[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20190627005628.GP17978@ZenIV.linux.org.uk>
Date: Thu, 27 Jun 2019 01:56:28 +0100
From: Al Viro <viro@...iv.linux.org.uk>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Alexei Starovoitov <alexei.starovoitov@...il.com>,
linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
tj@...nel.org
Subject: Re: misuse of fget_raw() in perf_event_get()
On Mon, Apr 15, 2019 at 11:04:06AM +0200, Peter Zijlstra wrote:
> On Sun, Apr 14, 2019 at 09:42:00PM -0700, Alexei Starovoitov wrote:
> > On Sat, Apr 13, 2019 at 10:02:42PM +0100, Al Viro wrote:
> > > What's the point of using fget_raw(), if you do
> > > _not_ accept O_PATH descriptors? That should be fget()...
> >
> > I think you're talking about commit e03e7ee34fdd ("perf/bpf: Convert perf_event_array to use struct file")
> > I don't really remember why we went with this instead of fget().
> > There was a bunch of back and forth back then with Peter.
>
> That was mostly on what refcount to use, you wanted to use the event
> refcount, and I suggested using the file refcount.
>
> If you look at:
>
> lkml.kernel.org/r/20160126161637.GF6357@...ns.programming.kicks-ass.net
>
> I too wondered about the fget_raw() at the time, whatever Al wants
> though, I never quite remember how that file stuff works :/
Anyway, fget() it becomes...
Powered by blists - more mailing lists