From: Mika Westerberg <mika.westerberg@linux.intel.com>
To: Basavaraj Natikar <Basavaraj.Natikar@amd.com>
Cc: andreas.noever@gmail.com, westeri@kernel.org,
YehezkelShB@gmail.com, linux-usb@vger.kernel.org,
Mario Limonciello <superm1@kernel.org>,
Sanath S <Sanath.S@amd.com>
Subject: Re: [PATCH v2] thunderbolt: Assert downstream port reset on shutdown
Date: Wed, 10 Jun 2026 08:27:34 +0200 [thread overview]
Message-ID: <20260610062734.GP2990@black.igk.intel.com> (raw)
In-Reply-To: <20260609104218.2770725-1-Basavaraj.Natikar@amd.com>
Hi,
On Tue, Jun 09, 2026 at 04:12:18PM +0530, Basavaraj Natikar wrote:
> On shutdown the connection manager tears down the router tree without
> signalling connected devices. A Thunderbolt 3 device directly connected
> to a USB4 host never receives a disconnect indication and during shutdown
> this can cause polling the dead link for up to 60 seconds. On some
> platforms this behavior leads to a warm reset instead of a shutdown due
> to this timeout.
>
> Fix this by asserting PORT_CS_19.DPR on each connected downstream port
> before tearing down the router tree. This drives SBTX low (USB4 spec
> section 6.9), causing the device to detect SBRX low and transition to
> Uninitialized Unplugged state immediately.
>
> Always do this on system shutdown/reboot by forcing tb->host_reset in the
> PCI ->shutdown callback. On plain driver unload only do it when the host
> router was actually reset on load (host_reset=1), since in that case the
> tunnels are not preserved across reload anyway; with host_reset=0 the
> tunnels are kept alive across unload/reload so the links are left intact.
> Restrict the reset to TBT3 devices.
>
> Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
> Co-developed-by: Sanath S <Sanath.S@amd.com>
> Signed-off-by: Sanath S <Sanath.S@amd.com>
> Signed-off-by: Basavaraj Natikar <Basavaraj.Natikar@amd.com>
> ---
> v2:
> - Add Reviewed-by tag from Mario Limonciello.
> - Restrict the DPR assertion to TBT3 devices, since the issue only
> affects TBT3 devices.
> - Only assert DPR on system shutdown/reboot (the PCI ->shutdown callback
> forces tb->host_reset) or when the host router was actually reset on
> load.
> - Reword comments and commit message to refer to the "router tree"
> instead of the "switch tree".
> ---
> drivers/thunderbolt/domain.c | 2 ++
> drivers/thunderbolt/nhi.c | 16 +++++++++++++++-
> drivers/thunderbolt/switch.c | 2 +-
> drivers/thunderbolt/tb.c | 20 ++++++++++++++++++++
> drivers/thunderbolt/tb.h | 1 +
> include/linux/thunderbolt.h | 6 ++++++
> 6 files changed, 45 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/thunderbolt/domain.c b/drivers/thunderbolt/domain.c
> --- a/drivers/thunderbolt/domain.c
> +++ b/drivers/thunderbolt/domain.c
> @@ -460,6 +460,8 @@ int tb_domain_add(struct tb *tb, bool reset)
> if (ret)
> goto err_ctl_stop;
>
> + tb->host_reset = reset;
> +
> /* Start the domain */
> if (tb->cm_ops->start) {
> ret = tb->cm_ops->start(tb, reset);
> diff --git a/drivers/thunderbolt/nhi.c b/drivers/thunderbolt/nhi.c
> --- a/drivers/thunderbolt/nhi.c
> +++ b/drivers/thunderbolt/nhi.c
> @@ -1436,6 +1436,20 @@ static void nhi_remove(struct pci_dev *pdev)
> nhi_shutdown(nhi);
> }
>
> +static void nhi_pci_shutdown(struct pci_dev *pdev)
> +{
> + struct tb *tb = pci_get_drvdata(pdev);
> +
> + /*
> + * On system shutdown/reboot force host_reset so the connection
> + * manager asserts DPR to signal disconnect before removing the
> + * router tree. Only TBT3 devices are reset. Plain driver unload
> + * uses ->remove (nhi_remove) and keeps the host_reset from load.
> + */
> + tb->host_reset = true;
> + nhi_remove(pdev);
> +}
There has been changes to support non-PCIe hosts so this does not apply on
top of my next branch.
> +
> /*
> * The tunneled pci bridges are siblings of us. Use resume_noirq to reenable
> * the tunnels asap. A corresponding pci quirk blocks the downstream bridges
> @@ -1558,7 +1572,7 @@ static struct pci_driver nhi_driver = {
> .id_table = nhi_ids,
> .probe = nhi_probe,
> .remove = nhi_remove,
> - .shutdown = nhi_remove,
> + .shutdown = nhi_pci_shutdown,
> .driver.pm = &nhi_pm_ops,
> };
>
> diff --git a/drivers/thunderbolt/switch.c b/drivers/thunderbolt/switch.c
> --- a/drivers/thunderbolt/switch.c
> +++ b/drivers/thunderbolt/switch.c
> @@ -704,7 +704,7 @@ int tb_port_disable(struct tb_port *port)
> return __tb_port_enable(port, false);
> }
>
> -static int tb_port_reset(struct tb_port *port)
Since you make it non-static please add kernel-doc as well.
