[libc] Implement (v|f)printf on the GPU

libc/config/gpu/entrypoints.txt

@@ -177,6 +177,10 @@ set(TARGET_LIBC_ENTRYPOINTS
     # stdio.h entrypoints
     libc.src.stdio.clearerr
     libc.src.stdio.fclose
+    libc.src.stdio.printf
+    libc.src.stdio.vprintf
+    libc.src.stdio.fprintf
+    libc.src.stdio.vfprintf
     libc.src.stdio.sprintf
     libc.src.stdio.snprintf
     libc.src.stdio.vsprintf

libc/include/llvm-libc-types/rpc_opcodes_t.h

@@ -34,6 +34,9 @@ typedef enum {
   RPC_PRINTF_TO_STDOUT,
   RPC_PRINTF_TO_STDERR,
   RPC_PRINTF_TO_STREAM,
+  RPC_PRINTF_TO_STDOUT_PACKED,
+  RPC_PRINTF_TO_STDERR_PACKED,
+  RPC_PRINTF_TO_STREAM_PACKED,
   RPC_REMOVE,
   RPC_LAST = 0xFFFF,
 } rpc_opcode_t;

libc/src/__support/arg_list.h

@@ -19,6 +19,11 @@
 namespace LIBC_NAMESPACE_DECL {
 namespace internal {
 
+template <typename V, typename A>
+LIBC_INLINE constexpr V align_up(V val, A align) {
+  return ((val + V(align) - 1) / V(align)) * V(align);
+}
+
 class ArgList {
   va_list vlist;
 
@@ -55,7 +60,34 @@ class MockArgList {
   }
 
   template <class T> LIBC_INLINE T next_var() {
-    ++arg_counter;
+    arg_counter++;
+    return T(arg_counter);
+  }
+
+  size_t read_count() const { return arg_counter; }
+};
+
+// Used by the GPU implementation to parse how many bytes need to be read from
+// the variadic argument buffer.
+template <bool packed> class DummyArgList {
+  size_t arg_counter = 0;
+
+public:
+  LIBC_INLINE DummyArgList() = default;
+  LIBC_INLINE DummyArgList(va_list) { ; }
+  LIBC_INLINE DummyArgList(DummyArgList &other) {
+    arg_counter = other.arg_counter;
+  }
+  LIBC_INLINE ~DummyArgList() = default;
+
+  LIBC_INLINE DummyArgList &operator=(DummyArgList &rhs) {
+    arg_counter = rhs.arg_counter;
+    return *this;
+  }
+
+  template <class T> LIBC_INLINE T next_var() {
+    arg_counter = packed ? arg_counter + sizeof(T)
+                         : align_up(arg_counter, alignof(T)) + sizeof(T);
     return T(arg_counter);
   }
 
@@ -64,7 +96,7 @@ class MockArgList {
 
 // Used for the GPU implementation of `printf`. This models a variadic list as a
 // simple array of pointers that are built manually by the implementation.
-class StructArgList {
+template <bool packed> class StructArgList {
   void *ptr;
   void *end;
 
@@ -86,15 +118,18 @@ class StructArgList {
   LIBC_INLINE void *get_ptr() const { return ptr; }
 
   template <class T> LIBC_INLINE T next_var() {
-    ptr = reinterpret_cast<void *>(
-        ((reinterpret_cast<uintptr_t>(ptr) + alignof(T) - 1) / alignof(T)) *
-        alignof(T));
-
+    if (!packed)
+      ptr = reinterpret_cast<void *>(
+          align_up(reinterpret_cast<uintptr_t>(ptr), alignof(T)));
     if (ptr >= end)
       return T(-1);
 
