Re: [PATCH 1/2] lib/intel_blt: BB index should be incremented in dwords

["Vivekanandan, Balasubramani" <balasubramani.vivekanandan@intel.com>]

On 10.11.2025 06:48, Dixit, Ashutosh wrote:
> On Mon, 10 Nov 2025 05:58:06 -0800, Balasubramani Vivekanandan wrote:
> >
> > Size in bytes in used to increment the BB index, which is incorrect.
> > Convert bytes to number of dwords.
> >
> > Signed-off-by: Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com>
> > ---
> >  lib/intel_blt.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/lib/intel_blt.c b/lib/intel_blt.c
> > index 7f7f5485e..9a0bcb96b 100644
> > --- a/lib/intel_blt.c
> > +++ b/lib/intel_blt.c
> > @@ -1714,7 +1714,7 @@ uint64_t emit_blt_fast_copy(int fd,
> >
> >	igt_assert(bb_pos + sizeof(data) < blt->bb.size);
> >	memcpy(bb + bb_pos, &data, sizeof(data));
> 
> What about here?
> 
> 
> > -	bb_pos += sizeof(data);
> > +	bb_pos += DIV_ROUND_UP(sizeof(data), sizeof(*bb));
> >
> >	if (emit_bbe) {
> >		igt_assert(bb_pos + sizeof(uint32_t) < blt->bb.size);
> 
> Here also has:
> 
> 	bb_pos += sizeof(uint32_t);
> 
> And the function is returning bb_pos and can be called multiple times with
> the returned bb_pos.

Right, fix is required at multiple places. I will fix those in the next
revision.

Regards,
Bala

> 
> > --
> > 2.43.0
> >
> 
> So there are multiple places where bb_pos is being added to bb. Maybe the
> fix should be to just make bb a uint8_t * rather than uint32 * ?