Re: [PATCH RFC v10 07/22] arm64: Initialize POR_EL1 register on cpu_resume()

From: Kevin Brodsky

Date: Thu Nov 12 2026 - 07:40:25 EST


On 10/11/2026 17:36, Yeoreum Yun wrote:
> On Tue, Nov 10, 2026 at 03:53:30PM +0000, Mark Brown wrote:
>> On Mon, Nov 09, 2026 at 03:58:32PM +0000, Kevin Brodsky wrote:
>>> From: Yeoreum Yun <yeoreum.yun@xxxxxxx>
>>>
>>> The POR_EL1 register is reset to an unknown value after cpu_suspend().
>>> Since POR_EL1 always holds POR_EL1_INIT when entering cpu_suspend(),
>>> initialize POR_EL1 with POR_EL1_INIT before cpu_do_resume().
>>>> +#ifdef CONFIG_ARM64_POE
>>> +alternative_if_not ARM64_HAS_S1POE
>>> + b .Lskip_por_set
>>> +alternative_else_nop_endif
>>> + mov_q x2, POR_EL1_INIT
>>> + msr_s SYS_POR_EL1, x2
>>> + /* isb can be skipped since cpu_do_resume() will do it. */
>>> +.Lskip_por_set:
>>> +#endif /* CONFIG_ARM64_POE */
>> cpu_do_resume() will update TCR2_EL1 to enable POE again, a sufficiently
>> unhelpful implementation could potentially decide to do that before this
>> POR_EL1 value takes effect AFAICT given that there's no barriers.
> Thanks. seems required to isb() after initialising SYS_POR_EL1.

Agreed (as Sashiko reported as well). Presumably that means we also need
an ISB after writing to POR_EL1 in cpu_enable_poe(), as that's basically
the same situation.

- Kevin