[dev.regabi] runtime/cgo: call setg_gcc in crosscall_amd64

Currently, when using cgo, the g pointer is set via a separate
call to setg_gcc or with inline assembly in threadentry. This CL
changes it to call setg_gcc in crosscall_amd64, like other g-
register platforms. When we have an actual g register on AMD64,
we'll need to set the register immediately before calling into
Go.

Change-Id: Ib1171e05cd0dabba3b7d12e072084d141051cf3d
Reviewed-on: https://go-review.googlesource.com/c/go/+/289192
Trust: Cherry Zhang <cherryyz@google.com>
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Reviewed-by: Than McIntosh <thanm@google.com>

src/runtime/cgo/gcc_amd64.S

@@ -30,9 +30,14 @@ EXT(crosscall_amd64):
 	pushq %r15
 
 #if defined(_WIN64)
+	movq %r8, %rdi	/* arg of setg_gcc */
+	call *%rdx	/* setg_gcc */
 	call *%rcx	/* fn */
 #else
-	call *%rdi	/* fn */
+	movq %rdi, %rbx
+	movq %rdx, %rdi	/* arg of setg_gcc */
+	call *%rsi	/* setg_gcc */
+	call *%rbx	/* fn */
 #endif
 
 	popq %r15

src/runtime/cgo/gcc_darwin_amd64.c

@@ -9,13 +9,16 @@
 #include "libcgo_unix.h"
 
 static void* threadentry(void*);
+static void (*setg_gcc)(void*);
 
 void
-x_cgo_init(G *g)
+x_cgo_init(G *g, void (*setg)(void*), void **tlsg, void **tlsbase)
 {
 	pthread_attr_t attr;
 	size_t size;
 
+	setg_gcc = setg;
+
 	pthread_attr_init(&attr);
 	pthread_attr_getstacksize(&attr, &size);
 	g->stacklo = (uintptr)&attr - size + 4096;
@@ -57,10 +60,6 @@ threadentry(void *v)
 	ts = *(ThreadStart*)v;
 	free(v);
 
-	// Move the g pointer into the slot reserved in thread local storage.
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
-	// Constant must match the one in cmd/link/internal/ld/sym.go.
-	asm volatile("movq %0, %%gs:0x30" :: "r"(ts.g));
-
-	crosscall_amd64(ts.fn);
 	return nil;
 }

src/runtime/cgo/gcc_dragonfly_amd64.c

@@ -61,11 +61,6 @@ threadentry(void *v)
 	ts = *(ThreadStart*)v;
 	free(v);
 
-	/*
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
-	 * Set specific keys.
-	 */
-	setg_gcc((void*)ts.g);
-
-	crosscall_amd64(ts.fn);
 	return nil;
 }

src/runtime/cgo/gcc_freebsd_amd64.c

@@ -69,11 +69,6 @@ threadentry(void *v)
 	free(v);
 	_cgo_tsan_release();
 
-	/*
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
-	 * Set specific keys.
-	 */
-	setg_gcc((void*)ts.g);
-
-	crosscall_amd64(ts.fn);
 	return nil;
 }

src/runtime/cgo/gcc_linux_amd64.c

@@ -89,11 +89,6 @@ threadentry(void *v)
 	free(v);
 	_cgo_tsan_release();
 
-	/*
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
-	 * Set specific keys.
-	 */
-	setg_gcc((void*)ts.g);
-
-	crosscall_amd64(ts.fn);
 	return nil;
 }

src/runtime/cgo/gcc_netbsd_amd64.c

@@ -62,11 +62,6 @@ threadentry(void *v)
 	ts = *(ThreadStart*)v;
 	free(v);
 
-	/*
-	 * Set specific keys.
-	 */
-	setg_gcc((void*)ts.g);
-
 	// On NetBSD, a new thread inherits the signal stack of the
 	// creating thread. That confuses minit, so we remove that
 	// signal stack here before calling the regular mstart. It's
@@ -78,6 +73,6 @@ threadentry(void *v)
 	ss.ss_flags = SS_DISABLE;
 	sigaltstack(&ss, nil);
 
-	crosscall_amd64(ts.fn);
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
 	return nil;
 }

src/runtime/cgo/gcc_openbsd_amd64.c

@@ -60,11 +60,6 @@ threadentry(void *v)
 	ts = *(ThreadStart*)v;
 	free(v);
 
-	/*
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
-	 * Set specific keys.
-	 */
-	setg_gcc((void*)ts.g);
-
-	crosscall_amd64(ts.fn);
 	return nil;
 }

src/runtime/cgo/gcc_solaris_amd64.c

@@ -72,11 +72,6 @@ threadentry(void *v)
 	ts = *(ThreadStart*)v;
 	free(v);
 
-	/*
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
-	 * Set specific keys.
-	 */
-	setg_gcc((void*)ts.g);
-
-	crosscall_amd64(ts.fn);
 	return nil;
 }

src/runtime/cgo/gcc_windows_amd64.c

@@ -12,10 +12,12 @@
 #include "libcgo_windows.h"
 
 static void threadentry(void*);
+static void (*setg_gcc)(void*);
 
 void
-x_cgo_init(G *g)
+x_cgo_init(G *g, void (*setg)(void*), void **tlsg, void **tlsbase)
 {
+	setg_gcc = setg;
 }
 
 
@@ -46,10 +48,8 @@ threadentry(void *v)
 	 */
 	asm volatile (
 	  "movq %0, %%gs:0x28\n"	// MOVL tls0, 0x28(GS)
-	  "movq %%gs:0x28, %%rax\n" // MOVQ 0x28(GS), tmp
+	  :: "r"(ts.tls)
-	  "movq %1, 0(%%rax)\n" // MOVQ g, 0(GS)
-	  :: "r"(ts.tls), "r"(ts.g) : "%rax"
 	);
 
-	crosscall_amd64(ts.fn);
+	crosscall_amd64(ts.fn, setg_gcc, (void*)ts.g);
 }

src/runtime/cgo/libcgo.h

@@ -66,7 +66,7 @@ uintptr_t _cgo_wait_runtime_init_done(void);
 /*
  * Call fn in the 6c world.
  */
-void crosscall_amd64(void (*fn)(void));
+void crosscall_amd64(void (*fn)(void), void (*setg_gcc)(void*), void *g);
 
 /*
  * Call fn in the 8c world.