[PATCH] Add -iglx & +iglx to Xserver.man

[Robert Morell]

On Tue, Nov 25, 2014 at 11:20:58PM -0800, Alan Coopersmith wrote:
> On 11/25/14 10:36 PM, Robert Morell wrote:
> > On Tue, Nov 25, 2014 at 07:26:14PM -0800, Alan Coopersmith wrote:
> >> Covers the current state after commits 99f0365b1fbdfd9238b9f,
> >> d0da0e9c3bb8fe0cd4879, & e3aa13b8d63ea2fba6eb4 were all applied.
> >>
> >> Signed-off-by: Alan Coopersmith <alan.coopersmith at oracle.com>
> >> ---
> >>   man/Xserver.man |   10 ++++++++++
> >>   1 file changed, 10 insertions(+)
> >>
> >> diff --git a/man/Xserver.man b/man/Xserver.man
> >> index c03830c..69ff626 100644
> >> --- a/man/Xserver.man
> >> +++ b/man/Xserver.man
> >> @@ -181,6 +181,16 @@ prints a usage message.
> >>   .B \-I
> >>   causes all remaining command line arguments to be ignored.
> >>   .TP 8
> >> +.B \-iglx
> >> +Prohibit creating indirect GLX contexts.  Indirect GLX is of limited use,
> >> +since it only supports up to OpenGL 1.4; it's slower than direct contexts;
> >
> > Technically, there exists protocol for much if not all of OpenGL 2.1,
> > and some extensions beyond that.  But not all server and client
> > implementations support it...
> 
> Sorry, I copied that from the comments in the commit that actually created the
> options, and took Eric's word for it:
> 
> http://cgit.freedesktop.org/xorg/xserver/commit/?id=99f0365b1fbdfd9238b9f5cc28491e4e6c7324f1
> 
> Would "only fully supports up to OpenGL 1.4, with partial support for
> OpenGL 2.1 and some later extensions" be more accurate?  Or is there
> a better wording someone can think of?

I'm not really sure myself what the extent of official
Khronos-sanctioned protocol support is -- it's spread out over a bunch
of extensions, and if there is any overall unifying documentation I'm
not aware of it.

Maybe a more future-proof statement appropriate for the man page would
be something like "lacks support for many modern OpenGL features and
extensions"?

> >> +and it opens a large attack surface for protocol parsing errors.
> >> +This is the default unless +iglx is specified, though it may not be honored
> >> +by GLX implementations other than the one provided with the X server.
> >
> > For what it's worth, in case anyone finds this thread in the archives,
> > I'll point out that NVIDIA's implementation does honor this option in
> > newer drivers, which contain the following entry in the changelog:
> >
> >      * Implemented support for disabling indirect GLX context creation using
> >        the -iglx option available on X.Org server release 1.16 and newer.  Note
> >        that future X.Org server releases may make the -iglx option the default.
> >        To re-enable support for indirect GLX on such servers, use the +iglx
> >        option.
> 
> That's good - perhaps we can drop that qualification if both NVIDIA & AMD
> will be supporting the check in their Xorg 1.17 ABI compatible versions.
> 
> I was thinking of older driver versions when I'd written it, but obviously,
> due to ABI changes, there shouldn't be many (any?) GLX implementations that
> can both be loaded by Xorg 1.17 and do not support this flag, so I don't
> know why I was thinking of that.

Yeah, good point.


- Robert

> > But your text above is still accurate, so other than perhaps softening
> > the OpenGL 1.4 limitation,
> >
> > Reviewed-by: Robert Morell <rmorell at nvidia.com>
> >
> >> +.TP 8
> >> +.B +iglx
> >> +Allow creating indirect GLX contexts.
> >> +.TP 8
> >>   .B \-maxbigreqsize \fIsize\fP
> >>   sets the maximum big request to
> >>   .I size
> >> --
> >> 1.7.9.2
> 
> 
> -- 
> 	-Alan Coopersmith-              alan.coopersmith at oracle.com
> 	 Oracle Solaris Engineering - http://blogs.oracle.com/alanc
>