Re: [PATCH v1 03/14] xen/riscv: introduce ioremap()

[Jan Beulich <jbeulich@suse.com>]

On 22.04.2025 10:40, Oleksii Kurochko wrote:
> 
> On 4/17/25 4:49 PM, Jan Beulich wrote:
>> On 17.04.2025 16:37, Oleksii Kurochko wrote:
>>> On 4/17/25 4:24 PM, Jan Beulich wrote:
>>>> On 17.04.2025 16:20, Oleksii Kurochko wrote:
>>>>> On 4/15/25 1:02 PM, Jan Beulich wrote:
>>>>>> On 15.04.2025 12:29, Oleksii Kurochko wrote:
>>>>>>> On 4/10/25 5:13 PM, Jan Beulich wrote:
>>>>>>>> On 08.04.2025 17:57, Oleksii Kurochko wrote:
>>>>>>>>> Based on RISC-V unpriviliged spec ( Version 20240411 ):
>>>>>>>>> ```
>>>>>>>>> For implementations that conform to the RISC-V Unix Platform Specification,
>>>>>>>>> I/O devices and DMA operations are required to access memory coherently and
>>>>>>>>> via strongly ordered I/O channels. Therefore, accesses to regular main memory
>>>>>>>>> regions that are concurrently accessed by external devices can also use the
>>>>>>>>> standard synchronization mechanisms. Implementations that do not conform
>>>>>>>>> to the Unix Platform Specification and/or in which devices do not access
>>>>>>>>> memory coherently will need to use mechanisms
>>>>>>>>> (which are currently platform-specific or device-specific) to enforce
>>>>>>>>> coherency.
>>>>>>>>>
>>>>>>>>> I/O regions in the address space should be considered non-cacheable
>>>>>>>>> regions in the PMAs for those regions. Such regions can be considered coherent
>>>>>>>>> by the PMA if they are not cached by any agent.
>>>>>>>>> ```
>>>>>>>>> and [1]:
>>>>>>>>> ```
>>>>>>>>> The current riscv linux implementation requires SOC system to support
>>>>>>>>> memory coherence between all I/O devices and CPUs. But some SOC systems
>>>>>>>>> cannot maintain the coherence and they need support cache clean/invalid
>>>>>>>>> operations to synchronize data.
>>>>>>>>>
>>>>>>>>> Current implementation is no problem with SiFive FU540, because FU540
>>>>>>>>> keeps all IO devices and DMA master devices coherence with CPU. But to a
>>>>>>>>> traditional SOC vendor, it may already have a stable non-coherency SOC
>>>>>>>>> system, the need is simply to replace the CPU with RV CPU and rebuild
>>>>>>>>> the whole system with IO-coherency is very expensive.
>>>>>>>>> ```
>>>>>>>>>
>>>>>>>>> and the fact that all known ( to me ) CPUs that support the H-extension
>>>>>>>>> and that ones is going to be supported by Xen have memory coherency
>>>>>>>>> between all I/O devices and CPUs, so it is currently safe to use the
>>>>>>>>> PAGE_HYPERVISOR attribute.
>>>>>>>>> However, in cases where a platform does not support memory coherency, it
>>>>>>>>> should support CMO extensions and Svpbmt. In this scenario, updates to
>>>>>>>>> ioremap will be necessary.
>>>>>>>>> For now, a compilation error will be generated to ensure that the need to
>>>>>>>>> update ioremap() is not overlooked.
>>>>>>>>>
>>>>>>>>> [1]https://patchwork.kernel.org/project/linux-riscv/patch/1555947870-23014-1-git-send-email-guoren@kernel.org/
>>>>>>>> But MMIO access correctness isn't just a matter of coherency. There may not
>>>>>>>> be any caching involved in most cases, or else you may observe significantly
>>>>>>>> delayed or even dropped (folded with later ones) writes, and reads may be
>>>>>>>> serviced from the cache instead of going to actual MMIO. Therefore ...
>>>>>>>>
>>>>>>>>> --- a/xen/arch/riscv/Kconfig
>>>>>>>>> +++ b/xen/arch/riscv/Kconfig
>>>>>>>>> @@ -15,6 +15,18 @@ config ARCH_DEFCONFIG
>>>>>>>>>      	string
>>>>>>>>>      	default "arch/riscv/configs/tiny64_defconfig"
>>>>>>>>>      
>>>>>>>>> +config HAS_SVPBMT
>>>>>>>>> +	bool
>>>>>>>>> +	help
