RE: schedule() or scheduler_tick() not called during qemu run

[Chan Kim]

Just to let you know I solved this boot problem by 
1. adding gic initialization in u-boot (using CONFIG_GICV3),
2. adding code to power up GICR (percpu) using GICR_PWRR register.
3. and enabling system counter(FREQ value, enable), and EL1 physical timer
(TVAL, enable, interrupt mask setting)
This made uart interrupt and timer interrupt working hence the scheduler by
timer and the shell program.
It turned out u-boot should do these things. (arm reference software's
arm-tf is a good reference and the newest u-boot didn't have these GICR_PWRR
things)
The boot is ok and shell runs ok in the FPGA board.
Hope this helps someone later.

Chan Kim

>-----Original Message-----
>From: Chan Kim <ckim@etri.re.kr>
>Sent: Monday, August 29, 2022 3:42 PM
>To: 'qemu-discuss@nongnu.org' <qemu-discuss@nongnu.org>
>Subject: schedule() or scheduler_tick() not called during qemu run
>
>Hello qemu experts,
>
>I'm trying to debug a situation where the /init script is processed at the
>end of the kernel_init function which is the final stage, but the last
>command in the /init script which is 'exec /bin/sh' hangs after the first
>call and return of schedule() function in my fpga test. The file system is
>initramfs.cpio.gz and was made using busybox. If I replace the /init scritp
>with a program which repeats printf("hello %d", cnt++), it stops after
>117~118 times and I can see schedule() was called and finished before it
>hangs.
>So I tried to follow what's happening using qemu (though the virtual
>machine is not exactly the same with the actual board), but when I set
>breakpoint in schedule() or scheduler_tick(), the breakpoint never kicks
>in(Of course other break points all work). Is this normal in qemu or is
>there something wrong in my qemu execution(I mean in my virtual machine)?
>BTW, this is an arm64 board.
>
>Chan Kim