2015-02-03 12:25:59

by Alexander Kuleshov

[permalink] [raw]
Subject: [x86_64] Question about early page tables initialization

Hello All,

I have a question about page tables initialization in the
arch/x86/boot/compressed/head_64.S

After we clear memory for page tables, there is code which
build PML4:

leal pgtable + 0(%ebx), %edi
leal 0x1007(%edi), %eax
movl %eax, 0(%edi)

Why there is offset 0x1007 instead just 0x7? 0x1007 is
4k + 7bit (PML4E) flags as i understand correctly. But
why we skip first 4k here?

Thank you.


2015-02-03 13:28:11

by Oren Twaig

[permalink] [raw]
Subject: Re: [x86_64] Question about early page tables initialization

Hi,

This is the corresponding C code which can help you understand:

u64 *pml4 = (u64*)pgtable;
u64 pdp = pgtable + 0x1000;
u64 pml4_entry = pdp | PTE_P | PTE_W | PTU; // present, write, userspace = 0x7
pml4[0] = pml4_entry;

The 0x1007 you see is just the calculation of the pml4_entry.

Oren Twaig.

On 02/03/2015 02:25 PM, Alex Kuleshov wrote:
> Hello All,
>
> I have a question about page tables initialization in the
> arch/x86/boot/compressed/head_64.S
>
> After we clear memory for page tables, there is code which
> build PML4:
>
> leal pgtable + 0(%ebx), %edi
> leal 0x1007(%edi), %eax
> movl %eax, 0(%edi)
>
> Why there is offset 0x1007 instead just 0x7? 0x1007 is
> 4k + 7bit (PML4E) flags as i understand correctly. But
> why we skip first 4k here?
>
> Thank you.
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>

2015-02-03 13:00:42

by Alexander Kuleshov

[permalink] [raw]
Subject: Re: [x86_64] Question about early page tables initialization

Hello Oren,

Ah yes, absolutely right. I thought almost the same, just overlooked
that pdp is right after pml4 in memory.

Oren thank you for help.

2015-02-03 18:42 GMT+06:00 Oren Twaig <[email protected]>:
> Hi,
>
> This is the corresponding C code which can help you understand:
>
> u64 *pml4 = (u64*)pgtable;
> u64 pdp = pgtable + 0x1000;
> u64 pml4_entry = pdp | PTE_P | PTE_W | PTU; // present, write, userspace = 0x7
> pml4[0] = pml4_entry;
>
> The 0x1007 you see is just the calculation of the pml4_entry.
>
> Oren Twaig.
>
> On 02/03/2015 02:25 PM, Alex Kuleshov wrote:
>> Hello All,
>>
>> I have a question about page tables initialization in the
>> arch/x86/boot/compressed/head_64.S
>>
>> After we clear memory for page tables, there is code which
>> build PML4:
>>
>> leal pgtable + 0(%ebx), %edi
>> leal 0x1007(%edi), %eax
>> movl %eax, 0(%edi)
>>
>> Why there is offset 0x1007 instead just 0x7? 0x1007 is
>> 4k + 7bit (PML4E) flags as i understand correctly. But
>> why we skip first 4k here?
>>
>> Thank you.
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>> the body of a message to [email protected]
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at http://www.tux.org/lkml/
>>
>
>



--
_________________________
0xAX