= m-1 : 2^(n+12) < l <= 2^(n+13)
Bucket m                : l > 2^(m+12)

Current solution:
_pos = fls64((l) >> 13)
[...]
_pos = (_pos < m) ? ((l == 1 << (_pos + 12)) ? _pos - 1 : _pos) : m

For l == 4096, _pos == -1, but should be _pos == 0.
For l == (1 << (m + 12)), _pos == m, but should be _pos == m - 1.

In order to avoid special treatment of these corner cases, the
calculation is adjusted. The new solution first subtracts the length by
one, and then calculates the correct bucket by shifting accordingly,
i.e. _pos = fls64((l - 1) >> 13), l > 0.
This not only fixes the issues named above, but also makes the whole
bucket assignment easier to follow.

Same is done for SMC_STAT_RMB_SIZE_SUB(_smc_stats, _tech, k, _len),
where the calculation of the bucket position is similar to the one
named above.

Fixes: e0e4b8fa5338 ("net/smc: Add SMC statistics support")
Suggested-by: Halil Pasic <pasic@linux.ibm.com>
Signed-off-by: Nils Hoppmann <niho@linux.ibm.com>
Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Wenjia Zhang <wenjia@linux.ibm.com>
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonet: nfc: fix races in nfc_llcp_sock_get() and nfc_llcp_sock_get_sn()
Eric Dumazet [Mon, 9 Oct 2023 12:31:10 +0000 (12:31 +0000)]
net: nfc: fix races in nfc_llcp_sock_get() and nfc_llcp_sock_get_sn()

[ Upstream commit 31c07dffafce914c1d1543c135382a11ff058d93 ]

Sili Luo reported a race in nfc_llcp_sock_get(), leading to UAF.

Getting a reference on the socket found in a lookup while
holding a lock should happen before releasing the lock.

nfc_llcp_sock_get_sn() has a similar problem.

Finally nfc_llcp_recv_snl() needs to make sure the socket
found by nfc_llcp_sock_from_sn() does not disappear.

Fixes: 8f50020ed9b8 ("NFC: LLCP late binding")
Reported-by: Sili Luo <rootlab@huawei.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Willy Tarreau <w@1wt.eu>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20231009123110.3735515-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonet/mlx5e: Again mutually exclude RX-FCS and RX-port-timestamp
Will Mortensen [Fri, 6 Oct 2023 05:37:06 +0000 (22:37 -0700)]
net/mlx5e: Again mutually exclude RX-FCS and RX-port-timestamp

[ Upstream commit da6192ca72d5ad913d109d43dc896290ad05d98f ]

Commit 1e66220948df8 ("net/mlx5e: Update rx ring hw mtu upon each rx-fcs
flag change") seems to have accidentally inverted the logic added in
commit 0bc73ad46a76 ("net/mlx5e: Mutually exclude RX-FCS and
RX-port-timestamp").

The impact of this is a little unclear since it seems the FCS scattered
with RX-FCS is (usually?) correct regardless.

Fixes: 1e66220948df8 ("net/mlx5e: Update rx ring hw mtu upon each rx-fcs flag change")
Tested-by: Charlotte Tan <charlotte@extrahop.com>
Reviewed-by: Charlotte Tan <charlotte@extrahop.com>
Cc: Adham Faris <afaris@nvidia.com>
Cc: Aya Levin <ayal@nvidia.com>
Cc: Tariq Toukan <tariqt@nvidia.com>
Cc: Moshe Shemesh <moshe@nvidia.com>
Cc: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Will Mortensen <will@extrahop.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20231006053706.514618-1-will@extrahop.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agoixgbe: fix crash with empty VF macvlan list
Dan Carpenter [Fri, 6 Oct 2023 12:53:09 +0000 (15:53 +0300)]
ixgbe: fix crash with empty VF macvlan list

[ Upstream commit 7b5add9af567c44e12196107f0fe106e194034fd ]

The adapter->vf_mvs.l list needs to be initialized even if the list is
empty.  Otherwise it will lead to crashes.

Fixes: a1cbb15c1397 ("ixgbe: Add macvlan support for VF")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Link: https://lore.kernel.org/r/ZSADNdIw8zFx1xw2@kadam
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonet/mlx5e: macsec: use update_pn flag instead of PN comparation
Radu Pirea (NXP OSS) [Thu, 5 Oct 2023 18:06:36 +0000 (21:06 +0300)]
net/mlx5e: macsec: use update_pn flag instead of PN comparation

[ Upstream commit fde2f2d7f23d39f2fc699ba6d91ac3f4a2e637ca ]

When updating the SA, use the new update_pn flags instead of comparing the
new PN with the initial one.

