I have a system with bad temperature sensors and the EC is
constantly shutting the system down, but provides no
visible indication that it is doing so.
I have serious concerns about the EC behavior in this case
as it seems to be doing things it shouldn't. However just
providing indication via the console about what it is doing
is at least useful for development and debug.
BUG=none
TEST=boot on system with bad temp sensors and see that
the EC indicates it is initiating a shutdown.
[14.008340 critical temperature; shutting down]
[14.008660 x86 power force shutdown]
[14.009153 LPC RESET# asserted]
Change-Id: I6beeb269a135bd8c245c0357670fe29648d48968
Signed-off-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27553
Reviewed-by: Randall Spangler <rspangler@chromium.org>
This is needed for flash pre-init to be able to hard reset to clear
uncommitted write protect flags without losing the reset flags.
BUG=chrome-os-partner:11368
TEST=manual
Use reboot and sysinfo commands...
1. reset with keyboard. flags -> reset-pin
2. 'reboot soft preserve' flags -> soft reset-pin power-on
3. 'reboot hard preserve' flags -> hard soft reset-pin power-on
4. 'reboot soft'. flags -> soft
5. 'reboot hard'. flags -> hard power-on
Change-Id: I6164a78d99c5c10330f90f651148c5795e7afdda
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27418
Reasons are really bitflags, not a single reason. This will make it
easier to implement flash protection on LM4, where hibernate is a
subset of power-on reasons.
Also added some additional flags we pass in a hibernate register so...
1) We don't recognize spurious RTC wake reasons
2) Hard reset via system_reset(1) is detected as a hard reset, not a RTC wake
BUG=chrome-os-partner:11368
TEST=manual
1. Keyboard reset = power-on reset-pin
2. Pull battery = power-on
3. reboot = soft
4. reboot hard = power-on rtc-alarm
5. hibernate 10 then push power button = power-on wake-pin
6. reboot 3 sec later = soft
7. hibernate 1 = power-on rtc-alarm
Change-Id: Icbbdbcf6dfd13c8a6a4f80a23f64cebebbfba26e
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27417
According to its datasheet, LM4 doesn't support vectreset, but does
support sysresetreq. Using sysresetreq properly resets all
peripherals and cleanly resets the LM4.
Auto-erase makes the flash command faster.
BUG=none
TEST=flash_link_ro, flash_link_a, flash_link, all with different images
Change-Id: I80b72de2ee0f42cee1b0f2f2a53062384d2c29aa
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27416
Link battery pack specification suggested a lower charging current
when voltage pass 8.0V. But the lowering the current in constant
current phase leads voltage drop. And the battery goes back to
high current zone, < 8.0V.
This CL adds a 10 seconds debounce time to prevent charging current
change too quickly.
Signed-off-by: Rong Chang <rongchang@chromium.org>
BUG=chrome-os-partner:9572
TEST=manual
watch battery V+ on oscilloscope when charging voltage cross 8.0V
Change-Id: I002f941e33b029e38f813ab2e292c6b73a054352
Reviewed-on: https://gerrit.chromium.org/gerrit/27275
Commit-Ready: Rong Chang <rongchang@chromium.org>
Reviewed-by: Rong Chang <rongchang@chromium.org>
Tested-by: Rong Chang <rongchang@chromium.org>
Kernel and U-boot define it elsewhere, but coreboot doesn't. So put
it here for greatest compatibility.
BUG=none
TEST=if it builds it works
Change-Id: I595910e9198e37bc97d23cd4c249454e1ed64cd8
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27375
Changed the parameters from int to uint32_t (which is how it was called
anyway).
BUG=chrome-os-partner:11045
TEST=manual
No visible change. Nothing should break.
Change-Id: I4fbe34f67df7d37f5039987a7a89e626916d6eb6
Signed-off-by: Bill Richardson <wfrichar@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27382
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Make sure that a negative 'num' param won't be accepted if passed by a
malicious caller.
BUG=chrome-os-partner:11048
TEST=manual
No visible changes, everything should continue to work.
Change-Id: I8128d24adc99e5ff954a6b8065e1bfa8bf20630e
Signed-off-by: Bill Richardson <wfrichar@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27386
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Make sure it doesn't run off the end of the I2C device list.
