linux-stable.git/drivers/thunderbolt/switch.c, branch v6.6.78

thunderbolt: Fix NULL pointer dereference in tb_port_update_credits()

2024-10-04T14:30:04+00:00

commit d3d17e23d1a0d1f959b4fa55b35f1802d9c584fa upstream.

Olliver reported that his system crashes when plugging in Thunderbolt 1
device:

 BUG: kernel NULL pointer dereference, address: 0000000000000020
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: 0000 [#1] PREEMPT SMP NOPTI
 RIP: 0010:tb_port_do_update_credits+0x1b/0x130 [thunderbolt]
 Call Trace:
  
  ? __die+0x23/0x70
  ? page_fault_oops+0x171/0x4e0
  ? exc_page_fault+0x7f/0x180
  ? asm_exc_page_fault+0x26/0x30
  ? tb_port_do_update_credits+0x1b/0x130
  ? tb_switch_update_link_attributes+0x83/0xd0
  tb_switch_add+0x7a2/0xfe0
  tb_scan_port+0x236/0x6f0
  tb_handle_hotplug+0x6db/0x900
  process_one_work+0x171/0x340
  worker_thread+0x27b/0x3a0
  ? __pfx_worker_thread+0x10/0x10
  kthread+0xe5/0x120
  ? __pfx_kthread+0x10/0x10
  ret_from_fork+0x31/0x50
  ? __pfx_kthread+0x10/0x10
  ret_from_fork_asm+0x1b/0x30
  

This is due the fact that some Thunderbolt 1 devices only have one lane
adapter. Fix this by checking for the lane 1 before we read its credits.

Reported-by: Olliver Schinagl 
Closes: https://lore.kernel.org/linux-usb/c24c7882-6254-4e68-8f22-f3e8f65dc84f@schinagl.nl/
Fixes: 81af2952e606 ("thunderbolt: Add support for asymmetric link")
Cc: stable@vger.kernel.org
Cc: Gil Fine 
Signed-off-by: Mika Westerberg 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Send uevent after asymmetric/symmetric switch

2024-10-04T14:30:04+00:00

commit 5391bcfa56c79a891734e4d22aa0ca3217b86491 upstream.

We should send uevent to userspace whenever the link speed or width
changes but tb_switch_asym_enable() and tb_switch_asym_disable() set the
sw->link_width already so tb_switch_update_link_attributes() never
noticed the change.

Fix this so that we let tb_switch_update_link_attributes() update the
fields accordingly.

Fixes: 81af2952e606 ("thunderbolt: Add support for asymmetric link")
Reported-by: Pengfei Xu 
Tested-by: Pengfei Xu 
Signed-off-by: Mika Westerberg 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Add support for asymmetric link

2024-10-04T14:30:01+00:00

[ Upstream commit 81af2952e60603d12415e1a6fd200f8073a2ad8b ]

USB4 v2 spec defines a Gen 4 link that can operate as an aggregated
symmetric (80/80G) or asymmetric (120/40G). When the link is asymmetric,
the USB4 port on one side of the link operates with three TX lanes and
one RX lane, while the USB4 port on the opposite side of the link
operates with three RX lanes and one TX lane.

Add support for the asymmetric link and provide functions that can be
used to transition the link to asymmetric and back.

Signed-off-by: Gil Fine 
Co-developed-by: Mika Westerberg 
Signed-off-by: Mika Westerberg 
Signed-off-by: Qin Wan 
Signed-off-by: Alexandru Gagniuc 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Make is_gen4_link() available to the rest of the driver

2024-10-04T14:30:00+00:00

[ Upstream commit aa673d606078da36ebc379f041c794228ac08cb5 ]

Rework the function to return the link generation, update the name to
tb_port_get_link_generation(), and make available to the rest of the
driver. This is needed in the subsequent patches.

Signed-off-by: Gil Fine 
Signed-off-by: Mika Westerberg 
Signed-off-by: Qin Wan 
Signed-off-by: Alexandru Gagniuc 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Mark XDomain as unplugged when router is removed

2024-08-29T15:33:12+00:00

commit e2006140ad2e01a02ed0aff49cc2ae3ceeb11f8d upstream.

I noticed that when we do discrete host router NVM upgrade and it gets
hot-removed from the PCIe side as a result of NVM firmware authentication,
if there is another host connected with enabled paths we hang in tearing
them down. This is due to fact that the Thunderbolt networking driver
also tries to cleanup the paths and ends up blocking in
tb_disconnect_xdomain_paths() waiting for the domain lock.

However, at this point we already cleaned the paths in tb_stop() so
there is really no need for tb_disconnect_xdomain_paths() to do that
anymore. Furthermore it already checks if the XDomain is unplugged and
bails out early so take advantage of that and mark the XDomain as
unplugged when we remove the parent router.

Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Fix wake configurations after device unplug

2024-04-27T15:11:38+00:00

commit c38fa07dc69f0b9e6f43ecab96dc7861a70c827c upstream.

Currently we don't configure correctly the wake events after unplug of device
router. What can happen is that the downstream ports of host router will be
configured to wake on: USB4-wake and wake-on-disconnect, but not on
wake-on-connect. This may cause the later plugged device not to wake the
domain and fail in enumeration. Fix this by clearing downstream port's "USB4
Port is Configured" bit, after unplug of a device router.

Signed-off-by: Gil Fine 
Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Avoid notify PM core about runtime PM resume

2024-04-27T15:11:38+00:00

commit dcd12acaf384c30437fa5a9a1f71df06fc9835fd upstream.

Currently we notify PM core about occurred wakes after any resume. This
is not actually needed after resume from runtime suspend. Hence, notify
PM core about occurred wakes only after resume from system sleep. Also,
if the wake occurred in USB4 router upstream port, we don't notify the
PM core about it since it is not actually needed and can cause
unexpected autowake (e.g. if /sys/power/wakeup_count is used).

While there add the missing kernel-doc for tb_switch_resume().

Signed-off-by: Gil Fine 
Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Make tb_switch_reset() support Thunderbolt 2, 3 and USB4 routers

2024-04-27T15:11:35+00:00

commit ec8162b3f0683ae08a21f20517cf49272b07ee0b upstream.

Currently tb_switch_reset() only did something for Thunderbolt 1
devices. Expand this to support all generations, including USB4, and
both host and device routers.

Signed-off-by: Sanath S 
Signed-off-by: Mika Westerberg 
Cc: Mario Limonciello 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Introduce tb_port_reset()

2024-04-27T15:11:35+00:00

commit 01da6b99d49f60b1edead44e33569b1a2e9f49b7 upstream.

Introduce a function that issues Downstream Port Reset to a USB4 port.
This supports Thunderbolt 2, 3 and USB4 routers.

Signed-off-by: Sanath S 
Signed-off-by: Mika Westerberg 
Cc: Mario Limonciello 
Signed-off-by: Greg Kroah-Hartman

thunderbolt: Set lane bonding bit only for downstream port

2023-12-03T06:33:06+00:00

commit 24d85bb3be373b5831699bddf698b392bd2b904d upstream.

Fix the lane bonding procedure to follow the steps described in USB4
Connection Manager guide. Hence, set the lane bonding bit only for
downstream port. This is needed for certain ASMedia device, otherwise
lane bonding fails and the device disconnects.

Cc: stable@vger.kernel.org
Signed-off-by: Gil Fine 
Signed-off-by: Mika Westerberg 
Signed-off-by: Greg Kroah-Hartman