[Qemu-devel] Re: [Bochs-developers] [PATCH] Read additional ACPI tables from a VM

[Gleb Natapov <gleb@redhat.com>]

On Mon, Feb 09, 2009 at 09:41:07PM +0100, Sebastian Herbszt wrote:
> Gleb Natapov wrote:
>
> Just a quick review.
>
>> diff --git a/bios/rombios32.c b/bios/rombios32.c
>> index f6ce225..29fd40a 100644
>> --- a/bios/rombios32.c
>> +++ b/bios/rombios32.c
>> @@ -455,6 +455,8 @@ unsigned long bios_table_end_addr;
>> #define QEMU_CFG_SIGNATURE  0x00
>> #define QEMU_CFG_ID         0x01
>> #define QEMU_CFG_UUID       0x02
>> +#define FW_CFG_ARCH_LOCAL     0x8000
>> +#define FW_CFG_ACPI_TABLES  (FW_CFG_ARCH_LOCAL + 0)
>
> No more QEMU_CFG prefix? Or is this different?
>
It is not. I'll change it to QEMU_CFG.

>> +
>> +static int acpi_load_table(int i, uint32_t addr, uint16_t *len)
>> +{
>> +    qemu_cfg_read((uint8_t*)len, sizeof(*len));
>> +
>> +    if (!*len)
>> +        return -1;
>> +
>> +    qemu_cfg_read((uint8_t*)addr, *len);
>> +    return 0;
>> +}
>> #endif
>>
>> void init_smp_msrs(void)
>
> Can you please diff against bochs cvs?
>
Will do. This patch was initially intended to go to qemu patch series so I
did it against patched bios source. And unpatched bochs bios does not
boot linux with qemu. IRQ routing problem or something.

>> +        rsdt->table_offset_entry[i+4] = cpu_to_le32(addr);
>> +        addr += len;
>> +    }
>> +#endif
>> +
>> +    /* RSDT */
>> +    memset(rsdt, 0, sizeof(*rsdt));
>
> This overwrites the "rsdt->table_offset_entry[i+4] = cpu_to_le32(addr);" assignment from above?
No since sizeof(*rstd) is less then that. But I'll move memset up, it
will be less confusing.

--
			Gleb.