-    T val = *reinterpret_cast<T *>(ptr);
-    ptr = reinterpret_cast<unsigned char *>(ptr) + sizeof(T);
+    // Memcpy because pointer alignment may be illegal given a packed struct.
+    T val;
+    __builtin_memcpy(&val, ptr, sizeof(T));
+
+    ptr =
+        reinterpret_cast<void *>(reinterpret_cast<uintptr_t>(ptr) + sizeof(T));
     return val;
   }
 };

libc/src/gpu/rpc_fprintf.cpp

@@ -30,6 +30,9 @@ int fprintf_impl(::FILE *__restrict file, const char *__restrict format,
   }
 
   port.send_n(format, format_size);
+  port.recv([&](rpc::Buffer *buffer) {
+    args_size = static_cast<size_t>(buffer->data[0]);
+  });
   port.send_n(args, args_size);
 
   uint32_t ret = 0;
@@ -51,7 +54,7 @@ int fprintf_impl(::FILE *__restrict file, const char *__restrict format,
   return ret;
 }
 
-// TODO: This is a stand-in function that uses a struct pointer and size in
+// TODO: Delete this and port OpenMP to use `printf`.
 // place of varargs. Once varargs support is added we will use that to
 // implement the real version.
 LLVM_LIBC_FUNCTION(int, rpc_fprintf,

libc/src/stdio/CMakeLists.txt

@@ -163,18 +163,6 @@ add_entrypoint_object(
     libc.src.stdio.printf_core.writer
 )
 
-add_entrypoint_object(
-  fprintf
-  SRCS
-    fprintf.cpp
-  HDRS
-    fprintf.h
-  DEPENDS
-    libc.hdr.types.FILE
-    libc.src.__support.arg_list
-    libc.src.stdio.printf_core.vfprintf_internal
-)
-
 add_entrypoint_object(
   vsprintf
   SRCS
@@ -197,18 +185,6 @@ add_entrypoint_object(
     libc.src.stdio.printf_core.writer
 )
 
-add_entrypoint_object(
-  vfprintf
-  SRCS
-    vfprintf.cpp
-  HDRS
-    vfprintf.h
-  DEPENDS
-    libc.hdr.types.FILE
-    libc.src.__support.arg_list
-    libc.src.stdio.printf_core.vfprintf_internal
-)
-
 add_subdirectory(printf_core)
 add_subdirectory(scanf_core)
 
@@ -258,6 +234,7 @@ add_stdio_entrypoint_object(fputc)
 add_stdio_entrypoint_object(putc)
 add_stdio_entrypoint_object(putchar)
 add_stdio_entrypoint_object(printf)
+add_stdio_entrypoint_object(fprintf)
 add_stdio_entrypoint_object(fgetc)
 add_stdio_entrypoint_object(fgetc_unlocked)
 add_stdio_entrypoint_object(getc)
@@ -270,3 +247,4 @@ add_stdio_entrypoint_object(stdin)
 add_stdio_entrypoint_object(stdout)
 add_stdio_entrypoint_object(stderr)
 add_stdio_entrypoint_object(vprintf)
+add_stdio_entrypoint_object(vfprintf)

libc/src/stdio/generic/CMakeLists.txt

@@ -363,19 +363,6 @@ add_entrypoint_object(
     libc.src.__support.File.platform_file
 )
 
-list(APPEND printf_deps
-      libc.src.__support.arg_list
-      libc.src.stdio.printf_core.vfprintf_internal
-)
-
-if(LLVM_LIBC_FULL_BUILD)
-  list(APPEND printf_deps
-      libc.src.__support.File.file
-      libc.src.__support.File.platform_file
-      libc.src.__support.File.platform_stdout
-  )
-endif()
-
 add_entrypoint_object(
   printf
   SRCS
@@ -396,6 +383,32 @@ add_entrypoint_object(
     ${printf_deps}
 )
 
+add_entrypoint_object(
+  fprintf
+  SRCS
+    fprintf.cpp
+  HDRS
+    ../fprintf.h
+  DEPENDS
+    libc.hdr.types.FILE
+    libc.src.__support.arg_list
+    libc.src.stdio.printf_core.vfprintf_internal
+    ${printf_deps}
+)
+
+add_entrypoint_object(
+  vfprintf
+  SRCS
+    vfprintf.cpp
+  HDRS
+    ../vfprintf.h
+  DEPENDS
+    libc.hdr.types.FILE
+    libc.src.__support.arg_list
+    libc.src.stdio.printf_core.vfprintf_internal
+    ${printf_deps}
+)
+
 add_entrypoint_object(
   fgets
   SRCS

libc/src/stdio/gpu/CMakeLists.txt

@@ -11,6 +11,14 @@ add_header_library(
     .stderr
 )
 
