memcpy() is meant to be used on normal memory and often implemented with
architecture-specific optimizations to make that as performant as
possible. MMIO registers often have special access restrictions that may
be incompatible with whatever memcpy() does. For example, on arm64 it
uses the LDP (load pair) to load 16 bytes at a time, which makes 4-byte
MMIO registers unhappy.
This patch removes the caching of the XHCI capreg registers and changes
it back to a pointer. The CAP_GET() macro is still accessing a full
(non-bitfield) uint32_t at the end so this should still generate a
4-byte access (which was the goal of the original change in CB:39838).
Signed-off-by: Julius Werner <jwerner@chromium.org>
Change-Id: Id058c8813087a8e8cb85f570399e07fb8a597108
Reviewed-on: https://review.coreboot.org/c/coreboot/+/40895
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
The QEMU XHCI controller does not support byte/word reads from the
capability register and it expects dword reads only.
In order to make this work move the access of the capability
register fields to use macros instead of a packed struct bitfield.
This issue was filed upstream:
https://bugs.launchpad.net/qemu/+bug/1693050
The original fix attempt in 2012 was not effective:
6ee021d410
With this change the controller is detected properly by the libpayload
USB drivers.
Change-Id: I048ed14921a4c9c0620c10b315b42476b6e5c512
Signed-off-by: Duncan Laurie <dlaurie@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/39838
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
This makes it easier to know what offset each register references.
BUG=b:76831439
TEST=none
Change-Id: I92dcbd463ceb4dd8edbbd97b51a4e9aa32a983a6
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Reviewed-on: https://review.coreboot.org/27474
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
1. Make the xHCI driver to support xHCI controller v1.1
2. And a new function xhci_ring_doorbell(), it aims to
add a memory barrier before ringing the doorbell, to ensure
all TRB changes are written to memory.
BRANCH=none
BUG=chrome-os-partner:52684
TEST=boot from USB on Kevin rk3399 platform
Change-Id: Ife1070d1265476d0f5b88e2acf3299fc84af5832
Signed-off-by: Martin Roth <martinroth@chromium.org>
Original-Commit-Id: 0c21e92
Original-Change-Id: I4e38e04dc3c7d32ee4bb424a473c70956a3c3ea9
Original-Signed-off-by: Liangfeng Wu <wulf@rock-chips.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/346831
Original-Commit-Ready: Brian Norris <briannorris@chromium.org>
Original-Tested-by: Douglas Anderson <dianders@chromium.org>
Original-Reviewed-by: Douglas Anderson <dianders@chromium.org>
Original-Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/15111
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Nico Huber <nico.h@gmx.de>
The 1.1 revision of the XHCI specification added an extra 5 bits to the
Max Scratchpad Bufs field of HCSPARAMS2 that newer controllers make use
of. Not honoring these bits means we're not allocating as many
scratchpad buffers as the controller expects, which means it will
interpret some uninitialized values from the end of the pointer array as
scratchpad buffer pointers, which obviously doesn't end well. Let's fix
that.
BRANCH=none
BUG=chrome-os-partner:42279
TEST=Makes a USB-related memory corruption issue disappear.
Original-Change-Id: I7c907492339262bda31cdd2b5c0b588de7df8544
Original-Signed-off-by: Julius Werner <jwerner@chromium.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/291681
Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Change-Id: Iba1007bfebffe1f564f78bb875fff9ba0fe11a38
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: http://review.coreboot.org/11189
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
This patch re-adds a few fixes that originally went into the
chromeos-2013.04 tree. I kinda seem to have slipped them into the
backport of Nico's original XHCI patch (crosreview.com/168097) instead
of making a new change, which was not very clever and caused them to be
forgotten in the later upstreaming wave.
Changing internal XHCI error numbers is just a cosmetic change to make
them uniquely identifyable in debug output. Bumping the timeout to 3
seconds is an actually important fix since we have seen mass storage
devices needing that much in the past.
BRANCH=None
BUG=None
TEST=Diffed payloads/libpayload/drivers/usb between chromeos-2013.04 and
chromeos-2015.07, confirmed that no serious differences remain.
