Simon Glass [Mon, 9 Nov 2015 06:47:48 +0000 (23:47 -0700)]
console: Add a console buffer
It is useful to be able to record console output and provide console input
via a buffer. This provides sandbox with the ability to run a command and
check its output. If the console is set to silent then no visible output
is generated.
This also provides a means to fix the problem where tests produce unwanted
output, such as errors or warnings. This can be confusing. We can instead
set the console to silent and record this output. It can be checked later
in the test if required.
It is possible that this may prove useful for non-test situations. For
example the console output may be suppressed for normal operations, but
recorded and stored for access by the OS. That feature is not implemented
at present.
Simon Glass [Mon, 9 Nov 2015 06:47:47 +0000 (23:47 -0700)]
Add a circular memory buffer implementation
This will be used to support console recording. It provides for a circular
buffer which can be written at the head and read from the tail. It supports
avoiding data copying by providing raw access to the data.
Simon Glass [Mon, 9 Nov 2015 06:47:45 +0000 (23:47 -0700)]
Move console definitions into a new console.h file
The console includes a global variable and several functions that are only
used by a small subset of U-Boot files. Before adding more functions, move
the definitions into their own header file.
Simon Glass [Mon, 9 Nov 2015 06:47:43 +0000 (23:47 -0700)]
sandbox: Add a way to skip time delays
Some tests are slow due to delays which are unnecessary on sandbox. The
worst offender is USB where we lose two seconds. Add a way to disable time
delays.
drivers/misc/cros_ec_sandbox.c: In function cros_ec_sandbox_packet:
drivers/misc/cros_ec_sandbox.c:483:5: warning: len may be used uninitialized in this function [-Wmaybe-uninitialized]
if (len < 0)
^
If the function process_cmd() is called with
req_hdr->command == EC_CMD_ENTERING_MODE, the value of len will be
returned uninitialized.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com> Acked-by: Simon Glass <sjg@chromium.org>
Bin Meng [Thu, 12 Nov 2015 13:33:03 +0000 (05:33 -0800)]
input: Change LED state bits to conform i8042 compatible keyboard
When sending LED update command to an i8042 compatible keyboard,
bit1 is 'Num Lock' and bit2 is 'Caps Lock' in the data byte. But
input library defines bit1 as 'Caps Lock' and bit2 as 'Num Lock'.
This causes a wrong LED to be set on an i8042 compatible keyboard.
Change the LED state bits to be i8042 compatible, and change the
keyboard flags as well.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Acked-by: Simon Glass <sjg@chromium.org>
Bin Meng [Thu, 12 Nov 2015 13:33:02 +0000 (05:33 -0800)]
input: Call keyboard's update_leds() method when the LEDs change
We should request keyboard to turn on/off its LED when detecting
any changes on the LEDs.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com> Acked-by: Simon Glass <sjg@chromium.org>
Minor changes to allow this to build without CONFIG_DM_KEYBOARD: Signed-off-by: Simon Glass <sjg@chromium.org>
Simon Glass [Wed, 11 Nov 2015 17:05:47 +0000 (10:05 -0700)]
video: input: Clean up after i8042 conversion
Now that i8042 uses driver model, adjust other mentions of it and remove old
code that is no-longer used. Update the README and unify the keyboard text
into one place.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Wed, 11 Nov 2015 17:05:43 +0000 (10:05 -0700)]
x86: Add an i8042 device for boards that have it
Some boards have an i8042 device. Enable the driver for all x86 boards, and
add a device tree node for those which may have this keyboard.
Also adjust the configuration so that i8042 is always separate from the VGA,
and rename the stdin driver accordingly. With this commit the keyboard will
not work, but it is fixed in the next commit.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Wed, 11 Nov 2015 17:05:41 +0000 (10:05 -0700)]
input: i8042: Convert to use the input library
At present the i8042 driver has its own logic and keymaps. In an effort to
unify the code, move it over to use the input library. This changes most of
the keycode-processing logic since it is now in that library. The main
responsibilities of the driver are now to handle the LEDs, deal with the
PS/2 extended keycodes and initialise the the keyboard.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com> Tested-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Mon, 19 Oct 2015 03:17:25 +0000 (21:17 -0600)]
input: Allow repeat filtering to be disabled
Generally the input library handles processing of a list of scanned keys.
