[PATCH] phy: qcom: qmp-combo: Leave the USB PCS alone in DP-only mode

[Joonhoe Kim]

On Thu, 22 Oct 2026 17:07:04 +0000, Sashiko AI review wrote:
> - [High] phy: qcom: qmp-combo: qmp_combo_typec_switch_set() powers USB in DP-only mode
[...]
> Would this overwrite the serdes, lane, and PCS tables, time out waiting
> for PHYSTATUS, and call clk_disable_unprepare(qmp->pipe_clk) on the
> error path while qmp->init_count remains non-zero?
[...]
> Should both qmp_combo_typec_switch_set() and the power-off path in
> qmp_combo_reconfigure_phy() also skip the USB PCS when qmp->qmpphy_mode is
> QMPPHY_MODE_DP_ONLY?

Yes. Since 89563a152973 ("phy: qualcomm: qmp-combo: Add preliminary
USB4 support") the mux keeps usb_init_count when it enters DP-only
mode, so an orientation change in that mode already re-ran the USB
PCS power-on over the DisplayPort configuration before this patch. v2
skips the USB PCS in qmp_combo_typec_switch_set() as well.

The power-off in qmp_combo_reconfigure_phy() when leaving DP-only mode
only asserts SW_RESET, clears the start bits and puts the PCS into
power-down, which it already is in DP-only mode, so it is harmless, but
v2 skips it too for consistency.

The pipe_clk handling on the error path of qmp_combo_usb_power_on() is
a separate issue: pipe_clk is enabled in qmp_combo_com_init() and
disabled in qmp_combo_com_exit(), so disabling it on a PHYSTATUS
timeout leaves it unbalanced whatever the mode. This patch does not
change that.

v2: https://lore.kernel.org/all/20261023123512.1912-1-26rote@gmail.com/

Thanks,
Joonhoe Kim