Re: CVS commit: src/sys/arch/amd64

[manu%netbsd.org@localhost (Emmanuel Dreyfus)]

Masanobu SAITOH <msaitoh%execsw.org@localhost> wrote:

>  I have a UEFI boot machine and it also doesn't boot well.
> 
>  - It hangs after attaching ioapic0, cpu0 or acpi0 (or something else).
>    The possibility is about 65%
>  - It sometimes panic in acpi_attach(), acpimcfg_probe or something else.
>    The possibility is about 10%
>  - It sometimes boot up.
>    The possibility is about 25%

I suspect this is a second bug that was undercovered by the multiboot
change. I get crashes like the one you report 100% reproductible with
qemu UEFI boot. I described the thing here:
http://mail-index.netbsd.org/tech-kern/2020/01/02/msg025911.html

Could you check with ddb the physical address accessed? Here is the
relevant excerpt in the message I posted:

db{0}> x/i $rip
netbsd:kmem_intr_alloc+0x64:    movq    %r12,0(%rax)
db{0}> print $rax
ffff920578600008
db{0}> call vtophys(ffff920578600008)
1000008

If you can add a #define DEBUG_MEMMAP 1 at the beginning of
src/sys/arch/x86/x86/efi.c and x86_machdep.c you will also have the
memory map provided by UEFI.

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu%netbsd.org@localhost