Hi Will,
On 2018/1/23 18:04, Will Deacon wrote:
> On Tue, Jan 23, 2018 at 04:28:45PM +0800, Yisheng Xie wrote:
>> On 2017/12/6 20:35, Will Deacon wrote:
>>> +#ifdef CONFIG_UNMAP_KERNEL_AT_EL0
>>> +static int __init map_entry_trampoline(void)
>>> +{
>>> + extern char __entry_tramp_text_start[];
>>> +
>>> + pgprot_t prot = rodata_enabled ? PAGE_KERNEL_ROX : PAGE_KERNEL_EXEC;
>>> + phys_addr_t pa_start = __pa_symbol(__entry_tramp_text_start);
>>> +
>>> + /* The trampoline is always mapped and can therefore be global */
>>> + pgprot_val(prot) &= ~PTE_NG;
>>> +
>>> + /* Map only the text into the trampoline page table */
>>> + memset(tramp_pg_dir, 0, PGD_SIZE);
>>> + __create_pgd_mapping(tramp_pg_dir, pa_start, TRAMP_VALIAS, PAGE_SIZE,
>>> + prot, pgd_pgtable_alloc, 0);
>>
>> How the tramp_pg_dir is used, should it be set to ttbr1 when exit kernel?
>> Sorry
>> for I do not find where it is used.
>
> Yes, that's what happens when we return to userspace. The code is a little
> convoluted, but the tramp_pg_dir is placed at a fixed offset from swapper
> (see the linker script) so the sub instruction in tramp_unmap_kernel is what
> gives us the ttbr1 value we need.
oh, I missed that. Maybe a comment inline is better to understand. Thanks once
more for your help and explain :)
Thanks
Yisheng
>
> Will
>
> .
>
