Hi Rick,
On Fri, Nov 01, 2024 at 01:57:50PM +0100, Rick wrote:
I compiled 6.12.0-rc5-00181-g6c52d4da1c74-dirty resulting in docking station not working.
Then I compiled 6.12.0-rc5-00181-g6c52d4da1c74-dirty without commit c6ca1ac9f472 (reverted), and now the docking station works correctly (as in screen output + USBs + Ethernet)
So it seems c6ca1ac9f472 is causing issues for my setup.
Okay, thanks for testing!
It indeed looks like there is no any kind of link issues anymore with that one reverted. So my suspect is that we are taking too long before we enumerate the device router which makes it to reset the link.
Can you try the below patch too on top of v6.12-rcX (without the revert) and see if that still keeps it working? This one cuts down the delay to 1ms which I'm hoping is sufficient for the device. Can you share dmesg+trace from that test as well?
diff --git a/drivers/thunderbolt/usb4.c b/drivers/thunderbolt/usb4.c index c6dcc23e8c16..1b740d7fc7da 100644 --- a/drivers/thunderbolt/usb4.c +++ b/drivers/thunderbolt/usb4.c @@ -48,7 +48,7 @@ enum usb4_ba_index {
/* Delays in us used with usb4_port_wait_for_bit() */ #define USB4_PORT_DELAY 50 -#define USB4_PORT_SB_DELAY 5000 +#define USB4_PORT_SB_DELAY 1000
static int usb4_native_switch_op(struct tb_switch *sw, u16 opcode, u32 *metadata, u8 *status,