Comparing the initial PN value with the new value will allow the user
to update the SA using the initial PN value as a parameter like this:
$ ip macsec add macsec0 tx sa 0 pn 1 on key 00 \
ead3664f508eb06c40ac7104cdae4ce5
$ ip macsec set macsec0 tx sa 0 pn 1 off

Fixes: 8ff0ac5be144 ("net/mlx5: Add MACsec offload Tx command support")
Fixes: aae3454e4d4c ("net/mlx5e: Add MACsec offload Rx command support")
Signed-off-by: Radu Pirea (NXP OSS) <radu-nicolae.pirea@oss.nxp.com>
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonet: phy: mscc: macsec: reject PN update requests
Radu Pirea (NXP OSS) [Thu, 5 Oct 2023 18:06:35 +0000 (21:06 +0300)]
net: phy: mscc: macsec: reject PN update requests

[ Upstream commit e0a8c918daa58700609ebd45e3fcd49965be8bbc ]

Updating the PN is not supported.
Return -EINVAL if update_pn is true.

The following command succeeded, but it should fail because the driver
does not update the PN:
ip macsec set macsec0 tx sa 0 pn 232 on

Fixes: 28c5107aa904 ("net: phy: mscc: macsec support")
Signed-off-by: Radu Pirea (NXP OSS) <radu-nicolae.pirea@oss.nxp.com>
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonet: macsec: indicate next pn update when offloading
Radu Pirea (NXP OSS) [Thu, 5 Oct 2023 18:06:33 +0000 (21:06 +0300)]
net: macsec: indicate next pn update when offloading

[ Upstream commit 0412cc846a1ef38697c3f321f9b174da91ecd3b5 ]

Indicate next PN update using update_pn flag in macsec_context.
Offloaded MACsec implementations does not know whether or not the
MACSEC_SA_ATTR_PN attribute was passed for an SA update and assume
that next PN should always updated, but this is not always true.

The PN can be reset to its initial value using the following command:
$ ip macsec set macsec0 tx sa 0 off #octeontx2-pf case

Or, the update PN command will succeed even if the driver does not support
PN updates.
$ ip macsec set macsec0 tx sa 0 pn 1 on #mscc phy driver case

Comparing the initial PN with the new PN value is not a solution. When
the user updates the PN using its initial value the command will
succeed, even if the driver does not support it. Like this:
$ ip macsec add macsec0 tx sa 0 pn 1 on key 00 \
ead3664f508eb06c40ac7104cdae4ce5
$ ip macsec set macsec0 tx sa 0 pn 1 on #mlx5 case

Signed-off-by: Radu Pirea (NXP OSS) <radu-nicolae.pirea@oss.nxp.com>
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Stable-dep-of: e0a8c918daa5 ("net: phy: mscc: macsec: reject PN update requests")
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agonet: refine debug info in skb_checksum_help()
Eric Dumazet [Fri, 6 Oct 2023 17:33:54 +0000 (17:33 +0000)]
net: refine debug info in skb_checksum_help()

[ Upstream commit 26c29961b142444cd99361644c30fa1e9b3da6be ]

syzbot uses panic_on_warn.

This means that the skb_dump() I added in the blamed commit are
not even called.

Rewrite this so that we get the needed skb dump before syzbot crashes.

Fixes: eeee4b77dc52 ("net: add more debug info in skb_checksum_help()")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/r/20231006173355.2254983-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agobpf: Fix verifier log for async callback return values
David Vernet [Mon, 9 Oct 2023 16:14:13 +0000 (11:14 -0500)]
bpf: Fix verifier log for async callback return values

[ Upstream commit 829955981c557c7fc7416581c4cd68a8a0c28620 ]

The verifier, as part of check_return_code(), verifies that async
callbacks such as from e.g. timers, will return 0. It does this by
correctly checking that R0->var_off is in tnum_const(0), which
effectively checks that it's in a range of 0. If this condition fails,
however, it prints an error message which says that the value should
have been in (0x0; 0x1). This results in possibly confusing output such
as the following in which an async callback returns 1:

  At async callback the register R0 has value (0x1; 0x0) should have been in (0x0; 0x1)

The fix is easy -- we should just pass the tnum_const(0) as the correct
range to verbose_invalid_scalar(), which will then print the following:

  At async callback the register R0 has value (0x1; 0x0) should have been in (0x0; 0x0)

Fixes: bfc6bb74e4f1 ("bpf: Implement verifier support for validation of async callbacks.")
Signed-off-by: David Vernet <void@manifault.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20231009161414.235829-1-void@manifault.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agodrm/vmwgfx: fix typo of sizeof argument
Konstantin Meskhidze [Tue, 5 Sep 2023 10:02:03 +0000 (18:02 +0800)]
drm/vmwgfx: fix typo of sizeof argument

[ Upstream commit 39465cac283702a7d4a507a558db81898029c6d3 ]

Since size of 'header' pointer and '*header' structure is equal on 64-bit
machines issue probably didn't cause any wrong behavior. But anyway,
fixing typo is required.

