1) When frequency changes, reload the watchdog timer right away, or it
may expire before the next reload. (Only matters when re-enabling the
PLL.)
2) Split out the timer/task debug output used by the watchdog into
their own routines, instead of assuming it's safe to call the command
handlers. Also make the flushes in those print routines safe to call
from interrupt level.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=none
TEST=waitms 1500; should print task dump again
Change-Id: I07e0ed24a526ae499566dab0bbeb0f5755cd5be6
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:9306
TEST=waitms 1600; see that timer info isn't all upscrewed
Change-Id: I7945f5114bbe0e9525cac76ce7376d4c32c4e654
This organizes the commands in the messaging protocol so that we have
a more clear distinction between commands which are intended to to
control or get status of the EC itself and those which are intended
to query information about peripherals, such as the obtaining the
keyboard state.
Note: This will require the mkbp code to be updated.
BUG=none
TEST=compile tested for now
Change-Id: I2d5c58fc794563d402da24e19fee146df817472a
Signed-off-by: David Hendricks <dhendrix@chromium.org>
(Or, if profiling is enabled.)
Also, track the number of task switches if profiling is enabled.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:9274
TEST=taskinfo
Also test by commenting out CONFIG_TASK_PROFILING in board.h; code
should still compile and run.
Change-Id: Ib3dbce74b6ccfba2bbe18c7309136412c30f364e
This change adds battery operating temperature check. Host and EC
will be turned off when overtemp. EC can be waked up by predefined
external signals like key press or AC_PRESENT gpio.
For safty reason, this change does not check battery temperature
after EC deep sleep.
Signed-off-by: Rong Chang <rongchang@chromium.org>
BUG=chrome-os-partner:8451,9108
TEST=none
Mock battery_temperature() fucntion to test.
Change-Id: I3203515b3df86192f690f9b98901020209ce49b3
This completes console output cleanup. The remaining calls to
uart_puts() and uart_printf() actually need to be that way.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:7464
TEST=manual
Change-Id: Ib1d6d370d30429017b3d11994894fece75fab6ea
This adds a 'ch' command which prints/sets which channels are active
This handles all the async output; the remaining debug commands will
be refactored to use ccprintf() / ccputs() in a followup CL.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:7464
TEST=manual
ch --> all channels active
ch 0x100 -> just port80 active
powerbtn -> system boots; only port 80 codes shown on console
Change-Id: I9efc43acec919b62b78c2c82c61946d32380adfe
Also tracks the distribution of IRQs, so we can see what's triggering
interrupts.
Task profiling is optional, enabled via CONFIG_TASK_PROFILING.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:7464
TEST=taskinfo
Change-Id: I266f2b49bff9648cda446210d5a302b460fec244
This saves power.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:8798
TEST=manual
Get a proto1 system modified with INA current sensor
1) From chroot:
dut-control i2c_mux_en:on i2c_mux:rem
dut-control pp3300_alw_mv pp3300_alw_ma
2) From EC console:
pll
(this should turn the PLL back on; it'll report clock frequency ~66MHz)
3) From chroot:
dut-control i2c_mux_en:on i2c_mux:rem
dut-control pp3300_alw_mv pp3300_alw_ma
Current (ma) should be bigger than in step 1
Change-Id: I806953684c57fd60bf481acb01dddffe2f2ad0ed
For those inputs from host, they are potential points fot cracker to stop
EC running. This patch removes these ASSERT() macro, but keeps those
for checking internal code logic.
BUG=none
TEST=build success
Change-Id: I91bf61f429a2387bb992b9518af60439f5592ea7
GPIO mode and pull-up/down registers do not all get initialized to
zero on reset. This patch ensures that all bits in the those registers
are set explicitly. An intermediate variable is used so that changes
are made atomically.
Note: output speed registers are also not all initialized to zero, but
we don't handle that in gpio_pre_init yet.
BUG=none
TEST=tested on newer daisy boards (which needed this patch to boot)
Change-Id: Ice2795197135dcee8f8484e4908dbfcf90fec2c9
Signed-off-by: David Hendricks <dhendrix@chromium.org>
This simply changes the constant '4' to MSG_PROTO_BYTES, which
includes the postamble. This helps reduce magic constant values
used in processing a packet.