Original-Change-Id: I03d865dbe536072d23374a49a0136e9f28568f8e
Original-Signed-off-by: Julius Werner <jwerner@chromium.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/290423
Original-Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Change-Id: I5d773d3a23683fb2164916cc046f4a711b8d259e
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: http://review.coreboot.org/11178
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
We have been trying to avoid reassigning previously used USB addresses
to different devices since CL:197420, because some devices seem to take
issue with that. Unfortunately, that patch doesn't affect XHCI: those
controllers insist on chosing addresses on their own. The only way to
prevent them from reusing a previously assigned address is to not
disable that slot at all.
This patch implements address reuse avoidance on XHCI by not disabling
slots when a device is detatched (which may occur both on physical
detachment or if we simply couldn't find a driver for that device).
Instead, we just release as many resources as we can for detached
devices (by dropping all endpoint contexts) and defer the final cleanup
until the point where the controller actually runs out of resources (a
point that we probably don't often reach in most firmware scenarios).
BRANCH=none
BUG=chrome-os-partner:42181
TEST=Booted an Oak plugged into a Servo without having a driver for the
SMSC network chip, observed that it could still enumerate the next
device afterwards. Kept unplugging/replugging stuff until the cleanup
triggered and made sure the controller still worked after that. Also
played around a bit on a Falco without issues.
Change-Id: Idfbab39abbc5bc5eff822bedf9c8d5bd4cad8cd2
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: 88c6bcbc41156729c3c38937c8a4adebc66f1ccb
Original-Change-Id: I0653a4f6a02c02498210a70ffdda9d986592813b
Original-Signed-off-by: Julius Werner <jwerner@chromium.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/284175
Original-Tested-by: Nicolas Boichat <drinkcat@chromium.org>
Original-Reviewed-by: Patrick Georgi <pgeorgi@chromium.org>
Reviewed-on: http://review.coreboot.org/10957
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
1. There is a mis-understanding to calculate the value of TD Size
in Normal TRB. For MTK's xHCI controller it defines a number of
packets that remain to be transferred for a TD after processing
all Max packets in all previous TRBs, that means don't include the
current TRB's.
2. To minimize the scheduling effort for synchronous endpoints in xHC,
the MTK architecture defines some extra SW scheduling parameters for
HW. According to these parameters provided by SW, the xHC can easily
decide whether a synchronous endpoint should be scheduled in a specific
uFrame. The extra SW scheduling parameters are put into reserved DWs
in Slot and Endpoint Context. But in coreboot synchronous transfer can
be ignored, so only two fields are set to a default value 1 to support
bulk and interrupt transfers, and others are set to zero.
3. For control transfer, it is better to read back doorbell register or add
a memory barrier after ringing the doorbell to flush posted write.
Otherwise the first command will be aborted on MTK's xHCI controller.
4. Before send commands to a port, the Port Power in PORTSC register should
be set to 1 on MTK's xHCI so a hook function of enable_port in
generic_hub_ops_t struct is provided.
Change-Id: Ie8878b50c048907ebf939b3f6657535a54877fde
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: 738609c11f16264c6e6429d478b2040cb391fe41
Original-Change-Id: Id9156892699e2e42a166c77fbf6690049abe953b
Original-Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
Original-Reviewed-on: https://chromium-review.googlesource.com/265362
Original-Reviewed-by: Julius Werner <jwerner@chromium.org>
Original-Commit-Queue: Yidi Lin <yidi.lin@mediatek.com>
Original-Tested-by: Yidi Lin <yidi.lin@mediatek.com>
Reviewed-on: http://review.coreboot.org/10389
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Tested-by: build bot (Jenkins)
If a TD is comprised of one or more Normal TRBs and terminated with an
Event Data TRB, then the transition to the Idle state (and associated
Stream state save) could occur after all the data for the TD has been
moved (e.g. after Transfer Event TRBs have been executed), but before the
Event Data TRB is executed. Under these conditions, the execution of the
Event Data TRB is necessary to complete the TD, otherwise it does not
occur until the next time the Stream is scheduled. This could lead to the
lock up.
The Evaluate Next TRB(ENT) flag provides a means of forcing the execution
of a terminating Event Data TRB. Setting ENT flag in last Normal TRB makes
the xHC to evaluate the Even Data TRB.
BUG=chrome-os-partner:29375
TEST=Verified kernel boot-up on storm from previously failing USB stick.
USB stick model: Sandisk Ultra USB 3.0 Pen Drive 32 GB
Strontium Jet USB 3.0 Pen Drive 32 GB
Change-Id: I092e2109c55c2274239c493cb67b47d730304ed2
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Original-Commit-Id: 7eefb3b2858c841165ae839d349d2a0be50fbcc8
Original-Change-Id: I4e123577ec5a5996d87d2fc52cb6cf5c571c9fae
Original-Signed-off-by: Sourabh Banerjee <sbanerje@codeaurora.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/220123
Original-Reviewed-by: Julius Werner <jwerner@chromium.org>
Original-Commit-Queue: Vadim Bendebury <vbendeb@chromium.org>
Original-Tested-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: http://review.coreboot.org/8736
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
xHCI Spec says TD Size (5 bits) field shall be forced to 31,
if the number of packets to be scheduled is greater than 31.
BUG=chrome-os-partner:27837
BRANCH=rambi,nyan
TEST=Manual: Ensure recovery boot with USB 2.0 media on Squawks
works fine without any babble errors.
Original-Change-Id: Iff14000e2a0ca1b28c49d0da921dbb2a350a1bbd
Original-Signed-off-by: Rajmohan Mani <rajmohan.mani@intel.com>
Original-Originally-Reviewed-on: https://chromium-review.googlesource.com/202297
Original-Reviewed-on: https://chromium-review.googlesource.com/202330
Original-Reviewed-by: Shawn Nematbakhsh <shawnn@chromium.org>
Original-Commit-Queue: Julius Werner <jwerner@chromium.org>
Tested-by: Julius Werner <jwerner@chromium.org>
(cherry picked from commit ae58b99370df3a86bf15d84b97db858a968b1dbd)
Signed-off-by: Marc Jones <marc.jones@se-eng.com>
Change-Id: I9668b947f676c109fad9297e5efde91bf7f796fd
Reviewed-on: http://review.coreboot.org/7913
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
This patch represents a major overhaul of the USB enumeration code in
order to make it cleaner and much more robust to weird or malicious
devices. The main improvement is that it correctly parses the USB
descriptors even if there are unknown descriptors interspersed within,
which is perfectly legal and in particular present on all SuperSpeed
devices (due to the SuperSpeed Endpoint Companion Descriptor).
In addition, it gets rid of the really whacky and special cased
get_descriptor() function, which would read every descriptor twice
whether it made sense or not. The new code makes the callers allocate
descriptor memory and only read stuff twice when it's really necessary
(i.e. the device and configuration descriptors).
Finally, it also moves some more responsibilities into the
controller-specific set_address() function in order to make sure things
are initialized at the same stage for all controllers. In the new model
it initializes the device entry (which zeroes the endpoint array), sets
up endpoint 0 (including MPS), sets the device address and finally
returns the whole usbdev_t structure with that address correctly set.
Note that this should make SuperSpeed devices work, but SuperSpeed hubs
are a wholly different story and would require a custom hub driver
(since the hub descriptor and port status formats are different for USB
3.0 ports, and the whole issue about the same hub showing up as two
different devices on two different ports might present additional
challenges). The stack currently just issues a warning and refuses to
initialize this part of the hub, which means that 3.0 devices connected
through a 3.0 hub may not work correctly.
Change-Id: Ie0b82dca23b7a750658ccc1a85f9daae5fbc20e1
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/170666
Reviewed-by: Kees Cook <keescook@chromium.org>
(cherry picked from commit ecec80e062f7efe32a9a17479dcf8cb678a4a98b)
Signed-off-by: Isaac Christensen <isaac.christensen@se-eng.com>
Reviewed-on: http://review.coreboot.org/6780
Tested-by: build bot (Jenkins)
This patch removes the confusing concept of a special "xhci_speed" with
a different numeric value from the usual speed used throughout the USB
core (except for the places directly interacting with the xHC, which are
explicitly marked). It also moves the MPS0 decoding function into the
core and moves some definitions around in preparation of later changes
that will make the stack SuperSpeed-ready. It makes both set_address
implementations share a constant for the specification-defined
SetAddress() recovery delay and removes pointless additional delays from
the non-XHCI version.
