From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: MIME-Version: 1.0 References: <155677652226.2336373.8700273400832001094.stgit@dwillia2-desk3.amr.corp.intel.com> <155677653785.2336373.11131100812252340469.stgit@dwillia2-desk3.amr.corp.intel.com> <20190502074803.GA3495@linux> In-Reply-To: <20190502074803.GA3495@linux> From: Dan Williams Date: Thu, 2 May 2019 07:03:45 -0700 Message-ID: Subject: Re: [PATCH v7 03/12] mm/sparsemem: Add helpers track active portions of a section at boot Content-Type: text/plain; charset="UTF-8" Sender: owner-linux-mm@kvack.org To: Oscar Salvador Cc: Andrew Morton , Michal Hocko , Vlastimil Babka , Logan Gunthorpe , Jane Chu , linux-nvdimm , Linux MM , Linux Kernel Mailing List List-ID: On Thu, May 2, 2019 at 12:48 AM Oscar Salvador wrote: > > On Wed, May 01, 2019 at 10:55:37PM -0700, Dan Williams wrote: > > Prepare for hot{plug,remove} of sub-ranges of a section by tracking a > > section active bitmask, each bit representing 2MB (SECTION_SIZE (128M) / > > map_active bitmask length (64)). If it turns out that 2MB is too large > > of an active tracking granularity it is trivial to increase the size of > > the map_active bitmap. > > > > The implications of a partially populated section is that pfn_valid() > > needs to go beyond a valid_section() check and read the sub-section > > active ranges from the bitmask. > > > > Cc: Michal Hocko > > Cc: Vlastimil Babka > > Cc: Logan Gunthorpe > > Tested-by: Jane Chu > > Signed-off-by: Dan Williams > > Unfortunately I did not hear back about the comments/questions I made for this > in the previous version. Apologies, yes, will incorporate. > > > --- > > include/linux/mmzone.h | 29 ++++++++++++++++++++++++++++- > > mm/page_alloc.c | 4 +++- > > mm/sparse.c | 48 ++++++++++++++++++++++++++++++++++++++++++++++++ > > 3 files changed, 79 insertions(+), 2 deletions(-) > > > > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h > > index 6726fc175b51..cffde898e345 100644 > > --- a/include/linux/mmzone.h > > +++ b/include/linux/mmzone.h > > @@ -1175,6 +1175,8 @@ struct mem_section_usage { > > unsigned long pageblock_flags[0]; > > }; > > > > +void section_active_init(unsigned long pfn, unsigned long nr_pages); > > + > > struct page; > > struct page_ext; > > struct mem_section { > > @@ -1312,12 +1314,36 @@ static inline struct mem_section *__pfn_to_section(unsigned long pfn) > > > > extern int __highest_present_section_nr; > > > > +static inline int section_active_index(phys_addr_t phys) > > +{ > > + return (phys & ~(PA_SECTION_MASK)) / SECTION_ACTIVE_SIZE; > > +} > > + > > +#ifdef CONFIG_SPARSEMEM_VMEMMAP > > +static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn) > > +{ > > + int idx = section_active_index(PFN_PHYS(pfn)); > > + > > + return !!(ms->usage->map_active & (1UL << idx)); > > section_active_mask() also converts the value to address/size. > Why do we need to convert the values and we cannot work with pfn/pages instead? > It should be perfectly possible unless I am missing something. > > The only thing required would be to export earlier your: > > +#define PAGES_PER_SUB_SECTION (SECTION_ACTIVE_SIZE / PAGE_SIZE) > +#define PAGE_SUB_SECTION_MASK (~(PAGES_PER_SUB_SECTION-1)) > > and change section_active_index to: > > static inline int section_active_index(unsigned long pfn) > { > return (pfn & ~(PAGE_SECTION_MASK)) / SUB_SECTION_ACTIVE_PAGES; > } > > In this way we do need to shift the values every time and we can work with them > directly. > Maybe you made it work this way because a reason I am missing. > > > +static unsigned long section_active_mask(unsigned long pfn, > > + unsigned long nr_pages) > > +{ > > + int idx_start, idx_size; > > + phys_addr_t start, size; > > + > > + if (!nr_pages) > > + return 0; > > + > > + start = PFN_PHYS(pfn); > > + size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION > > + - (pfn & ~PAGE_SECTION_MASK))); > > It seems to me that we already picked the lowest value back in > section_active_init, so we should be fine if we drop the min() here? > > Another thing is why do we need to convert the values to address/size, and we > cannot work with pfns/pages. > Unless I am missing something it should be possible. Right, I believe the physical address conversion was a holdover from a previous version and these helpers can be cleaned up to be pfn based, good catch. > > > + size = ALIGN(size, SECTION_ACTIVE_SIZE); > > + > > + idx_start = section_active_index(start); > > + idx_size = section_active_index(size); > > + > > + if (idx_size == 0) > > + return -1; > > Maybe we would be better off converting that -1 into something like "FULL_SECTION", > or at least dropping a comment there that "-1" means that the section is fully > populated. Agreed, I'll add a #define. Thanks Oscar.