Re: [PATCH v2] mm: disable kernelcore=mirror when no mirror memory

[Kefeng Wang <wangkefeng.wang@huawei.com>]

On 2023/8/4 1:53, Andrew Morton wrote:
> On Wed, 2 Aug 2023 15:23:28 +0800 Wupeng Ma <mawupeng1@huawei.com> wrote:
> 
>> From: Ma Wupeng <mawupeng1@huawei.com>
>>
>> For system with kernelcore=mirror enabled while no mirrored memory is
>> reported by efi. This could lead to kernel OOM during startup since
>> all memory beside zone DMA are in the movable zone and this prevents
>> the kernel to use it.
>>
>> Zone DMA/DMA32 initialization is independent of mirrored memory and
>> their max pfn is set in zone_sizes_init(). Since kernel can fallback
>> to zone DMA/DMA32 if there is no memory in zone Normal, these zones
>> are seen as mirrored memory no mather their memory attributes are.
>>
>> To solve this problem, disable kernelcore=mirror when there is no real
>> mirrored memory exists.
>>
>> ...
>>
>> diff --git a/mm/internal.h b/mm/internal.h
>> index a7d9e980429a..b8aaf9cac87e 100644
>> --- a/mm/internal.h
>> +++ b/mm/internal.h
>> @@ -1005,6 +1005,7 @@ static inline bool gup_must_unshare(struct vm_area_struct *vma,
>>   }
>>   
>>   extern bool mirrored_kernelcore;
>> +extern bool memblock_has_mirror(void);
>>   
>>   static inline bool vma_soft_dirty_enabled(struct vm_area_struct *vma)
>>   {
>> diff --git a/mm/memblock.c b/mm/memblock.c
>> index f9e61e565a53..913b2520a9a0 100644
>> --- a/mm/memblock.c
>> +++ b/mm/memblock.c
>> @@ -161,6 +161,11 @@ static int memblock_can_resize __initdata_memblock;
>>   static int memblock_memory_in_slab __initdata_memblock;
>>   static int memblock_reserved_in_slab __initdata_memblock;
>>   
>> +bool __init_memblock memblock_has_mirror(void)
>> +{
>> +	return system_has_some_mirror;
>> +}
>> +
>>   static enum memblock_flags __init_memblock choose_memblock_flags(void)
>>   {
>>   	return system_has_some_mirror ? MEMBLOCK_MIRROR : MEMBLOCK_NONE;
>> diff --git a/mm/mm_init.c b/mm/mm_init.c
>> index a1963c3322af..e8af1b628430 100644
>> --- a/mm/mm_init.c
>> +++ b/mm/mm_init.c
>> @@ -377,6 +377,11 @@ static void __init find_zone_movable_pfns_for_nodes(void)
>>   	if (mirrored_kernelcore) {
>>   		bool mem_below_4gb_not_mirrored = false;
>>   
>> +		if (!memblock_has_mirror()) {
>> +			pr_warn("The system has no mirror memory, ignore kernelcore=mirror.\n");
>> +			goto out;
>> +		}
>> +
>>   		for_each_mem_region(r) {
>>   			if (memblock_is_mirror(r))
>>   				continue;
> 
> I've already queued Levi's "mm/mm_init: ignore kernelcore=mirror boot
> option when no mirror memory presents."
> (https://lkml.kernel.org/r/20230802183614.15520-1-ppbuk5246@gmail.com)
> 
> The main difference appears to be that Levi's patch zeroes out
> zone_movable_pfn.
> 
> Please let's check each other's work and decide how to proceed?

We already know whether system has mirror memory or not, so no need 
recheck it again and no need to touch zone_movable_pfn.