Alan, Sarah,
This revision boils down the port power control fixes to the
bare minimum to get the implementation functional and reliable.
Data structure changes are constrained to struct usb_port and
gone are the clumsier attempts at wider reworks from v1 [1] and
v2 [2]. No device model changes to consider or changes to the
meaning of 'runtime_status' for port devices. Three disconnect
bugs are fixed:
1/ Superspeed devices downgrade to their hi-speed connection: fix this by
preventing superspeed poweroff until the peer port is suspended. See
patch 5.
2/ khubd taking disconnect action on ports that are in the process of
being recovered: khubd now ignores ports in the pm-runtime-suspended
state. Alan, per your comment [3] this effectively uses the pm_usage
counter and state as a lock against khubd. See patch 7.
3/ Superspeed devices fail to reconnect: Seen during repeated toggles of
the port power state. Fixed by forcing a full recovery cycle of the
device before allowing the next suspend, and blocking khubd while the
resume is in progress. See patch 9.
Patch overview:
[PATCH 01/10] usb: assign default peer ports for root hubs
[PATCH 02/10] usb: find external hub port peers
[PATCH 03/10] usb: find internal hub tier mismatch via acpi
[PATCH 04/10] usb: sysfs link peer ports
* Per our discussions of v1 these patches implement a
simple algorithm for associating peer ports across
internal and external hubs.
[PATCH 05/10] usb: defer suspension of superspeed port while peer is powered
* Fix case 1
[PATCH 06/10] usb: gate clearing FEAT_C_ENABLE to usb2 hubs
* Cleanup misuse of ClearPortFeature(PORT_C_ENABLE)
[PATCH 07/10] usb: synchronize port poweroff and khubd
* Fix case 2
[PATCH 08/10] usb: cleanup straggling C_PORT_RESET C_PORT_LINK_STATE notifications
* Handle some unexpected hub events encountered during testing
[PATCH 09/10] usb: make khubd and subsequent suspension wait for port recovery
* Fix case 3
[PATCH 10/10] usb: documentation for usb port power off mechanisms
These patches were tested by repeatedly power toggling all
35 ports in the following topology:
/: Bus 03.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/4p, 5000M
|__ Port 1: Dev 2, If 0, Class=hub, Driver=hub/4p, 5000M
|__ Port 1: Dev 3, If 0, Class=vend., Driver=ax88179_178a, 5000M
|__ Port 2: Dev 4, If 0, Class=stor., Driver=usb-storage, 5000M
|__ Port 4: Dev 5, If 0, Class=stor., Driver=usb-storage, 5000M
/: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/9p, 480M
|__ Port 1: Dev 2, If 0, Class=hub, Driver=hub/4p, 480M
|__ Port 3: Dev 4, If 0, Class=HID, Driver=usbhid, 12M
|__ Port 2: Dev 3, If 0, Class=hub, Driver=hub/4p, 480M
|__ Port 1: Dev 5, If 0, Class=HID, Driver=usbhid, 1.5M
|__ Port 2: Dev 6, If 0, Class=HID, Driver=usbhid, 1.5M
/: Bus 01.Port 1: Dev 1, Class=root_hub, Driver=ehci-pci/2p, 480M
|__ Port 1: Dev 2, If 0, Class=hub, Driver=hub/8p, 480M
Each iteration of the test verifies that no disconnects occur and that
all ports reach the 'suspended' state. To force device suspension
interface drivers are unbound for power-off and then rebound. Note that
since the hub drivers are never unbound their parent ports remain active
due to ->do_remote_wakeup for the hub device, but the the 30 other ports
reliably suspend and resume now with these patches. The proposed warm
reset changes [4] do not appear to be required as long as superspeed hub
parent ports remain powered.
[1] http://marc.info/?l=linux-usb&m=138260013707007&w=2
[2] http://marc.info/?l=linux-usb&m=138511124910669&w=2
[3] http://marc.info/?l=linux-usb&m=138775577717546&w=2
[4] http://marc.info/?l=linux-usb&m=138759482824618&w=2
---
Dan Williams (9):
usb: assign default peer ports for root hubs
usb: find external hub port peers
usb: find internal hub tier mismatch via acpi
usb: sysfs link peer ports
usb: don't suspend port while peer is powered
usb: gate clearing FEAT_C_ENABLE to usb2 hubs
usb: synchronize port poweroff and khubd
usb: cleanup straggling C_PORT_RESET C_PORT_LINK_STATE notifications
usb: make khubd and subsequent suspension wait for port recovery
Lan Tianyu (1):
USB: Documentation for USB port power off mechanisms
Documentation/usb/power-management.txt | 210 +++++++++++++++++++++++++++
drivers/usb/core/hub.c | 112 ++++++++++----
drivers/usb/core/hub.h | 14 ++
drivers/usb/core/port.c | 252 +++++++++++++++++++++++++++++++-
drivers/usb/core/usb-acpi.c | 35 ++++
drivers/usb/core/usb.h | 2
6 files changed, 574 insertions(+), 51 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
