Background:
Not much. Some C. Have been using Linux since pre-1.0 and s390 Linux
admin for the last seven years or so. I've been following kernel
janitors and lkml (lkml when i can) for awhile now. Also reading "Linux
Kernel Development" by Robert Love.
Now:
Saw in the janitor list about removing lock_kernel(). Created the
following patch for linux/fs/fat against vanilla 2.6.23, but I have some
questions.
1. In file.c, line 304:
lock_kernel();
fat_free(inode, nr_clusters);
unlock_kernel();
Why wrap fat_free with lock_kernel and not lock in the function
itself?
2. In inode.c, starting at lines 439 and 570, the two functions
fat_clear_inode() and fat_write_inode() lock_kernel and then spin lock.
Why the spin lock if it already has the bkl? Since the spin locks are
used, it makes me less sure of this. I didn't touch lock_kernel() in
inode.c at all. In fact I'm almost positive I'm missing something, but
I'm here to learn. So here it is:
--- linux-2.6.23/fs/fat/dir.c 2007-10-09 15:31:38.000000000 -0500
+++ linux-2.6.23-jcl/fs/fat/dir.c 2007-10-12 11:02:17.000000000
-0500
@@ -459,8 +459,9 @@
int chi, chl, i, i2, j, last, last_u, dotoffset = 0;
loff_t cpos;
int ret = 0;
+ rwlock_t mr_rwlock = RW_LOCK_UNLOCKED;
- lock_kernel();
+ read_lock(&mr_rwlock);
cpos = filp->f_pos;
/* Fake . and .. for the root directory. */
@@ -642,7 +643,7 @@
if (unicode)
free_page((unsigned long)unicode);
out:
- unlock_kernel();
+ read_unlock(&mr_rwlock);
return ret;
}
diff -ur linux-2.6.23/fs/fat/file.c linux-2.6.23-jcl/fs/fat/file.c
--- linux-2.6.23/fs/fat/file.c 2007-10-09 15:31:38.000000000 -0500
+++ linux-2.6.23-jcl/fs/fat/file.c 2007-10-12 11:37:47.000000000
-0500
@@ -160,8 +160,9 @@
struct msdos_sb_info *sbi = MSDOS_SB(dentry->d_sb);
struct inode *inode = dentry->d_inode;
int mask, error = 0;
+ rwlock_t mr_rwlock = RW_LOCK_UNLOCKED;
- lock_kernel();
+ write_lock(&mr_rwlock);
/*
* Expand the file. Since inode_setattr() updates ->i_size
@@ -206,7 +207,7 @@
mask = sbi->options.fs_fmask;
inode->i_mode &= S_IFMT | (S_IRWXUGO & ~mask);
out:
- unlock_kernel();
+ write_unlock(&mr_rwlock);
return error;
}
@@ -287,6 +288,7 @@
struct msdos_sb_info *sbi = MSDOS_SB(inode->i_sb);
const unsigned int cluster_size = sbi->cluster_size;
int nr_clusters;
+ rwlock_t mr_rwlock = RW_LOCK_UNLOCKED;
/*
* This protects against truncating a file bigger than it was
then
@@ -297,9 +299,9 @@
nr_clusters = (inode->i_size + (cluster_size - 1)) >>
sbi->cluster_bits;
- lock_kernel();
+ write_lock(&mr_rwlock);
fat_free(inode, nr_clusters);
- unlock_kernel();
+ write_lock(&mr_rwlock);
fat_flush_inodes(inode->i_sb, inode, NULL);
}
-
To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
