]> git.ipfire.org Git - thirdparty/kernel/stable.git/commitdiff
exfat: add cache option for __exfat_ent_get
authorChi Zhiling <chizhiling@kylinos.cn>
Wed, 14 Jan 2026 12:12:37 +0000 (20:12 +0800)
committerNamjae Jeon <linkinjeon@kernel.org>
Thu, 12 Feb 2026 12:21:47 +0000 (21:21 +0900)
When multiple entries are obtained consecutively, these entries are mostly
stored adjacent to each other. this patch introduces a "last" parameter to
cache the last opened buffer head, and reuse it when possible, which
reduces the number of sb_bread() calls.

When the passed parameter "last" is NULL, it means cache option is
disabled, the behavior unchanged as it was.

Signed-off-by: Chi Zhiling <chizhiling@kylinos.cn>
Reviewed-by: Yuezhang Mo <Yuezhang.Mo@sony.com>
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
fs/exfat/fatent.c

index c9c5f2e3a05ebc4af3c2c8d3edbf0c7328499205..0cfbc0b435bd8bcd079e046a1a9ba4c53712bd6a 100644 (file)
@@ -36,18 +36,23 @@ static int exfat_mirror_bh(struct super_block *sb, sector_t sec,
 }
 
 static int __exfat_ent_get(struct super_block *sb, unsigned int loc,
-               unsigned int *content)
+               unsigned int *content, struct buffer_head **last)
 {
        unsigned int off;
        sector_t sec;
-       struct buffer_head *bh;
+       struct buffer_head *bh = last ? *last : NULL;
 
        sec = FAT_ENT_OFFSET_SECTOR(sb, loc);
        off = FAT_ENT_OFFSET_BYTE_IN_SECTOR(sb, loc);
 
-       bh = sb_bread(sb, sec);
-       if (!bh)
-               return -EIO;
+       if (!bh || bh->b_blocknr != sec || !buffer_uptodate(bh)) {
+               brelse(bh);
+               bh = sb_bread(sb, sec);
+               if (last)
+                       *last = bh;
+               if (unlikely(!bh))
+                       return -EIO;
+       }
 
        *content = le32_to_cpu(*(__le32 *)(&bh->b_data[off]));
 
@@ -55,7 +60,8 @@ static int __exfat_ent_get(struct super_block *sb, unsigned int loc,
        if (*content > EXFAT_BAD_CLUSTER)
                *content = EXFAT_EOF_CLUSTER;
 
-       brelse(bh);
+       if (!last)
+               brelse(bh);
        return 0;
 }
 
@@ -95,7 +101,7 @@ int exfat_ent_get(struct super_block *sb, unsigned int loc,
                return -EIO;
        }
 
-       err = __exfat_ent_get(sb, loc, content);
+       err = __exfat_ent_get(sb, loc, content, NULL);
        if (err) {
                exfat_fs_error_ratelimit(sb,
                        "failed to access to FAT (entry 0x%08x, err:%d)",