On Thu, Nov 14, 2013 at 04:11:38PM +0100, Daniel Vetter wrote: > v2: Fix more. > > Cc: Ben Widawsky <ben@xxxxxxxxxxxx> > Signed-off-by: Daniel Vetter <daniel.vetter@xxxxxxxx> > --- > lib/intel_batchbuffer.c | 4 +++- > tests/gem_cs_prefetch.c | 3 --- > 2 files changed, 3 insertions(+), 4 deletions(-) > > diff --git a/lib/intel_batchbuffer.c b/lib/intel_batchbuffer.c > index e284d4870cd6..06a543728579 100644 > --- a/lib/intel_batchbuffer.c > +++ b/lib/intel_batchbuffer.c > @@ -239,7 +239,9 @@ intel_blt_copy(struct intel_batchbuffer *batch, > CHECK_RANGE(src_pitch) && CHECK_RANGE(dst_pitch)); > #undef CHECK_RANGE > > - BLIT_COPY_BATCH_START(batch->devid, cmd_bits); > + BEGIN_BATCH(intel_gen(batch->devid) >= 8 ? 10 : 8); > + OUT_BATCH(XY_SRC_COPY_BLT_CMD | cmd_bits | > + (intel_gen(batch->devid) >= 8 ? 8 : 6)); > OUT_BATCH((br13_bits) | > (0xcc << 16) | /* copy ROP */ > dst_pitch); I'd vote to just move this into the BLIT_COPY_BATCH_START macro. We already have the devid. > diff --git a/tests/gem_cs_prefetch.c b/tests/gem_cs_prefetch.c > index 229a385a2fcd..eb515eadca2e 100644 > --- a/tests/gem_cs_prefetch.c > +++ b/tests/gem_cs_prefetch.c > @@ -133,9 +133,6 @@ int main(int argc, char **argv) > /* copy the sample batch with the gpu to the new one, so that we > * also test the unmappable part of the gtt. */ > BLIT_COPY_BATCH_START(batch->devid, 0); > - OUT_BATCH(XY_SRC_COPY_BLT_CMD | > - XY_SRC_COPY_BLT_WRITE_ALPHA | > - XY_SRC_COPY_BLT_WRITE_RGB); > OUT_BATCH((3 << 24) | /* 32 bits */ > (0xcc << 16) | /* copy ROP */ > 4096); This hunk was already fixed upstream. > -- > 1.8.1.4 > -- Ben Widawsky, Intel Open Source Technology Center _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
