usb: xhci-mtk: Do not use xhci's virt_dev in drop_endpoint
authorIkjoon Jang <ikjn@chromium.org>
Thu, 5 Aug 2021 05:37:47 +0000 (13:37 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 5 Aug 2021 10:32:56 +0000 (12:32 +0200)
commitb8731209958a1dffccc2888121f4c0280c990550
tree7d73ff75bae433e74070fed10f8b84856348394d
parent177cd475e1f12d256c9bf7de274fef7bea30cddb
usb: xhci-mtk: Do not use xhci's virt_dev in drop_endpoint

xhci-mtk depends on xhci's internal virt_dev when it retrieves its
internal data from usb_host_endpoint both in add_endpoint and
drop_endpoint callbacks. But when setup packet was retired by
transaction errors in xhci_setup_device() path, a virt_dev for the slot
is newly created with real_port 0. This leads to xhci-mtks's NULL pointer
dereference from drop_endpoint callback as xhci-mtk assumes that virt_dev's
real_port is always started from one. The similar problems were addressed
by [1] but that can't cover the failure cases from setup_device.

This patch drops the usages of xhci's virt_dev in xhci-mtk's drop_endpoint
callback by adopting rhashtable for searching mtk's schedule entity
from a given usb_host_endpoint pointer instead of searching a linked list.
So mtk's drop_endpoint callback doesn't have to rely on virt_dev at all.

[1] https://lore.kernel.org/r/1617179142-2681-2-git-send-email-chunfeng.yun@mediatek.com

Signed-off-by: Ikjoon Jang <ikjn@chromium.org>
Link: https://lore.kernel.org/r/20210805133731.1.Icc0f080e75b1312692d4c7c7d25e7df9fe1a05c2@changeid
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/usb/host/xhci-mtk-sch.c
drivers/usb/host/xhci-mtk.h