Messages in this thread | | | Date | Sun, 17 Mar 2013 19:42:55 +0100 | From | Oleg Nesterov <> | Subject | Re: [PATCH] uprobes: Use file_inode() |
| |
On 03/17, Al Viro wrote: > > On Sun, Mar 17, 2013 at 07:00:36PM +0100, Oleg Nesterov wrote: > > Cleanup. Now that we have f_inode/file_inode() we can use it > > instead of ->f_mapping->host. > > No. This is *not* guaranteed to be the same thing in general; note that > e.g. for block devices ->f_mapping->host is *not* equal to file_inode().
Yes,
> It probably is valid in this particular case,
And yes (I think). In fact I think ->f_mode is "more correct" in this case. Say, if this uprobe was created by create_trace_uprobe() we use d_inode, and uprobe_mmap/etc uses file_inode() only to compare this pointer with uprobe->inode.
But I'll try to recheck, and:
> but at the very least you > need to explain that in commit message,
OK. Will do, thanks.
Oleg.
| |