Change-Id: I422379d05d4a502b12dae183504e5231add5466a
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/170664
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Commit-Queue: Ronald Minnich <rminnich@chromium.org>
(cherry picked from commit f160d4439c0d7cea1d2e6b97207935d61dcbb2f2)
Signed-off-by: Isaac Christensen <isaac.christensen@se-eng.com>
Reviewed-on: http://review.coreboot.org/6776
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This patch updates the libpayload XHCI stack to run on ARM CPUs (tested
with the DWC3 controller on an Exynos5420). Firstly, it adds support for
64-byte Slot/Endpoint Context sizes. Since the existing context handling
code represented the whole device context as a C struct (whose size has
to be known at compile time), it was necessary to refactor the input and
device context structures to consist of pointers to the actual contexts
instead.
Secondly, it moves all data structures that the xHC accesses through DMA
to cache-coherent memory. With a similar rationale as in the ARM patches
for EHCI, using explicit cache maintenance functions to correctly handle
the actual transfer buffers in all cases is presumably impossible.
Instead this patch also chooses to create a DMA bounce buffer in the
XHCI stack where transfer buffers which are not already cache-coherent
will be copied to/from.
Change-Id: I14e82fffb43b4d52d687b65415f2e33920e088de
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/169453
Reviewed-by: Stefan Reinauer <reinauer@google.com>
(cherry picked from commit 1fa9964063cce6cbd87ba68334806dde8aa2354c)
Signed-off-by: Isaac Christensen <isaac.christensen@se-eng.com>
Reviewed-on: http://review.coreboot.org/6643
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The current XHCI code only sets IOC on the last TRB of a TD, and
doesn't set ISP anywhere. On my Synopsys DesignWare3 controller, this
won't generate an event at all when we have a short transfer that is not
on the last TRB of a TD, resulting in event ring desync and everyone
having a bad time. However, just setting ISP on other TRBs doesn't
really make for a nice solution: we then need to do ugly special casing
to fish out the spurious second transfer event you get for short
packets, and we still need a way to figure out how many bytes were
transferred. Since the Short Packet transfer event only reports
untransferred bytes for the current TRB, we would have to manually walk
the rest of the unprocessed TRB chain and add up the bytes. Check out
U-Boot and the Linux kernel to see how complicated this looks in
practice.
Now what if we had a way to just tell the HC "I want an event at exactly
*this* point in the TD, I want it to have the right completion code for
the whole TD, and to contain the exact number of bytes written"? Enter
the Event Data TRB: this little gizmo really does pretty much exactly
what any sane XHCI driver would want, and I have no idea why it isn't
used more often. It solves both the short packet event generation and
counting the transferred bytes without requiring any special magic in
software.
Change-Id: Idab412d61edf30655ec69c80066bfffd80290403
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/170980
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
(cherry picked from commit e512c8bcaa5b8e05cae3b9d04cd4947298de999d)
Signed-off-by: Isaac Christensen <isaac.christensen@se-eng.com>
Reviewed-on: http://review.coreboot.org/6516
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
When libpayload header files are included in the payload itself, it's possible
that the payloads config settings will conflict with the ones in libpayload.
It's also possible for the libpayload config settings to conflict with the
payloads. To avoid that, the libpayload config settings have _LP_ (for
libpayload) added to them. The symbols themselves as defined in the Config.in files
are still the same, but the prefix added to them is now CONFIG_LP_ instead of just
CONFIG_.
Change-Id: Ib8a46d202e7880afdeac7924d69a949bfbcc5f97
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/65303
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Tested-by: Gabe Black <gabeblack@chromium.org>
Commit-Queue: Gabe Black <gabeblack@chromium.org>
(cherry picked from commit 23e866da20862cace0ed2a67d6fb74056bc9ea9a)
Signed-off-by: Isaac Christensen <isaac.christensen@se-eng.com>
Reviewed-on: http://review.coreboot.org/6427
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
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>
Improve scanning for USB controllers.
Limitations:
- OHCI doesn't support interrupt transfers yet (ie. no keyboards)
- xHCI just does initialization and device attach/detach so far
Signed-off-by: Patrick Georgi <patrick@georgi-clan.de>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5691 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Julius Werner