Fix a possible null-pointer dereference (hopefully) before anyone runs
into this. Also don't switch ports to xHCI if initialization failed.
Change-Id: I5dbaeb435a98ead0b50d27fde13c9f1433ea3e81
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/6245
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
As the controller structure is never fully cleared, this one wasn't
initialized for non-pci controllers (but checked for non-null later).
Change-Id: I852671c5f55650bdb6cd97f4ec74b1f95ee894c7
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/6246
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Using void* for physical addresses leads to much casting and confuses
developers when to convert from physical to virtual addresses or
the other way around. When using plain integers for physical addresses
and pointers for virtual addresses things become much cleaner and we
won't ever end up dereferencing a physical address.
Change-Id: I24cd53b81c7863b6d14f0cbb4ce8937728b37c1c
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/6244
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Remove a redundant phys_to_virt() that sneaked in the initialization of
PCI xHCI controllers. The use of casts from void* to u32 (and vice versa)
prompts for things going wrong here. That will be addressed in a later
commit.
Change-Id: Ibc71ed6ee7016529c0e3a51559aaec07aaaba315
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/6243
Tested-by: build bot (Jenkins)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
being a good citizen on the box, libpayload tries to return to EHCI
mode on shutdown, so a non-XHCI capable USB driver after it (eg. in
the OS) finds something to work with.
Change-Id: Id227d646e08a258b841c644263112f0815dd486c
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-on: http://review.coreboot.org/4547
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The EHCI driver defines a maximum transfer timeout of two seconds. The
comments state that during tests the maximum amount of required transfer
time was for the SCSI TEST_UNIT_READY command on certain devices. We
have now observed a USB device (Patriot Memory 13fe:3100) that can NAK
this command for slightly more than two seconds. It will also completely
fail if the timeout hits, since it gets confused by the subsequent CSW
retry/recovery mechanism and starts producing babble errors. This patch
increases the timeout to three seconds to circumvent this problem.
To test, boot a Falco from a red-black RageXT USB stick.
Change-Id: I3c4fef468fb16eacc5a487d76d025a78fb450e27
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/63095
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-by: Sameer Nanda <snanda@chromium.org>
Reviewed-on: http://review.coreboot.org/4379
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Mass storage devices such as card readers show up as
as USB devices. However the media not be inserted. In those
situations the previous code would just fake a disk and
call usbcreate_disk. This is inappropriate because it forms
a 1:1 mapping of USB device to disk leading to the inability
to remove the disk and/or handle "hot plug" card insertion
and removals.
To alleviate this issue introduce the notion of ready to the
usbmsc structure. It tracks detached, not ready, and ready
states. The polling routine is then used to track not ready
to ready transitions thereby creating and removing disks
appropriately. This handles the case of inserting and removing
a card that shows up as a new disk.
Booted recovery mode. Able to observe inerstion and removal
of sdcard. Also able to insert valid USB flash drive to boot
as well.
Change-Id: I3eefbe537ec1b9c975744b8984b06c17ae236f40
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/57948
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: http://review.coreboot.org/4226
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
There is currently a hard-coded 30 sec delay in the mass storage
driver while waiting for each device to become ready. However, mass
storage card readers that are empty return an error code on the
TEST UNIT READY command. A REQUEST SENSE command then needs to be
issued and interrogate the data to determine if no media is present.
If no media determination is found to be true the USB device is no
longer considered a candidate to be a disk.
This code does lead to the fact that the media card reader needs to be
populated at enumeration time. I suspect this is not an issue as it
appears the storage stack in libpayload can't handle removable media
coming online later.
Booted recovery and dev modes. Noted that removable mass storage
devices with no media were ignored without any boot delay.
Change-Id: Ida7a45614d97c6e6fbfc9bb099765aad4df550fd
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/57828
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: http://review.coreboot.org/4225
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Restructure USB stack to not depend on PCI, and
make PCI stub available on x86, but provide fixed
BARs for ARM (Exynos 5)
Change-Id: Iee7c8b134c22b661a9a515e24943470c9dbadd1f
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/49970
Reviewed-on: http://review.coreboot.org/4175
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Two structures in the USB EHCI stack were pointing
to hardware but not marked attribute((packed)) hence
leaving it to GCC to correctly align the data structures.
