>
> >
> > CLANG warning: "Value stored to 'ref_limit' is never read"
> >
> > Commenting out since there is a ToDo that refers to ref_limit
> >
> > Found by Covscan.
> >
> > Signed-off-by: Uri Lublin <uril@xxxxxxxxxx>
> > ---
> >
> > Should the comment be deleted too ?
> >
>
> I think would be time to revise the comment and maybe the code.
>
> The comment is (I suppose):
> while (ip < ip_bound) { // TODO: maybe separate a run from the same seg
> or from
> // different ones in order to spare ref <
> ref_limit
>
> but this came from commit:
> commit c1b79eb035fa158fb2ac3bc8e559809611070016
> Author: Yaniv Kamay <ykamay@xxxxxxxxxx>
> Date: Sat Sep 19 21:25:46 2009 +0300
>
> fresh start
>
> with no further history :-(
> Is the former history somewhere?
>
> Something is suggesting me that the loop was "while (ref < ref_limit)"
> instead of the current "while (ip < ip_bound)". Also wondering why
> ip_bound is ref_limit (seg->lines_end) minus 2 pixels and so why RLE
> check ignores last 2 pixels. But I have similar questions for ip_limit.
>
> > ---
> > server/glz-encode.tmpl.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/server/glz-encode.tmpl.c b/server/glz-encode.tmpl.c
> > index 48bab50bc..ad48c86c8 100644
> > --- a/server/glz-encode.tmpl.c
> > +++ b/server/glz-encode.tmpl.c
> > @@ -282,7 +282,7 @@ static void FNAME(compress_seg)(Encoder *encoder,
> > uint32_t seg_idx, PIXEL *from,
> >
> > ip += 3;
> > ref = anchor + 2;
> > - ref_limit = (PIXEL *)(seg->lines_end);
> > + //ref_limit = (PIXEL *)(seg->lines_end);
> > len = 3;
> >
> > x = *ref;
>
I realized that GLZ is just LZ with the dictionary shared between
multiple images. So I looked at LZ code (spice-common) to find the
missing clues.
In the LZ code that part is used for both RLE and dictionary. So the
loop in LZ code (common/lz_compress_tmpl.c) is
while ((ip < ip_bound) && (ref < ref_limit)) { // TODO: maybe separate a run from
// the same seg or from different
// ones in order to spare
// ref < ref_limit
so here the comment make sense. However this was copied verbatim
in GLZ but ref < ref_limit check was removed as the same segment is used.
I suppose that the LZ code having ref == ip - 1 in the RLE case is comparing
an additional pixel leaving out only 1 pixel but this is OT for this case.
I'll write a patch (will be more time consuming to write the comment!).
Frediano
_______________________________________________
Spice-devel mailing list
Spice-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/spice-devel