openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity
linux/drivers/usb/host/ehci-timer.c

435 lines
13 KiB
C
Raw Normal View History

USB: EHCI: introduce high-res timer This patch (as1572) begins the conversion of ehci-hcd over to using high-resolution timers rather than old-fashioned low-resolution kernel timers. This reduces overhead caused by timer roundoff on systems where HZ is smaller than 1000. Also, the new timer framework introduced here is much more logical and easily extended than the ad-hoc approach ehci-hcd currently uses for timers. An hrtimer structure is added to ehci_hcd, along with a bitflag array and an array of ktime_t values, to keep track of which timing events are pending and what their expiration times are. Only the infrastructure for the timing operations is added in this patch. Later patches will add routines for handling each of the various timing events the driver needs. In some cases the new hrtimer handlers will replace the existing handlers for ehci-hcd's kernel timers; as this happens the old timers will be removed. In other cases the new timing events will replace busy-wait loops. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:21:54 +08:00
/*
* Copyright (C) 2012 by Alan Stern
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the
* Free Software Foundation; either version 2 of the License, or (at your
* option) any later version.
*
* This program is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
* or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* for more details.
*/
/* This file is part of ehci-hcd.c */
/*-------------------------------------------------------------------------*/
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
/* Set a bit in the USBCMD register */
static void ehci_set_command_bit(struct ehci_hcd *ehci, u32 bit)
{
ehci->command |= bit;
ehci_writel(ehci, ehci->command, &ehci->regs->command);
/* unblock posted write */
ehci_readl(ehci, &ehci->regs->command);
}
/* Clear a bit in the USBCMD register */
static void ehci_clear_command_bit(struct ehci_hcd *ehci, u32 bit)
{
ehci->command &= ~bit;
ehci_writel(ehci, ehci->command, &ehci->regs->command);
/* unblock posted write */
ehci_readl(ehci, &ehci->regs->command);
}
/*-------------------------------------------------------------------------*/
USB: EHCI: introduce high-res timer This patch (as1572) begins the conversion of ehci-hcd over to using high-resolution timers rather than old-fashioned low-resolution kernel timers. This reduces overhead caused by timer roundoff on systems where HZ is smaller than 1000. Also, the new timer framework introduced here is much more logical and easily extended than the ad-hoc approach ehci-hcd currently uses for timers. An hrtimer structure is added to ehci_hcd, along with a bitflag array and an array of ktime_t values, to keep track of which timing events are pending and what their expiration times are. Only the infrastructure for the timing operations is added in this patch. Later patches will add routines for handling each of the various timing events the driver needs. In some cases the new hrtimer handlers will replace the existing handlers for ehci-hcd's kernel timers; as this happens the old timers will be removed. In other cases the new timing events will replace busy-wait loops. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:21:54 +08:00
/*
* EHCI timer support... Now using hrtimers.
*
* Lots of different events are triggered from ehci->hrtimer. Whenever
* the timer routine runs, it checks each possible event; events that are
* currently enabled and whose expiration time has passed get handled.
* The set of enabled events is stored as a collection of bitflags in
* ehci->enabled_hrtimer_events, and they are numbered in order of
* increasing delay values (ranging between 1 ms and 100 ms).
*
* Rather than implementing a sorted list or tree of all pending events,
* we keep track only of the lowest-numbered pending event, in
* ehci->next_hrtimer_event. Whenever ehci->hrtimer gets restarted, its
* expiration time is set to the timeout value for this event.
*
* As a result, events might not get handled right away; the actual delay
* could be anywhere up to twice the requested delay. This doesn't
* matter, because none of the events are especially time-critical. The
* ones that matter most all have a delay of 1 ms, so they will be
* handled after 2 ms at most, which is okay. In addition to this, we
* allow for an expiration range of 1 ms.
*/
/*
* Delay lengths for the hrtimer event types.
* Keep this list sorted by delay length, in the same order as
* the event types indexed by enum ehci_hrtimer_event in ehci.h.
*/
static unsigned event_delays_ns[] = {
USB: EHCI: use hrtimer for async schedule This patch (as1576) adds hrtimer support for managing ehci-hcd's async schedule. Just as with the earlier change to the periodic schedule management, two new hrtimer events take care of everything. One event polls at 1-ms intervals to see when the Asynchronous Schedule Status (ASS) flag matches the Asynchronous Schedule Enable (ASE) value; the schedule's state must not be changed until it does. The other event delays for 15 ms after the async schedule becomes empty before turning it off. The new events replace a busy-wait poll and a kernel timer usage. They also replace the rather illogical method currently used for indicating the async schedule should be turned off: attempting to unlink the dedicated QH at the head of the async list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:21 +08:00
1 * NSEC_PER_MSEC, /* EHCI_HRTIMER_POLL_ASS */
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
1 * NSEC_PER_MSEC, /* EHCI_HRTIMER_POLL_PSS */
USB: EHCI: use hrtimer for controller death This patch (as1578) adds an hrtimer event to handle the death of an EHCI controller. When a controller dies, it doesn't necessarily stop running right away. The new event polls at 1-ms intervals to see when all activity has safely stopped. This replaces a busy-wait polling loop in the current code. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:31 +08:00
1 * NSEC_PER_MSEC, /* EHCI_HRTIMER_POLL_DEAD */
USB: EHCI: use hrtimer for interrupt QH unlink This patch (as1577) adds hrtimer support for unlinking interrupt QHs in ehci-hcd. The current code relies on a fixed delay of either 2 or 55 us, which is not always adequate and in any case is totally bogus. Thanks to internal caching, the EHCI hardware may continue to access an interrupt QH for more than a millisecond after it has been unlinked. In fact, the EHCI spec doesn't say how long to wait before using an unlinked interrupt QH. The patch sets the delay to 9 microframes minimum, which ought to be adequate. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:26 +08:00
1125 * NSEC_PER_USEC, /* EHCI_HRTIMER_UNLINK_INTR */
USB: EHCI: use hrtimer for (s)iTD deallocation This patch (as1579) adds an hrtimer event to handle deallocation of iTDs and siTDs in ehci-hcd. Because of the frame-oriented approach used by the EHCI periodic schedule, the hardware can continue to access the Transfer Descriptor for isochronous (or split-isochronous) transactions for up to a millisecond after the transaction completes. The iTD (or siTD) must not be reused before then. The strategy currently used involves putting completed iTDs on a list of cached entries and every so often returning them to the endpoint's free list. The new strategy reduces overhead by putting completed iTDs back on the free list immediately, although they are not reused until it is safe to do so. When the isochronous endpoint stops (its queue becomes empty), the iTDs on its free list get moved to a global list, from which they will be deallocated after a minimum of 2 ms. This delay is what the new hrtimer event is for. Overall this may not be a tremendous improvement over the current code, but to me it seems a lot more clear and logical. In addition, it removes the need for each iTD to keep a reference to the ehci_iso_stream it belongs to, since the iTD never needs to be moved back to the stream's free list from the global list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:35 +08:00