Repeated keys need to be generated based on a timer in this case, since all
that is provided is a list of keys current depressed.
Keyboards which do their own scanning will resend codes when they want to
inject a repeating key. Provide a function which tells the input library to
accept repeating keys and not to try to second-guess the caller.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Mon, 19 Oct 2015 03:17:24 +0000 (21:17 -0600)]
input: Add a function to add a keycode to the existing set
Most keyboards can be scanned to produce a list of the keycodes which are
depressed. With the i8042 keyboard this scanning is done internally and
only the processed results are returned.
In this case, when a key is pressed, a 'make' code is sent. When the key
is released a 'break' code is sent. This means that the driver needs to
keep track of which keys are pressed. It also means that any protocol error
can lead to stuck keys.
In order to support this type of keyboard, add a function when can be used
to provide a single keycode and either add it to the list of what is pressed
or remove it from the list. Then the normal input_send_keycodes() function
can be used to actually do the decoding work.
Add debugging to display the ASCII characters written to the input queue
also.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Mon, 19 Oct 2015 03:17:21 +0000 (21:17 -0600)]
i8042: Adjust keyboard init to assume success
Modify i8042_kbd_init() so that the normal pass is sucessful init and
failure exits early. This will make the code easier to extend and is easier
to read.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Mon, 19 Oct 2015 03:17:19 +0000 (21:17 -0600)]
i8042: Use functions to handle register access
At present the register access in kbd_reset() is quite primitive. This makes
it hard to follow.
Create functions to read and write data, both to a single register, and via
the command/data approach.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-on: Intel Crown Bay and QEMU Tested-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Mon, 19 Oct 2015 03:17:18 +0000 (21:17 -0600)]
video: Drop unused console functions
CONFIG_CONSOLE_CURSOR, CONFIG_SYS_CONSOLE_BLINK_COUNT and
CONFIG_CONSOLE_TIME are not used by any board. The implementation is not
great and stands in the way of a refactor of i8042. Drop these for now.
They can be re-introduced quite easily later, perhaps with driver-model
real-time-clock (RTC) support.
When reintroducing, it might be useful to make a few changes:
- Blink time would be more useful than blink count
- The confusing #ifdefs should be avoided
- The time functions should support driver model
- It would be best keyed off console_tstc() or some similar idle loop
rather than a particular input driver (i8042 in this case)
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Mon, 19 Oct 2015 03:17:17 +0000 (21:17 -0600)]
dm: cros_ec: Convert cros_ec keyboard driver to driver model
Adjust the cros_ec keyboard driver to support driver model. Make this the
default for all Exynos boards so that those that use a keyboard will build
correctly with this driver.
Simon Glass [Mon, 19 Oct 2015 03:17:16 +0000 (21:17 -0600)]
dm: tegra: Convert keyboard driver to driver model
Adjust the tegra keyboard driver to support driver model, using the new
uclass. Make this the default for all Tegra boards so that those that use
a keyboard will build correctly with this driver.
Simon Glass [Mon, 19 Oct 2015 03:17:13 +0000 (21:17 -0600)]
input: Add the keycode translation tables separately
Require the caller to add the keycode translation tables separately so that
it can select which ones to use. In a later patch we will add the option to
add German tables.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Mon, 19 Oct 2015 03:17:11 +0000 (21:17 -0600)]
input: Add a device pointer to the input config
The read_keys() method in input is passed a struct input_config. Add a
device pointer there so that we can find out the device that is referred
to with driver model.
Once all drivers are converted we can update the input structure to use
driver model instead.
Signed-off-by: Simon Glass <sjg@chromium.org> Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Simon Glass [Tue, 8 Sep 2015 17:15:11 +0000 (11:15 -0600)]
dm: usb: Add support for USB keyboards with driver model
Switch USB keyboards over to use driver model instead of scanning with the
horrible usb_get_dev_index() function. This involves creating a new uclass
for keyboards, although so far there is no API.
rockchip: Turn on CONFIG_DEBUG_UART_NS16550 in defconfigs
No UART driver was specified in defconfig, thus
DEBUG_UART_ALTERA_JTAGUART was incorrectly selected by default since
commit 220e8021af96741bd7149ca9895e1f0c8a38d0bb added a new Altera UART
driver.
