Re: [RFC V2] mm: Enable generic pfn_valid() to handle early sections with memmap holes

[Mike Rapoport]

On Tue, May 25, 2021 at 11:30:15AM +0530, Anshuman Khandual wrote:
> 
> On 5/24/21 12:22 PM, Mike Rapoport wrote:
> > Hello Anshuman,
> > 
> > On Mon, May 24, 2021 at 10:28:32AM +0530, Anshuman Khandual wrote:
> >>
> >> On 4/22/21 1:20 PM, Anshuman Khandual wrote:
> >>> Platforms like arm and arm64 have redefined pfn_valid() because their early
> >>> memory sections might have contained memmap holes after freeing parts of it
> >>> during boot, which should be skipped while validating a pfn for struct page
> >>> backing. This scenario on certain platforms where memmap is not continuous,
> >>> could be captured with a new option CONFIG_HAVE_EARLY_SECTION_MEMMAP_HOLES.
> >>> Then the generic pfn_valid() can be improved to accommodate such platforms.
> >>> This reduces overall code footprint and also improves maintainability.
> >>>
> >>> free_unused_memmap() and pfn_to_online_page() have been updated to include
> >>> such cases. This also exports memblock_is_memory() for all drivers that use
> >>> pfn_valid() but lack required visibility. After the new config is in place,
> >>> drop CONFIG_HAVE_ARCH_PFN_VALID from arm64 platforms.
> >>>
> >>> Cc: Catalin Marinas <catalin.marinas@xxxxxxx>
> >>> Cc: Will Deacon <will@xxxxxxxxxx>
> >>> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> >>> Cc: Mike Rapoport <rppt@xxxxxxxxxx>
> >>> Cc: David Hildenbrand <david@xxxxxxxxxx>
> >>> Cc: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
> >>> Cc: linux-kernel@xxxxxxxxxxxxxxx
> >>> Cc: linux-mm@xxxxxxxxx
> >>> Suggested-by: David Hildenbrand <david@xxxxxxxxxx>
> >>> Signed-off-by: Anshuman Khandual <anshuman.khandual@xxxxxxx>
> >>> ---
> >>> This patch applies on the latest mainline kernel after Mike's series
> >>> regarding arm64 based pfn_valid().
> >>>
> >>> https://lore.kernel.org/linux-mm/20210422061902.21614-1-rppt@xxxxxxxxxx/T/#t
> >>>
> >>> Changes in RFC V2:
> >>>
> >>> - Dropped support for arm (32 bit)
> >>> - Replaced memblock_is_map_memory() check with memblock_is_memory()
> >>> - MEMBLOCK_NOMAP memory are no longer skipped for pfn_valid()
> >>> - Updated pfn_to_online_page() per David
> >>> - Updated free_unused_memmap() to preserve existing semantics per Mike
> >>> - Exported memblock_is_memory() instead of memblock_is_map_memory()
> >>>
> >>> Changes in RFC V1:
> >>>
> >>> - https://patchwork.kernel.org/project/linux-mm/patch/1615174073-10520-1-git-send-email-anshuman.khandual@xxxxxxx/
> >>>
> >>>  arch/arm64/Kconfig            |  2 +-
> >>>  arch/arm64/include/asm/page.h |  1 -
> >>>  arch/arm64/mm/init.c          | 41 -----------------------------------
> >>>  include/linux/mmzone.h        | 18 ++++++++++++++-
> >>>  mm/Kconfig                    |  9 ++++++++
> >>>  mm/memblock.c                 |  8 +++++--
> >>>  mm/memory_hotplug.c           |  5 +++++
> >>>  7 files changed, 38 insertions(+), 46 deletions(-)
> >>>
> >>> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> >>> index b4a9b493ce72..4cdc3570ffa9 100644
> >>> --- a/arch/arm64/Kconfig
> >>> +++ b/arch/arm64/Kconfig
> >>> @@ -144,7 +144,6 @@ config ARM64
> >>>  	select HAVE_ARCH_KGDB
> >>>  	select HAVE_ARCH_MMAP_RND_BITS
> >>>  	select HAVE_ARCH_MMAP_RND_COMPAT_BITS if COMPAT
> >>> -	select HAVE_ARCH_PFN_VALID
> >>>  	select HAVE_ARCH_PREL32_RELOCATIONS
> >>>  	select HAVE_ARCH_SECCOMP_FILTER
> >>>  	select HAVE_ARCH_STACKLEAK
> >>> @@ -167,6 +166,7 @@ config ARM64
> >>>  		if $(cc-option,-fpatchable-function-entry=2)
> >>>  	select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY \
> >>>  		if DYNAMIC_FTRACE_WITH_REGS
> >>> +	select HAVE_EARLY_SECTION_MEMMAP_HOLES
> >>>  	select HAVE_EFFICIENT_UNALIGNED_ACCESS
> >>>  	select HAVE_FAST_GUP
> >>>  	select HAVE_FTRACE_MCOUNT_RECORD
> >>> diff --git a/arch/arm64/include/asm/page.h b/arch/arm64/include/asm/page.h
> >>> index 75ddfe671393..fcbef3eec4b2 100644
> >>> --- a/arch/arm64/include/asm/page.h
> >>> +++ b/arch/arm64/include/asm/page.h
> >>> @@ -37,7 +37,6 @@ void copy_highpage(struct page *to, struct page *from);
> >>>  
> >>>  typedef struct page *pgtable_t;
> >>>  
> >>> -int pfn_valid(unsigned long pfn);
> >>>  int pfn_is_map_memory(unsigned long pfn);
> >>>  
> >>>  #include <asm/memory.h>
> >>> diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> >>> index f431b38d0837..5731a11550d8 100644
> >>> --- a/arch/arm64/mm/init.c
> >>> +++ b/arch/arm64/mm/init.c
> >>> @@ -217,47 +217,6 @@ static void __init zone_sizes_init(unsigned long min, unsigned long max)
> >>>  	free_area_init(max_zone_pfns);