Next, the number of reserved bytes in hc_op_t was wrong
(but implicitly aligned to the correct values on x86)
It seems this worked fine on x86, but on ARM it was doing
the wrong thing.
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Change-Id: I94bed4850ded7d3f7bbc7ff3079c103c6054c22d
Reviewed-on: https://gerrit.chromium.org/gerrit/55555
Commit-Queue: Stefan Reinauer <reinauer@google.com>
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Tested-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/4174
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
On Intel's Panther Point the xHCI ports are shared with an EHCI
controller. Our xHCI driver switches them to xHCI, naturally. But
we forgot to switch them back on shutdown, which left them
unusable by a non-xHCI aware operating system.
Change-Id: I70ef08655a603b42ee939935d50cf77ea97878a3
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/3791
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Well, it turned out to be more as some gaps ;)
but we finally have xHCI running. It's well tested against a QM77 Ivy
Bridge board.
We have no SuperSpeed support (yet). On Ivy Bridge, SuperSpeed is not
advertised and USB 3 devices will just work at HighSpeed.
There are still some bit fields in xhci_private.h, so this might need
little more work to run on ARM.
Change-Id: I7a2cb3f226d24573659142565db38b13acdc218c
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/3452
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This is mostly a rewrite, don't even try to read a diff.
Tested with an internal rate matching hub on a QM77 board and three hubs
integrated into DELL monitors.
Change-Id: Ib12fa2aa90af4e0f37143d2ed92c4a1705b6d774
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/3451
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The current drivers for external usb hubs and root hubs all follow
the same pattern. Before adding another one with 90% of the same code,
extract the common parts and rewrite them with a simple interface.
This also adds debouncing of new attachments. Current drivers just
waited 100ms before they reset the device. However, we should check
if the device becomes disconnected and reconnected during this period.
Porting of the current hub drivers will take place in separate
commits (when I have time to test the older HCIs).
Change-Id: I0c0ce0ac1b1cc51fb4cd009b3f9fcd1b9d2ba8fe
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/3450
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Read bInterval from endpoint descriptors and store it in our endpoint_t
struct. The interval is encoded dependently on the device' speed and the
endpoint's type. Therefore, it will be normalized to the binary logarithm
of the number of microframes, i.e.
t = 125us * 2^interval
The interval attribute will be used in the xHCI driver.
Change-Id: I65a8eda6145faf34666800789f0292e640a8141b
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/3449
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
xHCI requires special treatment of set_address since it determines
the device number itself (instead of the driver, as with the other
controllers). The controller also wants to validate a chosen device
configuration and we need to setup additional structures for the
device and the endpoints.
Therefore, we add three functions to the hci_t structure, namely:
set_address()
finish_device_config()
destroy_device()
Current implementation for the Set Address request moved into
generic_set_address() which is set_address() for the UHCI, OCHI and
EHCI drivers. The latter two are only provided as hooks for the xHCI
driver.
The Set Configuration request is moved after endpoint enumeration.
For all other controller drivers nothing changes, as there is no other
device communication between the lines where the set_configuration()
call moved.
Change-Id: I6127627b9367ef573aa1a1525782bc1304ea350d
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/3447
Tested-by: build bot (Jenkins)
Reviewed-by: Anton Kochkov <anton.kochkov@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
During device initialization, skip any non-endpoint descriptor before
reading the endpoint descriptors. By now, only HID descriptors were
skipped.
Change-Id: I190f3ae44b864aa71d5f32c3738097cf8f33a61b
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/3446
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Anton Kochkov <anton.kochkov@gmail.com>
This change modifies the code in libpayload that scans the PCI hierarchy for
USB controllers. Previously, if a devices primary function (function 0) was a
bridge, then none of the other functions, if any, would be looked at. If one
of the other functions was a bridge, that wouldn't be handled either. The new
version looks at each function that's present no matter what, and if it
discovers that it's a bridge it scans the other side.
Change-Id: I37f269a4fe505fd32d9594e2daf17ddd78609c15
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/2517
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
The function dump_qh() was added a while back but never used.
Hide it behind USB_DEBUG so it doesn't cause warnings when not
debugging the USB stack.
Change-Id: Idb3c7bb214895ef82676d181836a578bf161e8e0
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/2909
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martin.roth@se-eng.com>
The controller's shutdown function free()s the controller structure so
we shouldn't access it any more after calling shutdown.
As all controllers detach themself, i.e. unchain themself from usb_hcs,
just keep iterating over usb_hcs until it's NULL.
Change-Id: Ie85caba0f685494c3fe04c550a5a14bc4158a94e
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Reviewed-on: http://review.coreboot.org/2900
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins)
Reviewed-by: Anton Kochkov <anton.kochkov@gmail.com>
It shouldn't be used any more as we're about to free() the memory behind
the controller -- therefore detach it.
Change-Id: I875322a9940570c51d412a7f3bfb6af4ea3b3764
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Reviewed-on: http://review.coreboot.org/2899
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Nico Huber <nico.huber@secunet.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
And include the new, split out version in drivers/keyboard.c and
drivers/usb/usbhid.c. Those files were including curses.h just for those
definitions, but the include path was only fixed up to to point to the
libpayload versions of those files if one of the variants of curses was
compiled in. If neither was, gcc would fall back to the system version of that
header which is wrong.
Change-Id: I8c2ee0baf5f0702bd8c713c8dd4613a4bb269ce5
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/2762
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This function is static and not used in that file. To avoid the compiler
complaining about that fact, put the two functions and the call to dump_ed
(currently #if 0) behind #ifdef USB_DEBUG
Change-Id: Ic373313b5fff81f09800f286b32238350ab699c6
Signed-off-by: Gabe Black <gabeblack@google.com>
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: http://review.coreboot.org/2716
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
It might be useful to provide a USB driver in the payload itself instead of in
libpayload. For example there are multiple payloads being built and linked
against the same libpayload, and they might not need or even want to have the
same set of drivers installed.
This change adds two new functions, usb_generic_create and usb_generic_remove,
which behave like the usbdisk_create and usbdisk_remove functions which are
defined for USB mass storage devices. If a USB device isn't recognized and
claimed by one of the built in USB class drivers (currently hub, hid, and msc)
and the create function is defined, then it will be called to give the payload
a chance to use the device. Once it's removed, if usb_generic_remove is
defined it will be called, effectively giving the payload notice.
Built and booted depthcharge on Link. Built depthcharge for Daisy. Built
a netbooting payload, called usb_poll() with those functions implemented, and
verified that they were called and that the devices they were told about were
reasonable and the same as what was reported by lsusb in the booted system.
Change-Id: Ief7c0a513b60849fbf2986ef4ae5c9e7825fef16
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/2666
Tested-by: build bot (Jenkins)
Reviewed-by: Kimarie Hoot <kimarie.hoot@se-eng.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
EHCI controllers see transfers as a queue of transfer descriptors
(qTDs), each of which can represent an aligned area of up to 20KB. Each
qTD is processed separately, which means that a single USB packet cannot
span multiple qTDs.
While this should not be a problem according to the specification, some
USB storage devices seem to get confused when a packet in the middle of
a transfer is smaller than the maximum packet size (512 bytes) due to
falling on a qTD boundary. This patch aligns the total transfer length
per qTD to 512 bytes to avoid that problem (any excess bytes will simply
roll over to the next qTD).
Change-Id: I0b5db07507699a3861b30c1a5ee774c45dda7fdd
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: http://review.coreboot.org/2651
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Initialize the high part of the address
and use 64-bit compatible descriptors.
