[Intel-gfx] [PATCH 4/4] drm: Add decoding of DRM and KMS ioctls

Gabriel Laskar gabriel at lse.epita.fr
Tue Jun 9 07:38:40 PDT 2015


On Tue, 9 Jun 2015 16:29:31 +0200
Patrik Jakobsson <patrik.jakobsson at linux.intel.com> wrote:

> On Tue, Jun 09, 2015 at 03:51:08PM +0200, Gabriel Laskar wrote:
> > On Tue,  9 Jun 2015 13:26:44 +0200
> > Patrik Jakobsson <patrik.jakobsson at linux.intel.com> wrote:
> > 
> > > This patch adds many of the DRM and KMS ioctls. The rest can be added as
> > > needed.
> > > 
> > > Signed-off-by: Patrik Jakobsson <patrik.jakobsson at linux.intel.com>
> > > ---
> > >  drm.c | 519 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > >  1 file changed, 519 insertions(+)
> > > 
> > > diff --git a/drm.c b/drm.c
> > > index fa98fb7..e550c34 100644
> > > --- a/drm.c
> > > +++ b/drm.c
> > > @@ -82,6 +82,468 @@ int drm_is_driver(struct tcb *tcp, const char *name)
> > >  	return strcmp(name, drv) == 0;
> > >  }
> > >  
> > > +static int drm_version(struct tcb *tcp, const unsigned int code, long arg)
> > > +{
> > > +	struct drm_version ver;
> > > +	char *name, *date, *desc;
> > > +	int ret;
> > > +
> > > +	if (entering(tcp) || umove(tcp, arg, &ver))
> > > +		return 0;
> > > +
> > > +	name = calloc(ver.name_len + 1, 1);
> > 
> > We have some wrappers for that now, you can call xcalloc(), but it will
> > die if this does not work. Your version have the advantage of not kill
> > strace, and just not decode the argument.
> > 
> 
> If ok I'll keep it as calloc? As you say, dying here is not really neccessary.

Yeah, that was mostly me thinking out loud. Imho, I would keep the
calloc(), but I don't know what Dmitry will prefer. Maybe for
consistency, it should be better to have an xcalloc().

-- 
Gabriel Laskar


More information about the Intel-gfx mailing list