2001-12-05 23:28:58

by Simon Kirby

Subject: [FS] Why doesn't this patch work?

I'm attempting to write this little dinky patch to see who calls fsync()
or fdatasync(), but it's spitting out compiler warnings. I can't figure
out why, though. What did I do wrong?

buffer.c: In function `report_culprit':
buffer.c:409: warning: assignment from incompatible pointer type
buffer.c:410: warning: passing arg 2 of `d_path' from incompatible pointer type
buffer.c:420: warning: passing arg 1 of `mntput' from incompatible pointer type

--- linux/fs/buffer.c.orig Wed Dec 5 08:53:28 2001
+++ linux/fs/buffer.c Wed Dec 5 15:14:38 2001
@@ -396,6 +396,30 @@
return ret;
}

+static void report_culprit(struct dentry *dentry, struct file *file, char *action)
+{
+ struct vfsmnt * mnt;
+ char * buf = (char*)__get_free_page(GFP_KERNEL);
+ char * path;
+ int len;
+
+ if (!buf)
+ return;
+
+ mnt = mntget(file->f_vfsmnt);
+ path = d_path(dentry, mnt, buf, PAGE_SIZE - 1);
+ len = buf + PAGE_SIZE - 1 - path;
+ if (len >= PAGE_SIZE - 1)
+ len = PAGE_SIZE - 1;
+ path[len] = '\0';
+
+ printk("Process %u (%s) %s()ed \"%s\".\n",
+ current->pid, current->comm, action, path);
+
+ free_page((unsigned long)buf);
+ mntput(mnt);
+}
+
asmlinkage long sys_fsync(unsigned int fd)
{
struct file * file;
@@ -415,6 +439,8 @@
if (!file->f_op || !file->f_op->fsync)
goto out_putf;

+ report_culprit(dentry, file, "fsync");
+
/* We need to protect against concurrent writers.. */
down(&inode->i_sem);
filemap_fdatasync(inode->i_mapping);
@@ -446,6 +472,8 @@
err = -EINVAL;
if (!file->f_op || !file->f_op->fsync)
goto out_putf;
+
+ report_culprit(dentry, file, "fdatasync");

down(&inode->i_sem);
filemap_fdatasync(inode->i_mapping);

Simon-

[ Stormix Technologies Inc. ][ NetNation Communications Inc. ]
[ [email protected] ][ [email protected] ]
[ Opinions expressed are not necessarily those of my employers. ]

2001-12-05 23:46:48

by Robert Love

Subject: Re: [FS] Why doesn't this patch work?

On Wed, 2001-12-05 at 18:28, Simon Kirby wrote:
> I'm attempting to write this little dinky patch to see who calls fsync()
> or fdatasync(), but it's spitting out compiler warnings. I can't figure
> out why, though. What did I do wrong?
>
> buffer.c: In function `report_culprit':
> buffer.c:409: warning: assignment from incompatible pointer type
> buffer.c:410: warning: passing arg 2 of `d_path' from incompatible pointer type
> buffer.c:420: warning: passing arg 1 of `mntput' from incompatible pointer type

s/struct vfsmnt/struct vfsmount/

Robert Love

2001-12-05 23:54:58

by Dave Jones

Subject: Re: [FS] Why doesn't this patch work?

On Wed, 5 Dec 2001, Simon Kirby wrote:

> buffer.c:409: warning: assignment from incompatible pointer type

> + struct vfsmnt * mnt;

vfsmount perhaps ?

Dave.

--
| Dave Jones. http://www.codemonkey.org.uk
| SuSE Labs

2001-12-06 00:08:38

by Simon Kirby

Subject: Re: [FS] Why doesn't this patch work?

On Wed, Dec 05, 2001 at 06:46:21PM -0500, Robert Love wrote:

> On Wed, 2001-12-05 at 18:28, Simon Kirby wrote:
> > I'm attempting to write this little dinky patch to see who calls fsync()
> > or fdatasync(), but it's spitting out compiler warnings. I can't figure
> > out why, though. What did I do wrong?
> >
> > buffer.c: In function `report_culprit':
> > buffer.c:409: warning: assignment from incompatible pointer type
> > buffer.c:410: warning: passing arg 2 of `d_path' from incompatible pointer type
> > buffer.c:420: warning: passing arg 1 of `mntput' from incompatible pointer type
>
> s/struct vfsmnt/struct vfsmount/

Doh!

Simon-

[ Stormix Technologies Inc. ][ NetNation Communications Inc. ]
[ [email protected] ][ [email protected] ]
[ Opinions expressed are not necessarily those of my employers. ]