[PATCH libdrm 4/8] xf86drmSL: Fix neighbour printing

[Jan Vesely]

On Fri, 2015-03-20 at 17:38 +0000, Emil Velikov wrote:
> On 27/02/15 18:07, Jan Vesely wrote:
> > Signed-off-by: Jan Vesely <jan.vesely at rutgers.edu>
> > ---
> >  xf86drmSL.c | 7 +++++--
> >  1 file changed, 5 insertions(+), 2 deletions(-)
> > 
> > diff --git a/xf86drmSL.c b/xf86drmSL.c
> > index acddb54..2160bb8 100644
> > --- a/xf86drmSL.c
> > +++ b/xf86drmSL.c
> > @@ -266,11 +266,14 @@ int drmSLLookupNeighbors(void *l, unsigned long key,
> >      SkipListPtr   list = (SkipListPtr)l;
> >      SLEntryPtr    update[SL_MAX_LEVEL + 1];
> >      int           retcode = 0;
> > +    SLEntryPtr    entry;
> > +
> > +    entry = SLLocate(list, key, update);
> >  
> >      *prev_key   = *next_key   = key;
> >      *prev_value = *next_value = NULL;
> > -	
> > -    if (update[0]) {
> > +
> > +    if (entry && update[0]) {
> From a very brief look at git log, the entry check should not be needed.
> Must admit that I've not looked at all in the implementation of either
> SLLocate or drmSLLookupNeighbors.

SLLocate might return early and leave the array uninitialized. All other
calls to it check the return value. I guess the warning that the
previous commit tried to fix was "set-but-unused" variable.

> 
> That said it seems that none of the three files
> (xf86drm{SL,Hash,Random}) has been build as a program for a while. Maybe
> we could split it out as a standalone test and let it churn at make
> check time ?

Sounds like a good idea. I'll try to take a look when time permits, but
I'd leave that as a separate patch.

thanks,
jan

> 
> Cheers,
> Emil

-- 
Jan Vesely <jan.vesely at rutgers.edu>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: This is a digitally signed message part
URL: <http://lists.freedesktop.org/archives/dri-devel/attachments/20150320/79550235/attachment-0001.sig>