differ in their details with calls to a new function, ehci_hcreset(),
that performs the reset.
The original sequences either had no delay or a 1ms delay between
telling the controller to stop and asserting the controller reset
bit. One instance of the original reset sequence waited for the
controller to indicate that its reset was complete before continuing,
but the other two immediately let the subsequent code execute. The
latter is a problem on some hardware, because a read of the HCCPARAMS
register returns an incorrect value while the reset is in progress,
which triggers an infinite loop in ehci_pci_givecontroller(), which
hangs the system on shutdown.
The reset sequence in ehci_hcreset() starts with the most complete
instance from the original code, which contains a loop to wait for
the controller to indicate that its reset is complete. This appears
to be the correct thing to do according to "Enhanced Host Controller
Interface Specification for Universal Serial Bus" revision 1.0,
section 2.3.1. Add another loop to wait for the controller to
indicate that it has stopped before setting the HCRESET bit. This
is required by the section 2.3.1 in the specification, which says
that setting HCRESET before the controller has halted "will result
in undefined behaviour".
Reviewed by: imp (previous patch version without the extra wait loop)
Tested by: se (previous patch version without the extra wait loop)
Approved by: re (bmah)
MFC after: 1 week
tasks. Since the host controllers rely on tasks to process transfer
timeouts, if a synchronous transfer from a driver was invoked from
a task and timed out, it would never complete because the single
task thread was stuck performing the synchronous transfer so couldn't
process the timeout.
This affected the axe, udav and ural drivers.
Problem hardware provided by: guido
close and re-open the default pipe instead of relying on the host
controller driver to notice the changes. Remove the unreliable code
that attempted to update these fields while the pipe was active.
This fixes a case where the hardware could cache and continue to
use the old address, resulting in a "getting first desc failed"
error.
PR: usb/103167
s/device_ptr_t/device_t/g
No md5 changes in the .o's
# Note to the md5 tracking club: $FreeBSD$ changes md5 after every commit
# so you need to checkout -kk to get $FreeBSD$ instead of the actual value
# of the keyword.
USBD_FORCE_SHORT_XFER to ensure that we actually build and execute
a transfer. This means that the various alloc_sqtd_chain functions
will always construct a transfer, so it is safe to modify the
allocated descriptors on return. Previously there were cases where
a zero length transfer would cause a NULL dereference.
Reported by: bp
host controllers to avoid the need to allocate any multi-page
physically contiguous memory blocks. This makes it possible to use
USB devices reliably on low-memory systems or when memory is too
fragmented for contiguous allocations to succeed.
The USB subsystem now uses bus_dmamap_load() directly on the buffers
supplied by USB peripheral drivers, so this also avoids having to
copy data back and forth before and after transfers. The ehci and
ohci controllers support scatter/gather as long as the buffer is
contiguous in the virtual address space. For uhci the hardware
cannot handle a physical address discontinuity within a USB packet,
so it is necessary to copy small memory fragments at times.
EHCI spec for linking in new qTDs into an asynchronous QH. This
requires that there is a qTD marked as not active and not halted
at the start of the QH's list, and the hardware will know to re-fetch
the qTD on each pass rather than just looking at the overlay qTD:
"The host controller must be able to advance the queue from the
Fetch QH state in order to avoid all hardware/software race
conditions. This simple mechanism allows software to simply link
qTDs to the queue head and activate them, then the host controller
will always find them if/when they are reachable."
This is achieved by keeping an "inactivesqtd" entry on the QH list,
and re-using it each time as the start of the next transfer, and
allocating a new qTD to become the next inactivesqtd. Then a new
transfer can be activated by just setting its "active" flag, which
avoids all the previous messing with overlay qTD state in
ehci_set_qh_qtd().
transfers. This fixes some cases where the software toggle tracking
was not doing the right thing. For example, a short transfer that
transferred 0 bytes of the requested qTD transfer size does cause
a toggle change, but the existing code was assuming it didn't.
