On 2021/12/16 9:10, Baoquan He wrote:
> On 12/15/21 at 02:28pm, Borislav Petkov wrote:
>> On Fri, Dec 10, 2021 at 02:55:26PM +0800, Zhen Lei wrote:
>>> @@ -518,7 +519,7 @@ static void __init reserve_crashkernel(void)
>>> }
>>> }
>>>
>>> - if (crash_base >= (1ULL << 32) && reserve_crashkernel_low()) {
>>> + if (crash_base >= CRASH_ADDR_LOW_MAX && reserve_crashkernel_low()) {
>>> memblock_phys_free(crash_base, crash_size);
>>> return;
>>> }
>>
>> That's not a equivalent transformation on X86_32.
The original value (1ULL << 32) is inaccurate, and it enlarged the CRASH_ADDR_LOW
upper limit. This is because when the memory is allocated from the low end,
the address cannot exceed CRASH_ADDR_LOW_MAX, see "if (!high)" branch. If
the memory is allocated from the high end, 'crash_base' is greater than or
equal to (1ULL << 32), and naturally, it is greater than CRASH_ADDR_LOW_MAX.
I think I should update the description, thanks.
if (!high)
crash_base = memblock_phys_alloc_range(crash_size,
CRASH_ALIGN, CRASH_ALIGN,
CRASH_ADDR_LOW_MAX);
if (!crash_base)
crash_base = memblock_phys_alloc_range(crash_size,
CRASH_ALIGN, CRASH_ALIGN,
CRASH_ADDR_HIGH_MAX);
>
> reserve_crashkernel_low() always return 0 on x86_32, so the not equivalent
> transformation for x86_32 doesn't matter, I think.
>
> .
>
