]> git.ipfire.org Git - thirdparty/linux.git/commitdiff
mm,swap: remove struct vma_swap_readahead
authorHuang Ying <ying.huang@intel.com>
Fri, 31 May 2024 08:12:29 +0000 (16:12 +0800)
committerAndrew Morton <akpm@linux-foundation.org>
Thu, 4 Jul 2024 02:30:03 +0000 (19:30 -0700)
When VMA based swap readahead is introduced in commit ec560175c0b6 ("mm,
swap: VMA based swap readahead"), "struct vma_swap_readahead" is defined
to describe the readahead window.  Because we wanted to save the PTE
entries in the struct at that time.  But after commit 4f8fcf4ced0b
("mm/swap: swap_vma_readahead() do the pte_offset_map()"), we no longer
save PTE entries in the struct.  The size of the struct becomes so small,
that it's better to use the fields of the struct directly.  This can
simplify the code to improve the code readability.  The line number of
source code reduces too.

No functionality change is expected in this patch.

Link: https://lkml.kernel.org/r/20240531081230.310128-3-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
mm/swap_state.c

index 88a16969027d3a6b9fd6fc315749663b8961e057..eb7a306a24245dd840acda42eecb215e9f55b962 100644 (file)
@@ -42,6 +42,8 @@ struct address_space *swapper_spaces[MAX_SWAPFILES] __read_mostly;
 static unsigned int nr_swapper_spaces[MAX_SWAPFILES] __read_mostly;
 static bool enable_vma_readahead __read_mostly = true;
 
+#define SWAP_RA_ORDER_CEILING  5
+
 #define SWAP_RA_WIN_SHIFT      (PAGE_SHIFT / 2)
 #define SWAP_RA_HITS_MASK      ((1UL << SWAP_RA_WIN_SHIFT) - 1)
 #define SWAP_RA_HITS_MAX       SWAP_RA_HITS_MASK
@@ -739,16 +741,9 @@ void exit_swap_address_space(unsigned int type)
        swapper_spaces[type] = NULL;
 }
 
-#define SWAP_RA_ORDER_CEILING  5
-
-struct vma_swap_readahead {
-       unsigned short win;
-       unsigned short offset;
-       unsigned short nr_pte;
-};
-
-static void swap_ra_info(struct vm_fault *vmf,
-                        struct vma_swap_readahead *ra_info)
+static unsigned short swap_vma_ra_win(struct vm_fault *vmf,
+                                     unsigned short *offset,
+                                     unsigned short *nr_pte)
 {
        struct vm_area_struct *vma = vmf->vma;
        unsigned long ra_val;
@@ -758,10 +753,8 @@ static void swap_ra_info(struct vm_fault *vmf,
 
        max_win = 1 << min_t(unsigned int, READ_ONCE(page_cluster),
                             SWAP_RA_ORDER_CEILING);
-       if (max_win == 1) {
-               ra_info->win = 1;
-               return;
-       }
+       if (max_win == 1)
+               return 1;
 
        faddr = vmf->address;
        fpfn = PFN_DOWN(faddr);
@@ -769,12 +762,11 @@ static void swap_ra_info(struct vm_fault *vmf,
        pfn = PFN_DOWN(SWAP_RA_ADDR(ra_val));
        prev_win = SWAP_RA_WIN(ra_val);
        hits = SWAP_RA_HITS(ra_val);
-       ra_info->win = win = __swapin_nr_pages(pfn, fpfn, hits,
-                                              max_win, prev_win);
+       win = __swapin_nr_pages(pfn, fpfn, hits, max_win, prev_win);
        atomic_long_set(&vma->swap_readahead_info,
                        SWAP_RA_VAL(faddr, win, 0));
        if (win == 1)
-               return;
+               return 1;
 
        if (fpfn == pfn + 1) {
                lpfn = fpfn;
@@ -795,8 +787,10 @@ static void swap_ra_info(struct vm_fault *vmf,
        end = min3(rpfn, PFN_DOWN(vma->vm_end),
                   PFN_DOWN((faddr & PMD_MASK) + PMD_SIZE));
 
-       ra_info->nr_pte = end - start;
-       ra_info->offset = fpfn - start;
+       *nr_pte = end - start;
+       *offset = fpfn - start;
+
+       return win;
 }
 
 /**
@@ -827,19 +821,17 @@ static struct folio *swap_vma_readahead(swp_entry_t targ_entry, gfp_t gfp_mask,
        pgoff_t ilx;
        unsigned int i;
        bool page_allocated;
-       struct vma_swap_readahead ra_info = {
-               .win = 1,
-       };
+       unsigned short win, nr_pte, offset;
 
-       swap_ra_info(vmf, &ra_info);
-       if (ra_info.win == 1)
+       win = swap_vma_ra_win(vmf, &offset, &nr_pte);
+       if (win == 1)
                goto skip;
 
-       addr = vmf->address - (ra_info.offset * PAGE_SIZE);
-       ilx = targ_ilx - ra_info.offset;
+       addr = vmf->address - offset * PAGE_SIZE;
+       ilx = targ_ilx - offset;
 
        blk_start_plug(&plug);
-       for (i = 0; i < ra_info.nr_pte; i++, ilx++, addr += PAGE_SIZE) {
+       for (i = 0; i < nr_pte; i++, ilx++, addr += PAGE_SIZE) {
                if (!pte++) {
                        pte = pte_offset_map(vmf->pmd, addr);
                        if (!pte)
@@ -859,7 +851,7 @@ static struct folio *swap_vma_readahead(swp_entry_t targ_entry, gfp_t gfp_mask,
                        continue;
                if (page_allocated) {
                        swap_read_folio(folio, false, &splug);
-                       if (i != ra_info.offset) {
+                       if (i != offset) {
                                folio_set_readahead(folio);
                                count_vm_event(SWAP_RA);
                        }