Reported and tested by: pav
MFC after: 2 weeks
is closed and then reopened. This appears to be necessary now that
we no longer clear endpoint stalls every time a pipe is opened.
Previously we could assume an initial toggle value of zero because
the clear-stall operation resets the device's toggle state.
Reported by: Holger Kipp
MFC after: 3 days
ATI EHCI controllers exhibit simmilar stall issues and require
this dropped interrupts workaround. Be verbose about it.
ehci.c:
ehcivar.h:
Slight change in comments to note about issues surrounding both
VIA and ATI EHCI controllers.
Approved by: iedowse
an interrupt appears to occur before the transfer has been marked
as completed. This caused umass transfers to get stuck, especially
when writing large files. The workaround sets up a timer that
rechecks for missed completed transfers if some operations are still
pending. Other suggested workarounds, such as performing a PCI read
immediately after acknowledging the interrupts, do not appear to
help.
Obtained from: OpenBSD
transfer, which lead to panics or page faults. For example if a
transfer timed out, another thread could come along and attempt to
abort the same transfer while the timeout task was sleeping in
the *_abort_xfer() function.
Add an "aborting" flag to the private transfer state in each host
controller driver and use this to ensure that the abort is only
executed once. Also prioritise normal abort requests over timeouts
so that the callback is always given a status of USB_CANCELLED even
if the timeout-initiated abort began first.
The crashes caused by this bug were mainly reported in connection
with lpd printing to a USB printer.
PR: usb/78208, usb/78986
cleared if the host controller retries the transfer and is successful,
but we were interpreting these bits as indicating a fatal error.
Ignore these error bits, and instead use the HALTED bit to determine
if the transfer failed. Also update the USBD_STALLED detection to
ignore these bits.
Obtained from: OpenBSD
between passes over a QH. Previously the accesses to a QH were
bunched together in time, so the interval was often much longer
than intended. This now appears to match the diagrams in the EHCI
spec, so remove the XXX comment.
ever working correctly: the code was linking the QHs together but
then immediately overwriting the "next" pointers. Oops. Also
initialise qh_endphub, since the EHCI spec says that we should
always set the pipe multiplier field to something sensible.
This appears to make basic split transactions work, so enable split
transactions for control, bulk and interrupt pipes (split isochronous
transfers are not yet implemented). It should now be possible to
use USB1 devices even when they are connected through a USB2 hub.
1/ doesn't matter on most of our architectures
2/ will never happen unless we start queueing multiple trasactions
to a single endpoint at one time (which we do not allow yet).
If anyone has a big_endian machine with EHCI they might check this
if they are having problems with EHCI but it's unlikely even there..
Submitted by: Hans Petter Selasky <hselasky@c2i.net>
MFC after: 3 days
to remove a transaction from the async schedule. The previous method didn't
work well and led to the hardware writing to free'd buffers etc, as
it didn't always know that the transaction had been aborted.
Written after consultation with David Brownell who wrote the Linux
EHCI driver.
As part of this give the sqh structure a "previous" pointer.
MFC after: 1 week
backed out commits were trying to address: when cancelling the timeout
callout, also cancel the abort_task event, since it is possible that
the timeout has already fired and set up an abort_task.
reports of problems. The bug is probably that there are cases where
`xfer->timeout && !sc->sc_bus.use_polling' is not a suitable test
for an active timeout callout, so an explicit flag will be necessary.
Apologies for the breakage.
transfer timeouts that typically cause a transfer to be completed
twice, resulting in panics and page faults:
o A transfer completion interrupt could arrive while an abort_task
event was set up, so the transfer would be aborted after it had
completed. This is very easy to reproduce. Fix this by setting
the transfer status to USBD_TIMEOUT before scheduling the
abort_task so that the transfer completion code will ignore it.
o The transfer completion code could execute concurrently with the
timeout callout, leaving the callout blocked (e.g. waiting for
Giant) while the transfer completion code runs. In this case,
callout_stop() does not prevent the callout from running, so
again the timeout code would run after the transfer was complete.