BUG=chrome-os-partner:11051
TEST=manual
No visible changes. Everything should continue to work.
Change-Id: I43182361cda6be578e57acef01afb27260cea80a
Signed-off-by: Bill Richardson <wfrichar@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27390
Reviewed-by: Randall Spangler <rspangler@chromium.org>
The charger task relies on calc_next_state() performing a delay before
returning. My reading of the code suggests that this doesn't happen
always. For example:
When pre-charging:
if (battery_temperature(&batt_temp) == EC_SUCCESS)
return ST_CHARGING;
When discharging and capacity is low:
/* Check remaining charge % */
if (battery_state_of_charge(&capacity) == 0 && capacity < 10)
return notify_battery_low();
I would like to suggest that the code be refactored to more like:
int next_checkms = 5000; /* next time to check battery */
while (1) {
int action = ACTION_NONE;
err = get_state(&state);
if (!err) {
action = calculate_action(&state);
err = perform_action(action, &next_check_ms);
}
usleep(next_check_ms * 1000);
}
so that the delays are really clear, the state is all read at once,
there is no reliance on earlier state, and we always delay even on
error.
In the meantime, this CL inserts a mandatory 5 second delay in the
loop, which should prevent the charger task lockup.
BUG=chrome-os-partner:11285
TEST=manual
(please do this test before committing)
1. boot to kernel, see that battery can be seen
2. suspend and resume device
3. see that the charger loop does not cause an EC watchdog reset and
AP power off/reset. There should be no watchdog warning message on the
EC console.
Change-Id: I141e374933c4dc0ec60bcdccf96443f57067c585
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27353
Reviewed-by: Rong Chang <rongchang@chromium.org>
Tested-by: Rong Chang <rongchang@chromium.org>
Reviewed-by: Che-Liang Chiou <clchiou@chromium.org>
Tested-by: Katie Roberts-Hoffman <katierh@chromium.org>
Commit-Ready: Katie Roberts-Hoffman <katierh@chromium.org>
These commands are used in factory test. If system is locked, GPIO
commands are disabled for security reason.
BUG=chrome-os-partner:11164
TEST= - 'ectool gpioget enable_backlight' gives 1.
- 'ectool gpioset enable_backlight 0' turns off display.
- Lock system. Check these commands return error.
Change-Id: I3ea41285075ebe963ba7d30e4ae183cef9b1c105
Reviewed-on: https://gerrit.chromium.org/gerrit/27019
Commit-Ready: Vic Yang <victoryang@chromium.org>
Reviewed-by: Vic Yang <victoryang@chromium.org>
Tested-by: Vic Yang <victoryang@chromium.org>
This command should only be accessible when flash write protect is
unlocked.
BUG=chrome-os-partner:9716
TEST= - Check we can charge and discharge normally.
- Connect AC and force idle. Check battery current is 0 mA.
Change-Id: I74a318c1f5562d6a23b722736615fd1f883dc35a
Reviewed-on: https://gerrit.chromium.org/gerrit/27259
Commit-Ready: Vic Yang <victoryang@chromium.org>
Reviewed-by: Vic Yang <victoryang@chromium.org>
Tested-by: Vic Yang <victoryang@chromium.org>
This is set when the EC boots and sees the recovery key combo.
This is preferable to the memmap'd keyboard recovery switch, which the
host has no way of clearing. The idea is that the host RO firmware
reads this event and sets the recovery reason.
I will be removing the memmap'd keyboard recovery switch once U-boot
and coreboot check this new event.
BUG=chrome-os-partner:10034
TEST=manual
1. power+esc+refresh; should see this even in EC log
[0.000838 event set 0x00004000]
(time of event may vary)
2. reboot EC; should NOT see the event
Change-Id: Id0672749f63c5022624a72ec91d30dcadfea5ef8
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27328
Now both copies of the event state live in host_event_commands.c, and
lpc / memmap just shadows the main copy.
BUG=chrome-os-partner:11172
TEST=manual
Boot system. should see events 0x2000, 0x80, 0x08 get set and then cleared.
At U-boot prompt, type on keyboard. Should set event 0x1000 get set,
but only on the first keypress (because U-boot doesn't consume that
event).