2 * NSEC_PER_MSEC, /* EHCI_HRTIMER_FREE_ITDS */
USB: EHCI: improve interrupt qh unlink ehci-hcd currently unlinks an interrupt QH when it becomes empty, that is, after its last URB completes. This works well because in almost all cases, the completion handler for an interrupt URB resubmits the URB; therefore the QH doesn't become empty and doesn't get unlinked. When we start using tasklets for URB completion, this scheme won't work as well. The resubmission won't occur until the tasklet runs, which will be some time after the completion is queued with the tasklet. During that delay, the QH will be empty and so will be unlinked unnecessarily. To prevent this problem, this patch adds a 5-ms time delay before empty interrupt QHs are unlinked. Most often, during that time the interrupt URB will be resubmitted and thus we can avoid unlinking the QH. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Ming Lei <ming.lei@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-07-03 22:53:10 +08:00
5 * NSEC_PER_MSEC, /* EHCI_HRTIMER_START_UNLINK_INTR */
USB: EHCI: use hrtimer for unlinking empty async QHs This patch (as1583) changes ehci-hcd to use an hrtimer event for unlinking empty (unused) async QHs instead of using a kernel timer. The check for empty QHs is moved to a new routine, where it doesn't require going through an entire scan of both the async and periodic schedules. And it can unlink multiple QHs at once, unlike the current code. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:53 +08:00
6 * NSEC_PER_MSEC, /* EHCI_HRTIMER_ASYNC_UNLINKS */
USB: EHCI: use hrtimer for the IAA watchdog This patch (as1581) replaces the iaa_watchdog kernel timer used by ehci-hcd with an hrtimer event, in keeping with the general conversion to high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:44 +08:00
10 * NSEC_PER_MSEC, /* EHCI_HRTIMER_IAA_WATCHDOG */
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
10 * NSEC_PER_MSEC, /* EHCI_HRTIMER_DISABLE_PERIODIC */
USB: EHCI: use hrtimer for async schedule This patch (as1576) adds hrtimer support for managing ehci-hcd's async schedule. Just as with the earlier change to the periodic schedule management, two new hrtimer events take care of everything. One event polls at 1-ms intervals to see when the Asynchronous Schedule Status (ASS) flag matches the Asynchronous Schedule Enable (ASE) value; the schedule's state must not be changed until it does. The other event delays for 15 ms after the async schedule becomes empty before turning it off. The new events replace a busy-wait poll and a kernel timer usage. They also replace the rather illogical method currently used for indicating the async schedule should be turned off: attempting to unlink the dedicated QH at the head of the async list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:21 +08:00
15 * NSEC_PER_MSEC, /* EHCI_HRTIMER_DISABLE_ASYNC */
USB: EHCI: use hrtimer for the I/O watchdog This patch (as1586) replaces the kernel timer used by ehci-hcd as an I/O watchdog with an hrtimer event. Unlike in the current code, the watchdog event is now always enabled whenever any isochronous URBs are active. This will prevent bugs caused by the periodic schedule wrapping around with no completion interrupts; the watchdog handler is guaranteed to scan the isochronous transfers at least once during each iteration of the schedule. The extra overhead will be negligible: one timer interrupt every 100 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:23:04 +08:00
100 * NSEC_PER_MSEC, /* EHCI_HRTIMER_IO_WATCHDOG */
USB: EHCI: introduce high-res timer This patch (as1572) begins the conversion of ehci-hcd over to using high-resolution timers rather than old-fashioned low-resolution kernel timers. This reduces overhead caused by timer roundoff on systems where HZ is smaller than 1000. Also, the new timer framework introduced here is much more logical and easily extended than the ad-hoc approach ehci-hcd currently uses for timers. An hrtimer structure is added to ehci_hcd, along with a bitflag array and an array of ktime_t values, to keep track of which timing events are pending and what their expiration times are. Only the infrastructure for the timing operations is added in this patch. Later patches will add routines for handling each of the various timing events the driver needs. In some cases the new hrtimer handlers will replace the existing handlers for ehci-hcd's kernel timers; as this happens the old timers will be removed. In other cases the new timing events will replace busy-wait loops. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:21:54 +08:00
};
/* Enable a pending hrtimer event */
static void ehci_enable_event(struct ehci_hcd *ehci, unsigned event,
bool resched)
{
ktime_t *timeout = &ehci->hr_timeouts[event];
if (resched)
*timeout = ktime_add(ktime_get(),
ktime_set(0, event_delays_ns[event]));
ehci->enabled_hrtimer_events |= (1 << event);
/* Track only the lowest-numbered pending event */
if (event < ehci->next_hrtimer_event) {
ehci->next_hrtimer_event = event;
hrtimer_start_range_ns(&ehci->hrtimer, *timeout,
NSEC_PER_MSEC, HRTIMER_MODE_ABS);
}
}
USB: EHCI: use hrtimer for async schedule This patch (as1576) adds hrtimer support for managing ehci-hcd's async schedule. Just as with the earlier change to the periodic schedule management, two new hrtimer events take care of everything. One event polls at 1-ms intervals to see when the Asynchronous Schedule Status (ASS) flag matches the Asynchronous Schedule Enable (ASE) value; the schedule's state must not be changed until it does. The other event delays for 15 ms after the async schedule becomes empty before turning it off. The new events replace a busy-wait poll and a kernel timer usage. They also replace the rather illogical method currently used for indicating the async schedule should be turned off: attempting to unlink the dedicated QH at the head of the async list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:21 +08:00
/* Poll the STS_ASS status bit; see when it agrees with CMD_ASE */
static void ehci_poll_ASS(struct ehci_hcd *ehci)
{
unsigned actual, want;
/* Don't enable anything if the controller isn't running (e.g., died) */
if (ehci->rh_state != EHCI_RH_RUNNING)
return;
want = (ehci->command & CMD_ASE) ? STS_ASS : 0;
actual = ehci_readl(ehci, &ehci->regs->status) & STS_ASS;
if (want != actual) {
USB: EHCI: decrease schedule-status poll timeout This patch (as1657) decreases the timeout used by ehci-hcd for polling the async and periodic schedule statuses. The timeout is currently set to 20 ms, which is much too high. Controllers should always update the schedule status within one or two ms of being told to do so; if they don't then something is wrong. Furthermore, bug reports have shown that sometimes controllers (particularly those made by VIA) don't update the status bit at all, even when the schedule does change state. When this happens, polling for 20 ms would cause an unnecessarily long delay. The delay is reduced to somewhere between 2 and 4 ms, depending on the slop allowed by the kernel's high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-19 00:04:54 +08:00
/* Poll again later, but give up after about 2-4 ms */
if (ehci->ASS_poll_count++ < 2) {
USB: EHCI: revert "remove ASS/PSS polling timeout" This patch (as1649) reverts commit 55bcdce8a8228223ec4d17d8ded8134ed265d2c5 (USB: EHCI: remove ASS/PSS polling timeout). That commit was written under the assumption that some controllers may take a very long time to turn off their async and periodic schedules. It now appears that in fact the schedules do get turned off reasonably quickly, but some controllers occasionally leave the schedules' status bits turned on and consequently ehci-hcd can't tell that the schedules are off. VIA controllers in particular have this problem. ehci-hcd tells the hardware to turn off the async schedule, the schedule does get turned off, but the status bit remains on. Since the EHCI spec requires that the schedules not be re-enabled until the previous disable has taken effect, with an unlimited timeout the async schedule never gets turned back on. The resulting symptom is that the system is unable to communicate with USB devices. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Reported-and-tested-by: Ronald <ronald645@gmail.com> Reported-and-tested-by: Paul Hartman <paul.hartman@gmail.com> Reported-and-tested-by: Dieter Nützel <dieter@nuetzel-hh.de> Reported-and-tested-by: Jean Delvare <khali@linux-fr.org> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-02-27 02:43:41 +08:00
ehci_enable_event(ehci, EHCI_HRTIMER_POLL_ASS, true);
return;
}
ehci_dbg(ehci, "Waited too long for the async schedule status (%x/%x), giving up\n",
want, actual);
USB: EHCI: use hrtimer for async schedule This patch (as1576) adds hrtimer support for managing ehci-hcd's async schedule. Just as with the earlier change to the periodic schedule management, two new hrtimer events take care of everything. One event polls at 1-ms intervals to see when the Asynchronous Schedule Status (ASS) flag matches the Asynchronous Schedule Enable (ASE) value; the schedule's state must not be changed until it does. The other event delays for 15 ms after the async schedule becomes empty before turning it off. The new events replace a busy-wait poll and a kernel timer usage. They also replace the rather illogical method currently used for indicating the async schedule should be turned off: attempting to unlink the dedicated QH at the head of the async list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:21 +08:00
}
ehci->ASS_poll_count = 0;
/* The status is up-to-date; restart or stop the schedule as needed */
if (want == 0) { /* Stopped */
if (ehci->async_count > 0)
ehci_set_command_bit(ehci, CMD_ASE);
} else { /* Running */
if (ehci->async_count == 0) {
/* Turn off the schedule after a while */
ehci_enable_event(ehci, EHCI_HRTIMER_DISABLE_ASYNC,
true);
}
}
}
/* Turn off the async schedule after a brief delay */
static void ehci_disable_ASE(struct ehci_hcd *ehci)
{
ehci_clear_command_bit(ehci, CMD_ASE);
}
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
/* Poll the STS_PSS status bit; see when it agrees with CMD_PSE */
static void ehci_poll_PSS(struct ehci_hcd *ehci)
{
unsigned actual, want;
/* Don't do anything if the controller isn't running (e.g., died) */
if (ehci->rh_state != EHCI_RH_RUNNING)
return;
want = (ehci->command & CMD_PSE) ? STS_PSS : 0;
actual = ehci_readl(ehci, &ehci->regs->status) & STS_PSS;
if (want != actual) {
USB: EHCI: decrease schedule-status poll timeout This patch (as1657) decreases the timeout used by ehci-hcd for polling the async and periodic schedule statuses. The timeout is currently set to 20 ms, which is much too high. Controllers should always update the schedule status within one or two ms of being told to do so; if they don't then something is wrong. Furthermore, bug reports have shown that sometimes controllers (particularly those made by VIA) don't update the status bit at all, even when the schedule does change state. When this happens, polling for 20 ms would cause an unnecessarily long delay. The delay is reduced to somewhere between 2 and 4 ms, depending on the slop allowed by the kernel's high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-19 00:04:54 +08:00
/* Poll again later, but give up after about 2-4 ms */
if (ehci->PSS_poll_count++ < 2) {
USB: EHCI: revert "remove ASS/PSS polling timeout" This patch (as1649) reverts commit 55bcdce8a8228223ec4d17d8ded8134ed265d2c5 (USB: EHCI: remove ASS/PSS polling timeout). That commit was written under the assumption that some controllers may take a very long time to turn off their async and periodic schedules. It now appears that in fact the schedules do get turned off reasonably quickly, but some controllers occasionally leave the schedules' status bits turned on and consequently ehci-hcd can't tell that the schedules are off. VIA controllers in particular have this problem. ehci-hcd tells the hardware to turn off the async schedule, the schedule does get turned off, but the status bit remains on. Since the EHCI spec requires that the schedules not be re-enabled until the previous disable has taken effect, with an unlimited timeout the async schedule never gets turned back on. The resulting symptom is that the system is unable to communicate with USB devices. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Reported-and-tested-by: Ronald <ronald645@gmail.com> Reported-and-tested-by: Paul Hartman <paul.hartman@gmail.com> Reported-and-tested-by: Dieter Nützel <dieter@nuetzel-hh.de> Reported-and-tested-by: Jean Delvare <khali@linux-fr.org> Cc: stable <stable@vger.kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-02-27 02:43:41 +08:00
ehci_enable_event(ehci, EHCI_HRTIMER_POLL_PSS, true);
return;
}
ehci_dbg(ehci, "Waited too long for the periodic schedule status (%x/%x), giving up\n",
want, actual);
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
}
ehci->PSS_poll_count = 0;
/* The status is up-to-date; restart or stop the schedule as needed */
if (want == 0) { /* Stopped */
USB: EHCI: always scan each interrupt QH This patch (as1585) fixes a bug in ehci-hcd's scheme for scanning interrupt QHs. Currently a single routine takes care of scanning everything on the periodic schedule. Whenever an interrupt occurs, it scans all isochronous and interrupt URBs scheduled for frames that have elapsed since the last scan. This has two disadvantages. The first is relatively minor: An interrupt QH is likely to end up getting scanned multiple times, particularly if the last scan was not fairly recent. (The current code avoids this by maintaining a periodic_stamp in each interrupt QH.) The second is more serious. The periodic schedule wraps around. If the last scan occurred during frame N, and the next scan occurs when the schedule has gone through an entire cycle and is back at frame N, the scanning code won't look at any frames other than N. Consequently it won't see any QHs that completed during frame N-1 or earlier. The patch replaces the entire frame-based approach for scanning interrupt QHs with a new routine using a list-based approach, the same as for async QHs. This has a slight disadvantage, because it means that all interrupt QHs have to be scanned every time. But it is more robust than the current approach. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:23:00 +08:00
if (ehci->periodic_count > 0)
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
ehci_set_command_bit(ehci, CMD_PSE);
} else { /* Running */
if (ehci->periodic_count == 0) {
/* Turn off the schedule after a while */
ehci_enable_event(ehci, EHCI_HRTIMER_DISABLE_PERIODIC,
true);
}
}
}
/* Turn off the periodic schedule after a brief delay */
static void ehci_disable_PSE(struct ehci_hcd *ehci)
{
ehci_clear_command_bit(ehci, CMD_PSE);
}
USB: EHCI: use hrtimer for controller death This patch (as1578) adds an hrtimer event to handle the death of an EHCI controller. When a controller dies, it doesn't necessarily stop running right away. The new event polls at 1-ms intervals to see when all activity has safely stopped. This replaces a busy-wait polling loop in the current code. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:31 +08:00
/* Poll the STS_HALT status bit; see when a dead controller stops */
static void ehci_handle_controller_death(struct ehci_hcd *ehci)
{
if (!(ehci_readl(ehci, &ehci->regs->status) & STS_HALT)) {
/* Give up after a few milliseconds */
if (ehci->died_poll_count++ < 5) {
/* Try again later */
ehci_enable_event(ehci, EHCI_HRTIMER_POLL_DEAD, true);
return;
}
ehci_warn(ehci, "Waited too long for the controller to stop, giving up\n");
}
/* Clean up the mess */
ehci->rh_state = EHCI_RH_HALTED;
ehci_writel(ehci, 0, &ehci->regs->configured_flag);
ehci_writel(ehci, 0, &ehci->regs->intr_enable);
ehci_work(ehci);
USB: EHCI: unlink multiple async QHs together This patch (as1582) changes ehci-hcd's strategy for unlinking async QHs. Currently the driver never unlinks more than one QH at a time. This can be inefficient and cause unnecessary delays, since a QH cannot be reused while it is waiting to be unlinked. The new strategy unlinks all the waiting QHs at once. In practice the improvement won't be very big, because it's somewhat uncommon to have two or more QHs waiting to be unlinked at any time. But it does happen, and in any case, doing things this way makes more sense IMO. The change requires the async unlinking code to be refactored slightly. Now in addition to the routines for starting and ending an unlink, there are new routines for unlinking a single QH and starting an IAA cycle. This approach is needed because there are two separate paths for unlinking async QHs: When a transfer error occurs or an URB is cancelled, the QH must be unlinked right away; When a QH has been idle sufficiently long, it is unlinked to avoid consuming DMA bandwidth uselessly. In the first case we want the unlink to proceed as quickly as possible, whereas in the second case we can afford to batch several QHs together and unlink them all at once. Hence the division of labor. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:49 +08:00
end_unlink_async(ehci);
USB: EHCI: use hrtimer for controller death This patch (as1578) adds an hrtimer event to handle the death of an EHCI controller. When a controller dies, it doesn't necessarily stop running right away. The new event polls at 1-ms intervals to see when all activity has safely stopped. This replaces a busy-wait polling loop in the current code. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:31 +08:00
/* Not in process context, so don't try to reset the controller */
}
USB: EHCI: improve interrupt qh unlink ehci-hcd currently unlinks an interrupt QH when it becomes empty, that is, after its last URB completes. This works well because in almost all cases, the completion handler for an interrupt URB resubmits the URB; therefore the QH doesn't become empty and doesn't get unlinked. When we start using tasklets for URB completion, this scheme won't work as well. The resubmission won't occur until the tasklet runs, which will be some time after the completion is queued with the tasklet. During that delay, the QH will be empty and so will be unlinked unnecessarily. To prevent this problem, this patch adds a 5-ms time delay before empty interrupt QHs are unlinked. Most often, during that time the interrupt URB will be resubmitted and thus we can avoid unlinking the QH. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Ming Lei <ming.lei@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-07-03 22:53:10 +08:00
/* start to unlink interrupt QHs */
static void ehci_handle_start_intr_unlinks(struct ehci_hcd *ehci)
{
bool stopped = (ehci->rh_state < EHCI_RH_RUNNING);
/*
* Process all the QHs on the intr_unlink list that were added
* before the current unlink cycle began. The list is in
* temporal order, so stop when we reach the first entry in the
* current cycle. But if the root hub isn't running then
* process all the QHs on the list.
*/
while (!list_empty(&ehci->intr_unlink_wait)) {
struct ehci_qh *qh;
qh = list_first_entry(&ehci->intr_unlink_wait,
struct ehci_qh, unlink_node);
if (!stopped && (qh->unlink_cycle ==
ehci->intr_unlink_wait_cycle))
break;
list_del_init(&qh->unlink_node);
USB: EHCI: store reason for unlinking a QH This patch replaces the "exception" bitflag in the ehci_qh structure with a more explicit "unlink_reason" bitmask. This is for use in the following patch, where we will need to have a good idea of the reason for unlinking a QH, not just "something exceptional happened". Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Tested-by: Michael Reutman <mreutman@epiqsolutions.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-01-26 04:42:04 +08:00
qh->unlink_reason |= QH_UNLINK_QUEUE_EMPTY;
USB: EHCI: improve interrupt qh unlink ehci-hcd currently unlinks an interrupt QH when it becomes empty, that is, after its last URB completes. This works well because in almost all cases, the completion handler for an interrupt URB resubmits the URB; therefore the QH doesn't become empty and doesn't get unlinked. When we start using tasklets for URB completion, this scheme won't work as well. The resubmission won't occur until the tasklet runs, which will be some time after the completion is queued with the tasklet. During that delay, the QH will be empty and so will be unlinked unnecessarily. To prevent this problem, this patch adds a 5-ms time delay before empty interrupt QHs are unlinked. Most often, during that time the interrupt URB will be resubmitted and thus we can avoid unlinking the QH. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Ming Lei <ming.lei@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-07-03 22:53:10 +08:00
start_unlink_intr(ehci, qh);
}
/* Handle remaining entries later */
if (!list_empty(&ehci->intr_unlink_wait)) {
ehci_enable_event(ehci, EHCI_HRTIMER_START_UNLINK_INTR, true);
++ehci->intr_unlink_wait_cycle;
}
}
USB: EHCI: use hrtimer for controller death This patch (as1578) adds an hrtimer event to handle the death of an EHCI controller. When a controller dies, it doesn't necessarily stop running right away. The new event polls at 1-ms intervals to see when all activity has safely stopped. This replaces a busy-wait polling loop in the current code. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:31 +08:00
USB: EHCI: use hrtimer for interrupt QH unlink This patch (as1577) adds hrtimer support for unlinking interrupt QHs in ehci-hcd. The current code relies on a fixed delay of either 2 or 55 us, which is not always adequate and in any case is totally bogus. Thanks to internal caching, the EHCI hardware may continue to access an interrupt QH for more than a millisecond after it has been unlinked. In fact, the EHCI spec doesn't say how long to wait before using an unlinked interrupt QH. The patch sets the delay to 9 microframes minimum, which ought to be adequate. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:26 +08:00
/* Handle unlinked interrupt QHs once they are gone from the hardware */
static void ehci_handle_intr_unlinks(struct ehci_hcd *ehci)
{
bool stopped = (ehci->rh_state < EHCI_RH_RUNNING);
/*
* Process all the QHs on the intr_unlink list that were added
* before the current unlink cycle began. The list is in
* temporal order, so stop when we reach the first entry in the
* current cycle. But if the root hub isn't running then
* process all the QHs on the list.
