Re: [PATCH v4 22/23] perf annotate-arm64: Support per-cpu variable access tracking

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Incomplete offset calculation for multi-register memory operands causes inaccurate type resolution.
--

--- Patch [22]: [PATCH v4 22/23] perf annotate-arm64: Support per-cpu variable access tracking ---
commit d7e10ade80ba3983e6d311e1ab11f19b4af2d53e
Author: Tengda Wu <wutengda@huaweicloud.com>
Subject: perf annotate-arm64: Support per-cpu variable access tracking

This commit extends update_insn_state() for arm64 to handle per-cpu variable
addressing. It identifies per-cpu base initialization and propagates type
information to subsequent instructions. This allows the tracker to correctly
resolve the actual variable type in per-cpu flows.

> diff --git a/tools/perf/util/annotate-data.c b/tools/perf/util/annotate-data.c
> index 7fe5b979563f7..8b0584486cce9 100644
> --- a/tools/perf/util/annotate-data.c
> +++ b/tools/perf/util/annotate-data.c
> @@ -1253,11 +1253,33 @@ static enum type_match_result check_matching_type(struct type_state *state,
>  	}
>  
>  	if (state->regs[reg].kind == TSR_KIND_PERCPU_BASE) {
> -		u64 var_addr = dloc->op->offset;
> +		u64 var_addr;
>  		int var_offset;

[ ... ]

> +		var_addr = dloc->op->offset;
> +
>  		if (dloc->op->multi_regs) {
>  			int reg2 = dloc->op->reg2;

[Severity: Medium]
Does the TSR_KIND_GLOBAL_ADDR handling later in check_matching_type() need
similar logic for multi-register operands to ensure accurate type resolution?

Currently, for TSR_KIND_GLOBAL_ADDR, the logic does not consider multi-register
operands:

	if (state->regs[reg].kind == TSR_KIND_GLOBAL_ADDR) {
		u64 var_addr = state->regs[reg].imm_value + dloc->op->offset;
		int var_offset;

This calculation omits adding state->regs[reg2].imm_value for multi-register
operands, unlike the TSR_KIND_PERCPU_BASE logic in the same function. The
omission also appears to apply to TSR_KIND_TYPE and TSR_KIND_POINTER.

Can this incomplete offset calculation cause incorrect type resolution in
tools/perf/util/annotate-data.c:check_matching_type() when perf processes a
memory instruction where op->multi_regs is true, the primary register is
TSR_KIND_GLOBAL_ADDR (or TYPE/POINTER), and the secondary register is
TSR_KIND_CONST?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260808122400.2961238-1-wutengda@huaweicloud.com?part=22