From: Mika Westerberg <mika.westerberg@linux.intel.com>
To: Sven Peter <sven@kernel.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Rob Herring <robh@kernel.org>,
Krzysztof Kozlowski <krzk+dt@kernel.org>,
Conor Dooley <conor+dt@kernel.org>,
Heikki Krogerus <heikki.krogerus@linux.intel.com>,
Janne Grunau <j@jannau.net>, Neal Gompa <neal@gompa.dev>,
Andreas Noever <andreas.noever@gmail.com>,
Mika Westerberg <westeri@kernel.org>,
Yehezkel Bernat <YehezkelShB@gmail.com>,
Philipp Zabel <p.zabel@pengutronix.de>,
Konrad Dybcio <konradybcio@kernel.org>,
linux-usb@vger.kernel.org, devicetree@vger.kernel.org,
linux-kernel@vger.kernel.org, asahi@lists.linux.dev,
linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 14/19] thunderbolt: Add QUIRK_NO_USB3_BW_ALLOC
Date: Tue, 1 Sep 2026 11:12:05 +0200 [thread overview]
Message-ID: <20260901091205.GE106095@black.igk.intel.com> (raw)
In-Reply-To: <20260830-b4-apple-soc-tbt-v1-14-44bc9348683c@kernel.org>
Hi,
On Sun, Aug 30, 2026 at 10:19:32PM +0200, Sven Peter wrote:
> The host routers on Apple Silicon do not implement USB3 bandwidth
> allocation registers: the CMR bit set in ADP_USB3_CS_2 by
> usb4_usb3_port_cm_request() is never acknowledged and every USB3 tunnel
> bringup fails there after the timeout:
>
> [ 49.259325] thunderbolt-apple-nhi 501f00000.nhi: 1:16: available bandwidth for new USB3 tunnel 18000/18000 Mb/s
> [ 49.265869] thunderbolt-apple-nhi 501f00000.nhi: 1:16: maximum required bandwidth for USB3 tunnel 9000 Mb/s
> [ 49.267894] thunderbolt-apple-nhi 501f00000.nhi: 0:4 <-> 1:16 (USB3): activating
> [ 49.269410] thunderbolt-apple-nhi 501f00000.nhi: 0:4 <-> 1:16 (USB3): allocating initial bandwidth 9000/9000 Mb/s
> [ 50.771764] thunderbolt-apple-nhi 501f00000.nhi: 1:16: USB3 tunnel activation failed, aborting
Okay this kind of makes sense. Apple is doing this only via software and
this is also what Linux needs to do with USB3 Gen T tunneling so we should
prepare for this.
> Add a quirk to skip bandwidth allocation entirely but keep tracking
> the initial allocation in software for the consumed bandwidth
> calculation. For completeness, discovered tunnels book 90% of the
> maximum link rate in software, matching tb_tunnel_alloc_usb3(), although
> Apple Silicon cannot have any pre-existing tunnels.
>
> Signed-off-by: Sven Peter <sven@kernel.org>
> ---
> drivers/thunderbolt/nhi.h | 1 +
> drivers/thunderbolt/tb.c | 8 ++++++
> drivers/thunderbolt/tb.h | 3 +++
> drivers/thunderbolt/tunnel.c | 59 +++++++++++++++++++++++++++++---------------
> 4 files changed, 51 insertions(+), 20 deletions(-)
>
> diff --git a/drivers/thunderbolt/nhi.h b/drivers/thunderbolt/nhi.h
> index c2d16ab6994e..afd9a11489a9 100644
> --- a/drivers/thunderbolt/nhi.h
> +++ b/drivers/thunderbolt/nhi.h
> @@ -169,6 +169,7 @@ struct tb_nhi_ops {
> #define QUIRK_E2E BIT(1)
> #define QUIRK_RESET_DMA_ON_TEARDOWN BIT(2)
> #define QUIRK_NO_DMA_PORT BIT(3)
> +#define QUIRK_NO_USB3_BW_ALLOC BIT(4)
This is not NHI thing but lives inside the USB4 domain so the quirk should
end up in drivers/thunderbolt/quirks.c.
For now I think that's enough but in the future we should provide software
mechanism that takes over the hardware if present, and then get rid of the
quirk alltogether.
> /*
> * Minimal number of vectors when we use MSI-X. Two for control channel
> diff --git a/drivers/thunderbolt/tb.c b/drivers/thunderbolt/tb.c
> index c43cdd2c2e2e..7ab53a30fe30 100644
> --- a/drivers/thunderbolt/tb.c
> +++ b/drivers/thunderbolt/tb.c
> @@ -3044,6 +3044,14 @@ static int tb_start(struct tb *tb, bool reset)
> */
> tb->root_switch->no_dma_port = tb->nhi->quirks & QUIRK_NO_DMA_PORT;
>
> + /*
> + * Apple Silicon host routers do not implement the USB3 bandwidth
> + * allocation registers: the CMR/HCA handshake is never acked and
> + * times out. Bandwidth for USB3 tunnels is only booked in software
> + * on these machines.
