Badari Pulavarty wrote:

Hi Andrew & Chuck,

This patch looks good. As we discussed earlier, this breaks "raw" driver.

Yes, Chuck's patch is clearly the right thing to do.  I got caught
out cheating :(

The problem was caused by the consolidation of the raw driver - it 
does not have a file * for the target blockdev inode.

I can fix it. Please let me know, if you want me to work on it.

OK, thanks.  I guess we can either:

- Just create a temp struct file on the stack (bit risky)

- Change the raw device's userspace API and make raw bind to
  fd's, not majors/minors

- Change rw_raw_dev to call inode->i_mapping->aops->direct_IO
  directly.

The latter sounds suitable ;)  I should have done that on
day one.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/