+add_header_library(
+  vfprintf_utils
+  HDRS
+    vfprintf_utils.h
+  DEPENDS
+    .gpu_file
+)
+
 add_entrypoint_object(
   feof
   SRCS
@@ -246,6 +254,46 @@ add_entrypoint_object(
     .gpu_file
 )
 
+add_entrypoint_object(
+  printf
+  SRCS
+    printf.cpp
+  HDRS
+    ../printf.h
+  DEPENDS
+    .vfprintf_utils
+)
+
+add_entrypoint_object(
+  vprintf
+  SRCS
+    vprintf.cpp
+  HDRS
+    ../vprintf.h
+  DEPENDS
+    .vfprintf_utils
+)
+
+add_entrypoint_object(
+  fprintf
+  SRCS
+    fprintf.cpp
+  HDRS
+    ../fprintf.h
+  DEPENDS
+    .vfprintf_utils
+)
+
+add_entrypoint_object(
+  vfprintf
+  SRCS
+    vfprintf.cpp
+  HDRS
+    ../vfprintf.h
+  DEPENDS
+    .vfprintf_utils
+)
+
 add_entrypoint_object(
   stdin
   SRCS

libc/src/stdio/gpu/fprintf.cpp

@@ -0,0 +1,31 @@
+//===-- GPU Implementation of fprintf -------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/stdio/fprintf.h"
+
+#include "src/__support/CPP/string_view.h"
+#include "src/__support/arg_list.h"
+#include "src/errno/libc_errno.h"
+#include "src/stdio/gpu/vfprintf_utils.h"
+
+#include <stdio.h>
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(int, fprintf,
+                   (::FILE *__restrict stream, const char *__restrict format,
+                    ...)) {
+  va_list vlist;
+  va_start(vlist, format);
+  cpp::string_view str_view(format);
+  int ret_val = vfprintf_internal(stream, format, str_view.size() + 1, vlist);
+  va_end(vlist);
+  return ret_val;
+}
+
+} // namespace LIBC_NAMESPACE

libc/src/stdio/gpu/printf.cpp

@@ -0,0 +1,29 @@
+//===-- GPU Implementation of printf --------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/stdio/printf.h"
+
+#include "src/__support/CPP/string_view.h"
+#include "src/__support/arg_list.h"
+#include "src/errno/libc_errno.h"
+#include "src/stdio/gpu/vfprintf_utils.h"
+
+#include <stdio.h>
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(int, printf, (const char *__restrict format, ...)) {
+  va_list vlist;
+  va_start(vlist, format);
+  cpp::string_view str_view(format);
+  int ret_val = vfprintf_internal(stdout, format, str_view.size() + 1, vlist);
+  va_end(vlist);
+  return ret_val;
+}
+
+} // namespace LIBC_NAMESPACE

libc/src/stdio/gpu/vfprintf.cpp

@@ -0,0 +1,28 @@
+//===-- GPU Implementation of vfprintf ------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "src/stdio/vfprintf.h"
+
+#include "src/__support/CPP/string_view.h"
+#include "src/__support/arg_list.h"
+#include "src/errno/libc_errno.h"
+#include "src/stdio/gpu/vfprintf_utils.h"
+
+#include <stdio.h>
+
+namespace LIBC_NAMESPACE {
+
+LLVM_LIBC_FUNCTION(int, vfprintf,
+                   (::FILE *__restrict stream, const char *__restrict format,
+                    va_list vlist)) {
+  cpp::string_view str_view(format);
+  int ret_val = vfprintf_internal(stream, format, str_view.size() + 1, vlist);
+  return ret_val;
+}
+
+} // namespace LIBC_NAMESPACE