> >>>  }
> >>>  
> >>> -int pfn_valid(unsigned long pfn)
> >>> -{
> >>> -	phys_addr_t addr = PFN_PHYS(pfn);
> >>> -
> >>> -	/*
> >>> -	 * Ensure the upper PAGE_SHIFT bits are clear in the
> >>> -	 * pfn. Else it might lead to false positives when
> >>> -	 * some of the upper bits are set, but the lower bits
> >>> -	 * match a valid pfn.
> >>> -	 */
> >>> -	if (PHYS_PFN(addr) != pfn)
> >>> -		return 0;
> >>> -
> >>> -#ifdef CONFIG_SPARSEMEM
> >>> -{
> >>> -	struct mem_section *ms;
> >>> -
> >>> -	if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
> >>> -		return 0;
> >>> -
> >>> -	ms = __pfn_to_section(pfn);
> >>> -	if (!valid_section(ms))
> >>> -		return 0;
> >>> -
> >>> -	/*
> >>> -	 * ZONE_DEVICE memory does not have the memblock entries.
> >>> -	 * memblock_is_memory() check for ZONE_DEVICE based
> >>> -	 * addresses will always fail. Even the normal hotplugged
> >>> -	 * memory will never have MEMBLOCK_NOMAP flag set in their
> >>> -	 * memblock entries. Skip memblock search for all non early
> >>> -	 * memory sections covering all of hotplug memory including
> >>> -	 * both normal and ZONE_DEVICE based.
> >>> -	 */
> >>> -	if (!early_section(ms))
> >>> -		return pfn_section_valid(ms, pfn);
> >>> -}
> >>> -#endif
> >>> -	return memblock_is_memory(addr);
> >>> -}
> >>> -EXPORT_SYMBOL(pfn_valid);
> >>> -
> >>>  int pfn_is_map_memory(unsigned long pfn)
> >>>  {
> >>>  	phys_addr_t addr = PFN_PHYS(pfn);
> >>> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> >>> index 961f0eeefb62..18bf71665211 100644
> >>> --- a/include/linux/mmzone.h
> >>> +++ b/include/linux/mmzone.h
> >>> @@ -1421,10 +1421,22 @@ static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
> >>>   *
> >>>   * Return: 1 for PFNs that have memory map entries and 0 otherwise
> >>>   */
> >>> +bool memblock_is_memory(phys_addr_t addr);
> >>> +
> >>>  static inline int pfn_valid(unsigned long pfn)
> >>>  {
> >>> +	phys_addr_t addr = PFN_PHYS(pfn);
> >>>  	struct mem_section *ms;
> >>>  
> >>> +	/*
> >>> +	 * Ensure the upper PAGE_SHIFT bits are clear in the
> >>> +	 * pfn. Else it might lead to false positives when
> >>> +	 * some of the upper bits are set, but the lower bits
> >>> +	 * match a valid pfn.
> >>> +	 */
> >>> +	if (PHYS_PFN(addr) != pfn)
> >>> +		return 0;
> >>> +
> >>>  	if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
> >>>  		return 0;
> >>>  	ms = __nr_to_section(pfn_to_section_nr(pfn));
> >>> @@ -1434,7 +1446,11 @@ static inline int pfn_valid(unsigned long pfn)
> >>>  	 * Traditionally early sections always returned pfn_valid() for
> >>>  	 * the entire section-sized span.
> >>>  	 */
> >>> -	return early_section(ms) || pfn_section_valid(ms, pfn);
> >>> +	if (early_section(ms))
> >>> +		return IS_ENABLED(CONFIG_HAVE_EARLY_SECTION_MEMMAP_HOLES) ?
> >>> +			memblock_is_memory(pfn << PAGE_SHIFT) : 1;
> >>> +
> >>> +	return pfn_section_valid(ms, pfn);
> >>>  }
> >>>  #endif
> >>
> >> Hello David/Mike,
> >>
> >> Now that pfn_is_map_memory() usage has been decoupled from pfn_valid() and
> >> SPARSEMEM_VMEMMAP is only available memory model on arm64, wondering if we
> >> still need this HAVE_EARLY_SECTION_MEMMAP_HOLES proposal ? Please do kindly
> >> suggest. Thank you.
> > 
> > Even now arm64 still frees parts of the memory map and pfn_valid() should
> > be able to tell if a part of a section is freed or not.
> > 
> > For instance for the following memory configuration
> >     
> >         |<----section---->|<----hole---->|<----section---->|
> >         +--------+--------+--------------+--------+--------+
> >         | bank 0 | unused |              | bank 1 | unused |
> >         +--------+--------+--------------+--------+--------+
> > 
> > the memory map corresponding to the "unused" areas is freed, but the generic
> > pfn_valid() will still return 1 there.
> 
> But is not free_unused_memmap() return early when CONFIG_SPARSEMEM_VMEMMAP
> is enabled, which is the only option now on arm64. Then how can memmap have
> holes (from unused areas) anymore ? Am I missing something here.
 
Ah, you are right, I missed this detail myself :)

With CONFIG_SPARSEMEM_VMEMMAP as the only memory model for arm64, we can
simply rid of arm64::pfn_valid() without any changes to the generic
version.
 
> > So we either should stop freeing unused memory map on arm64, or keep
> > arm64::pfn_valid() or implement something along the lines of this patch.
> > 
> > I personally don't think that the memory savings from freeing the unused
> > memory map worth the pain of maintenance and bugs happening here and there.
> > 

-- 
Sincerely yours,
Mike.