----- Original Message ----- > From: Dave Airlie <[email protected]> > > Not sure if this makes a difference or not, but either its inconsistent > for some undocumented reason or its just a bug.
PIPE_TRANSFER_UNSYNCHRONIZED only makes sense when one maps for the 2nd time (or after). It really makes no sense to set UNSYNCRHONIZED with a buffer that was just created, as the map can't block (i.e, there is no drawing to synchronize with anyway). UNSYNCHRONIZED was actually set here before, and exposed a bug in svga pipe driver because it is such an unexpected thing to do. Of course, the svga pipe driver bug was fixed -- as apps could make the same mistake with mapbufferrange --, but we decided not to set the flag anymore just in case it caused problems elsewhere. But I suppose that the asymmetry causes suspicion, so maybe a comment would be a good idea indeed. Jose > Signed-off-by: Dave Airlie <[email protected]> > --- > src/gallium/auxiliary/util/u_upload_mgr.c | 3 ++- > 1 file changed, 2 insertions(+), 1 deletion(-) > > diff --git a/src/gallium/auxiliary/util/u_upload_mgr.c > b/src/gallium/auxiliary/util/u_upload_mgr.c > index 6859751..de448e0 100644 > --- a/src/gallium/auxiliary/util/u_upload_mgr.c > +++ b/src/gallium/auxiliary/util/u_upload_mgr.c > @@ -138,7 +138,8 @@ u_upload_alloc_buffer( struct u_upload_mgr *upload, > upload->map = pipe_buffer_map_range(upload->pipe, upload->buffer, > 0, size, > PIPE_TRANSFER_WRITE | > - PIPE_TRANSFER_FLUSH_EXPLICIT, > + PIPE_TRANSFER_FLUSH_EXPLICIT | > + PIPE_TRANSFER_UNSYNCHRONIZED, > &upload->transfer); > if (upload->map == NULL) { > upload->transfer = NULL; > -- > 1.8.1.2 > > _______________________________________________ > mesa-dev mailing list > [email protected] > http://lists.freedesktop.org/mailman/listinfo/mesa-dev > _______________________________________________ mesa-dev mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/mesa-dev