> +int tb_port_reset(struct tb_port *port)
> {
> if (tb_switch_is_usb4(port->sw))
> return port->cap_usb4 ? usb4_port_reset(port) : 0;
> diff --git a/drivers/thunderbolt/tb.c b/drivers/thunderbolt/tb.c
> --- a/drivers/thunderbolt/tb.c
> +++ b/drivers/thunderbolt/tb.c
> @@ -2935,6 +2935,7 @@ static void tb_stop(struct tb *tb)
> struct tb_cm *tcm = tb_priv(tb);
> struct tb_tunnel *tunnel;
> struct tb_tunnel *n;
> + struct tb_port *port;
>
> cancel_delayed_work(&tcm->remove_work);
> /* tunnels are only present after everything has been initialized */
> @@ -2948,6 +2949,25 @@ static void tb_stop(struct tb *tb)
> tb_tunnel_deactivate(tunnel);
> tb_tunnel_put(tunnel);
> }
> + /*
> + * Signal disconnect to connected devices before the router tree is
> + * removed below. A Thunderbolt 3 device directly connected to a USB4
> + * host otherwise never receives a disconnect indication, leaving
> + * firmware to poll the dead link for up to ~60 s which on some
> + * platforms turns the shutdown into a warm reset. Asserting
> + * PORT_CS_19.DPR drives SBTX low (USB4 spec section 6.9) so the device
> + * detects SBRX low and goes to Uninitialized Unplugged immediately.
> + */
> + if (tb->host_reset) {
> + tb_switch_for_each_port(tb->root_switch, port) {
> + if (!tb_port_is_null(port) || !tb_port_has_remote(port))
> + continue;
> + if (tb_switch_is_usb4(port->remote->sw))
> + continue;
> + if (tb_port_reset(port))
> + tb_port_dbg(port, "DPR failed, continuing\n");
"downstream port reset failed, continuing\n"
> + }
> + }
> tb_switch_remove(tb->root_switch);
> tcm->hotplug_active = false; /* signal tb_handle_hotplug to quit */
> }
> diff --git a/drivers/thunderbolt/tb.h b/drivers/thunderbolt/tb.h
> --- a/drivers/thunderbolt/tb.h
> +++ b/drivers/thunderbolt/tb.h
> @@ -1102,6 +1102,7 @@ int tb_port_clear_counter(struct tb_port *port, int counter);
> int tb_port_unlock(struct tb_port *port);
> int tb_port_enable(struct tb_port *port);
> int tb_port_disable(struct tb_port *port);
> +int tb_port_reset(struct tb_port *port);
> int tb_port_alloc_in_hopid(struct tb_port *port, int hopid, int max_hopid);
> void tb_port_release_in_hopid(struct tb_port *port, int hopid);
> int tb_port_alloc_out_hopid(struct tb_port *port, int hopid, int max_hopid);
> diff --git a/include/linux/thunderbolt.h b/include/linux/thunderbolt.h
> --- a/include/linux/thunderbolt.h
> +++ b/include/linux/thunderbolt.h
> @@ -76,5 +76,10 @@
> * @index: Linux assigned domain number
> * @security_level: Current security level
> * @nboot_acl: Number of boot ACLs the domain supports
> + * @host_reset: Host router was reset on driver load, or forced on system
> + * shutdown/reboot. When set, tb_stop() asserts DPR on connected
> + * downstream ports to signal disconnect before tearing down the
> + * router tree. Only TBT3 devices are reset; USB4 routers are
Thunderbolt 3
And I think this should be part of struct tb_nhi instead.
> + * skipped.
> * @privdata: Private connection manager specific data
> */
> @@ -89,5 +94,6 @@ struct tb {
> int index;
> enum tb_security_level security_level;
> size_t nboot_acl;
> + bool host_reset;
> unsigned long privdata[];
> };
> --
> 2.34.1
prev parent reply other threads:[~2026-06-10 6:27 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-09 10:42 [PATCH v2] thunderbolt: Assert downstream port reset on shutdown Basavaraj Natikar
2026-06-10 6:27 ` Mika Westerberg [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260610062734.GP2990@black.igk.intel.com \
--to=mika.westerberg@linux.intel.com \
--cc=Basavaraj.Natikar@amd.com \
--cc=Sanath.S@amd.com \
--cc=YehezkelShB@gmail.com \
--cc=andreas.noever@gmail.com \
--cc=linux-usb@vger.kernel.org \
--cc=superm1@kernel.org \
--cc=westeri@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox