This patch fixes the driver to allow both CONFIG_USB_EHCI_HCD_PPC_OF and
CONFIG_USB_ECHI_HCD_XILINX to be selected.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
CC: John Linn <john.linn@xilinx.com>
CC: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The usbcore headers: hcd.h and hub.h are shared between usbcore,
HCDs and a couple of other drivers (e.g. USBIP modules).
So, it makes sense to move them into a more public location and
to cleanup dependency of those modules on kernel internal headers.
This patch moves hcd.h from drivers/usb/core into include/linux/usb/
Signed-of-by: Eric Lescouet <eric@lescouet.org>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1369) fixes a problem in ehci-hcd. Some controllers
occasionally run into trouble when the driver reclaims siTDs too
quickly. This can happen while streaming audio; it causes the
controller to crash.
The patch changes siTD reclamation to work the same way as iTD
reclamation: Completed siTDs are stored on a list and not reused until
at least one frame has passed.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Nate Case <ncase@xes-inc.com>
CC: <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
The EHCI driver stores in usb_host_endpoint.hcpriv a pointer to either
an ehci_qh or an ehci_iso_stream structure, and uses the contents of the
hw_info1 field to distinguish the two cases.
After ehci_qh was split into hw and sw parts, ehci_iso_stream must also
be adjusted so that it again looks like an ehci_qh structure.
This fixes a NULL pointer access in ehci_endpoint_disable() when it
tries to access qh->hw->hw_info1.
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Reported-by: Colin Fletcher <colin.m.fletcher@googlemail.com>
Cc: stable <stable@kernel.org>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1330) fixes a bug in khbud's handling of remote
wakeups. When a device sends a remote-wakeup request, the parent hub
(or the host controller driver, for directly attached devices) begins
the resume sequence and notifies khubd when the sequence finishes. At
this point the port's SUSPEND feature is automatically turned off.
However the device needs an additional 10-ms resume-recovery time
(TRSMRCY in the USB spec). Khubd does not wait for this delay if the
SUSPEND feature is off, and as a result some devices fail to behave
properly following a remote wakeup. This patch adds the missing
delay to the remote-wakeup path.
It also extends the resume-signalling delay used by ehci-hcd and
uhci-hcd from 20 ms (the value in the spec) to 25 ms (the value we use
for non-remote-wakeup resumes). The extra time appears to help some
devices.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@kernel.org>
Cc: Rickard Bellini <rickard.bellini@ericsson.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1300) adds native scatter-gather support to ehci-hcd.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The Freescale MX27 and MX31 SoCs have a EHCI controller onboard.
The controller is capable of USB on the go. This patch adds
a driver to support all three of them.
Users have to pass details about serial interface configuration in the
platform data.
The USB OTG core used here is the ARC core, so the driver should
be renamed and probably be merged with ehci-fsl.c eventually.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Daniel Mack <daniel@caiaq.de>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The EHCI specification says that an EHCI host controller may cache part of
the isochronous schedule. The EHCI controller must advertise how much it
caches in the schedule through the HCCPARAMS register isochronous
scheduling threshold (IST) bits.
In theory, adding new iTDs within the IST should be harmless. The HW will
follow the old cached linked list and miss the new iTD. SW will notice HW
missed the iTD and return 0 for the transfer length.
However, Intel ICH9 chipsets (and some later chipsets) have issues when SW
attempts to schedule a split transaction within the IST. All transfers
will cease being sent out that port, and the drivers will see isochronous
packets complete with a length of zero. Start of frames may or may not
also disappear, causing the device to go into auto-suspend. This "bus
stall" will continue until a control or bulk transfer is queued to a
device under that roothub.
Most drivers will never cause this behavior, because they use multiple
URBs with multiple packets to keep the bus busy. If you limit the number
of URBs to one, you may be able to hit this bug.
Make sure the EHCI driver does not schedule full-speed transfers within
the IST under an Intel chipset. Make sure that when we fall behind the
current microframe plus IST, we schedule the new transfer at the next
periodic interval after the IST.
Don't change the scheduling for new transfers, since the schedule slop will
always be greater than the IST. Allow high speed isochronous transfers to
be scheduled within the IST, since this doesn't trigger the Intel chipset
bug.
Make sure that if the host caches the full frame, the EHCI driver's
internal isochronous threshold (ehci->i_thresh) is set to
8 microframes + 2 microframes wiggle room. This is similar to what is done in
the case where the host caches less than the full frame.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Clemens Ladisch <clemens@ladisch.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
this driver has been sitting in linux-omap tree for quite
some time. It adds support for omap's ehci controller.
Signed-off-by: Felipe Balbi <felipe.balbi@nokia.com>
Signed-off-by: Vikram Pandita <vikram.pandita@ti.com>
Signed-off-by: Ajay Kumar Gupta <ajay.gupta@ti.com>
Signed-off-by: Anand Gadiyar <gadiyar@ti.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add bus glue driver for Xilinx USB host controller. The controller can be
configured as HS only or HS/FS hybrid. The driver uses the device tree file
to configure the driver according to the setting in the hardware system.
This driver has been tested with usbtest using the NET2280 PCI card.
Signed-off-by: Julie Zhu <julie.zhu@xilinx.com>
Signed-off-by: John Linn <john.linn@xilinx.com>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
a quirky chipset needs periodic schedules to run for a minimum
time before they can be disabled again. This enforces the requirement
with a time stamp and a calculated delay
Signed-off-by: Oliver Neukum <oliver@neukum.org>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If the EHCI debug port is initialized and in use, the EHCI host
controller driver must follow two rules.
1) If the EHCI host driver issues a controller reset, the debug
controller driver re-initialization must get called after the reset
is completed.
2) The EHCI host driver should ignore any requests to the physical
EHCI debug port when the EHCI debug port is in use.
The code to check for the debug port was moved from ehci_pci_reinit()
to ehci_pci_setup because it must get called prior to ehci_reset()
which will clear the debug port registers.
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: dbrownell@users.sourceforge.net
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1281) changes the way ehci-hcd deschedules interrupt
QHs, copying the approach used for async QHs. The caller is no longer
responsible for rescheduling the QH if its queue is non-empty; instead
the reschedule is done directly by intr_deschedule(), after calling
qh_completions(). This is exactly the same as how end_unlink_async()
works.
ehci_urb_dequeue() and intr_deschedule() now correctly handle the case
where they are called while another interrupt URB for the same QH is
being given back. This was a surprisingly large blind spot. And
scan_periodic() now respects the new needs_rescan flag.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: David Brownell <david-b@pacbell.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1280) fixes an obscure bug in ehci-hcd's dequeuing logic
for async URBs. If a later URB is unlinked and the completion
routine unlinks an earlier URB, then the earlier URB won't be given
back in a timely manner because the endpoint queue isn't rescanned as
it should be.
Similar bugs occur if an endpoint is reset or a halt is cleared while
a completion routine is running, because the subroutines don't test
for the COMPLETING state.
All these problems are solved by adding a new needs_rescan flag to the
ehci_qh structure. If the flag is set while scanning through an idle
QH, the scan will be repeated. If the QH isn't idle then an unlink
cycle will be initiated, and the proper action will be taken when it
becomes idle.
Also, an unnecessary test is removed from qh_link_async(): That
routine is never called if the QH's state isn't IDLE.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: David Brownell <david-b@pacbell.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add host USB High speed driver for at91sam9g45 series.
The host driver is an EHCI with its companion OHCI. EHCI is
handled by the new ehci-atmel.c whereas the OHCI is always
handled by ohci-at91.c.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
The Intel Moorestown EHCI controller supports non-standard HOSTPCx register
extension. This register controls the LPM behaviour and controls the behaviour
of each USB port.
Signed-off-by: Jacob Pan <jacob.jun.pan@intel.com>
Signed-off-by: Alek Du <alek.du@intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The ehci_qh structure merged hw and sw together which is not good:
1. More and more items are being added into ehci_qh, the ehci_qh software
part are unnecessary to be allocated in DMA qh_pool.
2. If HCD has local SRAM, the sw part will consume it too, and it won't
bring any benefit.
3. For non-cache-coherence system, the entire ehci_qh is uncachable, actually
we only need the hw part to be uncacheable. Spliting them will let the sw
part to be cacheable.
Signed-off-by: Alek Du <alek.du@intel.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
CC: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Basically the io watchdog is only useful for those quirk HCDs. For most
good ones, it only brings unnecessary wakeups. At least, I know the
Intel EHCI HCDs should turn off the flag.
Signed-off-by: Alek Du <alek.du@intel.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add ehci support for w90p910 platform.
Signed-off-by: Wan ZongShun <mcuos.com@gmail.com>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1273) fixes two(!) bugs introduced by the new
Clear-TT-Buffer implementation in ehci-hcd.
It is now possible for an idle QH to have some URBs on its
queue -- this will happen if a Clear-TT-Buffer is pending for
the QH's endpoint. Consequently we should not issue a warning
when someone tries to unlink an URB from an idle QH; instead
we should process the request immediately.
The refcounts for QHs could get messed up, because
submit_async() would increment the refcount when calling
qh_link_async() and qh_link_async() would then refuse to link
the QH into the schedule if a Clear-TT-Buffer was pending.
Instead we should increment the refcount only when the QH
actually is added to the schedule. The current code tries to
be clever by leaving the refcount alone if an unlink is
immediately followed by a relink; the patch changes this to an
unconditional decrement and increment (although they occur in
the opposite order).
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: David Brownell <david-b@pacbell.net>
Tested-by: Manuel Lauss <manuel.lauss@gmail.com>
Tested-by: Matthijs Kooijman <matthijs@stdin.nl>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1256) changes ehci-hcd and all the other drivers in the
EHCI family to make use of the new clear_tt_buffer callbacks. When a
Clear-TT-Buffer request is in progress for a QH, the QH is not allowed
to be linked into the async schedule until the request is finished.
At that time, if there are any URBs queued for the QH, it is linked
into the async schedule.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is an update to the "usb-ehci-update-toggle-state-for-linked-qhs"
patch. Since an HCD's endpoint_reset method can be called in
interrupt context, it mustn't assume that interrupts are enabled or
that it can sleep.
So we revert to the original way of refreshing QHs' toggle bits. Now
the endpoint_reset method merely clears the toggle flag in the device
structure (as was done before) and starts an async QH unlink. When the
QH is linked again, after the unlink finishes and an URB is queued,
the qh_refresh() routine will update the QH's toggle bit.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: David <david@unsolicited.net>
CC: David Brownell <david-b@pacbell.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1245) fixes a bug in ehci-hcd. When an URB is queued
for an endpoint whose QH is already in the LINKED state, the QH
doesn't get refreshed. As a result, if usb_clear_halt() was called
during the time that the QH was linked but idle, the data toggle value
in the QH doesn't get reset.
The symptom is that after a clear_halt, data gets lost and transfers
time out. This problem is starting to show up now because the
"ehci-hcd unlink speedups" patch causes QHs with no queued URBs to
remain linked for a suitable time.
The patch utilizes the new endpoint_reset mechanism to fix the
problem. When an endpoint is reset, the new method forcibly unlinks
the QH (if necessary) and safely updates the toggle value. This
allows qh_update() to be simplified and avoids using usb_device's
toggle bits in a rather unintuitive way.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: David Brownell <david-b@pacbell.net>
Tested-by: David <david@unsolicited.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Replace all DMA_64BIT_MASK macro with DMA_BIT_MASK(64)
Signed-off-by: Yang Hongyang<yanghy@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch (as1205) moves timer_action() from ehci.h to ehci-hcd.c and
makes it out-of-line. Over the years it has grown too big to be inline
any more.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Currently ITDs are immediately recycled whenever their URB completes.
However, EHCI hardware can sometimes remember some ITD state. This
means that when the ITD is reused before end-of-frame it may sometimes
cause the hardware to reference bogus state.
This patch defers reusing such ITDs by moving them into a new ehci member
cached_itd_list. ITDs resting in cached_itd_list are moved back into their
stream's free_list once scan_periodic() detects that the active frame has
elapsed.
This makes the snd_usb_us122l driver (in kernel since .28) work right
when it's hooked up through EHCI.
[ dbrownell@users.sourceforge.net: comment fixups ]
Signed-off-by: Karsten Wiese <fzu@wemgehoertderstaat.de>
Tested-by: Philippe Carriere <philippe-f.carriere@wanadoo.fr>
Tested-by: Federico Briata <federicobriata@gmail.com>
Cc: stable <stable@kernel.org>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1165) makes a few small changes in the logic used by
ehci-hcd when it encounters a controller error:
Instead of printing out the masked status, it prints the
original status as read directly from the hardware.
It doesn't check for the STS_HALT status bit before taking
action. The mere fact that the STS_FATAL bit is set means
that something bad has happened and the controller needs to
be reset. With the old code this test could never succeed
because the STS_HALT bit was masked out from the status.
I anticipate that this will prevent the occasional "irq X: nobody cared"
problem people encounter when their EHCI controllers die.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: David Brownell <david-b@pacbell.net>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This gets rid of an annoying warning in ehci-hcd.c when DEBUG isn't
enabled:
warning: label 'err_debug' defined but not used
by moving it inside the already-existing #ifdef DEBUG, so that it
matches the goto. And now my regular build is warning-free again.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch (as1147) fixes the remote-wakeup support for EHCI
controllers using the ARC/TDI "embedded-TT" core. These controllers
turn off the RESUME bit by themselves when a port resume is complete;
hence we need to keep separate track of which ports are suspended or
in the process of resuming.
The patch also makes a couple of small improvements in ehci_irq(),
replacing reads of the command register with the value already stored
in a local variable.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Thomas Reitmayr <treitmayr@devbase.at>
CC: David Brownell <david-b@pacbell.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1139) adds a warning to the system log whenever ehci-hcd
is loaded after ohci-hcd or uhci-hcd. Nowadays most distributions are
pretty good about not doing this; maybe the warning will help convince
anyone still doing it wrong.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@kernel.org> [2.6.27]
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1145) removes the essentially useless driver-version
strings from ehci-hcd, ohci-hcd, and uhci-hcd. It also unifies the
form of the banner lines they display upon loading and adds a missing
test for usb_disabled() to ehci-hcd.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Now that arch/ppc is gone we don't need CONFIG_PPC_MERGE anymore remove
the dead code associated with !CONFIG_PPC_MERGE.
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch is based on the following ideas:
1. Some usb devices (such as usb video class) have endpoints of high
interval attribute, so reading "periodic" file need more debug buffer
to accommodate the qh or itd schedule information. For example, 4KB
buffer is not enough for a single interrupt qh of 2ms period.
2. print a %p need 16 byte buffer on 64-bits arch, but 8 byte on 32-bits
arch. Add a extra bonus for 64-bits arch.
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
I noticed that the "Refactor "if (handshake()) state = HC_STATE_HALT"
patch from earlier this year perpetuated a potential problem: it can
mark the controller as halted when it's still running (but not acting
as, perhaps wrongly, expected).
That caused some hangs and crashes, rather than more polite failure
modes of a truly halted controller. This patch forces a true halt,
and emits a (previously missing) diagnostic.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes some performance bugs observed with some workloads
when unlinking EHCI queue header (QH) descriptors from the async ring
(control/bulk schedule).
The mechanism intended to defer unlinking an empty QH (so there is no
penalty in common cases where it's quickly reused) was not working as
intended. Sometimes the unlink was scheduled:
- too quickly ... which can be a *strong* negative effect, since
that QH becomes unavailable for immediate re-use;
- too slowly ... wasting DMA cycles, usually a minor issue except
for increased bus contention and power usage;
Plus there was an extreme case of "too slowly": a logical error in the
IAA watchdog-timer conversion meant that sometimes the unlink never
got scheduled.
The fix replaces a simple counter with a timestamp derived from the
controller's 8 KHz microframe counter, and adjusts the timer usage
for some issues associated with HZ being less than 8K.
(Based on a patch originally by Alan Stern, and good troubleshooting
from Leonid.)
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Leonid <leonidv11@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch refactors some shutdown code so it can be shared between
ehci_stop() and ehci_shutdown().
This also fixes a couple potential bugs:
- ehci_shutdown() was not locking ehci->lock before halting the HC.
- ehci_shutdown() didn't disable the watchdog and IAA timers.
- ehci_stop() was resetting the host controller when it may have been
running, which the EHCI spec says "may result in undefined behavior".
ehci_stop() was calling port_power() to turn off the ports, which waited
20ms after applying the port change. The msleep was for the case where
the HC might take 20ms to turn the ports on; since we're shutting them
off, we can avoid the msleep and just use ehci_turn_off_ports().
ehci_stop() doesn't need to clear the intr_enable register or revert
ownership of the companion controllers to the BIOS, because the host
controller reset should have done that. There might be a buggy host
controller that doesn't follow the reset rules, but for now we assume
it's redundant code and remove it.
[ A subsequent patch will cancel the timers later ... this version
carries forward existing bugs where timers could get re-armed
after they're canceled. ]
Signed-off-by: Sarah Sharp <sarah.a.sharp@intel.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Some EHCI fault paths with large control transfers aren't coded. Avoid
problems by rejecting transfers that may need two qTDs (16+ KB). This is
mostly paranoia; even 4 KB transfers are rare, and most HCDs use lower
limits (so it's unlikely anyone would ever try such a thing).
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It was pointed out that we found and fixed the cause of the "bogus"
fatal IRQ reports some time ago ... this patch removes the code
which was working around that bug ("status" got clobbered), and a
comment which needlessly confused folk reading this code.
This also includes a minor cleanup to the code which fixed that bug.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Refactor the EHCI "if (handshake()) state = HC_STATE_HALT" idiom,
which appears 4 times, by replacing it with calls to a new function
called handshake_on_error_set_halt(). Saves a few bytes too.
Signed-off-by: Karsten Wiese <fzu@wemgehoertderstaat.de>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Minor cleanups to the EHCI code: revision history is what source
code repositories should have. Switch to a more standard way to
kick in verbose debugging -- don't be EHCI-specific.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The recent EHCI driver update to split the IAA watchdog timer out from
the other timers made several things work better, but not everything;
and it created a couple new issues in bugzilla. Ergo this patch:
- Handle a should-be-rare SMP race between the watchdog firing
and (very late) IAA interrupts;
- Remove a shouldn't-have-been-added WARN_ON() test;
- Guard against one observed OOPS;
- If this watchdog fires during clean HC shutdown, it should act
as a NOP instead of interfering with the shutdown sequence;
- Guard against silicon errata hypothesized by some vendors:
* IAA status latch broken, but IAAD cleared OK;
* IAAD wasn't cleared when IAA status got reported;
The WARN_ON is in bugzilla as 10168; the OOPS as 10078; these are
both regressions.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Tested-by: Gordon Farquharson <gordonfarquharson@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Currently, this setup:
CONFIG_USB_ARCH_HAS_EHCI=y
CONFIG_USB_EHCI_HCD=y
CONFIG_USB_EHCI_HCD_PPC_OF=y
Will fail to build:
CC drivers/usb/host/ehci-hcd.o
drivers/usb/host/ehci-hcd.c:1018:2: error: #error "missing bus glue for ehci-hcd"
make[3]: *** [drivers/usb/host/ehci-hcd.o] Error 1
ehci-hcd.c actually contains OF_PLATFORM_DRIVER glue, so error is bogus.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The previous fix for a "sparse" warning in ehci_urb_dequeue() was
incorrect. After rescheduling interrupt transfers it returned the
URB's completion status, not status for the dequeue operation itself.
This patch resolves that issue, cleans up the code in the reschedule
path, and shrinks the object code by a dozen bytes.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