Fixes: 7a73ba7469cb ("drm/vmwgfx: Use TTM handles instead of SIDs as user-space surface handles.")
Co-developed-by: Ivanov Mikhail <ivanov.mikhail1@huawei-partners.com>
Signed-off-by: Konstantin Meskhidze <konstantin.meskhidze@huawei.com>
Reviewed-by: Zack Rusin <zackr@vmware.com>
Signed-off-by: Zack Rusin <zackr@vmware.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230905100203.1716731-1-konstantin.meskhidze@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agoriscv, bpf: Sign-extend return values
Björn Töpel [Wed, 4 Oct 2023 12:07:05 +0000 (14:07 +0200)]
riscv, bpf: Sign-extend return values

[ Upstream commit 2f1b0d3d733169eb11680bfa97c266ae5e757148 ]

The RISC-V architecture does not expose sub-registers, and hold all
32-bit values in a sign-extended format [1] [2]:

  | The compiler and calling convention maintain an invariant that all
  | 32-bit values are held in a sign-extended format in 64-bit
  | registers. Even 32-bit unsigned integers extend bit 31 into bits
  | 63 through 32. Consequently, conversion between unsigned and
  | signed 32-bit integers is a no-op, as is conversion from a signed
  | 32-bit integer to a signed 64-bit integer.

While BPF, on the other hand, exposes sub-registers, and use
zero-extension (similar to arm64/x86).

This has led to some subtle bugs, where a BPF JITted program has not
sign-extended the a0 register (return value in RISC-V land), passed
the return value up the kernel, e.g.:

  | int from_bpf(void);
  |
  | long foo(void)
  | {
  |    return from_bpf();
  | }

Here, a0 would be 0xffff_ffff, instead of the expected
0xffff_ffff_ffff_ffff.

Internally, the RISC-V JIT uses a5 as a dedicated register for BPF
return values.

Keep a5 zero-extended, but explicitly sign-extend a0 (which is used
outside BPF land). Now that a0 (RISC-V ABI) and a5 (BPF ABI) differs,
a0 is only moved to a5 for non-BPF native calls (BPF_PSEUDO_CALL).

Fixes: 2353ecc6f91f ("bpf, riscv: add BPF JIT for RV64G")
Signed-off-by: Björn Töpel <bjorn@rivosinc.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://github.com/riscv/riscv-isa-manual/releases/download/riscv-isa-release-056b6ff-2023-10-02/unpriv-isa-asciidoc.pdf
Link: https://github.com/riscv-non-isa/riscv-elf-psabi-doc/releases/download/draft-20230929-e5c800e661a53efe3c2678d71a306323b60eb13b/riscv-abi.pdf
Link: https://lore.kernel.org/bpf/20231004120706.52848-2-bjorn@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agoriscv, bpf: Factor out emit_call for kernel and bpf context
Pu Lehui [Wed, 15 Feb 2023 13:52:03 +0000 (21:52 +0800)]
riscv, bpf: Factor out emit_call for kernel and bpf context

[ Upstream commit 0fd1fd0104954380477353aea29c347e85dff16d ]

The current emit_call function is not suitable for kernel function call as
it store return value to bpf R0 register. We can separate it out for common
use. Meanwhile, simplify judgment logic, that is, fixed function address
can use jal or auipc+jalr, while the unfixed can use only auipc+jalr.

Signed-off-by: Pu Lehui <pulehui@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Björn Töpel <bjorn@rivosinc.com>
Acked-by: Björn Töpel <bjorn@rivosinc.com>
Link: https://lore.kernel.org/bpf/20230215135205.1411105-3-pulehui@huaweicloud.com
Stable-dep-of: 2f1b0d3d7331 ("riscv, bpf: Sign-extend return values")
Signed-off-by: Sasha Levin <sashal@kernel.org>
9 months agoxen-netback: use default TX queue size for vifs
Roger Pau Monne [Thu, 5 Oct 2023 14:08:31 +0000 (16:08 +0200)]
xen-netback: use default TX queue size for vifs

[ Upstream commit 66cf7435a26917c0c4d6245ad9137e7606e84fdf ]

Do not set netback interfaces (vifs) default TX queue size to the ring size.
The TX queue size is not related to the ring size, and using the ring size (32)
as the queue size can lead to packet drops.  Note the TX side of the vif
interface in the netback domain is the one receiving packets to be injected
to the guest.

Do not explicitly set the TX queue length to any value when creating the
interface, and instead use the system default.  Note that the queue length can
also be adjusted at runtime.

Fixes: f942dc2552b8 ('xen network backend driver')
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
Reviewed-by: Ross Lagerwall <ross.lagerwall@citrix.com>
Acked-by: Wei Liu <wei.liu@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>