[RFT PATCH 0/3] xhci: avoid futile stop endpoint command and host teardown

[Mathias Nyman <mathias.nyman@linux.intel.com>]

This series is an attempt to avoid stop endpoint command failure
leading to xhci teardown with messages:

xHCI host not responding to stop endpoint command
xHCI host controller not responding, assume dead

Series is based on a discussion with Michal and Xu Rao:

https://lore.kernel.org/linux-usb/D9BA02889D046D23+20260617100957.2888108-1-raoxu@uniontech.com/

A Renesas xHC host with a device behind a disconnected hub failed to
complete a stop endoint command immediately after a endpoint soft
reset and restart. Michal had some plausible theories why stop endpoint
command issued immediately after endpoint restart fails after disconnect.

Many of these cases can be avoided as xhci driver knows the rootport
link state and can avoid futile transfer retry and recovery on link
error or disconnect

I don't have a easy way to reproduce this myself, all testing is
welcome

Thanks
Mathias

Mathias Nyman (3):
  xhci: include all root port children in recovery prevention on link
    error
  xhci: prevent endpoint recovery after roothub disconnect
  xhci: avoid xHC endpoint changes after disconnect or link error.

 drivers/usb/host/xhci-ring.c | 50 +++++++++++++++++++++++++-----------
 drivers/usb/host/xhci.c      |  9 ++++---
 drivers/usb/host/xhci.h      | 11 +++-----
 3 files changed, 43 insertions(+), 27 deletions(-)

-- 
2.43.0