BUG=chrome-os-partner:8975
TEST=tested on daisy (with kernel mkbp driver change)
Signed-off-by: David Hendricks <dhendrix@chromium.org>
Change-Id: Id4634076ad63f45783354179dfebea4fd450fc1e
This patch moves all I2C port initialization into configure_board
to ensure the alternate function gets set properly.
I2C ports should come up in their high-impedance state (Output,
open-drain, output set) and then get set as alternate function.
However, configure_board() runs before gpio_pre_init(), so the
mode register was getting set set back to general purpose output
instead of alt. function.
TODO: Fix gpio_pre_init() so we do not need to explicitly handle port
configuration in configure_board().
BUG=none
TEST=tested on daisy using keyboard
Change-Id: If837acd4f4204e467e7ed276f048b5b70ecbdb25
Signed-off-by: David Hendricks <dhendrix@chromium.org>
Source buffer was not being incremented properly, so the destination
buffer containing message payload was always msg[0].
BUG=none
TEST=tested on daisy (now keyboard protocol actually shows keys...)
Signed-off-by: David Hendricks <dhendrix@chromium.org>
Change-Id: If5a417291c9bff36bbeb2a87153de80300045257
Pass the actual msg_len instead of the maximum outbound message size.
BUG=none
TEST=tested on daisy using logic analyzer (notes below)
Beforehand, the output on a logic analyzer looked sane until the last
bits and stop condition, after which the bus appeared to go haywire
and the system would reset. I'm guessing that this is due to the EC
trying to drive SDA line while the host is holding SCL high.
Change-Id: I1afa520d858905a571540341fe02f374dcdf43c1
Signed-off-by: David Hendricks <dhendrix@chromium.org>
During the reboot_ec command, the keyboard state is lost after jump.
We need to restore info including:
- code set
- controller_ram[0]:
- XLATE
- KB/TP disabled
- KB/TP IRQ enabled
Remove the un-necessary keyboard_init() function.
BUG=chrome-os-partner:9102
TEST=tested on link.
EC runs on A
% ectool reboot_ec A
keyboard still working
% ectool reboot_ec RO
keybaord still working
% ectool reboot_ec RO
keybaord still working
ESC + power yo reset all system
repeat above steps and the keyboard keeps working.
Change-Id: I0fe21f7876459fc8047ff018fbfaaef5311cc49b
Proto1 has A3 silicon which is factory-trimmed.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:7693
TEST=boot and look for glitchy EC console. If it's not glitchy, it worked.
Change-Id: I56cb2458e600e76e458bce0f24832ef4f456ac14
These are faster than writing the entire EC, and safer if you're
mucking with things like the clock config. Of course, if you're
changing verified boot or something else that happens before
vboot_init(), these new commands won't help, and you should keep using
the existing flash_link command.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=none
TEST=manual from openocd
- Compile a new image
- From openocd, flash_link_b
- From ec console, note build compile date in image A: shouldn't have changed
- From ec consle, sysjump b; compile date should be for your new image
Change-Id: Ifd6aabe963948d3c5da5bdeb78c5f962a0f9f857
This also changes shared_mem to use all the remaining RAM, instead of
reserving a fixed-size buffer.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=chrome-os-partner:9161
TEST=manual
hostevent --> all masks should be 0
hostevent smi 0x12300000
hostevent --> should confirm SMI mask was set
sysjump b
hostevent --> should confirm SMI mask is still set
reboot
hostevent --> should confirm SMI mask is back to 0
Change-Id: Iccb6da6ccc93ee5036a3f478d24b717a462d9150
This covers modules which need to initialize before task_start(), but
don't particularly care in what order they're initialized.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=none
TEST=if it boots, it works
Change-Id: I69829aac8d1c3c14ee04916a794b84bbf03a09eb
More cleanup to come...
BUG=chrome-os-partner:7839
TEST=manual
Lights should blink in various ways, depending on the CPU state.
Also try the konami code on the keyboard.