> + */
> + tb->root_switch->no_usb3_bw_alloc = tb->nhi->quirks & QUIRK_NO_USB3_BW_ALLOC;
> +
> ret = tb_switch_configure(tb->root_switch);
> if (ret) {
> tb_switch_put(tb->root_switch);
> diff --git a/drivers/thunderbolt/tb.h b/drivers/thunderbolt/tb.h
> index 0977cd690d7e..530b0e6c3a11 100644
> --- a/drivers/thunderbolt/tb.h
> +++ b/drivers/thunderbolt/tb.h
> @@ -140,6 +140,8 @@ struct tb_switch_tmu {
> * @nvm: Pointer to the NVM if the switch has one (%NULL otherwise)
> * @no_nvm_upgrade: Prevent NVM upgrade of this switch
> * @no_dma_port: Prevent adding the DMA port of this switch
> + * @no_usb3_bw_alloc: Host router does not implement the USB3 bandwidth
> + * allocation registers (ADP_USB3_CS_1..4)
> * @safe_mode: The switch is in safe-mode
> * @boot: Whether the switch was already authorized on boot or not
> * @rpm: The switch supports runtime PM
> @@ -196,6 +198,7 @@ struct tb_switch {
> struct tb_nvm *nvm;
> bool no_nvm_upgrade;
> bool no_dma_port;
> + bool no_usb3_bw_alloc;
> bool safe_mode;
> bool boot;
> bool rpm;
> diff --git a/drivers/thunderbolt/tunnel.c b/drivers/thunderbolt/tunnel.c
> index 7e8284575dff..9178120a6b66 100644
> --- a/drivers/thunderbolt/tunnel.c
> +++ b/drivers/thunderbolt/tunnel.c
> @@ -2268,24 +2268,40 @@ struct tb_tunnel *tb_tunnel_discover_usb3(struct tb *tb, struct tb_port *down,
> if (!tb_route(down->sw)) {
> int ret;
>
> - /*
> - * Read the initial bandwidth allocation for the first
> - * hop tunnel.
> - */
> - ret = usb4_usb3_port_allocated_bandwidth(down,
> - &tunnel->allocated_up, &tunnel->allocated_down);
> - if (ret)
> - goto err_deactivate;
> + tunnel->consumed_bandwidth = tb_usb3_consumed_bandwidth;
> +
> + if (down->sw->no_usb3_bw_alloc) {
> + /*
> + * The host router does not implement the bandwidth
> + * allocation registers and nothing can be read back
> + * here. Book 90% of the maximum link rate in software
> + * instead.
> + */
> + ret = tb_usb3_max_link_rate(tunnel->dst_port, down);
> + if (ret < 0)
> + goto err_deactivate;
> +
> + tunnel->allocated_up = ret * 90 / 100;
> + tunnel->allocated_down = tunnel->allocated_up;
> + } else {
> + /*
> + * Read the initial bandwidth allocation for the first
> + * hop tunnel.
> + */
> + ret = usb4_usb3_port_allocated_bandwidth(down,
> + &tunnel->allocated_up, &tunnel->allocated_down);
> + if (ret)
> + goto err_deactivate;
> +
> + tunnel->pre_activate = tb_usb3_pre_activate;
> + tunnel->release_unused_bandwidth =
> + tb_usb3_release_unused_bandwidth;
> + tunnel->reclaim_available_bandwidth =
> + tb_usb3_reclaim_available_bandwidth;
> + }
>
> tb_tunnel_dbg(tunnel, "currently allocated bandwidth %d/%d Mb/s\n",
> tunnel->allocated_up, tunnel->allocated_down);
> -
> - tunnel->pre_activate = tb_usb3_pre_activate;
> - tunnel->consumed_bandwidth = tb_usb3_consumed_bandwidth;
> - tunnel->release_unused_bandwidth =
> - tb_usb3_release_unused_bandwidth;
> - tunnel->reclaim_available_bandwidth =
> - tb_usb3_reclaim_available_bandwidth;
> }
>
> tb_tunnel_dbg(tunnel, "discovered\n");
> @@ -2365,12 +2381,15 @@ struct tb_tunnel *tb_tunnel_alloc_usb3(struct tb *tb, struct tb_port *up,
> tunnel->allocated_up = min(max_rate, max_up);
> tunnel->allocated_down = min(max_rate, max_down);
>
> - tunnel->pre_activate = tb_usb3_pre_activate;
> tunnel->consumed_bandwidth = tb_usb3_consumed_bandwidth;
> - tunnel->release_unused_bandwidth =
> - tb_usb3_release_unused_bandwidth;
