[GIT PULL] runtime verification: Updates for 6.17

[Richard Weinberger]

On Sat, Aug 16, 2025 at 3:44 PM Russell King (Oracle)
<linux at armlinux.org.uk> wrote:
> > With this change:
> > [    2.943690] Kernel panic - not syncing: VFS: Unable to mount root
> > fs on unknown-block(0,0)
> > [    2.950101] CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted
> > 6.17.0-rc1 #176 NONE
> > [    2.950509] Hardware name: Generic DT based system
> > [    2.950865] Call trace:
> > [    2.951722]  unwind_backtrace from show_stack+0x18/0x1c
> > [    2.953111]  show_stack from dump_stack_lvl+0x54/0x68
> > [    2.953312]  dump_stack_lvl from vpanic+0xf8/0x388
> > [    2.953526]  vpanic from __do_trace_suspend_resume+0x0/0x50
>
> Please lookup the value of __do_trace_suspend_resume in System.map.

Hm, I see it twice in System.map, both times non-global.
$  grep __do_trace_suspend_resume System.map
c0303098 t __do_trace_suspend_resume
c03c5758 t __do_trace_suspend_resume

>
> >
> > With commit  3f045de7f5 "panic: Add vpanic()" reverted:
> > [    2.680077] Kernel panic - not syncing: VFS: Unable to mount root
> > fs on unknown-block(0,0)
> > [    2.690241] CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted
> > 6.17.0-rc1-dirty #177 NONE
> > [    2.690832] Hardware name: Generic DT based system
> > [    2.691339] Call trace:
> > [    2.692530]  unwind_backtrace from show_stack+0x18/0x1c
> > [    2.694283]  show_stack from dump_stack_lvl+0x54/0x68
> > [    2.694536]  dump_stack_lvl from panic+0x110/0x31c
> > [    2.694944]  panic from mount_root_generic+0x1fc/0x2ac
>
> Also please lookup mount_root_generic.

$ grep mount_root_generic System.map
c1c01814 T mount_root_generic

>
> > Maybe it changes the stack layout enough such that dump_stack() on
> > ARM32 is no longer
> > able to walk it? I'm using gcc 15.1.1, if this matters.

In the meanwhile I tried also with gcc 8.5, same problem.

> Also please check whether you're using frame pointers or the unwinder
> (CONFIG_UNWINDER_FRAME_POINTER or CONFIG_ARM_UNWIND).

With CONFIG_UNWINDER_FRAME_POINTER the stack trace is sane,
so only CONFIG_ARM_UNWIND is broken.
I kind of expected it the other way around...

-- 
Thanks,
//richard