[PATCH v3 1/4] aarch64: make explicit that CFI gnu_window_save is for Sparc, not AArch64

[Indu Bhagat]

On 1/13/25 3:22 AM, Matthieu Longo wrote:
> - add a detailed comment when parsing DW_CFA_GNU_window_save in SFrame to
>    explain why we are checking whether the targeted architecture is AArch64,
>    whereas this CFI is a Sparc extension.
> - replace .cfi_gnu_window_save by .cfi_negate_ra_state in existing AArch64
>    DWARF tests as this is the preferred directive since GCC 15.
> - add a new AARch64 test to check backward compatibility with old GCC

Nit: AARch64 -> AArch64

>    versions that emits .cfi_gnu_window_save.
> ---
>   gas/gen-sframe.c                              |  8 +++-
>   gas/testsuite/gas/aarch64/pac_ab_key.s        |  4 +-
>   .../gas/aarch64/pac_compat_cfi_window_save.d  | 44 +++++++++++++++++++
>   .../gas/aarch64/pac_compat_cfi_window_save.s  | 20 +++++++++
>   4 files changed, 73 insertions(+), 3 deletions(-)
>   create mode 100644 gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.d
>   create mode 100644 gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.s
> 
> diff --git a/gas/gen-sframe.c b/gas/gen-sframe.c
> index 71296f2f4fb..a3c40bdd735 100644
> --- a/gas/gen-sframe.c
> +++ b/gas/gen-sframe.c
> @@ -1273,7 +1273,13 @@ sframe_xlate_do_aarch64_negate_ra_state (struct sframe_xlate_ctx *xlate_ctx,
>   }
>   
>   /* Translate DW_CFA_GNU_window_save into SFrame context.
> -   DW_CFA_AARCH64_negate_ra_state is multiplexed with DW_CFA_GNU_window_save.
> +   DW_CFA_GNU_window_save is a DWARF Sparc extension, but is multiplexed with a
> +   directive of DWARF AArch64 extension: DW_CFA_AARCH64_negate_ra_state.
> +   The AArch64 backend of GCC 14 and older versions was emitting mistakenly the
> +   Sparc CFI directive (.cfi_window_save).  From GCC 15, the AArch64 backend
> +   only emits .cfi_negate_ra_state.  For backward compatibility, the handler for
> +   .cfi_window_save needs to check whether the directive was used in a AArch ABI

Nit: AArch -> AArch64 ?

In context of the SFrame changes here,  AArch64 should still make sense 
(although the multiplexing of directives is applicable to the 32-bit Arm 
as well).

OK for the SFrame file changes.

> +   context or not.
>      Return SFRAME_XLATE_OK if success.  */
>   
>   static int
> diff --git a/gas/testsuite/gas/aarch64/pac_ab_key.s b/gas/testsuite/gas/aarch64/pac_ab_key.s
> index 4b328e72ae4..3b81919409d 100644
> --- a/gas/testsuite/gas/aarch64/pac_ab_key.s
> +++ b/gas/testsuite/gas/aarch64/pac_ab_key.s
> @@ -7,7 +7,7 @@ _Z5foo_av:
>   .LFB0:
>   	.cfi_startproc
>   	hint	25 // paciasp
> -	.cfi_window_save
> +	.cfi_negate_ra_state
>   	stp	x29, x30, [sp, -16]!
>   	.cfi_def_cfa_offset 16
>   	.cfi_offset 29, -16
> @@ -23,7 +23,7 @@ _Z5foo_bv:
>   	.cfi_startproc
>   	.cfi_b_key_frame
>   	hint	27 // pacibsp
> -	.cfi_window_save
> +	.cfi_negate_ra_state
>   	stp	x29, x30, [sp, -16]!
>   	.cfi_def_cfa_offset 16
>   	.cfi_offset 29, -16
> diff --git a/gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.d b/gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.d
> new file mode 100644
> index 00000000000..8e59086c1b4
> --- /dev/null
> +++ b/gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.d
> @@ -0,0 +1,44 @@
> +#objdump: --dwarf=frames
> +# This test is only valid on ELF based ports.
> +#notarget: *-*-*coff *-*-pe *-*-wince *-*-*aout* *-*-netbsd
> +
> +## ARMv8.3 adds support for a new security feature named Pointer Authentication.
> +## The main idea behind this is to use the unused bits in the pointer values.
> +## Each pointer is patched with a PAC before writing to memory, and is verified
> +## before using it.
> +## When the pointers are mangled, the stack trace generator needs to know so it
> +## can mask off the PAC from the pointer value to recover the return address,
> +## and conversely, skip doing so if the pointers are not mangled.
> +##
> +## .cfi_negate_ra_state CFI directive is usually used to convey this information.
> +## .cfi_negate_ra_state and .cfi_window_save are both in the processor-specific
> +## numbering space, but use the same code value in the dwarf tables.
> +## In GCC 14 and older, the Sparc DWARF extension .cfi_window_save is emitted
> +## instead of .cfi_negate_ra_state, but it mapped to the same value. GCC 15 fixed
> +## this naming issue and there is no change to the object file created when the
> +## source is assembled. Nevertheless the support for the SPARC directive is
> +## preserved in binutils for backward compatibility with existing GCC releases,
> +## hence this test.
> +
> +.+:     file .+
> +
> +Contents of the .eh_frame section:
> +
> +0+ 0+10 0+ CIE
> +  Version:               1
> +  Augmentation:          "zR"
> +  Code alignment factor: 4
> +  Data alignment factor: -8
> +  Return address column: 30
> +  Augmentation data:     1b
> +  DW_CFA_def_cfa: r31 \(sp\) ofs 0
> +
> +0+14 0+18 0+18 FDE cie=0+ pc=0+\.\.0+8
> +  DW_CFA_advance_loc: 4 to 0+4
> +  DW_CFA_AARCH64_negate_ra_state
> +  DW_CFA_advance_loc: 4 to 0+8
> +  DW_CFA_def_cfa_offset: 16
> +  DW_CFA_offset: r29 \(x29\) at cfa-16
> +  DW_CFA_offset: r30 \(x30\) at cfa-8
> +  DW_CFA_nop
> +  DW_CFA_nop
> diff --git a/gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.s b/gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.s
> new file mode 100644
> index 00000000000..92a54f3a344
> --- /dev/null
> +++ b/gas/testsuite/gas/aarch64/pac_compat_cfi_window_save.s
> @@ -0,0 +1,20 @@
> +	.arch armv8-a
> +	.text
> +	.align	2
> +	.global	_Z5foo_av
> +	.type	_Z5foo_av, %function
> +_Z5foo_av:
> +.LFB0:
> +	.cfi_startproc
> +	hint	25 // paciasp
> +	.cfi_window_save // really .cfi_negate_ra_state
> +	stp	x29, x30, [sp, -16]!
> +	.cfi_def_cfa_offset 16
> +	.cfi_offset 29, -16
> +	.cfi_offset 30, -8
> +	.cfi_endproc
> +.LFE0:
> +	.size	_Z5foo_av, .-_Z5foo_av
> +	.align	2
> +	.global	_Z5foo_bv
> +	.type	_Z5foo_bv, %function