Signed-off-by: Ariel D'Alessandro <ariel@vanguardiasur.com.ar> Acked-by: Simon Glass <sjg@chromium.org>
Michal Simek [Tue, 8 Sep 2015 15:07:01 +0000 (17:07 +0200)]
net: zynq: Fix MDC setting for zynq
Based on spec:
"MDC must not exceed 2.5 MHz (MDC is only active during MDIO read and
write operations)"
Zynq is running on 111MHz. Current setting is 32 which is 111/32=3.47
which is above of 2.5MHz.
Using 48 divider will give us correct setting according spec
(111/48=2.31).
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Michal Simek [Tue, 8 Sep 2015 14:55:42 +0000 (16:55 +0200)]
net: zynq: Fix mdc clock division setting for 100Mbit/s
Using set and clear macro is incorrect because it is not overwritting
origin mdc clock division setup.
For example origin setup is 8(0b001) and new setup is 64(0b100) which
means 0b101 is setup which is 96 divider.
Using writel to rewrite all setting like for 1000Mbit/s case.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Zynq has no priority queues.
ZynqMP has one priority queue and this change is required
to get ethernet working.
This patch was not needed on ep108 for uknown reason even
it should be used.
Tested on Zynq and ZynqMP.
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Mon, 17 Aug 2015 07:58:54 +0000 (09:58 +0200)]
net: zynq: Add dummy packet to fix packet duplication issue
Target is duplicating packets. IP prefetches another BD and process it
when the first one is sent. Adding one dummy BD to the chain fix the
problem with packet duplication.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Michal Simek [Tue, 27 Oct 2015 12:52:47 +0000 (13:52 +0100)]
dm: sf: Add support for all targets which requires MANUAL_RELOC
It is follow up patch based on
"dm: Add support for all targets which requires MANUAL_RELOC"
(sha1: 484fdf5ba058b07be5ca82763aa2b72063540ef3)
to update function pointers for DM.
Using post_bind is not ideal but it is one on current option what can be
used. Variable reloc_done has to be used do not call relocation after
every bind. Maybe new core functions should be introduced for this case.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Adds new entries to struct driver and struct uclass_driver without
extending code for manual relocation. This patch fixes it for all
architectures which requires MANUAL_RELOC.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Simon Glass <sjg@chromium.org>
Michal Simek [Tue, 27 Oct 2015 12:36:42 +0000 (13:36 +0100)]
dm: spi: Add support for all targets which requires MANUAL_RELOC
It is follow up patch based on
"dm: Add support for all targets which requires MANUAL_RELOC"
(sha1: 484fdf5ba058b07be5ca82763aa2b72063540ef3)
to update function pointers for DM.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Simon Glass <sjg@chromium.org>
Nathan Rossi [Fri, 6 Nov 2015 09:25:06 +0000 (19:25 +1000)]
microblaze: Fix C99/gnu99 compatiblity for inline functions
'extern inline' is not portable across various C standards. To ensure
compatiblity with various standards/compilers change the functions to
static inline. This is a portable construct and serves as a comparable
definition to 'extern inline' from the gnu90 standard.
Additionally remove the function prototypes as they are not required due
to the functions being declared static and functions are correctly
ordered based on dependence.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com> Cc: Michal Simek <monstr@monstr.eu> Cc: Tom Rini <trini@konsulko.com> Acked-by: Mans Rullgard <mans@mansr.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Michal Simek [Wed, 7 Oct 2015 14:42:56 +0000 (16:42 +0200)]
net: zynq: Add support for different PHY interface types
MII is setup by default for all cases. The most of boards are using
RGMII but PHY drivers are not doing any specific setting that's why MII
setting was working fine. With TI DP83867 is necessary to setup
paramaters based on interface type.
Use one setting per board for it which is something what will be removed
when driver is moved to DM.
Signed-off-by: Michal Simek <michal.simek@xilinx.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Nathan Rossi [Tue, 17 Nov 2015 12:56:57 +0000 (22:56 +1000)]
ARM: zynq: Add target for building bootable SPL image for Zynq
Add a build target to generate 'boot.bin' which includes SPL. This is
used by the platforms BootROM to load SPL directly.
