Skip to contentSkip to conversation

fuse: respect FOPEN_KEEP_CACHE on opendir

PATCH Applied Amir Goldstein 5 participants 5 messages Jan 1, 2025 – Jan 6, 2025

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 fuse: respect FOPEN_KEEP_CACHE on opendir

The re-factoring of fuse_dir_open() missed the need to invalidate directory inode page cache with open flag FOPEN_KEEP_CACHE.

---

Miklos,

I verified the fix using:
passthrough_ll -d -o source=/src,cache=always /mnt

and watching debug prints from repeating 'ls /mnt' invocations.

With current upstream, dir cache is kept even though passthrough_ll never sets keep_cache in opendir.

passthrough_hp always set keep_cache together with cache_readdir, so it could not have noticed this regression.

I've modified passthrough_ll as follows to test the keep_cache flag:

        fi->fh = (uintptr_t) d;

< if (lo->cache == CACHE_ALWAYS)

Quoted 1 line
      if (lo->cache != CACHE_NEVER)
                fi->cache_readdir = 1;
Quoted 2 lines
      if (lo->cache == CACHE_ALWAYS)
              fi->keep_cache = 1;
        fuse_reply_open(req, fi);
        return;

Thanks,
Amir.

fs/fuse/dir.c +2 -0
index 494ac372ace07..e540d05549fff 100644
@@ -1681,6 +1681,8 @@ static int fuse_dir_open(struct inode *inode, struct file *file)
 		 */
 		if (ff->open_flags & (FOPEN_STREAM | FOPEN_NONSEEKABLE))
 			nonseekable_open(inode, file);
+		if (!(ff->open_flags & FOPEN_KEEP_CACHE))
+			invalidate_inode_pages2(inode->i_mapping);
 	}
 
 	return err;
Signature · 1 line
2.34.1
Reviewed-by on v1 1/1
#1 lore raw
Quoted 41 lines · Amir Goldstein

The re-factoring of fuse_dir_open() missed the need to invalidate directory inode page cache with open flag FOPEN_KEEP_CACHE.

---

Miklos,

I verified the fix using:
passthrough_ll -d -o source=/src,cache=always /mnt

and watching debug prints from repeating 'ls /mnt' invocations.

With current upstream, dir cache is kept even though passthrough_ll never sets keep_cache in opendir.

passthrough_hp always set keep_cache together with cache_readdir, so it could not have noticed this regression.

I've modified passthrough_ll as follows to test the keep_cache flag:

        fi->fh = (uintptr_t) d;

< if (lo->cache == CACHE_ALWAYS)

Quoted 1 line
      if (lo->cache != CACHE_NEVER)
                fi->cache_readdir = 1;
Quoted 2 lines
      if (lo->cache == CACHE_ALWAYS)
              fi->keep_cache = 1;
        fuse_reply_open(req, fi);
        return;

Thanks,
Amir.

fs/fuse/dir.c +0 -0
index 494ac372ace07..e540d05549fff 100644

fs/fuse/dir.c static int fuse_dir_open(struct inode *inode, struct file *file) · line 1688

Quoted 8 lines
fs/fuse/dir.c +2 -0
@@ -1681,6 +1681,8 @@ static int fuse_dir_open(struct inode *inode, struct file *file)
 		 */
 		if (ff->open_flags & (FOPEN_STREAM | FOPEN_NONSEEKABLE))
 			nonseekable_open(inode, file);
+		if (!(ff->open_flags & FOPEN_KEEP_CACHE))
+			invalidate_inode_pages2(inode->i_mapping);
 	}
 
 	return err;

LGTM. Thanks for the quick fix Amir, especially during holidays!

maintainer Applied
#2 lore raw
In reply to Amir Goldstein
Quoted 4 lines · Amir Goldstein

The re-factoring of fuse_dir_open() missed the need to invalidate directory inode page cache with open flag FOPEN_KEEP_CACHE.

Applied to the vfs.fixes branch of the vfs/vfs.git tree. Patches in the vfs.fixes branch should appear in linux-next soon.

Please report any outstanding bugs that were missed during review in a new review to the original patch series allowing us to drop it.

It's encouraged to provide Acked-bys and Reviewed-bys even though the patch has now been applied. If possible patch trailers will be updated.

Note that commit hashes shown below are subject to change due to rebase, trailer updates or similar. If in doubt, please check the listed branch.

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git
branch: vfs.fixes

[1/1] fuse: respect FOPEN_KEEP_CACHE on opendir

      https://git.kernel.org/vfs/vfs/c/03f275adb8fb
maintainer
#3 lore raw
In reply to Amir Goldstein
Quoted 25 lines · Amir Goldstein

The re-factoring of fuse_dir_open() missed the need to invalidate directory inode page cache with open flag FOPEN_KEEP_CACHE.

---

Miklos,

I verified the fix using:
passthrough_ll -d -o source=/src,cache=always /mnt

and watching debug prints from repeating 'ls /mnt' invocations.

With current upstream, dir cache is kept even though passthrough_ll never sets keep_cache in opendir.

passthrough_hp always set keep_cache together with cache_readdir, so it could not have noticed this regression.

I've modified passthrough_ll as follows to test the keep_cache flag:

Quoted 8 lines
        fi->fh = (uintptr_t) d;

< if (lo->cache == CACHE_ALWAYS)

Quoted 1 line
      if (lo->cache != CACHE_NEVER)
                fi->cache_readdir = 1;
Quoted 2 lines
      if (lo->cache == CACHE_ALWAYS)
              fi->keep_cache = 1;
        fuse_reply_open(req, fi);
        return;

Thanks for fixing this, Amir.

Miklos

#4 lore raw

Thanks everyone for the quick turnaround, especially Amir for the quick fix!!

Looking forward to the new release with this fix.

Regards,
Prince Kumar.

Quoted 39 lines · Miklos Szeredi
Quoted 25 lines · Amir Goldstein

The re-factoring of fuse_dir_open() missed the need to invalidate directory inode page cache with open flag FOPEN_KEEP_CACHE.

---

Miklos,

I verified the fix using:
passthrough_ll -d -o source=/src,cache=always /mnt

and watching debug prints from repeating 'ls /mnt' invocations.

With current upstream, dir cache is kept even though passthrough_ll never sets keep_cache in opendir.

passthrough_hp always set keep_cache together with cache_readdir, so it could not have noticed this regression.

I've modified passthrough_ll as follows to test the keep_cache flag:

Quoted 8 lines
        fi->fh = (uintptr_t) d;

< if (lo->cache == CACHE_ALWAYS)

Quoted 1 line
      if (lo->cache != CACHE_NEVER)
                fi->cache_readdir = 1;
Quoted 2 lines
      if (lo->cache == CACHE_ALWAYS)
              fi->keep_cache = 1;
        fuse_reply_open(req, fi);
        return;

Thanks for fixing this, Amir.

Miklos