On 9/13/18 6:24 PM, Thomas Gleixner wrote:
> On Thu, 13 Sep 2018, Brijesh Singh wrote:
>>
>> +void __weak mem_encrypt_free_decrypted_mem(void) { }
>> +
>> void __ref free_initmem(void)
>> {
>> e820__reallocate_tables();
>>
>> + mem_encrypt_free_decrypted_mem();
>> +
>> free_kernel_image_pages(&__init_begin, &__init_end);
>> }
>>
>> diff --git a/arch/x86/mm/mem_encrypt.c b/arch/x86/mm/mem_encrypt.c
>> index b2de398..718acdf 100644
>> --- a/arch/x86/mm/mem_encrypt.c
>> +++ b/arch/x86/mm/mem_encrypt.c
>> @@ -348,6 +348,16 @@ bool sev_active(void)
>> EXPORT_SYMBOL(sev_active);
>>
>> /* Architecture __weak replacement functions */
>> +void __init mem_encrypt_free_decrypted_mem(void)
>> +{
>> + if (mem_encrypt_active())
>> + return;
> Why?
Hmm, I think we should be able to free the unused memory when memory
encryption is active.
In that case, since memory range was mapped C=0 so we should do
set_memory_encryptyed()
to change from C=0 -> C=1 before freeing it.
>> +
>> + free_init_pages("unused decrypted",
>> + (unsigned long)__start_bss_decrypted_unused,
>> + (unsigned long)__end_bss_decrypted);
> Everything _AFTER_ __start_bss_decrypted_unused _IS_ unused and can be
> freed. No conditional nothing. It's unused in any case.
>
> Thanks,
>
> tglx
>
>
