diff options
| author | Alec Berg <alecaberg@chromium.org> | 2015-10-23 16:37:45 -0700 |
|---|---|---|
| committer | chrome-bot <chrome-bot@chromium.org> | 2015-10-23 22:49:29 -0700 |
| commit | 9acd84460e953dd793c9594b141ca7e38568820d (patch) | |
| tree | 5e12ae87e6f7c94fb49d74381adccabcdfb8b2e6 /board/glados_pd/board.c | |
| parent | 71e12124e5e5e54e921a6940ed93599a90b3035a (diff) | |
| download | chrome-ec-9acd84460e953dd793c9594b141ca7e38568820d.tar.gz | |
glados: oak: reboot EC if PD MCU crashes
If PD MCU crashes, it will go back to RO code and stay there
until the next AP boot. So, if EC detects PD has crashed, then
EC should panic reboot with debug message that it detected
a PD crash.
PD MCU crash is detected by EC by seeing the PD MCU transition
from RW to RO, without it setting the flag that it got there
from a sysjump.
This CL also makes minor changes to oak_pd and glados_pd board.c
files to make them identical, other than the few minor real
differences between them.
BUG=none
BRANCH=none
TEST=tested on glados using pdcmd console command on EC to test
sysjumps and reboots:
sysjump to RW: pdcmd 0xd2 0 2 0
sysjump to RO: pdcmd 0xd2 0 1 0
cold reboot: pdcmd 0xd2 0 4 0
Verified that PD can jump back and forth between RO and RW
without EC panicing. Verified that if PD MCU is in RW and
reboots, then the EC will panic and print 'PD crash'. Verify
if PD MCU reboots while in RO, without ever going to RW first,
then EC does not panic.
Change-Id: Id3191f4005e70a6c61a9322bf535b4374e85eb9a
Signed-off-by: Alec Berg <alecaberg@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/308586
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Diffstat (limited to 'board/glados_pd/board.c')
| -rw-r--r-- | board/glados_pd/board.c | 14 |
1 files changed, 11 insertions, 3 deletions
diff --git a/board/glados_pd/board.c b/board/glados_pd/board.c index d1f4568842..791d29a097 100644 --- a/board/glados_pd/board.c +++ b/board/glados_pd/board.c @@ -21,13 +21,15 @@ #define CPRINTS(format, args...) cprints(CC_USBCHARGE, format, ## args) +/* Indicate which source is driving the ec_int line. */ static uint32_t ec_int_status; +static uint32_t pd_status_flags; + void pd_send_ec_int(void) { - /* If any of 3 sources are active, then drive the line low */ + /* If any sources are active, then drive the line low */ gpio_set_level(GPIO_EC_INT, !ec_int_status); - } void vbus0_evt(enum gpio_signal signal) @@ -61,6 +63,12 @@ static void board_init(void) /* Enable interrupts on VBUS transitions. */ gpio_enable_interrupt(GPIO_USB_C0_VBUS_WAKE_L); gpio_enable_interrupt(GPIO_USB_C1_VBUS_WAKE_L); + + /* Set PD MCU system status bits */ + if (system_jumped_to_this_image()) + pd_status_flags |= PD_STATUS_JUMPED_TO_IMAGE; + if (system_get_image_copy() == SYSTEM_IMAGE_RW) + pd_status_flags |= PD_STATUS_IN_RW; } DECLARE_HOOK(HOOK_INIT, board_init, HOOK_PRIO_DEFAULT); @@ -136,7 +144,7 @@ static int ec_status_host_cmd(struct host_cmd_handler_args *args) * ec_int_status is used to store state for HOST_EVENT, * TCPC 0 Alert, and TCPC 1 Alert bits. */ - r->status = ec_int_status; + r->status = ec_int_status | pd_status_flags; args->response_size = sizeof(*r); /* Have the PD follow the EC into hibernate. */ |