This change also conditionally changes what the 'boot.bin' target
generates depending on the SoC. Leaving the behaviour unchanged for the
AT91 targets.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com> Cc: Tom Rini <trini@konsulko.com> Cc: Andreas Bießmann <andreas.devel@googlemail.com> Cc: Heiko Schocher <hs@denx.de> Reviewed-by: Tom Rini <trini@konsulko.com>
Nathan Rossi [Tue, 17 Nov 2015 12:56:56 +0000 (22:56 +1000)]
tools: zynqimage: Add Xilinx Zynq boot header generation to mkimage
As with other platforms vendors love to create their own boot header
formats. Xilinx is no different and for the Zynq platform/SoC there
exists the "boot.bin" which is read by the platforms bootrom. This
format is described to a useful extent within the Xilinx Zynq TRM.
This implementation adds support for the 'zynqimage' to mkimage. The
implementation only considers the most common boot header which is
un-encrypted and packed directly after the boot header itself (no
XIP, etc.). However this implementation does take into consideration the
other fields of the header for image dumping use cases (vector table and
register initialization).
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com> Cc: Michal Simek <michal.simek@xilinx.com> Cc: Tom Rini <trini@konsulko.com> Reviewed-by: Tom Rini <trini@konsulko.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
net: phy: Add support for Texas Instruments DP83867
Code is taken from Linux kernel driver (v4.2).
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com> Signed-off-by: Michal Simek <michal.simek@xilinx.com> Reviewed-by: Tom Rini <trini@konsulko.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Kevin Smith [Fri, 23 Oct 2015 17:51:47 +0000 (17:51 +0000)]
env_ubi.c: Correct pointer error in env load
The variable "buf" in this function is a char array, and the
function ubi_volume_read is expecting a char *. In the call, the
address of the pointer is being taken, incorrectly passing a
char **. The compiler warning was being silenced by the cast.
Remove the address operator and the cast.
Signed-off-by: Kevin Smith <kevin.smith@elecsyscorp.com> Cc: Joe Hershberger <joe.hershberger@ni.com> Cc: Tom Rini <trini@konsulko.com> Acked-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Heiko Schocher <hs@denx.de>
Stephen Warren [Tue, 17 Nov 2015 17:29:08 +0000 (10:29 -0700)]
itest: add missing break statements to evalexp()
The commit mentioned below replaced return statements inside a switch so
that other code could be called after the switch. However, it didn't add
any break statements, causing the cases to run together. Fix this.
Reported-by: Coverity (CID 132282, 132283) Fixes: 7861204c9af7 ("itest: make memory access work under sandbox") Signed-off-by: Stephen Warren <swarren@nvidia.com>
Stephen Warren [Tue, 17 Nov 2015 17:29:07 +0000 (10:29 -0700)]
test: fat: add error-checking to non-contig test
Check the result code of all command that are executed. Without this,
if the fallocate invocation fails (this feature is not supported on ext3
filesystems for example) then a zero-length output file will be created,
and subsequent the mkfs and mount invocations will fail, which will cause
the subsequent dd invocation to attempt to fill up the host's entire free
disk space. That's not a nice user experience!
Related, if fallocate does fail, try to create the test disk image using
dd instead. That should work everywhere.
Fixes: 4a28274227d0 ("test: fat: add test of non-contiguous file reads") Signed-off-by: Stephen Warren <swarren@nvidia.com>
Stephen Warren [Sun, 15 Nov 2015 06:53:49 +0000 (23:53 -0700)]
Fix sandbox build on Ubuntu 10.04
gcc 4.4.3 (which is the default native compiler on x86-64 Ubuntu 10.04)
doesn't seem to like initializers for sub-fields of anonymous unions.
Solve this by replacing the initialization with an assignment. This
fixes:
lib/lz4_wrapper.c: In function ‘ulz4fn’:
lib/lz4_wrapper.c:97: error: unknown field ‘raw’ specified in initializer
Signed-off-by: Stephen Warren <swarren@wwwdotorg.org> Reviewed-by: Tom Rini <trini@konsulko.com> Acked-by: Simon Glass <sjg@chromium.org>
This board uses the same CPU (8309) as VECT1. The memory however is
different since it has NAND Flash, the NOR Flash partitioning is
different and of course the FPGAs as well.
