mirror of
https://github.com/Telecominfraproject/OpenCellular.git
synced 2026-01-06 15:01:35 +00:00
fc783fb585
The existing state machine does the following: - when power button is pressed it 1. powers on the AP 2. sets a timer of 1 sec and then 3. waits for power button to be released When the timer fires it checks xpshold is set by the AP and if so it clears the pwron signal (which is used by the AP to detect power button is pressed). The problem occurs when the user holds the power button for more than a second. The AP turns on xpshold, then notices that pwron is still on and subsequently powers down because it thinks the power button is pressed. When the button is finally released, since it was held down for more than a second, the timer routine notices that xpshold is not on and therefore shuts down the system. Another problem found while analysing this state machine is that loop checking for poweroff only triggers on the rising edge of xpshold. This means that if the AP powers down the EC might miss a possible power event. Here is the proposed fix: When the power button is pressed the EC will: 1. power on the AP 2. Check for xpshold to be asserted with a 1 sec timeout 3. If uboot is healthy xpshold should come on pretty quickly; the EC then waits for the power button to be released in less than 8 seconds 4. If the power button is released then the EC waits for power off events. 5. If the power button is not released it waits for upto 8 seconds before turning off the AP. The added wrinkle is how to address a borked uboot case. In the case where xpshold doesn't come on in < 1 second, the EC will allow the AP to stay on for upto 16 seconds so that USB boot can finish. The user must hold the power button down until uboot boots and sets xpshold. The assumption here is that USB boot takes < 16 seconds. BUG=chrome-os-partner:12748 TEST="follow instructions in bug report" Change-Id: I5b582a6c3ae3449238e2813e4a581bd8f92dd846 Signed-off-by: Puneet Kumar <puneetster@chromium.org> Reviewed-on: https://gerrit.chromium.org/gerrit/31291 Reviewed-by: Vincent Palatin <vpalatin@chromium.org> Reviewed-by: David Hendricks <dhendrix@chromium.org>
- EC Lib
This wraps Blizzard driverlib and implements the EC chip interface defined
by Google. See below diagram for architecture.
+--------------------+
| Host BIOS/OS |
+--------------------+
---- host interface ----
+--------------------+
| Google EC features |
+--------------------+
---- chip interface ---- The interface is defined in
src/platform/ec/chip_interface/*.
+--------------------+ But the real implementation is in EC Lib.
| EC Lib |
+--------------------+
| Blizzard low level |
| driver, the |
| driverlib. |
+--------------------+
Build Options
=============
- CONFIG_WATCHDOG_HELP
Try to detect a watchdog that is about to fire, and print a trace.
This is needed on STM32, where the independent watchdog has no early
warning feature and the windowed watchdog has a very short period.
- CONFIG_PANIC_NEW_STACK
When reporting a panic, change to a completely new stack. This might
help get a useful trace out a situation where the stack or stack
pointer has been corrupted.
- CONFIG_PANIC_HELP
Report extra information about a panic, such as the fault address,
here shown as bfar. This shows the reason for the fault and may help
to determine the cause.
=== EXCEPTION: 03 ====== xPSR: 01000000 ===========
r0 :0000000b r1 :00000047 r2 :60000000 r3 :200013dd
r4 :00000000 r5 :080053f4 r6 :200013d0 r7 :00000002
r8 :00000000 r9 :200013de r10:00000000 r11:00000000
r12:00000000 sp :200009a0 lr :08002b85 pc :08003a8a
Precise data bus error, Forced hard fault, Vector catch, bfar = 60000000
mmfs = 00008200, shcsr = 00000000, hfsr = 40000000, dfsr = 00000008
- CONFIG_ASSERT_HELP
Report assertion failures in a vebose manner to aid debugging. When
enabled an ASSERT() which fails will produce message in the form:
ASSERTION FAILURE '<expr>' in function() at file:line