Re: bcm2711_thermal: Kernel panic - not syncing: Asynchronous SError Interrupt

[Florian Fainelli <f.fainelli@gmail.com>]

On 7/28/2022 2:06 AM, Juerg Haefliger wrote:
> On Wed, 27 Jul 2022 14:51:24 -0700
> Florian Fainelli <f.fainelli@gmail.com> wrote:
> 
>> On 7/27/22 01:05, Juerg Haefliger wrote:
>>> On Wed, 10 Feb 2021 14:59:45 -0800
>>> Florian Fainelli <f.fainelli@gmail.com> wrote:
>>>    
>>>> On 2/10/2021 8:55 AM, Nicolas Saenz Julienne wrote:
>>>>> Hi Robin,
>>>>>
>>>>> On Wed, 2021-02-10 at 16:25 +0000, Robin Murphy wrote:
>>>>>> On 2021-02-10 13:15, Nicolas Saenz Julienne wrote:
>>>>>>> [ Add Robin, Catalin and Florian in case they want to chime in ]
>>>>>>>
>>>>>>> Hi Juerg, thanks for the report!
>>>>>>>
>>>>>>> On Wed, 2021-02-10 at 11:48 +0100, Juerg Haefliger wrote:
>>>>>>>> Trying to dump the BCM2711 registers kills the kernel:
>>>>>>>>
>>>>>>>> # cat /sys/kernel/debug/regmap/dummy-avs-monitor\@fd5d2000/range
>>>>>>>> 0-efc
>>>>>>>> # cat /sys/kernel/debug/regmap/dummy-avs-monitor\@fd5d2000/registers
>>>>>>>>
>>>>>>>> [   62.857661] SError Interrupt on CPU1, code 0xbf000002 -- SError
>>>>>>>
>>>>>>> So ESR's IDS (bit 24) is set, which means it's an 'Implementation Defined
>>>>>>> SError,' hence IIUC the rest of the error code is meaningless to anyone outside
>>>>>>> of Broadcom/RPi.
>>>>>>
>>>>>> It's imp-def from the architecture's PoV, but the implementation in this
>>>>>> case is Cortex-A72, where 0x000002 means an attributable, containable
>>>>>> Slave Error:
>>>>>>
>>>>>> https://developer.arm.com/documentation/100095/0003/system-control/aarch64-register-descriptions/exception-syndrome-register--el1-and-el3?lang=en
>>>>>>
>>>>>> In other words, the thing at the other end of an interconnect
>>>>>> transaction said "no" :)
>>>>>>
>>>>>> (The fact that Cortex-A72 gets too far ahead of itself to take it as a
>>>>>> synchronous external abort is a mild annoyance, but hey...)
>>>>>
>>>>> Thanks for both your clarifications! Reading arm documentation is a skill on
>>>>> its own.
>>>>
>>>> Yes it is.
>>>>   
>>>>>      
>>>>>>> The regmap is created through the following syscon device:
>>>>>>>
>>>>>>> 	avs_monitor: avs-monitor@7d5d2000 {
>>>>>>> 		compatible = "brcm,bcm2711-avs-monitor",
>>>>>>> 			     "syscon", "simple-mfd";
>>>>>>> 		reg = <0x7d5d2000 0xf00>;
>>>>>>>
>>>>>>> 		thermal: thermal {
>>>>>>> 			compatible = "brcm,bcm2711-thermal";
>>>>>>> 			#thermal-sensor-cells = <0>;
>>>>>>> 		};
>>>>>>> 	};
>>>>>>>
>>>>>>> I've done some tests with devmem, and the whole <0x7d5d2000 0xf00> range is
>>>>>>> full of addresses that trigger this same error. Also note that as per Florian's
>>>>>>> comments[1]: "AVS_RO_REGISTERS_0: 0x7d5d2200 - 0x7d5d22e3." But from what I can
>>>>>>> tell, at least 0x7d5d22b0 seems to be faulty too.
>>>>>>>
>>>>>>> Any ideas/comments? My guess is that those addresses are marked somehow as
>>>>>>> secure, and only for VC4 to access (VC4 is RPi4's co-processor). Ultimately,
>>>>>>> the solution is to narrow the register range exposed by avs-monitor to whatever
>>>>>>> bcm2711-thermal needs (which is ATM a single 32bit register).
>>>>>>
>>>>>> When a peripheral decodes a region of address space, nobody says it has
>>>>>> to accept accesses to *every* address in that space; registers may be
>>>>>> sparsely populated, and although some devices might be "nice" and make
>>>>>> unused areas behave as RAZ/WI, others may throw slave errors if you poke
>>>>>> at the wrong places. As you note, in a TrustZone-aware device some
>>>>>> registers may only exist in one or other of the Secure/Non-Secure
>>>>>> address spaces.
>>>>>>
>>>>>> Even when there is a defined register at a given address, it still
>>>>>> doesn't necessarily accept all possible types of access; it wouldn't be
>>>>>> particularly friendly, but a device *could* have, say, some registers
>>>>>> that support 32-bit accesses and others that only support 16-bit
>>>>>> accesses, and thus throw slave errors if you do the wrong thing in the
>>>>>> wrong place.
>>>>>>
>>>>>> It really all depends on the device itself.
>>>>>
>>>>> All in all, assuming there is no special device quirk to apply, the feeling I'm
>>>>> getting is to just let the error be. As you hint, firmware has no blame here,
>>>>> and debugfs is a 'best effort, zero guarantees' interface after all.
>>>>
>>>> We should probably fill a regmap_access_table to deny reading registers
>>>> for which there is no address decoding and possibly another one to deny
>>>> writing to the read-only registers.
>>>
>>>
>>> Below is a patch that adds a read access table but it seems wrong to include
>>> 'internal.h' and add the table in the thermal driver. Shouldn't this happen
>>> in a higher layer, somehow between syscon and the thermal node?
>>
>> What is the purpose of doing doing this though that cannot already be done using devmem/devmem2 if the point is explore the address space?
> 
> The goal is to prevent a kernel crash when doing
> $ cat /sys/kernel/debug/regmap/dummy-avs-monitor\@fd5d2000/registers

Fair enough, but that really does not scale across drivers nor across 
power management decisions being made to various drivers.

The thermal sensor is unlikely to ever be clock gated by the time Linux 
runs, but if you were to do the same thing for any other type of 
peripheral, chances are the same outcome would be produced.

So this really begs the question as to how to address this globally 
short of disabling regmap debugfs support which is likely what is 
happening in a production environment anyway.
-- 
Florian