*/
ehci->intr_unlinking = true;
USB: EHCI: convert singly-linked lists to list_heads This patch (as1664) converts ehci-hcd's async_unlink, async_iaa, and intr_unlink from singly-linked lists to standard doubly-linked list_heads. Originally it didn't seem necessary to use list_heads, because items are always added to and removed from these lists in FIFO order. But now with more list processing going on, it's easier to use the standard routines than continue with a roll-your-own approach. I don't know if the code ends up being notably shorter, but the patterns will be more familiar to any kernel hacker. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-23 01:31:45 +08:00
while (!list_empty(&ehci->intr_unlink)) {
struct ehci_qh *qh;
USB: EHCI: use hrtimer for interrupt QH unlink This patch (as1577) adds hrtimer support for unlinking interrupt QHs in ehci-hcd. The current code relies on a fixed delay of either 2 or 55 us, which is not always adequate and in any case is totally bogus. Thanks to internal caching, the EHCI hardware may continue to access an interrupt QH for more than a millisecond after it has been unlinked. In fact, the EHCI spec doesn't say how long to wait before using an unlinked interrupt QH. The patch sets the delay to 9 microframes minimum, which ought to be adequate. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:26 +08:00
USB: EHCI: convert singly-linked lists to list_heads This patch (as1664) converts ehci-hcd's async_unlink, async_iaa, and intr_unlink from singly-linked lists to standard doubly-linked list_heads. Originally it didn't seem necessary to use list_heads, because items are always added to and removed from these lists in FIFO order. But now with more list processing going on, it's easier to use the standard routines than continue with a roll-your-own approach. I don't know if the code ends up being notably shorter, but the patterns will be more familiar to any kernel hacker. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-23 01:31:45 +08:00
qh = list_first_entry(&ehci->intr_unlink, struct ehci_qh,
unlink_node);
USB: EHCI: use hrtimer for interrupt QH unlink This patch (as1577) adds hrtimer support for unlinking interrupt QHs in ehci-hcd. The current code relies on a fixed delay of either 2 or 55 us, which is not always adequate and in any case is totally bogus. Thanks to internal caching, the EHCI hardware may continue to access an interrupt QH for more than a millisecond after it has been unlinked. In fact, the EHCI spec doesn't say how long to wait before using an unlinked interrupt QH. The patch sets the delay to 9 microframes minimum, which ought to be adequate. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:26 +08:00
if (!stopped && qh->unlink_cycle == ehci->intr_unlink_cycle)
break;
USB: EHCI: improve interrupt qh unlink ehci-hcd currently unlinks an interrupt QH when it becomes empty, that is, after its last URB completes. This works well because in almost all cases, the completion handler for an interrupt URB resubmits the URB; therefore the QH doesn't become empty and doesn't get unlinked. When we start using tasklets for URB completion, this scheme won't work as well. The resubmission won't occur until the tasklet runs, which will be some time after the completion is queued with the tasklet. During that delay, the QH will be empty and so will be unlinked unnecessarily. To prevent this problem, this patch adds a 5-ms time delay before empty interrupt QHs are unlinked. Most often, during that time the interrupt URB will be resubmitted and thus we can avoid unlinking the QH. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Ming Lei <ming.lei@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-07-03 22:53:10 +08:00
list_del_init(&qh->unlink_node);
USB: EHCI: use hrtimer for interrupt QH unlink This patch (as1577) adds hrtimer support for unlinking interrupt QHs in ehci-hcd. The current code relies on a fixed delay of either 2 or 55 us, which is not always adequate and in any case is totally bogus. Thanks to internal caching, the EHCI hardware may continue to access an interrupt QH for more than a millisecond after it has been unlinked. In fact, the EHCI spec doesn't say how long to wait before using an unlinked interrupt QH. The patch sets the delay to 9 microframes minimum, which ought to be adequate. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:26 +08:00
end_unlink_intr(ehci, qh);
}
/* Handle remaining entries later */
USB: EHCI: convert singly-linked lists to list_heads This patch (as1664) converts ehci-hcd's async_unlink, async_iaa, and intr_unlink from singly-linked lists to standard doubly-linked list_heads. Originally it didn't seem necessary to use list_heads, because items are always added to and removed from these lists in FIFO order. But now with more list processing going on, it's easier to use the standard routines than continue with a roll-your-own approach. I don't know if the code ends up being notably shorter, but the patterns will be more familiar to any kernel hacker. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-23 01:31:45 +08:00
if (!list_empty(&ehci->intr_unlink)) {
USB: EHCI: use hrtimer for interrupt QH unlink This patch (as1577) adds hrtimer support for unlinking interrupt QHs in ehci-hcd. The current code relies on a fixed delay of either 2 or 55 us, which is not always adequate and in any case is totally bogus. Thanks to internal caching, the EHCI hardware may continue to access an interrupt QH for more than a millisecond after it has been unlinked. In fact, the EHCI spec doesn't say how long to wait before using an unlinked interrupt QH. The patch sets the delay to 9 microframes minimum, which ought to be adequate. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:26 +08:00
ehci_enable_event(ehci, EHCI_HRTIMER_UNLINK_INTR, true);
++ehci->intr_unlink_cycle;
}
ehci->intr_unlinking = false;
}
USB: EHCI: use hrtimer for (s)iTD deallocation This patch (as1579) adds an hrtimer event to handle deallocation of iTDs and siTDs in ehci-hcd. Because of the frame-oriented approach used by the EHCI periodic schedule, the hardware can continue to access the Transfer Descriptor for isochronous (or split-isochronous) transactions for up to a millisecond after the transaction completes. The iTD (or siTD) must not be reused before then. The strategy currently used involves putting completed iTDs on a list of cached entries and every so often returning them to the endpoint's free list. The new strategy reduces overhead by putting completed iTDs back on the free list immediately, although they are not reused until it is safe to do so. When the isochronous endpoint stops (its queue becomes empty), the iTDs on its free list get moved to a global list, from which they will be deallocated after a minimum of 2 ms. This delay is what the new hrtimer event is for. Overall this may not be a tremendous improvement over the current code, but to me it seems a lot more clear and logical. In addition, it removes the need for each iTD to keep a reference to the ehci_iso_stream it belongs to, since the iTD never needs to be moved back to the stream's free list from the global list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:35 +08:00
