[Mesa-dev] Bug in 48157b904a9 found by Coverity

Damien Lespiau damien.lespiau at intel.com
Wed Nov 19 10:05:42 PST 2014


On Wed, Nov 19, 2014 at 12:42:45PM -0500, Ilia Mirkin wrote:
> On Wed, Nov 19, 2014 at 12:28 PM, Damien Lespiau
> <damien.lespiau at intel.com> wrote:
> > On Wed, Nov 19, 2014 at 12:13:41PM -0500, Ilia Mirkin wrote:
> >> Hey guys,
> >>
> >> Just got around to looking at the Coverity email about newly
> >> introduced bugs, and this seems like a legit issue. Based on the
> >> context of the code, I'm guessing you meant to do (input_index - 16)
> >> << 1.
> >
> > Oh my. The proposed fix looks good, mind crafting a patch?
> 
> Probably best done by someone with access to the hardware/simulator or
> who has read the specs (are the gen9 specs even public yet? haven't
> checked).

Nop.

> As an aside, it seems like there can be 32 var varyings, and there are
> a bunch of additional slots on top of that... not sure if you meant to
> start at VARYING_SLOT_VAR0 or if you wanted all the varying slots, and
> there's some sort of additional mechanism that makes sure that at most
> 32 can be set. [I guess this is for FS, so it can't have a whole lot
> of other inputs, but gl_FragCoord comes to mind. Perhaps that counts
> against the varying max.]

It does look fishy. That's probably a question for Ken, a real mesa
person though.

-- 
Damien


More information about the mesa-dev mailing list