> - tunnel->reclaim_available_bandwidth =
> - tb_usb3_reclaim_available_bandwidth;
> +
> + if (!down->sw->no_usb3_bw_alloc) {
> + tunnel->pre_activate = tb_usb3_pre_activate;
> + tunnel->release_unused_bandwidth =
> + tb_usb3_release_unused_bandwidth;
> + tunnel->reclaim_available_bandwidth =
> + tb_usb3_reclaim_available_bandwidth;
> + }
> }
>
> return tunnel;
>
> --
> 2.55.0
>
next prev parent reply other threads:[~2026-09-01 9:12 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-30 20:19 [PATCH 00/19] Initial USB4/Thunderbolt support for Apple M1/M2/M3 SoCs Sven Peter
2026-08-30 20:19 ` [PATCH 01/19] dt-bindings: usb: Add thunderbolt-switch property Sven Peter
2026-08-30 20:19 ` [PATCH 02/19] usb: typec: Add thunderbolt switch Sven Peter
2026-09-01 11:17 ` Heikki Krogerus
2026-09-01 18:53 ` Sven Peter
2026-08-30 20:19 ` [PATCH 03/19] usb: typec: tipd: Hook up Thunderbolt switch for CD321x Sven Peter
2026-08-30 20:19 ` [PATCH 04/19] dt-bindings: thunderbolt: Add Apple USB4/Thunderbolt NHI Sven Peter
2026-08-30 20:19 ` [PATCH 05/19] dt-bindings: thunderbolt: Add Apple USB4/Thunderbolt ACIO block Sven Peter
2026-08-30 20:19 ` [PATCH 06/19] thunderbolt: Try reading host DROM from device tree first Sven Peter
2026-09-01 8:48 ` Mika Westerberg
2026-08-30 20:19 ` [PATCH 07/19] thunderbolt: Don't read the UID if we already know it Sven Peter
2026-08-30 20:19 ` [PATCH 08/19] thunderbolt: Allocate ring HopID before requesting the ring interrupt Sven Peter
2026-08-30 20:19 ` [PATCH 09/19] thunderbolt: Add ring_interrupt_active to tb_nhi_ops Sven Peter
2026-08-30 20:19 ` [PATCH 10/19] thunderbolt: Make the ring register layout configurable Sven Peter
2026-09-01 8:58 ` Mika Westerberg
2026-09-01 18:56 ` Sven Peter
2026-08-30 20:19 ` [PATCH 11/19] thunderbolt: Add ring_interrupt_mask to tb_nhi_ops Sven Peter
2026-08-30 20:19 ` [PATCH 12/19] thunderbolt: Add ring_configure " Sven Peter
2026-08-30 20:19 ` [PATCH 13/19] thunderbolt: Add QUIRK_NO_DMA_PORT Sven Peter
2026-09-01 9:04 ` Mika Westerberg
2026-09-01 17:06 ` Sven Peter
2026-08-30 20:19 ` [PATCH 14/19] thunderbolt: Add QUIRK_NO_USB3_BW_ALLOC Sven Peter
2026-09-01 9:12 ` Mika Westerberg [this message]
2026-08-30 20:19 ` [PATCH 15/19] thunderbolt: Export symbols required by the Apple Silicon driver Sven Peter
2026-08-30 20:19 ` [PATCH 16/19] thunderbolt: Add Apple Silicon support Sven Peter
2026-09-01 10:09 ` Mika Westerberg
2026-09-01 19:06 ` Sven Peter
2026-08-30 20:19 ` [PATCH 17/19] arm64: dts: apple: t8103: Add USB4 ACIO and NHI Sven Peter
2026-09-01 10:20 ` Mika Westerberg
2026-08-30 20:19 ` [PATCH 18/19] arm64: dts: apple: t8112: " Sven Peter
2026-08-30 20:19 ` [PATCH 19/19] arm64: dts: apple: t60xx: " Sven Peter
2026-08-31 17:44 ` [PATCH 00/19] Initial USB4/Thunderbolt support for Apple M1/M2/M3 SoCs Joshua Peisach
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=20260901091205.GE106095@black.igk.intel.com \
--to=mika.westerberg@linux.intel.com \
--cc=YehezkelShB@gmail.com \
--cc=andreas.noever@gmail.com \
--cc=asahi@lists.linux.dev \
--cc=conor+dt@kernel.org \
--cc=devicetree@vger.kernel.org \
--cc=gregkh@linuxfoundation.org \
--cc=heikki.krogerus@linux.intel.com \
--cc=j@jannau.net \
--cc=konradybcio@kernel.org \
--cc=krzk+dt@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=neal@gompa.dev \
--cc=p.zabel@pengutronix.de \
--cc=robh@kernel.org \
--cc=sven@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