/* Start another free-iTDs/siTDs cycle */
static void start_free_itds(struct ehci_hcd *ehci)
{
if (!(ehci->enabled_hrtimer_events & BIT(EHCI_HRTIMER_FREE_ITDS))) {
ehci->last_itd_to_free = list_entry(
ehci->cached_itd_list.prev,
struct ehci_itd, itd_list);
ehci->last_sitd_to_free = list_entry(
ehci->cached_sitd_list.prev,
struct ehci_sitd, sitd_list);
ehci_enable_event(ehci, EHCI_HRTIMER_FREE_ITDS, true);
}
}
/* Wait for controller to stop using old iTDs and siTDs */
static void end_free_itds(struct ehci_hcd *ehci)
{
struct ehci_itd *itd, *n;
struct ehci_sitd *sitd, *sn;
if (ehci->rh_state < EHCI_RH_RUNNING) {
ehci->last_itd_to_free = NULL;
ehci->last_sitd_to_free = NULL;
}
list_for_each_entry_safe(itd, n, &ehci->cached_itd_list, itd_list) {
list_del(&itd->itd_list);
dma_pool_free(ehci->itd_pool, itd, itd->itd_dma);
if (itd == ehci->last_itd_to_free)
break;
}
list_for_each_entry_safe(sitd, sn, &ehci->cached_sitd_list, sitd_list) {
list_del(&sitd->sitd_list);
dma_pool_free(ehci->sitd_pool, sitd, sitd->sitd_dma);
if (sitd == ehci->last_sitd_to_free)
break;
}
if (!list_empty(&ehci->cached_itd_list) ||
!list_empty(&ehci->cached_sitd_list))
start_free_itds(ehci);
}
USB: EHCI: use hrtimer for the IAA watchdog This patch (as1581) replaces the iaa_watchdog kernel timer used by ehci-hcd with an hrtimer event, in keeping with the general conversion to high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:44 +08:00
/* Handle lost (or very late) IAA interrupts */
static void ehci_iaa_watchdog(struct ehci_hcd *ehci)
{
USB: EHCI: reorganize ehci_iaa_watchdog() This patch (as1635) rearranges the control-flow logic in ehci_iaa_watchdog() slightly to agree better with the comments. It also changes a verbose-debug message to a regular debug message. Expiration of the IAA watchdog is an unusual event and can lead to problems; we need to know about it if it happens during debugging. It should not be necessary to set a "verbose" compilation option. No behavioral changes other than the debug message. Lots of apparent changes to the source text, though, because the indentation level was decreased. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-19 00:05:19 +08:00
u32 cmd, status;
USB: EHCI: use hrtimer for the IAA watchdog This patch (as1581) replaces the iaa_watchdog kernel timer used by ehci-hcd with an hrtimer event, in keeping with the general conversion to high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:44 +08:00
USB: EHCI: fix up incorrect merge resolution This patch (as1671) fixes up an incorrect resolution of a merge conflict between Greg KH's usb-linus branch and his usb-next branch. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-22 00:48:42 +08:00
/*
* Lost IAA irqs wedge things badly; seen first with a vt8235.
* So we need this watchdog, but must protect it against both
* (a) SMP races against real IAA firing and retriggering, and
* (b) clean HC shutdown, when IAA watchdog was pending.
*/
USB: EHCI: improve end_unlink_async() This patch (as1665) changes the way ehci-hcd's end_unlink_async() routine works in order to avoid recursive execution and to be more efficient: Now when an IAA cycle ends, a new one gets started up right away (if it is needed) instead of waiting until the just-unlinked QH has been processed. The async_iaa list is renamed to async_idle, which better expresses its new purpose: It is now the list of QHs which are now completely idle and are waiting to be processed by end_unlink_async(). A new flag is added to track whether an IAA cycle is in progress, because the list formerly known as async_iaa no longer stores the QHs waiting for the IAA to finish. The decision about how many QHs to process when an IAA cycle ends is now made at the end of the cycle, when we know the current state of the hardware, rather than at the beginning. This means a bunch of logic got moved from start_iaa_cycle() to end_unlink_async(). Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-23 01:31:58 +08:00
if (!ehci->iaa_in_progress || ehci->rh_state != EHCI_RH_RUNNING)
USB: EHCI: fix up incorrect merge resolution This patch (as1671) fixes up an incorrect resolution of a merge conflict between Greg KH's usb-linus branch and his usb-next branch. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-22 00:48:42 +08:00
return;
USB: EHCI: reorganize ehci_iaa_watchdog() This patch (as1635) rearranges the control-flow logic in ehci_iaa_watchdog() slightly to agree better with the comments. It also changes a verbose-debug message to a regular debug message. Expiration of the IAA watchdog is an unusual event and can lead to problems; we need to know about it if it happens during debugging. It should not be necessary to set a "verbose" compilation option. No behavioral changes other than the debug message. Lots of apparent changes to the source text, though, because the indentation level was decreased. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-19 00:05:19 +08:00
/* If we get here, IAA is *REALLY* late. It's barely
* conceivable that the system is so busy that CMD_IAAD
* is still legitimately set, so let's be sure it's
* clear before we read STS_IAA. (The HC should clear
* CMD_IAAD when it sets STS_IAA.)
*/
cmd = ehci_readl(ehci, &ehci->regs->command);
USB: EHCI: use hrtimer for the IAA watchdog This patch (as1581) replaces the iaa_watchdog kernel timer used by ehci-hcd with an hrtimer event, in keeping with the general conversion to high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:44 +08:00
USB: EHCI: reorganize ehci_iaa_watchdog() This patch (as1635) rearranges the control-flow logic in ehci_iaa_watchdog() slightly to agree better with the comments. It also changes a verbose-debug message to a regular debug message. Expiration of the IAA watchdog is an unusual event and can lead to problems; we need to know about it if it happens during debugging. It should not be necessary to set a "verbose" compilation option. No behavioral changes other than the debug message. Lots of apparent changes to the source text, though, because the indentation level was decreased. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-19 00:05:19 +08:00
/*
* If IAA is set here it either legitimately triggered
* after the watchdog timer expired (_way_ late, so we'll
* still count it as lost) ... or a silicon erratum:
* - VIA seems to set IAA without triggering the IRQ;
* - IAAD potentially cleared without setting IAA.
