[Mesa-dev] [PATCH v4 1/1] clover: Wait for requested operation if blocking flag is set
Aaron Watry
awatry at gmail.com
Tue Sep 19 17:04:39 UTC 2017
On Fri, Sep 15, 2017 at 8:46 PM, Jan Vesely <jan.vesely at rutgers.edu> wrote:
> On Thu, 2017-09-14 at 14:39 -0700, Francisco Jerez wrote:
>> Jan Vesely <jan.vesely at rutgers.edu> writes:
>>
>> > On Mon, 2017-09-04 at 13:23 -0700, Francisco Jerez wrote:
>> > > Jan Vesely <jan.vesely at rutgers.edu> writes:
>> > >
>> > > > v2: wait in map_buffer and map_image as well
>> > > > v3: use event::wait instead of wait (skips fence wait for hard_event)
>> > > > v4: use wait_signalled()
>> > > >
>> > > > Signed-off-by: Jan Vesely <jan.vesely at rutgers.edu>
>> > > > ---
>> > > > Hi Francisco,
>> > > >
>> > > > once again sorry for the delay, and thanks for you patience.
>> > > > This patch applies on top of the two you attached during our email
>> > > > discussion.
>> > > > From what I can tell, the functionality is identical to v3 after your
>> > > > two patches are applied ("event:wait()" calls "wait_signalled()"), but I
>> > > > suppose calling non-virtual function is preferrable. if not, feel free
>> > > > to use v3.
>> > > >
>> > >
>> > > Yeah, I find v4 more readable than calling the base class'
>> > > implementation of wait(). Patch is:
>> > >
>> > > Reviewed-by: Francisco Jerez <currojerez at riseup.net>
>> >
>> > thanks. will you include with the other 2 patches, or should I push it
>> > separately after those 2 are in?
>> >
>>
>> I don't have review tags for the other two, but assuming they get your
>> R-b feel free to push all the three patches yourself.
>
> they look good to me, now that I understand how they work.
> although I'd prefer external review given how long it took me.
>
> @Aaron, do you mind taking a look at the two attached patches?
I had taken a look at them previously when I tried to implement a
semi-efficient blocking enqueue[Map|Read|Write]. They look ok to me.
I'd at least say add my Acked-by on Francisco's patches, and
Reviewed-by for yours (this one).
--Aaron
>
> thanks,
> Jan
>
>>
>> > regards,
>> > Jan
>> >
>> > >
>> > > Thanks.
>> > >
>> > > > thanks,
>> > > > Jan
>> > > >
>> > > > src/gallium/state_trackers/clover/api/transfer.cpp | 30 ++++++++++++++++++++--
>> > > > 1 file changed, 28 insertions(+), 2 deletions(-)
>> > > >
>> > > > diff --git a/src/gallium/state_trackers/clover/api/transfer.cpp b/src/gallium/state_trackers/clover/api/transfer.cpp
>> > > > index f7046253be..34559042ae 100644
>> > > > --- a/src/gallium/state_trackers/clover/api/transfer.cpp
>> > > > +++ b/src/gallium/state_trackers/clover/api/transfer.cpp
>> > > > @@ -295,6 +295,9 @@ clEnqueueReadBuffer(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > > &mem, obj_origin, obj_pitch,
>> > > > region));
>> > > >
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > ret_object(rd_ev, hev);
>> > > > return CL_SUCCESS;
>> > > >
>> > > > @@ -325,6 +328,9 @@ clEnqueueWriteBuffer(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > > ptr, {}, obj_pitch,
>> > > > region));
>> > > >
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > ret_object(rd_ev, hev);
>> > > > return CL_SUCCESS;
>> > > >
>> > > > @@ -362,6 +368,9 @@ clEnqueueReadBufferRect(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > > &mem, obj_origin, obj_pitch,
>> > > > region));
>> > > >
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > ret_object(rd_ev, hev);
>> > > > return CL_SUCCESS;
>> > > >
>> > > > @@ -399,6 +408,9 @@ clEnqueueWriteBufferRect(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > > ptr, host_origin, host_pitch,
>> > > > region));
>> > > >
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > ret_object(rd_ev, hev);
>> > > > return CL_SUCCESS;
>> > > >
>> > > > @@ -504,6 +516,9 @@ clEnqueueReadImage(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > > &img, src_origin, src_pitch,
>> > > > region));
>> > > >
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > ret_object(rd_ev, hev);
>> > > > return CL_SUCCESS;
>> > > >
>> > > > @@ -538,6 +553,9 @@ clEnqueueWriteImage(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > > ptr, {}, src_pitch,
>> > > > region));
>> > > >
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > ret_object(rd_ev, hev);
>> > > > return CL_SUCCESS;
>> > > >
>> > > > @@ -667,7 +685,11 @@ clEnqueueMapBuffer(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > >
>> > > > void *map = mem.resource(q).add_map(q, flags, blocking, obj_origin, region);
>> > > >
>> > > > - ret_object(rd_ev, create<hard_event>(q, CL_COMMAND_MAP_BUFFER, deps));
>> > > > + auto hev = create<hard_event>(q, CL_COMMAND_MAP_BUFFER, deps);
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > + ret_object(rd_ev, hev);
>> > > > ret_error(r_errcode, CL_SUCCESS);
>> > > > return map;
>> > > >
>> > > > @@ -695,7 +717,11 @@ clEnqueueMapImage(cl_command_queue d_q, cl_mem d_mem, cl_bool blocking,
>> > > >
>> > > > void *map = img.resource(q).add_map(q, flags, blocking, origin, region);
>> > > >
>> > > > - ret_object(rd_ev, create<hard_event>(q, CL_COMMAND_MAP_IMAGE, deps));
>> > > > + auto hev = create<hard_event>(q, CL_COMMAND_MAP_IMAGE, deps);
>> > > > + if (blocking)
>> > > > + hev().wait_signalled();
>> > > > +
>> > > > + ret_object(rd_ev, hev);
>> > > > ret_error(r_errcode, CL_SUCCESS);
>> > > > return map;
>> > > >
>> > > > --
>> > > > 2.13.5
>>
>> _______________________________________________
>> mesa-dev mailing list
>> mesa-dev at lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
More information about the mesa-dev
mailing list