[PATCH v1 1/1] riscv: mm: add missing memcpy in kasan_init

[Palmer Dabbelt]

On Sun, 09 Oct 2022 23:49:36 PDT (-0700), atishp at atishpatra.org wrote:
> On Sun, Oct 9, 2022 at 1:31 AM <panqinglin2020 at iscas.ac.cn> wrote:
>>
>> From: Qinglin Pan <panqinglin2020 at iscas.ac.cn>
>>
>> Hi Atish,
>>
>> It seems that the panic is due to the missing memcpy during kasan_init.
>> Could you please check whether this patch is helpful?
>>
>> When doing kasan_populate, the new allocated base_pud/base_p4d should
>> contain kasan_early_shadow_{pud, p4d}'s content. Add the missing memcpy
>> to avoid page fault when read/write kasan shadow region.
>>
>> Tested on:
>>  - qemu with sv57 and CONFIG_KASAN on.
>>  - qemu with sv48 and CONFIG_KASAN on.
>>
>> Signed-off-by: Qinglin Pan <panqinglin2020 at iscas.ac.cn>
>> ---
>>  arch/riscv/mm/kasan_init.c | 7 ++++++-
>>  1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/riscv/mm/kasan_init.c b/arch/riscv/mm/kasan_init.c
>> index a22e418dbd82..e1226709490f 100644
>> --- a/arch/riscv/mm/kasan_init.c
>> +++ b/arch/riscv/mm/kasan_init.c
>> @@ -113,6 +113,8 @@ static void __init kasan_populate_pud(pgd_t *pgd,
>>                 base_pud = pt_ops.get_pud_virt(pfn_to_phys(_pgd_pfn(*pgd)));
>>         } else if (pgd_none(*pgd)) {
>>                 base_pud = memblock_alloc(PTRS_PER_PUD * sizeof(pud_t), PAGE_SIZE);
>> +               memcpy(base_pud, (void *)kasan_early_shadow_pud,
>> +                       sizeof(pud_t) * PTRS_PER_PUD);
>>         } else {
>>                 base_pud = (pud_t *)pgd_page_vaddr(*pgd);
>>                 if (base_pud == lm_alias(kasan_early_shadow_pud)) {
>> @@ -173,8 +175,11 @@ static void __init kasan_populate_p4d(pgd_t *pgd,
>>                 base_p4d = pt_ops.get_p4d_virt(pfn_to_phys(_pgd_pfn(*pgd)));
>>         } else {
>>                 base_p4d = (p4d_t *)pgd_page_vaddr(*pgd);
>> -               if (base_p4d == lm_alias(kasan_early_shadow_p4d))
>> +               if (base_p4d == lm_alias(kasan_early_shadow_p4d)) {
>>                         base_p4d = memblock_alloc(PTRS_PER_PUD * sizeof(p4d_t), PAGE_SIZE);
>> +                       memcpy(base_p4d, (void *)kasan_early_shadow_p4d,
>> +                               sizeof(p4d_t) * PTRS_PER_P4D);
>> +               }
>>         }
>>
>>         p4dp = base_p4d + p4d_index(vaddr);
>> --
>> 2.35.1
>>
>
> Yes. This patch fixes the boot issue for me with Kasan enabled on v6.0.
>
> Tested-by: Atish Patra <atishp at rivosinc.com>
>
> Thanks for the patch. Few nit comments:
>
> You can drop the message addressed to me in the commit text.
> Usually, that should be after the last sign off between two  "---"
>
> As conor suggested, there should be a Fixes tag[1] in the commit text.
>
> [1] https://www.kernel.org/doc/html/v4.10/process/submitting-patches.html#:~:text=A%20Fixes%3A%20tag%20indicates%20that,versions%20should%20receive%20your%20fix.

I've got this on not-for-next in a cleaned up state, so no reason to 
resend just for these.  It also fixes my boot issues, but it's not quite 
passing the smell test right now.  I've looked at this a few times and 
every time I do I manage to convince myself that there's a bunch of 
issues in these kasan initialization routines, but my attempts to clean 
them up just result in more breakages.  So I think there's something I 
don't quite get about this yet.

I'll try and find some time this weekend to dig into this further, 
hopefully it's just all the merge window interrupts that are preventing 
me from getting anywhere.