// // Copyright 2012 Francisco Jerez // // Permission is hereby granted, free of charge, to any person obtaining a // copy of this software and associated documentation files (the "Software"), // to deal in the Software without restriction, including without limitation // the rights to use, copy, modify, merge, publish, distribute, sublicense, // and/or sell copies of the Software, and to permit persons to whom the // Software is furnished to do so, subject to the following conditions: // // The above copyright notice and this permission notice shall be included in // all copies or substantial portions of the Software. // // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR // IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, // FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL // THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, // WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF // OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE // SOFTWARE. // #include <cstring> #include "api/util.hpp" #include "core/event.hpp" #include "core/resource.hpp" using namespace clover; namespace { typedef resource::point point; /// /// Common argument checking shared by memory transfer commands. /// void validate_base(cl_command_queue q, cl_uint num_deps, const cl_event *deps) { if (!q) throw error(CL_INVALID_COMMAND_QUEUE); if (bool(num_deps) != bool(deps) || any_of(is_zero<cl_event>(), deps, deps + num_deps)) throw error(CL_INVALID_EVENT_WAIT_LIST); if (any_of([&](const cl_event ev) { return &ev->ctx != &q->ctx; }, deps, deps + num_deps)) throw error(CL_INVALID_CONTEXT); } /// /// Memory object-specific argument checking shared by most memory /// transfer commands. /// void validate_obj(cl_command_queue q, cl_mem obj) { if (!obj) throw error(CL_INVALID_MEM_OBJECT); if (&obj->ctx != &q->ctx) throw error(CL_INVALID_CONTEXT); } /// /// Class that encapsulates the task of mapping an object of type /// \a T. The return value of get() should be implicitly /// convertible to \a void *. /// template<typename T> struct __map; template<> struct __map<void *> { static void * get(cl_command_queue q, void *obj, cl_map_flags flags, size_t offset, size_t size) { return (char *)obj + offset; } }; template<> struct __map<const void *> { static const void * get(cl_command_queue q, const void *obj, cl_map_flags flags, size_t offset, size_t size) { return (const char *)obj + offset; } }; template<> struct __map<memory_obj *> { static mapping get(cl_command_queue q, memory_obj *obj, cl_map_flags flags, size_t offset, size_t size) { return { *q, obj->resource(q), flags, true, { offset }, { size }}; } }; /// /// Software copy from \a src_obj to \a dst_obj. They can be /// either pointers or memory objects. /// template<typename T, typename S> std::function<void (event &)> soft_copy_op(cl_command_queue q, T dst_obj, const point &dst_orig, const point &dst_pitch, S src_obj, const point &src_orig, const point &src_pitch, const point ®ion) { return [=](event &) { auto dst = __map<T>::get(q, dst_obj, CL_MAP_WRITE, dst_pitch(dst_orig), dst_pitch(region)); auto src = __map<S>::get(q, src_obj, CL_MAP_READ, src_pitch(src_orig), src_pitch(region)); point p; for (p[2] = 0; p[2] < region[2]; ++p[2]) { for (p[1] = 0; p[1] < region[1]; ++p[1]) { std::memcpy(static_cast<char *>(dst) + dst_pitch(p), static_cast<const char *>(src) + src_pitch(p), src_pitch[0] * region[0]); } } }; } /// /// Hardware copy from \a src_obj to \a dst_obj. /// template<typename T, typename S> std::function<void (event &)> hard_copy_op(cl_command_queue q, T dst_obj, const point &dst_orig, S src_obj, const point &src_orig, const point ®ion) { return [=](event &) { dst_obj->resource(q).copy(*q, dst_orig, region, src_obj->resource(q), src_orig); }; } } PUBLIC cl_int clEnqueueReadBuffer(cl_command_queue q, cl_mem obj, cl_bool blocking, size_t offset, size_t size, void *ptr, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { validate_base(q, num_deps, deps); validate_obj(q, obj); if (!ptr || offset > obj->size() || offset + size > obj->size()) throw error(CL_INVALID_VALUE); hard_event *hev = new hard_event( *q, CL_COMMAND_READ_BUFFER, { deps, deps + num_deps }, soft_copy_op(q, ptr, { 0 }, { 1 }, obj, { offset }, { 1 }, { size, 1, 1 })); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueWriteBuffer(cl_command_queue q, cl_mem obj, cl_bool blocking, size_t offset, size_t size, const void *ptr, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { validate_base(q, num_deps, deps); validate_obj(q, obj); if (!ptr || offset > obj->size() || offset + size > obj->size()) throw error(CL_INVALID_VALUE); hard_event *hev = new hard_event( *q, CL_COMMAND_WRITE_BUFFER, { deps, deps + num_deps }, soft_copy_op(q, obj, { offset }, { 1 }, ptr, { 0 }, { 1 }, { size, 1, 1 })); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueReadBufferRect(cl_command_queue q, cl_mem obj, cl_bool blocking, const size_t *obj_origin, const size_t *host_origin, const size_t *region, size_t obj_row_pitch, size_t obj_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, void *ptr, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { validate_base(q, num_deps, deps); validate_obj(q, obj); if (!ptr) throw error(CL_INVALID_VALUE); hard_event *hev = new hard_event( *q, CL_COMMAND_READ_BUFFER_RECT, { deps, deps + num_deps }, soft_copy_op(q, ptr, host_origin, { 1, host_row_pitch, host_slice_pitch }, obj, obj_origin, { 1, obj_row_pitch, obj_slice_pitch }, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueWriteBufferRect(cl_command_queue q, cl_mem obj, cl_bool blocking, const size_t *obj_origin, const size_t *host_origin, const size_t *region, size_t obj_row_pitch, size_t obj_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, const void *ptr, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { validate_base(q, num_deps, deps); validate_obj(q, obj); if (!ptr) throw error(CL_INVALID_VALUE); hard_event *hev = new hard_event( *q, CL_COMMAND_WRITE_BUFFER_RECT, { deps, deps + num_deps }, soft_copy_op(q, obj, obj_origin, { 1, obj_row_pitch, obj_slice_pitch }, ptr, host_origin, { 1, host_row_pitch, host_slice_pitch }, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueCopyBuffer(cl_command_queue q, cl_mem src_obj, cl_mem dst_obj, size_t src_offset, size_t dst_offset, size_t size, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { validate_base(q, num_deps, deps); validate_obj(q, src_obj); validate_obj(q, dst_obj); hard_event *hev = new hard_event( *q, CL_COMMAND_COPY_BUFFER, { deps, deps + num_deps }, hard_copy_op(q, dst_obj, { dst_offset }, src_obj, { src_offset }, { size, 1, 1 })); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueCopyBufferRect(cl_command_queue q, cl_mem src_obj, cl_mem dst_obj, const size_t *src_origin, const size_t *dst_origin, const size_t *region, size_t src_row_pitch, size_t src_slice_pitch, size_t dst_row_pitch, size_t dst_slice_pitch, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { validate_base(q, num_deps, deps); validate_obj(q, src_obj); validate_obj(q, dst_obj); hard_event *hev = new hard_event( *q, CL_COMMAND_COPY_BUFFER_RECT, { deps, deps + num_deps }, soft_copy_op(q, dst_obj, dst_origin, { 1, dst_row_pitch, dst_slice_pitch }, src_obj, src_origin, { 1, src_row_pitch, src_slice_pitch }, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueReadImage(cl_command_queue q, cl_mem obj, cl_bool blocking, const size_t *origin, const size_t *region, size_t row_pitch, size_t slice_pitch, void *ptr, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { image *img = dynamic_cast<image *>(obj); validate_base(q, num_deps, deps); validate_obj(q, img); if (!ptr) throw error(CL_INVALID_VALUE); hard_event *hev = new hard_event( *q, CL_COMMAND_READ_IMAGE, { deps, deps + num_deps }, soft_copy_op(q, ptr, {}, { 1, row_pitch, slice_pitch }, obj, origin, { 1, img->row_pitch(), img->slice_pitch() }, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueWriteImage(cl_command_queue q, cl_mem obj, cl_bool blocking, const size_t *origin, const size_t *region, size_t row_pitch, size_t slice_pitch, const void *ptr, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { image *img = dynamic_cast<image *>(obj); validate_base(q, num_deps, deps); validate_obj(q, img); if (!ptr) throw error(CL_INVALID_VALUE); hard_event *hev = new hard_event( *q, CL_COMMAND_WRITE_IMAGE, { deps, deps + num_deps }, soft_copy_op(q, obj, origin, { 1, img->row_pitch(), img->slice_pitch() }, ptr, {}, { 1, row_pitch, slice_pitch }, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueCopyImage(cl_command_queue q, cl_mem src_obj, cl_mem dst_obj, const size_t *src_origin, const size_t *dst_origin, const size_t *region, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { image *src_img = dynamic_cast<image *>(src_obj); image *dst_img = dynamic_cast<image *>(dst_obj); validate_base(q, num_deps, deps); validate_obj(q, src_img); validate_obj(q, dst_img); hard_event *hev = new hard_event( *q, CL_COMMAND_COPY_IMAGE, { deps, deps + num_deps }, hard_copy_op(q, dst_obj, dst_origin, src_obj, src_origin, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueCopyImageToBuffer(cl_command_queue q, cl_mem src_obj, cl_mem dst_obj, const size_t *src_origin, const size_t *region, size_t dst_offset, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { image *src_img = dynamic_cast<image *>(src_obj); validate_base(q, num_deps, deps); validate_obj(q, src_img); validate_obj(q, dst_obj); hard_event *hev = new hard_event( *q, CL_COMMAND_COPY_IMAGE_TO_BUFFER, { deps, deps + num_deps }, soft_copy_op(q, dst_obj, { dst_offset }, { 0, 0, 0 }, src_obj, src_origin, { 1, src_img->row_pitch(), src_img->slice_pitch() }, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC cl_int clEnqueueCopyBufferToImage(cl_command_queue q, cl_mem src_obj, cl_mem dst_obj, size_t src_offset, const size_t *dst_origin, const size_t *region, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { image *dst_img = dynamic_cast<image *>(src_obj); validate_base(q, num_deps, deps); validate_obj(q, src_obj); validate_obj(q, dst_img); hard_event *hev = new hard_event( *q, CL_COMMAND_COPY_BUFFER_TO_IMAGE, { deps, deps + num_deps }, soft_copy_op(q, dst_obj, dst_origin, { 1, dst_img->row_pitch(), dst_img->slice_pitch() }, src_obj, { src_offset }, { 0, 0, 0 }, region)); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); } PUBLIC void * clEnqueueMapBuffer(cl_command_queue q, cl_mem obj, cl_bool blocking, cl_map_flags flags, size_t offset, size_t size, cl_uint num_deps, const cl_event *deps, cl_event *ev, cl_int *errcode_ret) try { validate_base(q, num_deps, deps); validate_obj(q, obj); if (offset > obj->size() || offset + size > obj->size()) throw error(CL_INVALID_VALUE); void *map = obj->resource(q).add_map( *q, flags, blocking, { offset }, { size }); ret_object(ev, new hard_event(*q, CL_COMMAND_MAP_BUFFER, { deps, deps + num_deps })); ret_error(errcode_ret, CL_SUCCESS); return map; } catch (error &e) { ret_error(errcode_ret, e); return NULL; } PUBLIC void * clEnqueueMapImage(cl_command_queue q, cl_mem obj, cl_bool blocking, cl_map_flags flags, const size_t *origin, const size_t *region, size_t *row_pitch, size_t *slice_pitch, cl_uint num_deps, const cl_event *deps, cl_event *ev, cl_int *errcode_ret) try { image *img = dynamic_cast<image *>(obj); validate_base(q, num_deps, deps); validate_obj(q, img); void *map = obj->resource(q).add_map( *q, flags, blocking, origin, region); ret_object(ev, new hard_event(*q, CL_COMMAND_MAP_IMAGE, { deps, deps + num_deps })); ret_error(errcode_ret, CL_SUCCESS); return map; } catch (error &e) { ret_error(errcode_ret, e); return NULL; } PUBLIC cl_int clEnqueueUnmapMemObject(cl_command_queue q, cl_mem obj, void *ptr, cl_uint num_deps, const cl_event *deps, cl_event *ev) try { validate_base(q, num_deps, deps); validate_obj(q, obj); hard_event *hev = new hard_event( *q, CL_COMMAND_UNMAP_MEM_OBJECT, { deps, deps + num_deps }, [=](event &) { obj->resource(q).del_map(ptr); }); ret_object(ev, hev); return CL_SUCCESS; } catch (error &e) { return e.get(); }