syscall: implement rawVforkSyscall for remaining linux platforms

This allows the use of CLONE_VFORK and CLONE_VM for fork/exec, preventing
'fork/exec ...: cannot allocate memory' failures from occuring when attempting
to execute commands from a Go process that has a large memory footprint.
Additionally, this should reduce the latency of fork/exec on these platforms.

Fixes #31936

Change-Id: I4e28cf0763173145cacaa5340680dca9ff449305
Reviewed-on: https://go-review.googlesource.com/c/go/+/295849
Trust: Joel Sing <joel@sing.id.au>
Run-TryBot: Joel Sing <joel@sing.id.au>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

src/syscall/asm_linux_386.s

@@ -110,6 +110,26 @@ ok2:
 	MOVL	$0, err+36(FP)
 	RET
 
+// func rawVforkSyscall(trap, a1 uintptr) (r1, err uintptr)
+TEXT ·rawVforkSyscall(SB),NOSPLIT|NOFRAME,$0-16
+	MOVL	trap+0(FP), AX	// syscall entry
+	MOVL	a1+4(FP), BX
+	MOVL	$0, CX
+	MOVL	$0, DX
+	POPL	SI // preserve return address
+	INVOKE_SYSCALL
+	PUSHL	SI
+	CMPL	AX, $0xfffff001
+	JLS	ok
+	MOVL	$-1, r1+8(FP)
+	NEGL	AX
+	MOVL	AX, err+12(FP)
+	RET
+ok:
+	MOVL	AX, r1+8(FP)
+	MOVL	$0, err+12(FP)
+	RET
+
 // func rawSyscallNoError(trap uintptr, a1, a2, a3 uintptr) (r1, r2 uintptr);
 TEXT ·rawSyscallNoError(SB),NOSPLIT,$0-24
 	MOVL	trap+0(FP), AX	// syscall entry

src/syscall/asm_linux_amd64.s

@@ -108,7 +108,7 @@ ok2:
 	RET
 
 // func rawVforkSyscall(trap, a1 uintptr) (r1, err uintptr)
-TEXT ·rawVforkSyscall(SB),NOSPLIT,$0-32
+TEXT ·rawVforkSyscall(SB),NOSPLIT|NOFRAME,$0-32
 	MOVQ	a1+8(FP), DI
 	MOVQ	$0, SI
 	MOVQ	$0, DX

src/syscall/asm_linux_arm.s

@@ -154,6 +154,27 @@ ok1:
 	MOVW	R0, err+24(FP)
 	RET
 
+// func rawVforkSyscall(trap, a1 uintptr) (r1, err uintptr)
+TEXT ·rawVforkSyscall(SB),NOSPLIT|NOFRAME,$0-16
+	MOVW	trap+0(FP), R7	// syscall entry
+	MOVW	a1+4(FP), R0
+	MOVW	$0, R1
+	MOVW	$0, R2
+	SWI	$0
+	MOVW	$0xfffff001, R1
+	CMP	R1, R0
+	BLS	ok
+	MOVW	$-1, R1
+	MOVW	R1, r1+8(FP)
+	RSB	$0, R0, R0
+	MOVW	R0, err+12(FP)
+	RET
+ok:
+	MOVW	R0, r1+8(FP)
+	MOVW	$0, R0
+	MOVW	R0, err+12(FP)
+	RET
+
 // func rawSyscallNoError(trap uintptr, a1, a2, a3 uintptr) (r1, r2 uintptr);
 TEXT ·rawSyscallNoError(SB),NOSPLIT,$0-24
 	MOVW	trap+0(FP), R7	// syscall entry

src/syscall/asm_linux_arm64.s

@@ -125,7 +125,6 @@ ok:
 	MOVD	ZR, err+24(FP)	// errno
 	RET
 
-
 // func rawSyscallNoError(trap uintptr, a1, a2, a3 uintptr) (r1, r2 uintptr);
 TEXT ·rawSyscallNoError(SB),NOSPLIT,$0-48
 	MOVD	a1+8(FP), R0

