linux-kernel - Re: [PATCH] arm64/hugetlb: Fix building errors in huge_ptep_clear

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <ded8ff72-3938-15ce-577b-42ea2daa3c9c@arm.com>
Date:   Fri, 27 May 2022 14:25:54 +0530
From:   Anshuman Khandual <anshuman.khandual@....com>
To:     Baolin Wang <baolin.wang@...ux.alibaba.com>,
        Gavin Shan <gshan@...hat.com>, torvalds@...ux-foundation.org
Cc:     catalin.marinas@....com, will@...nel.org,
        akpm@...ux-foundation.org, naresh.kamboju@...aro.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] arm64/hugetlb: Fix building errors in
 huge_ptep_clear_flush()



On 5/27/22 14:12, Baolin Wang wrote:
> 
> 
> On 5/27/2022 3:12 PM, Gavin Shan wrote:
>> Hi Baolin,
>>
>> On 5/27/22 12:51 PM, Baolin Wang wrote:
>>> Fix below building errors which was caused by commit ae07562909f3
>>> ("mm: change huge_ptep_clear_flush() to return the original pte")
>>> interacting with commit fb396bb459c1 ("arm64/hugetlb: Drop TLB flush
>>> from get_clear_flush()").
>>>
>>> Due to the new get_clear_contig() has dropped TLB flush, we should
>>> add an explicit TLB flush in huge_ptep_clear_flush() to keep original
>>> semantics when changing to use new get_clear_contig().
>>>
>>> "
>>> arch/arm64/mm/hugetlbpage.c: In function ‘huge_ptep_clear_flush’:
>>> arch/arm64/mm/hugetlbpage.c:515:9: error: implicit declaration of
>>> function ‘get_clear_flush’; did you mean ‘ptep_clear_flush’?
>>> [-Werror=implicit-function-declaration]
>>>    515 |  return get_clear_flush(vma->vm_mm, addr, ptep, pgsize, ncontig);
>>>        |         ^~~~~~~~~~~~~~~
>>>        |         ptep_clear_flush
>>> "
>>>
>>> Reported-by: Linux Kernel Functional Testing <lkft@...aro.org>
>>> Suggested-by: Catalin Marinas <catalin.marinas@....com>
>>> Signed-off-by: Baolin Wang <baolin.wang@...ux.alibaba.com>
>>> Cc: Catalin Marinas <catalin.marinas@....com>
>>> Cc: Anshuman Khandual <anshuman.khandual@....com>
>>> Cc: Andrew Morton <akpm@...ux-foundation.org>
>>> ---
>>>   arch/arm64/mm/hugetlbpage.c | 5 ++++-
>>>   1 file changed, 4 insertions(+), 1 deletion(-)
>>>
>>
>> I ran to the compiling failure either and it would be caused by
>> commit fb396bb459c1 ("arm64/hugetlb: Drop TLB flush from get_clear_flush()").
>> It's worthy to have a "Fixes" tag. With those fixed:
> 
> Thanks for reminding. IMHO, better to add 2 related commits' fix tag. Linus, could you help to add them when applying this patch, or need a resend? Thanks.
> 
> Fixes: ae07562909f3 ("mm: change huge_ptep_clear_flush() to return the original pte")
> Fixes: fb396bb459c1 ("arm64/hugetlb: Drop TLB flush from get_clear_flush()").

Although this merge conflict happened in flight, the earlier commit ae07562909f3
came afterwards by which time get_clear_contig() has been added without required
TLB flush. Hence seems like 'Fixes:' tag applies to the earlier commit itself.

Nonetheless LGTM.

Reviewed-by: Anshuman Khandual <anshuman.khandual@....com>

> 
>>
>> Reviewed-by: Gavin Shan <gshan@...hat.com>
>>
>>
>>> diff --git a/arch/arm64/mm/hugetlbpage.c b/arch/arm64/mm/hugetlbpage.c
>>> index 0f0c17dfeb9c..e2a5ec9fdc0d 100644
>>> --- a/arch/arm64/mm/hugetlbpage.c
>>> +++ b/arch/arm64/mm/hugetlbpage.c
>>> @@ -507,12 +507,15 @@ pte_t huge_ptep_clear_flush(struct vm_area_struct *vma,
>>>   {
>>>       size_t pgsize;
>>>       int ncontig;
>>> +    pte_t orig_pte;
>>>       if (!pte_cont(READ_ONCE(*ptep)))
>>>           return ptep_clear_flush(vma, addr, ptep);
>>>       ncontig = find_num_contig(vma->vm_mm, addr, ptep, &pgsize);
>>> -    return get_clear_flush(vma->vm_mm, addr, ptep, pgsize, ncontig);
>>> +    orig_pte = get_clear_contig(vma->vm_mm, addr, ptep, pgsize, ncontig);
>>> +    flush_tlb_range(vma, addr, addr + pgsize * ncontig);
>>> +    return orig_pte;
>>>   }
>>>   static int __init hugetlbpage_init(void)
>>>
>>
>> Thanks,
>> Gavin