Change-Id: I90be9aabb611278ed509493fbab4d5faff74e24c
Signed-off-by: Bill Richardson <wfrichar@chromium.org>
We can clear the reset cause in system pre-init now because of a
previous change which preserves it across a sysjump.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=none
TEST=if it boots, it works
Change-Id: I1d8b99df5a0be0de9545d22ad1a6b7fb3140f813
The commands in ectool.c look like this:
int cmd_foo(int argc, char *argv[]) { ... }
but unlike normal C convention, argv[0] is NOT the command. This change
makes argv[0] be the command name, so it's just like main().
BUG=chrome-os-partner:8748
TEST=manual
Run ectool as usual. You should see no difference in behavior.
Change-Id: Ia92784d46a287ab08f279f6255487817b620f200
Signed-off-by: Bill Richardson <wfrichar@chromium.org>
This will be used for sleep/wake/sysjump/etc. For now it's just wired
up to clock frequency changing.
Signed-off-by: Randall Spangler <rspangler@chromium.org>
BUG=none
TEST=manual: use nopll command, should still work
Change-Id: Iedcea5830bc18eacfd955c29b8f793aba8905dd8
Adding trickle charging mode to precharge batteries with
voltage lower than minimal design value. This CL adds
control to charger voltage to track battery input current
change.
To prevent battery from deeply discharging, this CL preserves
3% of the design capacity.
Minor bug fixes include error state check and charger control
logic.
Signed-off-by: Rong Chang <rongchang@chromium.org>
BUG=chrome-os-partner:8660,8661
TEST=manual
Plug AC power, the power adapter led should be
'yellow'. On the EC serial console, type 'battery'
and 'charger' commands.
Battery input current should staid close to its
desired current.
A deeply discharged battery (5.5V) should be revived to
a healthy state after 30 minutes ~ 4 hours.
Change-Id: Ibaa2396c6b751639d98db32f5919b1e8ec700e40
For legacy issue, the keyboard controller defaults to turn on XLATE
(translate) bit (which means EC generates scan codeset 1 althought
it internally supports codeset 2). In normal case, the BIOS/OS would clear
the XLATE bit to ask EC to generate codeset 2.
However, when EC jumps happens, the internal keyboard state doesn't know
this and always reset XLATE as on. This makes the EC generate garbage to OS.
So, this patch fixes would clear the XLATE if the EC reboot is a warmboot
based on the assumption that moderm OSes clear the XLATE bit.
BUG=chrome-os-partner:9102
TEST=on link
% ectool reboot_ec A (from r438 to r438)
Expect keyboard is hang (r438 is still buggy)
% chromeos-firmwareupdater --mode=recovery
Expect fail at gec_need_2nd_pass()
% ectool reboot_ec RO
Keyboard is still working! Bug fixed!
% ectool reboot_ec RO
Again. Still working.
Change-Id: If47bd8d7bbbb03b810d3b464ba3d92f8ff548237
The reboot_ec command could warm boot the EC while the host is still
running. However, this resets the internal state so that the keyboard
module is disabled on the EC side.
Check the reset cause during the keyboard init code. If it is wrm boot,
enable the keyboard (assume the host is on).
BUG=chrome-os-partner:9102
TEST=on link 1.0
% ectool version
Firmware copy: RO
% ectool reboot_ec RO
the keyboard keeps working.
Change-Id: I0009c561e2cd88789e50f9129b494538e50ee00e
This adds I2C2 support for Daisy:
- Initializes I2C2 GPIO lines
- Adds CONFIG_I2C so main() will call init function
- Adds work task for I2C2
BUG=chromium-os:28925
TEST=build on daisy and discovery; run on daisy
Change-Id: I147e3781b8bcac87ff248fb45c9978b614a24b89
Signed-off-by: Simon Glass <sjg@chromium.org>
Signed-off-by: David Hendricks <dhendrix@chromium.org>
Set up signal to audio codec as open-drain output. This should
be benign on older revisions of Daisy.
The signal CODEC_INT is driven low whenever a keypress has been
processed. For Daisy, a delay of about ~1ms from the time a key
is pressed until the time the codec is signaled is okay. This
gives us time to scan the columns to see if a key was actually
pressed so that we don't cause spurious codec interrupts.
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=None
TEST=Tested on Daisy using oscilloscope
Change-Id: Id3564f4aacbf7294b7151b082075f3c3ec8b1eb2