*/
status = ehci_readl(ehci, &ehci->regs->status);
if ((status & STS_IAA) || !(cmd & CMD_IAAD)) {
COUNT(ehci->stats.lost_iaa);
ehci_writel(ehci, STS_IAA, &ehci->regs->status);
USB: EHCI: use hrtimer for the IAA watchdog This patch (as1581) replaces the iaa_watchdog kernel timer used by ehci-hcd with an hrtimer event, in keeping with the general conversion to high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:44 +08:00
}
USB: EHCI: reorganize ehci_iaa_watchdog() This patch (as1635) rearranges the control-flow logic in ehci_iaa_watchdog() slightly to agree better with the comments. It also changes a verbose-debug message to a regular debug message. Expiration of the IAA watchdog is an unusual event and can lead to problems; we need to know about it if it happens during debugging. It should not be necessary to set a "verbose" compilation option. No behavioral changes other than the debug message. Lots of apparent changes to the source text, though, because the indentation level was decreased. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-03-19 00:05:19 +08:00
ehci_dbg(ehci, "IAA watchdog: status %x cmd %x\n", status, cmd);
end_unlink_async(ehci);
USB: EHCI: use hrtimer for the IAA watchdog This patch (as1581) replaces the iaa_watchdog kernel timer used by ehci-hcd with an hrtimer event, in keeping with the general conversion to high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:44 +08:00
}
USB: EHCI: use hrtimer for the I/O watchdog This patch (as1586) replaces the kernel timer used by ehci-hcd as an I/O watchdog with an hrtimer event. Unlike in the current code, the watchdog event is now always enabled whenever any isochronous URBs are active. This will prevent bugs caused by the periodic schedule wrapping around with no completion interrupts; the watchdog handler is guaranteed to scan the isochronous transfers at least once during each iteration of the schedule. The extra overhead will be negligible: one timer interrupt every 100 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:23:04 +08:00
/* Enable the I/O watchdog, if appropriate */
static void turn_on_io_watchdog(struct ehci_hcd *ehci)
{
/* Not needed if the controller isn't running or it's already enabled */
if (ehci->rh_state != EHCI_RH_RUNNING ||
(ehci->enabled_hrtimer_events &
BIT(EHCI_HRTIMER_IO_WATCHDOG)))
return;
/*
* Isochronous transfers always need the watchdog.
* For other sorts we use it only if the flag is set.
*/
if (ehci->isoc_count > 0 || (ehci->need_io_watchdog &&
ehci->async_count + ehci->intr_count > 0))
ehci_enable_event(ehci, EHCI_HRTIMER_IO_WATCHDOG, true);
}
USB: EHCI: introduce high-res timer This patch (as1572) begins the conversion of ehci-hcd over to using high-resolution timers rather than old-fashioned low-resolution kernel timers. This reduces overhead caused by timer roundoff on systems where HZ is smaller than 1000. Also, the new timer framework introduced here is much more logical and easily extended than the ad-hoc approach ehci-hcd currently uses for timers. An hrtimer structure is added to ehci_hcd, along with a bitflag array and an array of ktime_t values, to keep track of which timing events are pending and what their expiration times are. Only the infrastructure for the timing operations is added in this patch. Later patches will add routines for handling each of the various timing events the driver needs. In some cases the new hrtimer handlers will replace the existing handlers for ehci-hcd's kernel timers; as this happens the old timers will be removed. In other cases the new timing events will replace busy-wait loops. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:21:54 +08:00
/*
* Handler functions for the hrtimer event types.
* Keep this array in the same order as the event types indexed by
* enum ehci_hrtimer_event in ehci.h.
*/
static void (*event_handlers[])(struct ehci_hcd *) = {
USB: EHCI: use hrtimer for async schedule This patch (as1576) adds hrtimer support for managing ehci-hcd's async schedule. Just as with the earlier change to the periodic schedule management, two new hrtimer events take care of everything. One event polls at 1-ms intervals to see when the Asynchronous Schedule Status (ASS) flag matches the Asynchronous Schedule Enable (ASE) value; the schedule's state must not be changed until it does. The other event delays for 15 ms after the async schedule becomes empty before turning it off. The new events replace a busy-wait poll and a kernel timer usage. They also replace the rather illogical method currently used for indicating the async schedule should be turned off: attempting to unlink the dedicated QH at the head of the async list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:21 +08:00
ehci_poll_ASS, /* EHCI_HRTIMER_POLL_ASS */
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
ehci_poll_PSS, /* EHCI_HRTIMER_POLL_PSS */
USB: EHCI: use hrtimer for controller death This patch (as1578) adds an hrtimer event to handle the death of an EHCI controller. When a controller dies, it doesn't necessarily stop running right away. The new event polls at 1-ms intervals to see when all activity has safely stopped. This replaces a busy-wait polling loop in the current code. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:31 +08:00
ehci_handle_controller_death, /* EHCI_HRTIMER_POLL_DEAD */
USB: EHCI: use hrtimer for interrupt QH unlink This patch (as1577) adds hrtimer support for unlinking interrupt QHs in ehci-hcd. The current code relies on a fixed delay of either 2 or 55 us, which is not always adequate and in any case is totally bogus. Thanks to internal caching, the EHCI hardware may continue to access an interrupt QH for more than a millisecond after it has been unlinked. In fact, the EHCI spec doesn't say how long to wait before using an unlinked interrupt QH. The patch sets the delay to 9 microframes minimum, which ought to be adequate. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:26 +08:00
ehci_handle_intr_unlinks, /* EHCI_HRTIMER_UNLINK_INTR */
USB: EHCI: use hrtimer for (s)iTD deallocation This patch (as1579) adds an hrtimer event to handle deallocation of iTDs and siTDs in ehci-hcd. Because of the frame-oriented approach used by the EHCI periodic schedule, the hardware can continue to access the Transfer Descriptor for isochronous (or split-isochronous) transactions for up to a millisecond after the transaction completes. The iTD (or siTD) must not be reused before then. The strategy currently used involves putting completed iTDs on a list of cached entries and every so often returning them to the endpoint's free list. The new strategy reduces overhead by putting completed iTDs back on the free list immediately, although they are not reused until it is safe to do so. When the isochronous endpoint stops (its queue becomes empty), the iTDs on its free list get moved to a global list, from which they will be deallocated after a minimum of 2 ms. This delay is what the new hrtimer event is for. Overall this may not be a tremendous improvement over the current code, but to me it seems a lot more clear and logical. In addition, it removes the need for each iTD to keep a reference to the ehci_iso_stream it belongs to, since the iTD never needs to be moved back to the stream's free list from the global list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:35 +08:00
