On Thu, Sep 21, 2017 at 03:36:10PM +0200, Christoph Hellwig wrote:

On Wed, Sep 20, 2017 at 04:08:24PM -0700, Darrick J. Wong wrote:

quoted

quoted

I'm still wondering why we allocate a potentially large getbmapx buffer,
fill it out, and only then format the results to userspace?  I think
getbmap (the ioctl) is now the only user of these functions, so can't
we just call the formatter directly from _getbmap_report_one and
_getbmap_report_hole, like what getfsmap does?

(I also feel like I've asked this before, so apologies if I'm merely
forgetting the answer.)

Oh right, it's because we have the inode locked, and copying things to
userspace could incur a page fault, which we can't risk with the inode
locked because some malicious person could create a fragmented file with
a bmap request header at the start of the file, mmap the file, and call
bmap on the fragmented file with the pointer being the mmap region.

Yes.

Can I get a Reviewed-by: tag now? :)

Reviewed-by: Darrick J. Wong <redacted>

--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html