Then from EC console,
hostevent clear 0x1000 -> see event 0x1000 clear
hostevent clear 0x1000 -> no debug output (it's already clear)
hostevent clearb 0x1000 -> see event copy B 0x1000 clear
hostevent clearb 0x1000 -> no debug output (copy B is already clear)
Change-Id: I855c035865649ba1490cd9027157d5bcdcc9895f
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27321
U-boot and ectool need this to see what events have occurred, in a way
which doesn't conflict with ACPI/SCI/SMI.
BUG=chrome-os-partner:11172
TEST=manual
- boot EC
- look at event set / event clear debug output; that's for copy A
- from ec console,
hostevent -> events=0, events-B = 0x2088
hostevent clearb 0x2000 -> events-B = 0x88
Change-Id: If842b157914426df593d82af2bfb08a923caa34b
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27317
Both EC and ectool are still backwards-compatible to the old interface.
BUG=chrome-os-partner:11275
TEST=manual
From U-boot prompt: mkbp hash // test old host talking to new EC
From root shell: ectool echash // test new host talking to new EC
You can also update just the OS and use an old EC, and verify that
'ectool echash' still works, which tests a new host talking to an old
EC.
Change-Id: I2afbb208cb16836f842ba119b74b1ab6a38ce5d5
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27313
Saves 2 params being passed around needlessly.
BUG=chrome-os-partner:11275
TEST=mkbp hash from u-boot console should still work
Change-Id: I958e4a09f16413e4d051e278dc0384aa9b791aa4
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27312
And retain compatibility for old requests.
BUG=chrome-os-partner:11275
TEST=from u-boot prompt, 'mkbp hash'
from root shell, 'ectool flashread 0 68084 /tmp/foo'
then compare to first 68084 bytes of ec.bin
Change-Id: Id82068773703543febde79fc820af7486502e01f
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27226
These commands are only available when system is unlocked.
BUG=chrome-os-partner:9716
TEST=When system is unlocked, check we can read/write I2C bus.
When system is locked, check the commands return error.
Change-Id: I8c4a8a42ab13d62c7e47bf33e2a7abc6f61e43fa
Reviewed-on: https://gerrit.chromium.org/gerrit/27254
Reviewed-by: Rong Chang <rongchang@chromium.org>
Commit-Ready: Vic Yang <victoryang@chromium.org>
Tested-by: Vic Yang <victoryang@chromium.org>
The spec of link internal battery pack changed. This CL adds
parameter for new pack, and renames vendor specific source file
accordingly.
The new trickle charging current limit should fix the slow
pre-charging issue.
Signed-off-by: Rong Chang <rongchang@chromium.org>
BUG=chrome-os-partner:11298,10201
TEST=manual
The minimum trickle charging current should be very close to
0.01 C (85mA). Previous battery pack firmware uses 5mA in pre-
charging.
Change-Id: I0bad679db7dd087894297e6eb0e85c9b12fdf444
Reviewed-on: https://gerrit.chromium.org/gerrit/27256
Reviewed-by: Vic Yang <victoryang@chromium.org>
Commit-Ready: Rong Chang <rongchang@chromium.org>
Tested-by: Rong Chang <rongchang@chromium.org>
This is necessary for an imminent change which passes version data
from the host bus (LPC/I2C/SPI) into the host command handler.
BUG=chrome-os-partner:11275
TEST=from u-boot prompt, 'mkbp hash'
Change-Id: If34d0d7c6dc320ad5632becf512c30900fd61aca
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27190
Reviewed-by: Bill Richardson <wfrichar@chromium.org>
Added version mask field to DECLARE_HOST_COMMAND() because it's
convenient to do so when I'm touching all host command
implementations, but all commands simply declare version 0 and nothing
checks it yet. Will add version support in a followup CL.
This change is internal to the EC; it does not change the data sent
over the host interface.
BUG=chrome-os-partner:11275
TEST=manual
ectool version && ectool echash; should get sane data from both
ectool flashread 0x80 0x40 /tmp/foo && od -tx1 /tmp/foo
should match data from offset 0x80 of ec.bin (od -j128 -n64 -tx1 ec.bin)
Change-Id: I5699f72b8d5e1ac23929353c9a34158d76c44206
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27172
Reviewed-by: Bill Richardson <wfrichar@chromium.org>
This change shorten the timeout period to 10ms.
