Re: [PATCH] x86/boot: minor improvement in kaslr

[Cao jin]

On 2/1/19 4:20 PM, Kees Cook wrote:
> On Fri, Feb 1, 2019 at 6:51 PM Cao jin <caoj.fnst@xxxxxxxxxxxxxx> wrote:
>>
>> comments fix: input_size is ZO image size which just don't count .bss
>> in, but has .text, .data, etc;
>> drop unecessary alignment: minimum is either 512M or output, both are
>> CONFIG_PHYSICAL_ALIGN aligned(output is aligned in head_32/64.S). But
>> mention it in earlier comments.
>>
>> Signed-off-by: Cao jin <caoj.fnst@xxxxxxxxxxxxxx>
>> ---
>>  arch/x86/boot/compressed/kaslr.c | 9 +++------
>>  1 file changed, 3 insertions(+), 6 deletions(-)
>>
>> diff --git a/arch/x86/boot/compressed/kaslr.c b/arch/x86/boot/compressed/kaslr.c
>> index 9ed9709d9947..a947c5aba34e 100644
>> --- a/arch/x86/boot/compressed/kaslr.c
>> +++ b/arch/x86/boot/compressed/kaslr.c
>> @@ -360,7 +360,7 @@ static void handle_mem_options(void)
>>   * (i.e. it does not include its run size). This range must be avoided
>>   * because it contains the data used for decompression.
>>   *
>> - * [input+input_size, output+init_size) is [_text, _end) for ZO. This
>> + * [input+input_size, output+init_size) is [_bss, _end) for ZO. This
> 
> This isn't right. The comment was correct before. See
> arch/x86/boot/compressed/vmlinux.lds.S for the layout of the ZO image:
> after the compressed image is _text, _rodata, _got, _data, _bss,
> _pgtable, and _end. "[_text, _end)" correctly identifies the span
> used.
> 

I am confused, doesn't input_size = ZO image size = .head.text +
.rodata..compressed + .rodata + .got + .data + .pgtable ? As I know,
.bss don't occupy any space in file, and ZO's full run size is against
the end of buffer, so I think the tiny gap here is just .bss, which is
also the stack and heap. Do I get it wrong?

>>   * range includes ZO's heap and stack, and must be avoided since it
>>   * performs the decompression.
>>   *
>> @@ -763,9 +763,6 @@ static unsigned long find_random_phys_addr(unsigned long minimum,
>>                 return 0;
>>         }
>>
>> -       /* Make sure minimum is aligned. */
>> -       minimum = ALIGN(minimum, CONFIG_PHYSICAL_ALIGN);
>> -
> 
> I would prefer to keep this runtime calculation since it enforces the
> requirement instead of making leaving it in a comment. When this goes
> wrong, you get an unbootable kernel, which is very frustrating to
> debug.
> 

I find that I maybe wrong here. It is said that CONFIG_PHYSICAL_ALIGN
must be a multiple of 0x200000 on 64-bit, so it could be 2M, 4M, 6M, 8M,
12M, 14M, 16M, while 512M can't be divided by 6, 10, 12, 14 without
remainder.

-- 
Sincerely,
Cao jin