Handle this case by checking the return value from callout_stop(),
and ignoring the transfer if the callout could not be removed.
o Finally, protect against a timeout callout occurring while a
transfer is being aborted by another process. Here we arrange
for the timeout processing to ignore the transfer, and use
callout_drain() to ensure that the callout has really gone before
completing the transfer.
This was tested by repeatedly performing USB transfers with a timeout
set to approximately the same as the normal transfer completion
time. In the PR below, apparently this occurred by accident with a
particular printer and the default timeout.
PR: kern/71491
just a convenience function to be called from debuggers that gets
compiled in when EHCI_DEBUG is defined. Move its declaration to
make this more obvious.
o Reduce the interrupt delay to 2 microframes.
o Follow the spec more closely when updating the overlay qTD in the QH.
o No need to generate an interrupt at the data part of a control
transfer, it's generated by the status transfer.
o Make sure to update the data toggle on short transfers.
o Turn the printf about needing toggle update into a DPRINTF.
o Keep track of what high speed port (if any) a device belongs to
so we can set the transaction translator fields for the transfer.
o Verbosely refuse to open low/full speed pipes that depend on
unimplemented split transaction support.
o Fix various typos in comments.
Obtained from: NetBSD
system BIOS to disable legacy device emulation as per the "EHCI
Extended Capability: Pre-OS to OS Handoff Synchronisation" section
of the EHCI spec. BIOSes that implement legacy emulation using SMIs
are supposed to disable the emulation when this procedure is performed.
to be particularly correct or optimal, but it seems to be enough
to allow the attachment of USB2 hubs and USB2 devices connected via
USB2 hubs. None of the split transaction support is implemented in
our USB stack, so USB1 peripherals will definitely not work when
connected via USB2 hubs.
rev. 1.67, author: mycroft
Fix a byte order error.
rev. 1.68, author: mycroft
Adjust some silliness that was causing us to do extra work for
"frame list rollover" interrupts, which we pretty much ignore.
Obtained from: NetBSD
ehci.c (1.55), ehcireg.h (1.16); author: mycroft
Set the data toggle correctly, and use EHCI_QTD_DTC. This fixes
problems with my ALi-based drive enclosure (it works now, rather
than failing to attach). Also seems to work with a GL811-based
enclosure and an ASUS enclosure with a CD-RW, on both Intel and
NEC controllers.
Note: The ALi enclosure is currently very SLOW, due to some issue
with taking too long to notice that the QTD is complete. This
requires more investigation.
ehci.c (1.56); author: mycroft
Failure to properly mask off UE_DIR_IN from the endpoint address
was causing OHCI_ED_FORMAT_ISO and EHCI_QH_HRECL to get set
spuriously, causing rather interesting lossage.
Suddenly I get MUCH better performance with ehci...
ehci.c (1.58); author: mycroft
Fix a stupid bug in ehci_check_intr() that caused use to try to
complete a transaction that was still running. Now ehci can
handle multiple devices being active at once.
ehci.c (1.59); author: enami
As the ehci_idone() now uses the variable `epipe'
unconditionally, always declare it (in other words, make this
file compile w/o EHCI_DEBUG).
ehci.c (1.60); author: mycroft
Remove comment about the data toggle being borked.
ehci.c (1.61); author: mycroft
Update comment.
ehci.c (1.62); author: mycroft
Adjust a couple of comments to make it clear WTF is going on.
ehci.c (1.63); author: mycroft
Fix an error in a debug printf().
ehci.c (1.64), ehcireg.h (1.17); author: mycroft
Further cleanup of toggle handling. Now that we use EHCI_QH_DTC,
we don't need to fiddle with the TOGGLE bit in the overlay
descriptor, so minimize how much we fuss with it.
Obtained from: NetBSD
