> -----Original Message-----
> From: Coly Li <colyli@xxxxxxx>
> Sent: Wednesday, December 16, 2020 6:30 PM
> To: Ren, Qiaowei <qiaowei.ren@xxxxxxxxx>
> Cc: linux-bcache@xxxxxxxxxxxxxxx; Ma, Jianpeng <jianpeng.ma@xxxxxxxxx>
> Subject: Re: [RFC PATCH 3/8] bcache: initialization of the buddy
>
> On 12/3/20 6:53 PM, Qiaowei Ren wrote:
> > This nvm pages allocator will implement the simple buddy to manage the
> > nvm address space. This patch initializes this buddy for new namespace.
> >
> > the unit of alloc/free of the buddy is page. DAX device has their
> > struct page(in dram or PMEM).
> >
> > struct { /* ZONE_DEVICE pages */
> > /** @pgmap: Points to the hosting device page map. */
> > struct dev_pagemap *pgmap;
> > void *zone_device_data;
> > /*
> > * ZONE_DEVICE private pages are counted as being
> > * mapped so the next 3 words hold the mapping, index,
> > * and private fields from the source anonymous or
> > * page cache page while the page is migrated to device
> > * private memory.
> > * ZONE_DEVICE MEMORY_DEVICE_FS_DAX pages also
> > * use the mapping, index, and private fields when
> > * pmem backed DAX files are mapped.
> > */
> > };
> >
> > ZONE_DEVICE pages only use pgmap. Other 4 words[16/32 bytes] don't use.
> > So the second/third word will be used as 'struct list_head ' which
> > list in buddy. The fourth word(that is normal struct page::index)
> > store pgoff which the page-offset in the dax device. And the fifth
> > word (that is normal struct page::private) store order of buddy.
> > page_type will be used to store buddy flags.
> >
> > Signed-off-by: Jianpeng Ma <jianpeng.ma@xxxxxxxxx>
> > Signed-off-by: Qiaowei Ren <qiaowei.ren@xxxxxxxxx>
> > ---
> > drivers/md/bcache/nvm-pages.c | 68
> > ++++++++++++++++++++++++++++++++++-
> > drivers/md/bcache/nvm-pages.h | 3 ++
> > 2 files changed, 70 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/md/bcache/nvm-pages.c
> > b/drivers/md/bcache/nvm-pages.c index 841616ea3267..7ffbfbacaf3f
> > 100644
> > --- a/drivers/md/bcache/nvm-pages.c
> > +++ b/drivers/md/bcache/nvm-pages.c
> > @@ -84,6 +84,17 @@ static void *nvm_pgoff_to_vaddr(struct
> nvm_namespace *ns, pgoff_t pgoff)
> > return ns->kaddr + ns->pages_offset + (pgoff << PAGE_SHIFT); }
> >
> > +static struct page *nvm_vaddr_to_page(struct nvm_namespace *ns, void
> > +*addr) {
> > + return virt_to_page(addr);
> > +}
> > +
> > +static inline void remove_owner_space(struct nvm_namespace *ns,
> > + pgoff_t pgoff, u32 nr)
> > +{
> > + bitmap_set(ns->pages_bitmap, pgoff, nr); }
> > +
> > static void init_owner_info(struct nvm_namespace *ns) {
> > struct owner_list_head *owner_list_head; @@ -126,6 +137,8 @@
> static
> > void init_owner_info(struct nvm_namespace *ns)
> > extent->nr = rec->nr;
> > list_add_tail(&extent->list, &extents-
> >extent_head);
> >
> > + remove_owner_space(extents->ns, rec->pgoff,
> rec->nr);
> > +
> > extents->ns->free -= rec->nr;
> > }
> > extents->size += nvm_pgalloc_recs->size; @@ -
> 143,6 +156,54 @@
> > static void init_owner_info(struct nvm_namespace *ns)
> > mutex_unlock(&only_set->lock);
> > }
> >
> > +static void init_nvm_free_space(struct nvm_namespace *ns) {
> > + unsigned int start, end, i;
> > + struct page *page;
> > + unsigned int pages;
> > + pgoff_t pgoff_start;
> > +
> > + bitmap_for_each_clear_region(ns->pages_bitmap, start, end, 0, ns-
> >pages_total) {
> > + pgoff_start = start;
> > + pages = end - start;
> > +
> > + while (pages) {
> > + for (i = MAX_ORDER - 1; i >= 0 ; i--) {
> > + if ((start % (1 << i) == 0) && (pages >= (1 << i)))
> > + break;
> > + }
> > +
> > + page = nvm_vaddr_to_page(ns, nvm_pgoff_to_vaddr(ns,
> pgoff_start));
> > + page->index = pgoff_start;
> > + page->private = i;
> > + __SetPageBuddy(page);
> > + list_add((struct list_head *)&page->zone_device_data,
> > +&ns->free_area[i]);
> > +
> > + pgoff_start += 1 << i;
> > + pages -= 1 << i;
> > + }
> > + }
> > +
> > + bitmap_for_each_set_region(ns->pages_bitmap, start, end, 0, ns-
> >pages_total) {
> > + pages = end - start;
> > + pgoff_start = start;
> > +
> > + while (pages) {
> > + for (i = MAX_ORDER - 1; i >= 0 ; i--) {
> > + if ((start % (1 << i) == 0) && (pages >= (1 << i)))
> > + break;
> > + }
> > +
> > + page = nvm_vaddr_to_page(ns, nvm_pgoff_to_vaddr(ns,
> pgoff_start));
> > + page->index = pgoff_start;
> > + page->private = i;
> > +
> > + pgoff_start += 1 << i;
> > + pages -= 1 << i;
> > + }
> > + }
> > +}
> > +
>
>
> The buddy structure should be initialized from the owner lists, we cannot
> assume the name space is empty. Because the user space may also allocate
> space from NVDIMM even before the first time it is attached by kernel driver.
>
In func init_owner_info, we already remove the alloced space.
Thanks!
Jianpeng.
>
> [snipped]
>
> Coly Li
