Re: with kernel 6.4 update, boot fails drops into maintenance

Hi Riccardo!

On Mon, 2023-07-31 at 12:21 +0200, Riccardo Mottola wrote:
> I have a Fire T2000 which has been stable for my usage for months.
> Now I upgraded kernel (and userland) in debian to:
> 
> Yesterday, when hooking up a serial console, i was getting asked for 
> root password and dropped into maintenance mode.
> Today I hooked up better through a console with minicom to be able to 
> copy the error, but I get these errors again and again:
> 
> 
> [   17.134833] OOPS: RPC [000000000042ce10]
> [   17.134843] OOPS: RPC <arch_cpu_idle+0x70/0xc0>
> [   17.134875] OOPS: Fault was to vaddr[bc0]
> [   17.134885] CPU: 25 PID: 0 Comm: swapper/25 Tainted: G D            
> 6.4.0-1-sparc64-smp #1  Debian 6.4.4-2
> [   17.134908] Call Trace:
> [   17.134915] [<0000000000de9614>] dump_stack+0x8/0x18
> [   17.134948] [<0000000000df547c>] do_sparc64_fault+0x4bc/0xb00
> [   17.134979] [<0000000000407714>] sparc64_realfault_common+0x10/0x20
> [   17.135017] [<0000000000000bc0>] 0xbc0
> [   17.135034] [<0000000000dea050>] default_idle_call+0x30/0x120
> [   17.135067] [<00000000004c32f8>] do_idle+0xd8/0x1a0
> [   17.135095] [<00000000004c3684>] cpu_startup_entry+0x24/0x40
> [   17.135123] [<0000000000440f28>] smp_callin+0xe8/0x120
> [   17.135145] [<00000000010d8c14>] 0x10d8c14
> [   17.135162] [<0000000000000000>] 0x0
> [   17.135179] Unable to handle kernel NULL pointer dereference
> [   17.135187] tsk->{mm,active_mm}->context = 0000000000000041
> [   17.135197] tsk->{mm,active_mm}->pgd = ffff800017270000
> [   17.135207]               \|/ ____ \|/
> [   17.135207]               "@'/ .. \`@"
> [   17.135207]               /_| \__/ |_\
> [   17.135207]                  \__U_/
> [   17.135216] swapper/25(0): Oops [#49]
> [   17.135230] CPU: 25 PID: 0 Comm: swapper/25 Tainted: G D            
> 6.4.0-1-sparc64-smp #1  Debian 6.4.4-2
> [   17.135253] TSTATE: 0000004411001605
> 
> 
> so something worsened! darn! this looks as back when all kernels failed 
> to boot on the Niagara.
> I set up minicom to log to a file and see see if I can catch something 
> before the crashes.

Kernel regressions should be reported to the sparclinux kernel mailing list [1].

Also, a bisect would be helpful to determine which commit broke the kernel.

Thanks,
Adrian

> [1] http://vger.kernel.org/vger-lists.html#sparclinux

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913