Re: [Bochs-developers] IOAPIC - vector 0x31 stuck?

[Sebastian H. <he...@gm...>]

Sebastian Herbszt wrote:
> Stanislav wrote:
>> I fixed the I/O APIC reset issue in CVS but this is not enough.
>> Now, after second reset you enter SMM and never exit ...
>> Looks like perfect BIOS issue, you should be an expert !
> 
> This problem is not related to the IO-APIC fix.
> On a reset we do enter smm_init() in rombios32.c, but since
> APMC_EN is still set we exit and not enter SMM. This is
> somewhat described in my post "CPU reset and SMBASE",
> but i will send a new email with a proper description. I think
> DEVACTB should get reset on bx_acpi_ctrl_c::reset().

Just to clarify this:

Bochs start
...
00000000000i[SYS  ] bx_pc_system_c::Reset(HARDWARE) called
00000000000i[CPU0 ] cpu hardware reset
...
00001492788i[BIOS ] smm_init
00001493236i[BIOS ] smm_init for real
00001493273i[PCI  ] setting SMRAM control register to 0x4a
00001657421i[CPU0 ] Enter to System Management Mode
...
Linux start
...
00272971621i[CPU0 ] Enter to System Management Mode
00272971631i[CPU0 ] RSM: Resuming from System Management Mode
...
00335940000i[WGUI ] system RESET callback
00335940000i[SYS  ] bx_pc_system_c::Reset(HARDWARE) called
00335940000i[CPU0 ] cpu hardware reset
...
00337442748i[BIOS ] smm_init
...
Linux start
...
00624897299i[CPU0 ] Enter to System Management Mode

and there is no return from SMM.
Since there is no "smm_init for real" after the reset SMBASE is
not relocated and when SMM is entered code is run from
0x38000 and this is not smm_code from rombios32start.S.

- Sebastian