On Tue 02-04-19 14:15:20, Yafang Shao wrote:
> We found that some latency spike was caused by page cache miss on our
> database server.
> So we decide to measure the page cache miss.
> Currently the kernel is lack of this facility for measuring it.
What are you going to use this information for?
> This patch introduces a new vm counter PGCACHEMISS for this purpose.
> This counter will be incremented in bellow scenario,
> - page cache miss in generic file read routine
> - read access page cache miss in mmap
> - read access page cache miss in swapin
>
> NB, readahead routine is not counted because it won't stall the
> application directly.
Doesn't this partially open the side channel we have closed for mincore
just recently?
> Signed-off-by: Yafang Shao <laoar.shao@xxxxxxxxx>
> ---
> include/linux/pagemap.h | 7 +++++++
> include/linux/vm_event_item.h | 1 +
> mm/filemap.c | 2 ++
> mm/memory.c | 1 +
> mm/shmem.c | 9 +++++----
> mm/vmstat.c | 1 +
> 6 files changed, 17 insertions(+), 4 deletions(-)
>
> diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
> index f939e00..8355b51 100644
> --- a/include/linux/pagemap.h
> +++ b/include/linux/pagemap.h
> @@ -233,6 +233,13 @@ pgoff_t page_cache_next_miss(struct address_space *mapping,
> pgoff_t page_cache_prev_miss(struct address_space *mapping,
> pgoff_t index, unsigned long max_scan);
>
> +static inline void page_cache_read_miss(struct vm_fault *vmf)
> +{
> + if (!vmf || (vmf->flags & (FAULT_FLAG_USER | FAULT_FLAG_WRITE)) ==
> + FAULT_FLAG_USER)
> + count_vm_event(PGCACHEMISS);
> +}
> +
> #define FGP_ACCESSED 0x00000001
> #define FGP_LOCK 0x00000002
> #define FGP_CREAT 0x00000004
> diff --git a/include/linux/vm_event_item.h b/include/linux/vm_event_item.h
> index 47a3441..d589f05 100644
> --- a/include/linux/vm_event_item.h
> +++ b/include/linux/vm_event_item.h
> @@ -29,6 +29,7 @@ enum vm_event_item { PGPGIN, PGPGOUT, PSWPIN, PSWPOUT,
> PGFREE, PGACTIVATE, PGDEACTIVATE, PGLAZYFREE,
> PGFAULT, PGMAJFAULT,
> PGLAZYFREED,
> + PGCACHEMISS,
> PGREFILL,
> PGSTEAL_KSWAPD,
> PGSTEAL_DIRECT,
> diff --git a/mm/filemap.c b/mm/filemap.c
> index 4157f85..fc12c2d 100644
> --- a/mm/filemap.c
> +++ b/mm/filemap.c
> @@ -2256,6 +2256,7 @@ static ssize_t generic_file_buffered_read(struct kiocb *iocb,
> goto out;
>
> no_cached_page:
> + page_cache_read_miss(NULL);
> /*
> * Ok, it wasn't cached, so we need to create a new
> * page..
> @@ -2556,6 +2557,7 @@ vm_fault_t filemap_fault(struct vm_fault *vmf)
> fpin = do_async_mmap_readahead(vmf, page);
> } else if (!page) {
> /* No page in the page cache at all */
> + page_cache_read_miss(vmf);
> count_vm_event(PGMAJFAULT);
> count_memcg_event_mm(vmf->vma->vm_mm, PGMAJFAULT);
> ret = VM_FAULT_MAJOR;
> diff --git a/mm/memory.c b/mm/memory.c
> index bd157f2..63bcd41 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -2754,6 +2754,7 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> ret = VM_FAULT_MAJOR;
> count_vm_event(PGMAJFAULT);
> count_memcg_event_mm(vma->vm_mm, PGMAJFAULT);
> + page_cache_read_miss(vmf);
> } else if (PageHWPoison(page)) {
> /*
> * hwpoisoned dirty swapcache pages are kept for killing
> diff --git a/mm/shmem.c b/mm/shmem.c
> index 3a4b74c..47e33a4 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -127,7 +127,7 @@ static int shmem_replace_page(struct page **pagep, gfp_t gfp,
> static int shmem_swapin_page(struct inode *inode, pgoff_t index,
> struct page **pagep, enum sgp_type sgp,
> gfp_t gfp, struct vm_area_struct *vma,
> - vm_fault_t *fault_type);
> + struct vm_fault *vmf, vm_fault_t *fault_type);
> static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
> struct page **pagep, enum sgp_type sgp,
> gfp_t gfp, struct vm_area_struct *vma,
> @@ -1159,7 +1159,7 @@ static int shmem_unuse_swap_entries(struct inode *inode, struct pagevec pvec,
> error = shmem_swapin_page(inode, indices[i],
> &page, SGP_CACHE,
> mapping_gfp_mask(mapping),
> - NULL, NULL);
> + NULL, NULL, NULL);
> if (error == 0) {
> unlock_page(page);
> put_page(page);
> @@ -1614,7 +1614,7 @@ static int shmem_replace_page(struct page **pagep, gfp_t gfp,
> static int shmem_swapin_page(struct inode *inode, pgoff_t index,
> struct page **pagep, enum sgp_type sgp,
> gfp_t gfp, struct vm_area_struct *vma,
> - vm_fault_t *fault_type)
> + struct vm_fault *vmf, vm_fault_t *fault_type)
> {
> struct address_space *mapping = inode->i_mapping;
> struct shmem_inode_info *info = SHMEM_I(inode);
> @@ -1636,6 +1636,7 @@ static int shmem_swapin_page(struct inode *inode, pgoff_t index,
> *fault_type |= VM_FAULT_MAJOR;
> count_vm_event(PGMAJFAULT);
> count_memcg_event_mm(charge_mm, PGMAJFAULT);
> + page_cache_read_miss(vmf);
> }
> /* Here we actually start the io */
> page = shmem_swapin(swap, gfp, info, index);
> @@ -1758,7 +1759,7 @@ static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
> page = find_lock_entry(mapping, index);
> if (xa_is_value(page)) {
> error = shmem_swapin_page(inode, index, &page,
> - sgp, gfp, vma, fault_type);
> + sgp, gfp, vma, vmf, fault_type);
> if (error == -EEXIST)
> goto repeat;
>
> diff --git a/mm/vmstat.c b/mm/vmstat.c
> index 36b56f8..c49ecba 100644
> --- a/mm/vmstat.c
> +++ b/mm/vmstat.c
> @@ -1188,6 +1188,7 @@ int fragmentation_index(struct zone *zone, unsigned int order)
> "pgfault",
> "pgmajfault",
> "pglazyfreed",
> + "pgcachemiss",
>
> "pgrefill",
> "pgsteal_kswapd",
> --
> 1.8.3.1
>
--
Michal Hocko
SUSE Labs