>>>>>>>>> +	  This config enables usage of Svpbmt ISA-extension ( Supervisor-mode:
>>>>>>>>> +	  page-based memory types).
>>>>>>>>> +
>>>>>>>>> +	  The memory type for a page contains a combination of attributes
>>>>>>>>> +	  that indicate the cacheability, idempotency, and ordering
>>>>>>>>> +	  properties for access to that page.
>>>>>>>>> +
>>>>>>>>> +	  The Svpbmt extension is only available on 64-bit cpus.
>>>>>>>> ... I kind of expect this extension (or anything else that there might be) will need
>>>>>>>> making use of.
>>>>>>> In cases where the Svpbmt extension isn't available, PMA (Physical Memory Attributes)
>>>>>>> is used to control which memory regions are cacheable, non-cacheable, readable, writable,
>>>>>>> etc. PMA is configured in M-mode by the firmware (e.g., OpenSBI), as is done in Andes
>>>>>>> cores, or it can be fixed at design time, as in SiFive cores.
>>>>>> How would things work if there was a need to map a RAM page uncacheable (via
>>>>>> ioremap() or otherwise)?
>>>>> My understanding is that Svpbmt is only needed when someone wants to change the memory
>>>>> attribute of a page set by PMA.
>>>>>
>>>>> The question is if non-cacheable RAM page is really needed if we have a coherency?
>>>> Aiui coherency here is among CPUs.
>>> ```
>>> For implementations that conform to the RISC-V Unix Platform Specification,
>>> I/O devices and DMA operations are required to access memory coherently and
>>> via strongly ordered I/O channels. Therefore, accesses to regular main memory
>>> regions that are concurrently accessed by external devices can also use the
>>> standard synchronization mechanisms. Implementations that do not conform
>>> to the Unix Platform Specification and/or in which devices do not access
>>> memory coherently will need to use mechanisms
>>> (which are currently platform-specific or device-specific) to enforce
>>> coherency.
>>> ```
>>> Based on this from the spec, coherency here is not only among CPUs.
>>>
>>>
>>>> Properties of devices in the system are
>>>> largely unknown?
>>> Yes, but still not sure what kind of property requires ioremap() which won't work
>>> without Svpmbt. Could you please tell me an example?
>> Well, above you said they all need to access memory coherently. That's the
>> "property" I was referring to.
> 
> Do you mean that device could have a property which tell that it would like to have non-cachable
> region used for that? I haven't seen such property in device tree files.
> 
> Do we have in Xen cases when Xen wants to have map part of RAM as non-cachebale and it is only the
> one option?

On x86 we have the case that IOMMUs may access memory non-coherently. This is
particular means that IOMMU page table updates (which necessarily live in RAM)
need to be done quite carefully. As it's all our code, we deal with the
situation by issuing cache flushes, avoiding the need for UC mappings.

Graphics engines may have similar constraints, aiui. With the driver code not
being part of Xen, we wouldn't be able to use a similar "simplification" there.
UC mappings would be pretty much unavoidable.

> I am also thinking why it can't be used cachable region + barrier (if we don't have memory coherency
> for everything).

Not sure what exactly you're asking here (if anything). An answer would very
likely depend on the specific kind of barrier you're thinking about. The
question would be what, if any, effect a barrier would have on the cache(s).

> Anyway, if it isn't an option to have mapped cacheble region + barrier then there is no any choice
> and the support of Svpmbt is required.

Quite possible.

Jan