[PATCH v5 16/16] riscv/cfi: Add __INDIRECT_RETURN attribute to swapcontext

[Gabriel Ravier]

On 6/28/26 9:02 AM, Jesse Huang wrote:
> ---
>   sysdeps/riscv/bits/indirect-return.h | 36 ++++++++++++++++++++++++++++
>   1 file changed, 36 insertions(+)
>   create mode 100644 sysdeps/riscv/bits/indirect-return.h
>
> diff --git a/sysdeps/riscv/bits/indirect-return.h b/sysdeps/riscv/bits/indirect-return.h
> new file mode 100644
> index 0000000000..1d8f658cc8
> --- /dev/null
> +++ b/sysdeps/riscv/bits/indirect-return.h
> @@ -0,0 +1,36 @@
> +/* Definition of __INDIRECT_RETURN.  RISC-V version.
> +   Copyright (C) 2026 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library; if not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +#ifndef _UCONTEXT_H
> +# error "Never include <bits/indirect-return.h> directly; use <ucontext.h> instead."
> +#endif
> +
> +/* __INDIRECT_RETURN indicates that swapcontext may return via
> +   an indirect branch.  This happens when GCS is enabled, so
> +   add the attribute if available, otherwise returns_twice has
> +   a similar effect, but it prevents some code transformations
> +   that can cause build failures in some rare cases so it is
> +   only used when GCS is enabled.  */
> +#if __glibc_has_attribute (__indirect_return__)
> +# define __INDIRECT_RETURN __attribute__ ((__indirect_return__))
> +#elif __glibc_has_attribute (__returns_twice__) \
> +      && defined __ARM_FEATURE_GCS_DEFAULT
> +# define __INDIRECT_RETURN __attribute__ ((__returns_twice__))
> +#else
> +# define __INDIRECT_RETURN
> +#endif


I would be rather surprised if RISC-V could ever define 
__ARM_FEATURE_GCS_DEFAULT under any circumstances whatsoever.