On 2026-02-09 19:11, Xi Ruoyao wrote:
> On Mon, 2026-02-02 at 19:57 -0800, H. Peter Anvin wrote:
>> That hack dates back from before the signal frame extension. It is no
>> longer necessary.
>
> Unfortunately at least it seems libgcc unwinder does not handle the
> signal frame extension properly. The code reads:
>
> fde = _Unwind_Find_FDE (context->ra + _Unwind_IsSignalFrame (context) - 1,
> &context->bases);
> if (fde == NULL)
> {
> #ifdef MD_FALLBACK_FRAME_STATE_FOR
> /* Couldn't find frame unwind info for this function. Try a
> target-specific fallback mechanism. This will necessarily
> not provide a personality routine or LSDA. */
> return MD_FALLBACK_FRAME_STATE_FOR (context, fs);
> #else
> return _URC_END_OF_STACK;
> #endif
> }
>
> fs->pc = context->bases.func;
>
> cie = get_cie (fde);
> insn = extract_cie_info (cie, context, fs);
>
> Thus, it indeed attempts to avoid subtracting 1 for a signal frame, but
> ... _Unwind_IsSignalFrame (context) actually extracts a flag in context
> which will only be raised up by extract_cie_info.
>
> Or am I missing something here?
>
Well, it looks like this might be a non-manifest bug on x86 *IN THIS CASE*,
because the fallback handler looks explicitly for the byte sequences:
popl %eax
movl $__NR_ia32_sigreturn, %eax
int $0x80
movl $__NR_ia32_rt_sigreturn, %eax
int $0x80
movq $__NR_x64_rt_sigreturn, %rax /* !!!! */
syscall
... so the vdso patchset is still OK.
Hard-coding a reliance on movq in the 64-bit sequence some serious brain
damage...
Could you file a gcc bug on this?
-hpa