src/syscall/asm_linux_mips64x.s

@@ -102,6 +102,26 @@ ok2:
 	MOVV	R0, err+72(FP)	// errno
 	RET
 
+// func rawVforkSyscall(trap, a1 uintptr) (r1, err uintptr)
+TEXT ·rawVforkSyscall(SB),NOSPLIT|NOFRAME,$0-32
+	MOVV	a1+8(FP), R4
+	MOVV	R0, R5
+	MOVV	R0, R6
+	MOVV	R0, R7
+	MOVV	R0, R8
+	MOVV	R0, R9
+	MOVV	trap+0(FP), R2	// syscall entry
+	SYSCALL
+	BEQ	R7, ok
+	MOVV	$-1, R1
+	MOVV	R1, r1+16(FP)	// r1
+	MOVV	R2, err+24(FP)	// errno
+	RET
+ok:
+	MOVV	R2, r1+16(FP)	// r1
+	MOVV	R0, err+24(FP)	// errno
+	RET
+
 TEXT ·rawSyscallNoError(SB),NOSPLIT,$0-48
 	MOVV	a1+8(FP), R4
 	MOVV	a2+16(FP), R5

src/syscall/asm_linux_mipsx.s

@@ -139,6 +139,23 @@ ok2:
 	MOVW	R0, err+36(FP)	// errno
 	RET
 
+// func rawVforkSyscall(trap, a1 uintptr) (r1, err uintptr)
+TEXT ·rawVforkSyscall(SB),NOSPLIT|NOFRAME,$0-16
+	MOVW	a1+4(FP), R4
+	MOVW	R0, R5
+	MOVW	R0, R6
+	MOVW	trap+0(FP), R2	// syscall entry
+	SYSCALL
+	BEQ	R7, ok
+	MOVW	$-1, R1
+	MOVW	R1, r1+8(FP)	// r1
+	MOVW	R2, err+12(FP)	// errno
+	RET
+ok:
+	MOVW	R2, r1+8(FP)	// r1
+	MOVW	R0, err+12(FP)	// errno
+	RET
+
 TEXT ·rawSyscallNoError(SB),NOSPLIT,$20-24
 	MOVW	a1+4(FP), R4
 	MOVW	a2+8(FP), R5

src/syscall/exec_linux.go

@@ -208,18 +208,12 @@ func forkAndExecInChild1(argv0 *byte, argv, envv []*byte, chroot, dir *byte, att
 		}
 	}
 
-	var hasRawVforkSyscall bool
-	switch runtime.GOARCH {
-	case "amd64", "arm64", "ppc64", "riscv64", "s390x":
-		hasRawVforkSyscall = true
-	}
-
 	// About to call fork.
 	// No more allocation or calls of non-assembly functions.
 	runtime_BeforeFork()
 	locked = true
 	switch {
-	case hasRawVforkSyscall && (sys.Cloneflags&CLONE_NEWUSER == 0 && sys.Unshareflags&CLONE_NEWUSER == 0):
+	case sys.Cloneflags&CLONE_NEWUSER == 0 && sys.Unshareflags&CLONE_NEWUSER == 0:
 		r1, err1 = rawVforkSyscall(SYS_CLONE, uintptr(SIGCHLD|CLONE_VFORK|CLONE_VM)|sys.Cloneflags)
 	case runtime.GOARCH == "s390x":
 		r1, _, err1 = RawSyscall6(SYS_CLONE, 0, uintptr(SIGCHLD)|sys.Cloneflags, 0, 0, 0, 0)

src/syscall/syscall_linux_386.go

@@ -387,6 +387,4 @@ func (cmsg *Cmsghdr) SetLen(length int) {
 	cmsg.Len = uint32(length)
 }
 
-func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno) {
-	panic("not implemented")
-}
+func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno)

src/syscall/syscall_linux_arm.go

@@ -236,6 +236,4 @@ func (cmsg *Cmsghdr) SetLen(length int) {
 	cmsg.Len = uint32(length)
 }
 
-func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno) {
-	panic("not implemented")
-}
+func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno)

src/syscall/syscall_linux_mips64x.go

@@ -214,6 +214,4 @@ func (cmsg *Cmsghdr) SetLen(length int) {
 	cmsg.Len = uint64(length)
 }
 
-func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno) {
-	panic("not implemented")
-}
+func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno)

src/syscall/syscall_linux_mipsx.go

@@ -224,6 +224,4 @@ func (cmsg *Cmsghdr) SetLen(length int) {
 	cmsg.Len = uint32(length)
 }
 
-func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno) {
-	panic("not implemented")
-}
+func rawVforkSyscall(trap, a1 uintptr) (r1 uintptr, err Errno)