This patch makes the EHCI driver work on ARM platforms which usually do
not support automatic cache snooping. It uses the new DMA memory
mechanism (which needs to be correctly set up in the Coreboot mainboard
code) to allocate all EHCI-internal communication structures in
cache-coherent memory, and cleans/invalidates the externally supplied
transfer buffers in Bulk and Control functions with explicit calls as
necessary.
Old-Change-Id: Ie8a62545d905b7a4fdd2a56b9405774be69779e5
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/167339
(cherry picked from commit 322338934add36a5372ffe7d2a45e61a4fdd4a54)
libpayload: ehci: Cache management is hard, let's go copying...
It turns out that my previous commit to make the EHCI stack cache aware
on ARM devices wasn't quite correct, and the problem is actually much
trickier than I thought. After having some fun with more weird transfer
problems that appear/disappear based on stack alignment, this is my
current worst-case threat model that any cache managing implementation
would need to handle correctly:
Some upper layer calls ehci_bulk() with a transfer buffer on its stack.
Due to stack alignment, it happens to start just at the top of a cache
line, so up to 64 - 4 bytes of ehci_bulk's stack will share that line.
ehci_bulk() calls dcache_clean() and initializes the USB transfer.
Between that point and the call to dcache_invalidate() at the end of
ehci_bulk(), any access to the stack variables in that cache line (even
a speculative prefetch) will refetch the line into the cache. Afterwards
any other access to a random memory location that just happens to get
aliased to the same cache line may evict it again, causing the processor
to write out stale data to the transfer buffer and possibly overwrite
data that has already been received over USB.
In short, any dcache_clean/dcache_invalidate-based implementation that
preserves correctness while allowing any arbitrary (non cache-aligned)
memory location as a transfer buffer is presumed to be impossible.
Instead, this patch causes all transfer data to be copied to/from a
cache-coherent bounce buffer. It will still transfer directly if the
supplied buffer is already cache-coherent, which can be used by callers
to optimize their transfers (and is true by default on x86).
Old-Change-Id: I112908410bdbc8ca028d44f2f5d388c529f8057f
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/169231
Reviewed-by: Stefan Reinauer <reinauer@chromium.org>
(cherry picked from commit 702dc50f1d56fe206442079fa443437f4336daed)
Squashed the initial commit and a follow up fix.
Change-Id: Idf7e5aa855b4f0221f82fa380a76049f273e4c88
Signed-off-by: Isaac Christensen <isaac.christensen@se-eng.com>
Reviewed-on: http://review.coreboot.org/6633
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
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>
Added ehci_reset() function to do a full reset of
the host controller
Change-Id: Ia48db8462ebbb8f260813eb6ba8349d002c4678b
Signed-off-by: Anton Kochkov <a.kochkov@securitycode.ru>
Reviewed-on: http://review.coreboot.org/1814
Tested-by: build bot (Jenkins)
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
This introduces a dummy queue head in the interrupt frame list of the
EHCI host controller. It's a workaround for broken controllers which
follow pointers from this list even if the terminate bit is set.
Fortunately, they do honor the bit in queue heads and having an empty
QH in the list doesn't violate the standard.
The linux kernel has a similar workaround for AMD SB700, SB800, and
Hudson-2/3 platforms. We observed this bug with an AMD SB600.
Change-Id: Ibbb66dea5fddc89c7995a24d746bedf6bfa887be
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1124
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: build bot (Jenkins)
This adds support for usb interrupt transfers in the EHCI driver. Split
transactions are supported, so this enables support for HID keyboards
devices over hubs in high-speed mode.
Change-Id: I9eb08f12b12c67ece10814952cb8651278b02f9d
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1083
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
With split transactions, the EHCI host controller can handle full- and
low-speed devices on hubs in high-speed mode. This adds support for split
transactions for control and bulk transfers.
Change-Id: I30fa1ce25757f33b1e6ed34207949c9255f05d49
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1081
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Enable power on EHCI root hub ports only if the controller supports it.
Wait 20ms for the power to become stable.
Change-Id: I8897756ed2bfcb88408fe5e9f9e3f8af5dd900ac
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: http://review.coreboot.org/1078
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <patrick@georgi-clan.de>
When converting EHCI to not use bitfields, two offsets were converted
incorrectly.
Change-Id: I0bb4bad0eee42e54ad4fd53d6c35b107e227c41a
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/593
Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marcj303@gmail.com>
We agreed that bitfields are a Bad Idea[tm].
Change-Id: If4c4cb748af340e2721b89fea8e035da0632971f
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/480
Tested-by: build bot (Jenkins)
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Peter Stuge <peter@stuge.se>
Some gcc versions seem to honor volatile at different places in a
struct declaration.
Change-Id: I0df2a3fb2eff4cee8cc1b8ac15d9cd9b86178752
Signed-off-by: Patrick Georgi <patrick.georgi@secunet.com>
Reviewed-on: http://review.coreboot.org/155
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marcj303@gmail.com>
Interrupt transfer support is missing (ie. no keyboard),
bulk and control transfers work (ie. mass storage).
Signed-off-by: Patrick Georgi <patrick.georgi@coresystems.de>
Acked-by: Peter Stuge <peter@stuge.se>
git-svn-id: svn://svn.coreboot.org/coreboot/trunk@5845 2b7e53f0-3cfb-0310-b3e9-8179ed1497e1
Julius Werner