Fix: CPU power down #197
Fix: CPU power down #197
Conversation
This commit fixes issue introduced in commit 6df80f4 by ensuring the CPU jumps to the CPU message handler (if received any CPU message) after waking from power down. This allows the target CPU to update the `psci_ctx.state` field to `ON`, enabling the vCPU to resume execution. Signed-off-by: João Peixoto <[email protected]>
Currently, Arm Trusted Firmware does not support power-down suspend calls. This is evident in the implementation of the `qemu_pwr_domain_suspend` function, which contains an `assert(0)` statement, causing the CPU to halt indefinitely without recovery. Signed-off-by: João Peixoto <[email protected]>
| @@ -127,6 +127,11 @@ void cpu_standby_wakeup(void) | |||
|
|
|||
| void cpu_powerdown_wakeup(void) | |||
There was a problem hiding this comment.
I think we can merge this with cpu_standby_wakeup and have a single wake up function
There was a problem hiding this comment.
I think we can merge this with
cpu_standby_wakeupand have a single wake up function
My question is: how can we invoke the cpu_standby() and cpu_powerdown() functions if cpu()->vcpu is NULL? Should we pass an argument to the new cpu_wakeup() function and determine the specific method to call based on that argument?
There was a problem hiding this comment.
For me it's ok how it is currently
IMO in this case it is better to have the two separate but similar functions
| @@ -127,6 +127,11 @@ void cpu_standby_wakeup(void) | |||
|
|
|||
| void cpu_powerdown_wakeup(void) | |||
There was a problem hiding this comment.
For me it's ok how it is currently
IMO in this case it is better to have the two separate but similar functions
No description provided.