Experiments showed that most master mode communication status bit
changed within 2000us. Hence 10ms timeout is much reasonable than
the original value.
Signed-off-by: Rong Chang <rongchang@chromium.org>
BUG=none
TEST=manual, with i2c scope
Change-Id: Ib26795bd2e4877b5c33719e96a2c2c7c11333eab
Reviewed-on: https://gerrit.chromium.org/gerrit/27145
Tested-by: Rong Chang <rongchang@chromium.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Commit-Ready: Rong Chang <rongchang@chromium.org>
We choose PE2/USB1_CTL1 for BOOTCFG. If we somehow brick the EC, we can
pull this signal to ground and boot EC into boot loader to recover it.
BUG=chrome-os-partner:8769
TEST=On link:
- Factory reset EC to clear BOOTCFG
- Flash new EC binary and reboot
- rw 0x400fe1d0 to check BOOTCFG value is correctly written
- Boot normally and check EC is operating correctly
- Pull the signal to ground and reset EC, check EC stays in boot
loader. Check in this case we can program EC.
- Attempt to write different value to BOOTCFG and check BOOTCFG
doesn't change.
Change-Id: Ia6705114d495b18bd7ee4afc1e61e84a21b51198
Reviewed-on: https://gerrit.chromium.org/gerrit/27000
Reviewed-by: Vic Yang <victoryang@chromium.org>
Tested-by: Vic Yang <victoryang@chromium.org>
Commit-Ready: Vic Yang <victoryang@chromium.org>
Now that ACPI events are handled directly in the LPC interrupt
handler, we can simplify the host event code.
BUG=chrome-os-partner:11240
TEST=boot system; should boot
close lid; should send SMI and suspend system
Change-Id: I8c73ea31a66e94310e4460a008635a103220413e
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27100
Reviewed-by: Bill Richardson <wfrichar@chromium.org>
And only support it for port 62/66.
Also remove 'ectool queryec', because it can't touch port 62/66 once
the kernel/ACPI owns it, and query-event isn't supported on the user
command port.
BUG=chrome-os-partner:11240
TEST=boot system and check EC console output; should see event clears between
host commands 0x23, 0x8e, but no hostcmd 0x84.
[0.396780 LPC RESET# deasserted]
[0.486953 Port 80: 0x29]
[0.487415 hostcmd1 0x23]
[0.764407 Port 80: 0x88]
[0.764579 event clear 0x00000008 -> 00002080]
[0.764928 event clear 0x00000080 -> 00002000]
[0.765224 event clear 0x00002000 -> 00000000]
[0.765578 hostcmd1 0x8e]
[0.765868 hostcmd1 0x06]
Change-Id: I8ed161dbccd396d685ddf6829a27dfef87d919fb
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27095
Reviewed-by: Bill Richardson <wfrichar@chromium.org>
BUG=chrome-os-partner:9922
TEST=manual
Press power+refresh+d.
From ec console, 'optget'. No reference to fake dev switch
From host, 'ectool vboot'. Should see either 'fake_dev=0' or no mention of fake dev switch at all.
Change-Id: I66bc5e926d6e639b206563e764bcc730cce9227c
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/27061
Reviewed-by: Bill Richardson <wfrichar@chromium.org>
This fixes a small bug introduced in a previous refactoring CL where
by the data buffer offset wasn't being incremented, thus causing
writes to repeatedly write the first two bytes of the incoming data
to each chunk of memory.
BUG=chrome-os-partner:11097
TEST=successfully wrote and verified data on Snow
Signed-off-by: David Hendricks <dhendrix@chromium.org>
Change-Id: I8dae00652f4d74bc6ef449297b6e78b13b21255f
Reviewed-on: https://gerrit.chromium.org/gerrit/26985
Reviewed-by: Katie Roberts-Hoffman <katierh@chromium.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Commit-Ready: David Hendricks <dhendrix@chromium.org>
Reviewed-by: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Currently host_command_reboot() casts the supplied buffer and access
it directly. This works until about half-way thru the function when
the same buffer potentially gets overwritten when sending the response
code to the host.
This CL gets around that by copying the reboot parameters into a
separate buffer.
