Bug fix: events: wrongly decoded ~external_ flag

lukstafi · Nov 21, 2024 · 9480850 · 9480850
1 parent 7689555
commit 9480850
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 5 deletions.
diff --git a/CHANGES.md b/CHANGES.md
@@ -1,12 +1,19 @@
-## [0.6.0] 2024-10-31
+## [0.6.1] current
+
+### Fixed
+
+- Docu-comment typo.
+- The flags `cu_event_wait_external` and `cu_event_wait_default` were switched around for `record ?external_` and `wait ?external_` event functions.
+
+## [0.6.0] 2024-11-01
 
 ### Added
 
 - `get_free_and_total_mem`.
 - Multiple missing `sexp_of` conversions.
 - `cuda_call_hook` to help in debugging.
 - `is_success` functions.
-- `mem_alloc_async` and `mem_free_async`. `mem_free_async` is attached as a finalizer by `mem_alloc_async` (with stream capture).
+- `mem_alloc_async` i.e. `Stream.mem_alloc`, and `mem_free_async` i.e. `Stream.mem_free`. `Stream.mem_free` is attached as a finalizer by `Stream.mem_alloc` (with stream capture).
 
 ### Changed
 

diff --git a/cudajit.ml b/cudajit.ml
@@ -1810,7 +1810,7 @@ module Event = struct
     let open Cuda_ffi.Types_generated in
     let flags =
       Unsigned.UInt.of_int64
-      @@ if external_ then cu_event_record_default else cu_event_record_external
+      @@ if external_ then cu_event_record_external else cu_event_record_default
     in
     check "cu_event_record_with_flags" @@ Cuda.cu_event_record_with_flags event stream.stream flags
 
@@ -1819,7 +1819,7 @@ module Event = struct
   let wait ?(external_ = false) stream event =
     let open Cuda_ffi.Types_generated in
     let flags =
-      Unsigned.UInt.of_int64 @@ if external_ then cu_event_wait_default else cu_event_wait_external
+      Unsigned.UInt.of_int64 @@ if external_ then cu_event_wait_external else cu_event_wait_default
     in
     check "cu_stream_wait_event" @@ Cuda.cu_stream_wait_event stream.stream event flags
 end

diff --git a/cudajit.mli b/cudajit.mli
@@ -679,7 +679,9 @@ module Stream : sig
     t ->
     unit
   (** Copies from the device memory into the bigarray (or its interval) asynchronously.
-      [host_offset] and [length] are in numbers of elements. See {!memcpy_D_to_H_async_unsafe}. *)
+      [host_offset] and [length] are in numbers of elements. See {!memcpy_D_to_H_unsafe} and
+      {{:https://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__MEM.html#group__CUDA__MEM_1g56f30236c7c5247f8e061b59d3268362}
+        cuMemcpyDtoHAsync}. *)
 
   val memcpy_D_to_D :
     ?kind:('a, 'b) Bigarray.kind ->