[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Mon, 12 Dec 2016 19:23:25 -0200
From: Gustavo Padovan <gustavo.padovan@...labora.com>
To: Chris Wilson <chris@...is-wilson.co.uk>,
Gustavo Padovan <gustavo@...ovan.org>,
dri-devel@...ts.freedesktop.org,
Gustavo Padovan <gustavo.padovan@...labora.co.uk>,
"open list:DRM DRIVER FOR MSM ADRENO GPU"
<linux-arm-msm@...r.kernel.org>,
open list <linux-kernel@...r.kernel.org>,
"open list:DRM DRIVER FOR MSM ADRENO GPU"
<freedreno@...ts.freedesktop.org>
Subject: Re: [PATCH] drm/msm: return fence_fd = -1 if gem_submit fails
2016-12-12 Chris Wilson <chris@...is-wilson.co.uk>:
> On Mon, Dec 12, 2016 at 05:41:08PM -0200, Gustavo Padovan wrote:
> > From: Gustavo Padovan <gustavo.padovan@...labora.co.uk>
> >
> > Previously we were returning garbage here, fix it by setting it to -1
> > before the first possible point of failure.
>
> The convention is that on error paths you do not modify user inputs. In
> particular, consider EINTR where the usual pattern (e.g. drmIoctl) is
>
> do {
> err = ioctl(fd, SUBMIT, arg);
> } while (err == -EINTR);
>
> If you modify the in fence before you consume it, you can't recreate it
> after handling the signal.
Right. I didn't know about that convention. So maybe we let it as is. :)
Gustavo
Powered by blists - more mailing lists