vfs: do (nearly) lockless generic_file_llseek

The i_mutex lock use of generic _file_llseek hurts.  Independent processes
accessing the same file synchronize over a single lock, even though
they have no need for synchronization at all.

Under high utilization this can cause llseek to scale very poorly on larger
systems.

This patch does some rethinking of the llseek locking model:

First the 64bit f_pos is not necessarily atomic without locks
on 32bit systems. This can already cause races with read() today.
This was discussed on linux-kernel in the past and deemed acceptable.
The patch does not change that.

Let's look at the different seek variants:

SEEK_SET: Doesn't really need any locking.
If there's a race one writer wins, the other loses.

For 32bit the non atomic update races against read()
stay the same. Without a lock they can also happen
against write() now.  The read() race was deemed
acceptable in past discussions, and I think if it's
ok for read it's ok for write too.

=> Don't need a lock.

SEEK_END: This behaves like SEEK_SET plus it reads
the maximum size too. Reading the maximum size would have the
32bit atomic problem. But luckily we already have a way to read
the maximum size without locking (i_size_read), so we
can just use that instead.

Without i_mutex there is no synchronization with write() anymore,
however since the write() update is atomic on 64bit it just behaves
like another racy SEEK_SET.  On non atomic 32bit it's the same
as SEEK_SET.

=> Don't need a lock, but need to use i_size_read()

SEEK_CUR: This has a read-modify-write race window
on the same file. One could argue that any application
doing unsynchronized seeks on the same file is already broken.
But for the sake of not adding a regression here I'm
using the file->f_lock to synchronize this. Using this
lock is much better than the inode mutex because it doesn't
synchronize between processes.

=> So still need a lock, but can use a f_lock.

This patch implements this new scheme in generic_file_llseek.
I dropped generic_file_llseek_unlocked and changed all callers.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

1 files changed, 41 insertions, 44 deletions

diff --git a/fs/read_write.c b/fs/read_write.c
index 179f1c33ea57..672b187def62 100644
--- a/fs/read_write.c
+++ b/fs/read_write.c
@@ -35,23 +35,45 @@ static inline int unsigned_offsets(struct file *file)
         return file->f_mode & FMODE_UNSIGNED_OFFSET;
 }
 
+static loff_t lseek_execute(struct file *file, struct inode *inode,
+                loff_t offset, loff_t maxsize)
+{
+        if (offset < 0 && !unsigned_offsets(file))
+                return -EINVAL;
+        if (offset > maxsize)
+                return -EINVAL;
+
+        if (offset != file->f_pos) {
+                file->f_pos = offset;
+                file->f_version = 0;
+        }
+        return offset;
+}
+
 /**
- * generic_file_llseek_unlocked - lockless generic llseek implementation
+ * generic_file_llseek - generic llseek implementation for regular files
  * @file:       file structure to seek on
  * @offset:     file offset to seek to
  * @origin:     type of seek
  *
- * Updates the file offset to the value specified by @offset and @origin.
- * Locking must be provided by the caller.
+ * This is a generic implemenation of ->llseek usable for all normal local
+ * filesystems.  It just updates the file offset to the value specified by
+ * @offset and @origin under i_mutex.
+ *
+ * Synchronization:
+ * SEEK_SET is unsynchronized (but atomic on 64bit platforms)
+ * SEEK_CUR is synchronized against other SEEK_CURs, but not read/writes.
+ * read/writes behave like SEEK_SET against seeks.
+ * SEEK_END
  */
 loff_t
-generic_file_llseek_unlocked(struct file *file, loff_t offset, int origin)
+generic_file_llseek(struct file *file, loff_t offset, int origin)
 {
         struct inode *inode = file->f_mapping->host;
 
         switch (origin) {
         case SEEK_END:
-                offset += inode->i_size;
+                offset += i_size_read(inode);
                 break;
         case SEEK_CUR:
                 /*
@@ -62,14 +84,22 @@ generic_file_llseek_unlocked(struct file *file, loff_t offset, int origin)
                  */
                 if (offset == 0)
                         return file->f_pos;
-                offset += file->f_pos;
-                break;
+                /*
+                 * f_lock protects against read/modify/write race with other
+                 * SEEK_CURs. Note that parallel writes and reads behave
+                 * like SEEK_SET.
+                 */
+                spin_lock(&file->f_lock);
+                offset = lseek_execute(file, inode, file->f_pos + offset,
+                                       inode->i_sb->s_maxbytes);
+                spin_unlock(&file->f_lock);
+                return offset;
         case SEEK_DATA:
                 /*
                  * In the generic case the entire file is data, so as long as
                  * offset isn't at the end of the file then the offset is data.
                  */
-                if (offset >= inode->i_size)
+                if (offset >= i_size_read(inode))
                         return -ENXIO;
                 break;
         case SEEK_HOLE:
@@ -77,46 +107,13 @@ generic_file_llseek_unlocked(struct file *file, loff_t offset, int origin)
                  * There is a virtual hole at the end of the file, so as long as
                  * offset isn't i_size or larger, return i_size.
                  */
-                if (offset >= inode->i_size)
+                if (offset >= i_size_read(inode))
                         return -ENXIO;
-                offset = inode->i_size;
+                offset = i_size_read(inode);
                 break;
         }
 
-        if (offset < 0 && !unsigned_offsets(file))
-                return -EINVAL;
-        if (offset > inode->i_sb->s_maxbytes)
-                return -EINVAL;
-
-        /* Special lock needed here? */
-        if (offset != file->f_pos) {
-                file->f_pos = offset;
-                file->f_version = 0;
-        }
-
-        return offset;
-}
-EXPORT_SYMBOL(generic_file_llseek_unlocked);
-
-/**
- * generic_file_llseek - generic llseek implementation for regular files
- * @file:       file structure to seek on
- * @offset:     file offset to seek to
- * @origin:     type of seek
- *
- * This is a generic implemenation of ->llseek useable for all normal local
- * filesystems.  It just updates the file offset to the value specified by
- * @offset and @origin under i_mutex.
- */
-loff_t generic_file_llseek(struct file *file, loff_t offset, int origin)
-{
-        loff_t rval;
-
-        mutex_lock(&file->f_dentry->d_inode->i_mutex);
-        rval = generic_file_llseek_unlocked(file, offset, origin);
-        mutex_unlock(&file->f_dentry->d_inode->i_mutex);
-
-        return rval;
+        return lseek_execute(file, inode, offset, inode->i_sb->s_maxbytes);
 }
 EXPORT_SYMBOL(generic_file_llseek);