Re: [PATCH v16 07/20] unwind_user/sframe: Wire up unwind_user to sframe

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Missing instruction pointer adjustment for return addresses during SFrame unwinding of non-topmost frames.
--

commit 9e76c3f974e7ef0d88d0ca8c60fa84ef664ebee7
Author: Josh Poimboeuf <jpoimboe@kernel.org>

unwind_user/sframe: Wire up unwind_user to sframe

Now that the sframe infrastructure is fully in place, make it work by hooking
it up to the unwind_user interface.

> diff --git a/kernel/unwind/user.c b/kernel/unwind/user.c
> index 90ab3c1a205ef..1fb2724197332 100644
> --- a/kernel/unwind/user.c
> +++ b/kernel/unwind/user.c

[ ... ]

> @@ -82,6 +83,16 @@ static int unwind_user_next_fp(struct unwind_user_state *state)
>  	return unwind_user_next_common(state, &fp_frame);
>  }
>  
> +static int unwind_user_next_sframe(struct unwind_user_state *state)
> +{
> +	struct unwind_user_frame frame;
> +
> +	/* sframe expects the frame to be local storage */
> +	if (sframe_find(state->ip, &frame))

[Severity: Medium]
For non-topmost stack frames, state->ip holds the return address, which points
to the instruction after the call. If a call is the very last instruction of a
function (like a call to a noreturn function), the return address will point to
the first instruction of the next contiguous function in memory.

Could sframe_find() incorrectly match the next function's FDE and apply its
unwind rules to the current frame? Other table-driven unwinders handle this
by subtracting 1 from the instruction pointer for non-topmost frames.

> +		return -ENOENT;
> +	return unwind_user_next_common(state, &frame);
> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260521142546.3908498-1-jremus@linux.ibm.com?part=7