Signed-off-by: Valentin Longchamp <valentin.longchamp@keymile.com> Signed-off-by: Christoph Dietrich <christoph.dietrich@keymile.com>
It should be after the u-boot reserved sectors and before the env
sectors, since the solution used for kmvect1 (tell the linker to put the
firmware into the u-boot produced binary, at the end of the area) should
be the exception.
The #define is only "conditional" so that we can still support kmvect1.
board/keymile/km82xx: setting dip_switch 3 and 4 will run bootloader
Setting dip_switch 3 and 4 also will run bootloader in COGE3 and COGE6
It is required remove local mgmt IP address, when DIP Switch PIN3 is
enabled. DIP Switch 4 also enabled to avoid u-boot update in future
for DIP switch enhancements.
The ODT parameters for km8360 set the ODT_WR_ACS bit in u-boot KM-2011.09
that is used in the release bootpackage for kmcoge5ne. During the
transition from the kmeter1 to km8360 this was changed to
ODT_RD_ONLY_CURRENT, which is uncorrect and causes faulty RAM accesses at
low temperatures.
This is now changed to ODT_WR_ONLY_CURRENT which is the equivalent of
ODT_WR_ACS.
On the km8321 boards is CONFIG_SYS_DDRCDR not defined, which leads to
the DDRCDR not being configured at startup and still containing the
reset value.
The required settings for our km8321 hardware designs are different than
the reset value and must be set with CONFIG_SYS_DDRCDR, that is used
by mpc83xx's cpu_init_f function at early CPU initialization.
The important settings are the DDR2 internal voltage level and the
half-strength "drivers".
In our case where the DRAM chips are soldered on board and the routing
for these signals under control, half-strength is sufficient as a few
measurements done in the lasts have shown. Since all the hardware
qualification tests have been performed with half strength, the nominal
strength settings are removed in favor of the default reset half
strength settings.
Holger Brunck [Tue, 17 Nov 2015 09:53:28 +0000 (10:53 +0100)]
km/powerpc: move open firmware defines to km-powerpc.h
We use the same settings for open firmware defines on all our powerpc
targets, so move them from the CPU specific headers to the common
powerpc header.
Holger Brunck [Tue, 17 Nov 2015 09:53:27 +0000 (10:53 +0100)]
km/powerpc: increase space for kernel imange and FDT blob
128kByte and 3,986MB may be in the future too little for kernel the fdt
blob respectively the kernel image. So increase the reserved areas here,
we have the space for this.
Holger Brunck [Tue, 17 Nov 2015 09:53:26 +0000 (10:53 +0100)]
powerpc/82xx/km: add testpin detection for mgcoge3ne
On mgcoge3ne we also want to start the test application if the testpin
is asserted. But we don't have a full POST test support yet. So simply
add a function to read the testpin value.
Holger Brunck [Tue, 17 Nov 2015 09:53:25 +0000 (10:53 +0100)]
powerpc/82xx: make set/get_pin for km82xx more flexible
The get_pin and set_pin funciton was only used for pins on Port D and
therefore the value was hard coded in the function. Enhance this with a
parameter, that we are able to use this functions for other ports too.
This board is similar to TUXX1, but it has differend FPGAs.
Signed-off-by: Christoph Dietrich <christoph.dietrich@keymile.com> Signed-off-by: Andreas Huber <andreas.huber@keymile.com> Signed-off-by: Valentin Longchamp <valentin.longchamp@keymile.com>
Holger Brunck [Fri, 13 Nov 2015 15:15:21 +0000 (16:15 +0100)]
km/scripts: load fdt_bid_kwkey.dtb when working with tftp
When loading the dtb file via tftp we should load the one which matches
boardId and hwKey and not a common one for the boardname. We have boards
were different hwKeys are used and then we may load an incorrect dtb
file. If no fdt_bid_kwkey.dtb file is not a fallback to boardname.dtb is used.
If a DTB is found with cramfsls, the bootscript continues as expected.
If none is found, the cramfsloadfdt and boot subbootcmds are updated to
not load the DTB from cramfs and not pass it to the kernel. The kernel
thus must have an appended DTB otherwise the boot will fail.
This is required for the km_kirkwood boards that must support .esw where
the DTB sometimes is appended (for backwards compatibility) and sometimes
is passed correctly (as we do now for all newer boards).
projects / u-boot / log