(waste a few bytes on 32-bit but should be harmless)
Read USB stick on a SandyBridge system which has 64-bit EHCI.
Change-Id: I59cc842459acecdde8f8bdd4795ebfeccb842c8f
Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/2650
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Kimarie Hoot <kimarie.hoot@se-eng.com>
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
That way when it's treated as a u32 when its value is extracted for numblocks
and blocksize below, it doesn't make the compiler unhappy, and it ensures that
the buffer will be properly aligned on architectures where that sort of thing
matters.
Built and saw warnings about type punning go away.
Change-Id: I254e0b5e70847112d660675b7df0ac9cb52e4051
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/2653
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
This function was using mdelay in a loop to check for the completion of an USB
controller operation. Since we're busy waiting anyway, we might as well wait
only 1 us before checking again and potentially seeing the completion 999 us
earlier than we would otherwise.
Change-Id: I177b303c5503a0078c608d5f945c395691d4bd8a
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/2522
Tested-by: build bot (Jenkins)
Reviewed-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-by: Anton Kochkov <anton.kochkov@gmail.com>
When checking to see if a PCI device exists at a particular bus/dev/func,
libpayload was checking the vendor and device id fields together against a 16
bit 0xffff. The two fields together are 32 bits, however, so the check was
never true, and all dev/func combinations on a particular bus would be
checked. That was slightly wasteful, but had relatively small impact.
Change-Id: Iad537295c33083243940b18e7a99af92857e1ef2
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: http://review.coreboot.org/2521
Tested-by: build bot (Jenkins)
Reviewed-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-by: Anton Kochkov <anton.kochkov@gmail.com>
Improve USB debugging for EHCI by adding dump_qh
and enhacing dump_td to dump all queue chain and information.
Change-Id: Ia8ecf19c6dac085cf9558bdf659a5e74ce332714
Signed-off-by: Anton Kochkov <anton.kochkov@gmail.com>
Reviewed-on: http://review.coreboot.org/2053
Tested-by: build bot (Jenkins)
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Nico Huber <nico.huber@secunet.com>
A "far" modifier sneaked into the USB driver, but gcc
doesn't understand it.
Change-Id: I5c67bd55eabce467e1aa107c95c1db2518af7b0e
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Reviewed-on: http://review.coreboot.org/2059
Tested-by: build bot (Jenkins)
Reviewed-by: Anton Kochkov <anton.kochkov@gmail.com>
Improved USB debugging for EHCI by enhacing dump_td
to dump all chain information
Change-Id: I8c667b43e09c39ff12aafbd779474efd652bd80f
Signed-off-by: Anton Kochkov <anton.kochkov@gmail.com>
Reviewed-on: http://review.coreboot.org/2054
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.huber@secunet.com>
Improving USB debugging for OHCI by enhacing dump_td
and adding dump_ed function to dump all chain information
Change-Id: Ia8b2a9b53e79b1f280fd12ea0d9233fc875e0b57
Signed-off-by: Anton Kochkov <anton.kochkov@gmail.com>
Reviewed-on: http://review.coreboot.org/2056
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.huber@secunet.com>
Previously printf()'s were used to show USB messages
which results in lots of USB information being shown
when it isn't needed. This will now use the usb_debug()
printing funtion that already exists in usb.h.
Change-Id: I2199814de3327417417eb2e26a660f4a5557cb9f
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/2044
Tested-by: build bot (Jenkins)
Reviewed-by: Anton Kochkov <anton.kochkov@gmail.com>
Slightly more complete keymap
Change-Id: I4fef6b8f75ab07cb20a3a8ccd7eaad81c9fe719f
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/1922
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
We have to free TDs more carefully if they have been processed by the
controller yet. The current code tries to force the controller to post
them back to the done queue, but that seems wrong. We can't be sure,
when they get written back. This resulted in leaking TDs with an invalid
reference to a freed interrupt queue.
The new approach: Mark the interrupt queue to be destroyed and handle
the freeing later, when the controller posted the last TD to the done
queue.
