Re: [PATCH] pinctrl: intel: zero-initialize capability_offset[] in probe

[Andy Shevchenko <andriy.shevchenko@intel.com>]

On Fri, May 15, 2026 at 08:00:49PM +0500, Stepan Ionichev wrote:
> intel_pinctrl_probe() declares a per-community capability_offset[]
> array on the stack and only writes the slots whose CAPLIST entries
> the device actually advertises:

Thanks, this is very long, but unfortunately bogus report.
So, here basically the summary is "Check if the HW is broken".
If the answer to that "yes", we have much bigger issues than
some parameter being precalculated. Besides that, have you checked
the code generation? Does compiler is stupid enough to calculate
base before any checks? It's quite clear that the first use of base
is the devm_pwm_lpss_probe(). Instead of this patch I can accept
one that splits definition and assignment and be like

	void __iomem *base;
	...
	base = community->regs + capability_offset + 4;
	chip = devm_pwm_lpss_probe(pctrl->dev, base, &info);

(without any Fixes tag).

-- 
With Best Regards,
Andy Shevchenko