Re: [PATCH 3/3] mm/vmalloc: fix vmalloc_to_page for huge vmap mappings

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Anshuman Khandual's on June 24, 2019 4:52 pm:
> 
> 
> On 06/23/2019 03:14 PM, Nicholas Piggin wrote:
>> vmalloc_to_page returns NULL for addresses mapped by larger pages[*].
>> Whether or not a vmap is huge depends on the architecture details,
>> alignments, boot options, etc., which the caller can not be expected
>> to know. Therefore HUGE_VMAP is a regression for vmalloc_to_page.
>> 
>> This change teaches vmalloc_to_page about larger pages, and returns
>> the struct page that corresponds to the offset within the large page.
>> This makes the API agnostic to mapping implementation details.
>> 
>> [*] As explained by commit 029c54b095995 ("mm/vmalloc.c: huge-vmap:
>>     fail gracefully on unexpected huge vmap mappings")
>> 
>> Signed-off-by: Nicholas Piggin <npiggin@xxxxxxxxx>
>> ---
>>  include/asm-generic/4level-fixup.h |  1 +
>>  include/asm-generic/5level-fixup.h |  1 +
>>  mm/vmalloc.c                       | 37 +++++++++++++++++++-----------
>>  3 files changed, 26 insertions(+), 13 deletions(-)
>> 
>> diff --git a/include/asm-generic/4level-fixup.h b/include/asm-generic/4level-fixup.h
>> index e3667c9a33a5..3cc65a4dd093 100644
>> --- a/include/asm-generic/4level-fixup.h
>> +++ b/include/asm-generic/4level-fixup.h
>> @@ -20,6 +20,7 @@
>>  #define pud_none(pud)			0
>>  #define pud_bad(pud)			0
>>  #define pud_present(pud)		1
>> +#define pud_large(pud)			0
>>  #define pud_ERROR(pud)			do { } while (0)
>>  #define pud_clear(pud)			pgd_clear(pud)
>>  #define pud_val(pud)			pgd_val(pud)
>> diff --git a/include/asm-generic/5level-fixup.h b/include/asm-generic/5level-fixup.h
>> index bb6cb347018c..c4377db09a4f 100644
>> --- a/include/asm-generic/5level-fixup.h
>> +++ b/include/asm-generic/5level-fixup.h
>> @@ -22,6 +22,7 @@
>>  #define p4d_none(p4d)			0
>>  #define p4d_bad(p4d)			0
>>  #define p4d_present(p4d)		1
>> +#define p4d_large(p4d)			0
>>  #define p4d_ERROR(p4d)			do { } while (0)
>>  #define p4d_clear(p4d)			pgd_clear(p4d)
>>  #define p4d_val(p4d)			pgd_val(p4d)
>> diff --git a/mm/vmalloc.c b/mm/vmalloc.c
>> index 4c9e150e5ad3..4be98f700862 100644
>> --- a/mm/vmalloc.c
>> +++ b/mm/vmalloc.c
>> @@ -36,6 +36,7 @@
>>  #include <linux/rbtree_augmented.h>
>>  
>>  #include <linux/uaccess.h>
>> +#include <asm/pgtable.h>
>>  #include <asm/tlbflush.h>
>>  #include <asm/shmparam.h>
>>  
>> @@ -284,26 +285,36 @@ struct page *vmalloc_to_page(const void *vmalloc_addr)
>>  
>>  	if (pgd_none(*pgd))
>>  		return NULL;
>> +
>>  	p4d = p4d_offset(pgd, addr);
>>  	if (p4d_none(*p4d))
>>  		return NULL;
>> -	pud = pud_offset(p4d, addr);
>> +	if (WARN_ON_ONCE(p4d_bad(*p4d)))
>> +		return NULL;
> 
> The warning here is a required addition but it needs to be moved after p4d_large()
> check. Please see the next comment below.
> 
>> +#ifdef CONFIG_HAVE_ARCH_HUGE_VMAP
>> +	if (p4d_large(*p4d))
>> +		return p4d_page(*p4d) + ((addr & ~P4D_MASK) >> PAGE_SHIFT);
>> +#endif
>>  
>> -	/*
>> -	 * Don't dereference bad PUD or PMD (below) entries. This will also
>> -	 * identify huge mappings, which we may encounter on architectures
>> -	 * that define CONFIG_HAVE_ARCH_HUGE_VMAP=y. Such regions will be
>> -	 * identified as vmalloc addresses by is_vmalloc_addr(), but are
>> -	 * not [unambiguously] associated with a struct page, so there is
>> -	 * no correct value to return for them.
>> -	 */
>> -	WARN_ON_ONCE(pud_bad(*pud));
>> -	if (pud_none(*pud) || pud_bad(*pud))
>> +	pud = pud_offset(p4d, addr);
>> +	if (pud_none(*pud))
>> +		return NULL;
>> +	if (WARN_ON_ONCE(pud_bad(*pud)))
>>  		return NULL;
>> +#ifdef CONFIG_HAVE_ARCH_HUGE_VMAP
>> +	if (pud_large(*pud))
>> +		return pud_page(*pud) + ((addr & ~PUD_MASK) >> PAGE_SHIFT);
>> +#endif
>> +
> 
> pud_bad() on arm64 returns true when the PUD does not point to a next page
> table page implying the fact that it might be a large/huge entry. I am not
> sure if the semantics holds good for other architectures too. But on arm64
> if pud_large() is true, then pud_bad() will be true as well. So pud_bad()
> check must happen after pud_large() check. So the sequence here should be
> 
> 1. pud_none()	--> Nothing is in here, return NULL
> 2. pud_large()	--> Return offset page address from the huge page mapping
> 3. pud_bad()	--> Return NULL as there is no more page table level left
> 
> Checking pud_bad() first can return NULL for a valid huge mapping.
> 
>>  	pmd = pmd_offset(pud, addr);
>> -	WARN_ON_ONCE(pmd_bad(*pmd));
>> -	if (pmd_none(*pmd) || pmd_bad(*pmd))
>> +	if (pmd_none(*pmd))
>> +		return NULL;
>> +	if (WARN_ON_ONCE(pmd_bad(*pmd)))
>>  		return NULL;
>> +#ifdef CONFIG_HAVE_ARCH_HUGE_VMAP
>> +	if (pmd_large(*pmd))
>> +		return pmd_page(*pmd) + ((addr & ~PMD_MASK) >> PAGE_SHIFT);
>> +#endif
> 
> Ditto.
> 
> I see that your previous proposal had this right which got changed in this
> manner after my comments. Sorry about it.
> 
> It was recently when I learned (correctly) that expected semantics of pxx_bad()
> is that - It does not point to the next page table page.  Hence I wonder why is
> this not renamed as pxx_table() instead to make it absolutely clear.
> 

Okay, I'll change it and resend. It worked okay on powerpc but it
looks like the usual precedent is testing for large before bad so we
will have to go with that.

Thanks,
Nick





[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux