This patchset attempts to fix a particularly nasty bug that causes the
xHCI host to be marked as 'dying' and all USB devices to be
disconnected.
The bug happens when some USB device (say a VIA USB 3.0 hub) refuses to
answer to the Set Address command in a timely manner, while other
commands are queued. See patch three for a longer (and funnier)
description of the issue.
Eric,
I would appreciate you looking over my fixes to the command cancellation
code you wrote.
Sarah Sharp
The following changes since commit 8af096f15978784e83c24ca36f52453e681945c5:
USB: xhci - fix bit definitions for IMAN register (2013-02-26 17:35:54 -0800)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/sarah/xhci.git ful-command-queue-fixes
for you to fetch changes up to 4a1df0d67f20cc014be93933ff37b4d0d0546af1:
NOT for merge: xhci command cancellation stress test. (2013-03-07 15:01:00 -0800)
----------------------------------------------------------------
Sarah Sharp (10):
xhci: Use command ring enqueue in cancellation.
xhci: Don't hold spinlock while stopping the command ring.
xhci: Handle command stalls.
xhci: Unconditionally complete canceled commands
xhci: Handle command ring stop between commands.
xhci: Refactor code to find the next enqueue pointer.
xhci: Warn if the command ring dequeue is hosed.
xhci: Add more debugging for command cancellation.
xhci: More command debugging.
NOT for merge: xhci command cancellation stress test.
drivers/usb/host/xhci-ring.c | 157 ++++++++++++++++++++++++++++++------------
drivers/usb/host/xhci.c | 108 +++++++++++++++++++++--------
drivers/usb/host/xhci.h | 3 +
3 files changed, 194 insertions(+), 74 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
