P33M

On a locally-compiled 5.15.45 the issue is not present. On a fresh rpi-update kernel (5.15.56) the issue is present. Bisecting...

Good: 93baedb93739 configs: Rengenerate Bad: 4ee9d0f86680 Merge remote-tracking branch 'stable/linux-5.15.y' into rpi-5.15.y The bad commit is the 5.15.50 upstream merge. Reverting b761f6dd3619fd9cf6f193a8363aab8600676e3f makes 5.15.50 a good commit. ``` commit b761f6dd3619fd9cf6f193a8363aab8600676e3f...

b8e68d153d429de337d173e6bfc295f1d8ba557d incorrectly adds the ID for BCM54213PE to the bcm_ptp_probe function. As you're building locally, can you try reverting it on top of latest rpi-5.15.y and test?

I think https://github.com/raspberrypi/linux/commit/b8e68d153d429de337d173e6bfc295f1d8ba557d needs to be reverted. BCM54213PE does not support PTP and including its device ID for PTP is a bug.

Yes, it looks like it just aligns downstream with the new upstream phy features.

What version of VL805 firmware is in use? `rpi-eeprom-update` will tell you. Hub status events are the sole determinant in updating the USB bus hierarchy. So if the issue is...

The hub status endpoint for device 3 (daisy-chained Genesys hub connected to another Genesys hub) stops getting polled after transaction 2266866. There was no bus activity aside from other endpoint...

> If I understand correctly, your test is to check whether the Hub interrupt end point polling is being stopped in-between irrespective of USB device unplug right ? Correct me...

From the TRB dump I can't see anything obviously wrong. As there are two hubs (devices/02 and devices/03) you would expect the different state of the status endpoint (polled or...

I've been attempting to replicate the issue with a LAN7500 dongle and a RTL8152 dongle but without success. Using either an ASmedia hub (two 4-port hubs in cascade) or a...