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
TEST=Tested using 'ectool reboot_ec' on Snow (see below)
Before, EC console displayed "[Executing host reboot command]" and
ectool showed no actual change (reboot command was being overwritten
with EC_REBOOT_CANCEL):
localhost ~ # ectool version | grep 'Firmware copy'
Firmware copy: RO
localhost ~ # ectool reboot_ec A
localhost ~ # ectool version | grep 'Firmware copy'
Firmware copy: RO
With the patch applied, the EC console shows the EC boot-up messages
and ectool confirms that the jump took place:
localhost ~ # ectool version | grep 'Firmware copy'
Firmware copy: RO
localhost ~ # ectool reboot_ec A
localhost ~ # ectool version | grep 'Firmware copy'
Firmware copy: A
Change-Id: I8aca8d468d1125c3fb8d320ec0fb79d2822b0b20
Reviewed-on: https://gerrit.chromium.org/gerrit/26998
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Commit-Ready: David Hendricks <dhendrix@chromium.org>
When debug port is left unconnected, EC get garbage keypress input from
both UART ports due to UART pin floating. By clearing UART receive FIFO
when initializing UART module, this can be prevented.
BUG=chrome-os-partner:10235
TEST=Reset EC with debug port unconnected. Check u-boot doesn't get any
keypress.
Change-Id: I3e84db1f2f95a0173c11180f6afb3edb705d0ae5
Reviewed-on: https://gerrit.chromium.org/gerrit/26908
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Tested-by: Rong Chang <rongchang@chromium.org>
Commit-Ready: Vic Yang <victoryang@chromium.org>
If the 'percent' passed in is lower than 0, then set it to 0. If it is
higher than 100, set to 100.
BUG=chrome-os-partner:11052
TEST=Check with 'kblight', 'fanduty', and 'faninfo'.
Change-Id: If84ab12658bf136eaaf1adecc0522a977c94f98d
Reviewed-on: https://gerrit.chromium.org/gerrit/26904
Reviewed-by: Randall Spangler <rspangler@chromium.org>
Commit-Ready: Vic Yang <victoryang@chromium.org>
Tested-by: Vic Yang <victoryang@chromium.org>
BUG=chrome-os-partner:11046
TEST=manual
1. Try 'ectool kbpress' command from root shell with a few params when system is unlocked (WP disabled). Should work.
2. Try again when WP is enabled and locked ('flashwp lock' then 'reboot' from EC console). Should fail.
2b. Alternately, use 'syslock' to lock system for just the current boot.
Change-Id: Ic271b13b1973ff3cb83a113274d4e33655c0cf85
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/26936
Reviewed-by: Bill Richardson <wfrichar@chromium.org>
Allows 'reboot cold' as an alias for 'reboot hard', since I keep
mis-typing that anyway. Returns error for any other option.
('reboot' by itself still does a warm reboot).
BUG=none
TEST=manual
reboot -> does warm reboot
reboot hard -> does hard reboot (reason = rtc alarm)
reboot cold -> ditto
reboot foo -> prints error
Change-Id: I183714d4ba09abee3bd9a6f0d5df82389becf410
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/26924
This allows the console or AP to keep the EC in its RO code.
Previously, the EC could jump from RO to RW even if the system was
locked in pre-init.
Also, sysjump console command doesn't need to check if system is
disabled before calling system_run_image_copy(), because that function
also checks. This now matches how the host command works.
BUG=chrome-os-partner:11147
TEST=manual
syslock
sysjump A -> works
reboot
syslock
sysjump disable
sysjump A -> fails
Repeat, using 'ectool reboot_ec disable-jump' at root shell instead of
'sysjump disable' at EC console.
Change-Id: I0b168a93e97802ba30e7c225b01d70ea66e8db58
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/26898
EC software sync still needs to be able to sysjump later after the AP
decides which image the EC should be running.
BUG=chrome-os-partner:11087
TEST=manual
sysjump A
sysjump RO
sysjump A
sysjump RO
Change-Id: I74eaf10dacf6eaa6aeabae2ff3b21ad387517605
Signed-off-by: Randall Spangler <rspangler@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/26896
Reviewed-by: Bill Richardson <wfrichar@chromium.org>