Change-Id: I79d80a9dc89e1ca79dc125c4bbccbf23664227b3
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1905
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
The connection state detection in the OHCI root hub driver was broken if
you used more than one device per root hub.
Change-Id: Ica5c735426beac45ef6f591ce68a72d8283a00f5
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1904
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
If somethings goes wrong during an interrupt transfer, drop the
transfer.
Change-Id: I450c08a7a0bf23fbee74237e0355d4a726ace114
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1901
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
If usb_poll() isn't called fast enough, the UHCI controller marks an
underrun interrupt queue as done (terminating the queue at the head).
We can recover from this situation, when usb_poll() gets called again,
and the queue is processed.
Change-Id: Id56c9df44d6dbd53cd30ad89dfb5bf5977799829
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1898
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
The linking of interrupt queues into UHCI controller's framelist (in
uhci_create_intr_queue()) was incomplete. The implementation of
uhci_destroy_intr_queue() was even worse, looking like it wanted to
clean up more than uhci_create_intr_queue() did.
This patch follows the simple approach that we used for OHCI and EHCI:
Each slot in the framelist holds only one interrupt queue. Therefore, we
have to look for free slots each time we want to link an interrupt queue
into the framelist. In return, we have a much simpler structured
framelist.
With this, USB devices using interrupt transfers (e.g. keyboards) can be
detached cleanly from UHCI controllers. Also, more than one of such
devices can be attached without further risk.
Change-Id: I07b81a3b6f2cb3ff69515c973b3ae6321ad969aa
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1897
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
The USB HID driver had some static variables with keyboard state. This
moves them to the driver's instance, so multiple attached keyboards
don't effect each other.
Change-Id: I3f1ccfdea95062b443cebe510abf2f72fdeb1916
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1907
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
Prevent race conditions, when an interrupt-queue underrun occurred and
the controller is currently working on our queue head or a transfer is
still in progress.
Change-Id: Ia14f80a08071306ee5d1349780be081bfacb206a
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1902
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
When a USB hub got removed, we should also remove all devices that
were attached to it.
Change-Id: I73c0da1b7570f1af9726925ca222781b3d752557
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1903
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
If an endpoint gets stalled by an MSC device, after successful
transmission of a command (CBW), we should still ask for the status
(CSW). Otherwise, the driver and the device get desynchronized on the
command tags.
Change-Id: I53167f22c43b3a237cb4539b3affe37799378b93
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1900
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
Stalled transfers are not fatal, so don't spew on the console on every
tiny failure.
Change-Id: I175c1e83a6af09c1abbd43d045ed6dbf0c79f871
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1899
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
dump_td() is orphaned but looks useful => commented out.
The delay identifier shadowed the global one => renamed to total_delay.
Change-Id: I4f3766a07db9194b2552ebf9302bd7ef8a66371f
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1895
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
We can trust free() and memset() to work correctly on volatile
references, so cast volatile pointers to (void *) when calling them.
Change-Id: Ieff7f78133b72f303349cca0a0ca3bbf37ec52bb
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1896
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
Tested-by: build bot (Jenkins)
usb_controller_initialize() is not declared in any header file nor
called from outside of usbinit.c, so make it static.
set_configuration() looks like beeing non-static on purpose (like the
other helpers around it in usb.c), so put a prototype into usb.h.
Change-Id: I08d93b3769d8398bb43462d9afdfeec81fef93ec
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1894
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
uhci_reset() differs in semantics compared to the other HCI's reset()
implementations. uhci_reset() does some initialization work after a
controller reset. So move the initialization part to a new function,
uhci_reinit(), which get's exported through a new entry in hci_t:
hci_t.init().
Warning: This breaks code that relies on the current, special,
counterintuitive behaviour of uhci_reset(). If one wants a working host
controller after calling hci_t.reset(), he should call hci_t.init()
afterwards.
Change-Id: Ia7ce80865d12d11157645ce251f77f349f8e3c34
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1851
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Nico Huber