end_free_itds, /* EHCI_HRTIMER_FREE_ITDS */
USB: EHCI: improve interrupt qh unlink ehci-hcd currently unlinks an interrupt QH when it becomes empty, that is, after its last URB completes. This works well because in almost all cases, the completion handler for an interrupt URB resubmits the URB; therefore the QH doesn't become empty and doesn't get unlinked. When we start using tasklets for URB completion, this scheme won't work as well. The resubmission won't occur until the tasklet runs, which will be some time after the completion is queued with the tasklet. During that delay, the QH will be empty and so will be unlinked unnecessarily. To prevent this problem, this patch adds a 5-ms time delay before empty interrupt QHs are unlinked. Most often, during that time the interrupt URB will be resubmitted and thus we can avoid unlinking the QH. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Ming Lei <ming.lei@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2013-07-03 22:53:10 +08:00
ehci_handle_start_intr_unlinks, /* EHCI_HRTIMER_START_UNLINK_INTR */
USB: EHCI: use hrtimer for unlinking empty async QHs This patch (as1583) changes ehci-hcd to use an hrtimer event for unlinking empty (unused) async QHs instead of using a kernel timer. The check for empty QHs is moved to a new routine, where it doesn't require going through an entire scan of both the async and periodic schedules. And it can unlink multiple QHs at once, unlike the current code. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:53 +08:00
unlink_empty_async, /* EHCI_HRTIMER_ASYNC_UNLINKS */
USB: EHCI: use hrtimer for the IAA watchdog This patch (as1581) replaces the iaa_watchdog kernel timer used by ehci-hcd with an hrtimer event, in keeping with the general conversion to high-res timers. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:44 +08:00
ehci_iaa_watchdog, /* EHCI_HRTIMER_IAA_WATCHDOG */
USB: EHCI: use hrtimer for the periodic schedule This patch (as1573) adds hrtimer support for managing ehci-hcd's periodic schedule. There are two issues to deal with. First, the schedule's state (on or off) must not be changed until the hardware status has caught up with the current command. This is handled by an hrtimer event that polls at 1-ms intervals to see when the Periodic Schedule Status (PSS) flag matches the Periodic Schedule Enable (PSE) value. Second, the schedule should not be turned off as soon as it becomes empty. Turning the schedule on and off takes time, so we want to wait until the schedule has been empty for a suitable period before turning it off. This is handled by an hrtimer event that gets set to expire 10 ms after the periodic schedule becomes empty. The existing code polls (for up to 1125 us and with interrupts disabled!) to check the status, and doesn't implement a delay before turning off the schedule. Furthermore, if the polling fails then the driver decides that the controller has died. This has caused problems for several people; some controllers can take 10 ms or more to turn off their periodic schedules. This patch fixes these issues. It also makes the "broken_periodic" workaround unnecessary; there is no longer any danger of turning off the periodic schedule after it has been on for less than 1 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:05 +08:00
ehci_disable_PSE, /* EHCI_HRTIMER_DISABLE_PERIODIC */
USB: EHCI: use hrtimer for async schedule This patch (as1576) adds hrtimer support for managing ehci-hcd's async schedule. Just as with the earlier change to the periodic schedule management, two new hrtimer events take care of everything. One event polls at 1-ms intervals to see when the Asynchronous Schedule Status (ASS) flag matches the Asynchronous Schedule Enable (ASE) value; the schedule's state must not be changed until it does. The other event delays for 15 ms after the async schedule becomes empty before turning it off. The new events replace a busy-wait poll and a kernel timer usage. They also replace the rather illogical method currently used for indicating the async schedule should be turned off: attempting to unlink the dedicated QH at the head of the async list. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:22:21 +08:00
ehci_disable_ASE, /* EHCI_HRTIMER_DISABLE_ASYNC */
USB: EHCI: use hrtimer for the I/O watchdog This patch (as1586) replaces the kernel timer used by ehci-hcd as an I/O watchdog with an hrtimer event. Unlike in the current code, the watchdog event is now always enabled whenever any isochronous URBs are active. This will prevent bugs caused by the periodic schedule wrapping around with no completion interrupts; the watchdog handler is guaranteed to scan the isochronous transfers at least once during each iteration of the schedule. The extra overhead will be negligible: one timer interrupt every 100 ms. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:23:04 +08:00
ehci_work, /* EHCI_HRTIMER_IO_WATCHDOG */
USB: EHCI: introduce high-res timer This patch (as1572) begins the conversion of ehci-hcd over to using high-resolution timers rather than old-fashioned low-resolution kernel timers. This reduces overhead caused by timer roundoff on systems where HZ is smaller than 1000. Also, the new timer framework introduced here is much more logical and easily extended than the ad-hoc approach ehci-hcd currently uses for timers. An hrtimer structure is added to ehci_hcd, along with a bitflag array and an array of ktime_t values, to keep track of which timing events are pending and what their expiration times are. Only the infrastructure for the timing operations is added in this patch. Later patches will add routines for handling each of the various timing events the driver needs. In some cases the new hrtimer handlers will replace the existing handlers for ehci-hcd's kernel timers; as this happens the old timers will be removed. In other cases the new timing events will replace busy-wait loops. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-07-11 23:21:54 +08:00
};
static enum hrtimer_restart ehci_hrtimer_func(struct hrtimer *t)
{
struct ehci_hcd *ehci = container_of(t, struct ehci_hcd, hrtimer);
ktime_t now;
unsigned long events;
unsigned long flags;
unsigned e;
spin_lock_irqsave(&ehci->lock, flags);
events = ehci->enabled_hrtimer_events;
ehci->enabled_hrtimer_events = 0;
ehci->next_hrtimer_event = EHCI_HRTIMER_NO_EVENT;
/*
* Check each pending event. If its time has expired, handle
* the event; otherwise re-enable it.
*/
now = ktime_get();
for_each_set_bit(e, &events, EHCI_HRTIMER_NUM_EVENTS) {
if (now.tv64 >= ehci->hr_timeouts[e].tv64)
event_handlers[e](ehci);
else
ehci_enable_event(ehci, e, false);
}
spin_unlock_irqrestore(&ehci->lock, flags);
return HRTIMER_NORESTART;
}