Re: [PATCH v8 1/7] iommu: enhance IOMMU default DMA mode build options

[Leizhen (ThunderTown)]

On 2019/5/30 20:20, John Garry wrote:
> On 30/05/2019 04:48, Zhen Lei wrote:
>> First, add build option IOMMU_DEFAULT_{LAZY|STRICT}, so that we have the
>> opportunity to set {lazy|strict} mode as default at build time. Then put
>> the three config options in an choice, make people can only choose one of
>> the three at a time.
>>
> 
> Since this was not picked up, but modulo (somtimes same) comments below:
> 
> Reviewed-by: John Garry <john.garry@xxxxxxxxxx>
> 
>> Signed-off-by: Zhen Lei <thunder.leizhen@xxxxxxxxxx>
>> ---
>> Âdrivers/iommu/Kconfig | 42 +++++++++++++++++++++++++++++++++++-------
>> Âdrivers/iommu/iommu.c |Â 3 ++-
>> Â2 files changed, 37 insertions(+), 8 deletions(-)
>>
>> diff --git a/drivers/iommu/Kconfig b/drivers/iommu/Kconfig
>> index 83664db5221df02..d6a1a45f80ffbf5 100644
>> --- a/drivers/iommu/Kconfig
>> +++ b/drivers/iommu/Kconfig
>> @@ -75,17 +75,45 @@ config IOMMU_DEBUGFS
>> ÂÂÂÂÂÂ debug/iommu directory, and then populate a subdirectory with
>> ÂÂÂÂÂÂ entries as required.
>>
>> -config IOMMU_DEFAULT_PASSTHROUGH
>> -ÂÂÂ bool "IOMMU passthrough by default"
>> +choice
>> +ÂÂÂ prompt "IOMMU default DMA mode"
>> ÂÂÂÂ depends on IOMMU_API
>> -ÂÂÂÂÂÂÂ help
>> -ÂÂÂÂÂ Enable passthrough by default, removing the need to pass in
>> -ÂÂÂÂÂ iommu.passthrough=on or iommu=pt through command line. If this
>> -ÂÂÂÂÂ is enabled, you can still disable with iommu.passthrough=off
>> -ÂÂÂÂÂ or iommu=nopt depending on the architecture.
>> +ÂÂÂ default IOMMU_DEFAULT_STRICT
>> +ÂÂÂ help
>> +ÂÂÂÂÂ This option allows IOMMU DMA mode to be chose at build time, to
> 
> As before:
> /s/chose/chosen/, /s/allows IOMMU/allows an IOMMU/
I'm sorry that the previous version was not modified.

> 
>> +ÂÂÂÂÂ override the default DMA mode of each ARCHs, removing the need to
> 
> Again, as before:
> ARCHs should be singular
OK

> 
>> +ÂÂÂÂÂ pass in kernel parameters through command line. You can still use
>> +ÂÂÂÂÂ ARCHs specific boot options to override this option again.
>> +
>> +config IOMMU_DEFAULT_PASSTHROUGH
>> +ÂÂÂ bool "passthrough"
>> +ÂÂÂ help
>> +ÂÂÂÂÂ In this mode, the DMA access through IOMMU without any addresses
>> +ÂÂÂÂÂ translation. That means, the wrong or illegal DMA access can not
>> +ÂÂÂÂÂ be caught, no error information will be reported.
>>
>> ÂÂÂÂÂÂ If unsure, say N here.
>>
>> +config IOMMU_DEFAULT_LAZY
>> +ÂÂÂ bool "lazy"
>> +ÂÂÂ help
>> +ÂÂÂÂÂ Support lazy mode, where for every IOMMU DMA unmap operation, the
>> +ÂÂÂÂÂ flush operation of IOTLB and the free operation of IOVA are deferred.
>> +ÂÂÂÂÂ They are only guaranteed to be done before the related IOVA will be
>> +ÂÂÂÂÂ reused.
> 
> why no advisory on how to set if unsure?
Because the LAZY and STRICT have their own advantages and disadvantages.

Should I say: If unsure, keep the defaultã

> 
>> +
>> +config IOMMU_DEFAULT_STRICT
>> +ÂÂÂ bool "strict"
>> +ÂÂÂ help
>> +ÂÂÂÂÂ For every IOMMU DMA unmap operation, the flush operation of IOTLB and
>> +ÂÂÂÂÂ the free operation of IOVA are guaranteed to be done in the unmap
>> +ÂÂÂÂÂ function.
>> +
>> +ÂÂÂÂÂ This mode is safer than the two above, but it maybe slower in some
>> +ÂÂÂÂÂ high performace scenarios.
> 
> and here?
> 
>> +
>> +endchoice
>> +
>> Âconfig OF_IOMMU
>> ÂÂÂÂÂÂÂ def_bool y
>> ÂÂÂÂÂÂÂ depends on OF && IOMMU_API
>> diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
>> index 67ee6623f9b2a4d..56bce221285b15f 100644
>> --- a/drivers/iommu/iommu.c
>> +++ b/drivers/iommu/iommu.c
>> @@ -43,7 +43,8 @@
>> Â#else
>> Âstatic unsigned int iommu_def_domain_type = IOMMU_DOMAIN_DMA;
>> Â#endif
>> -static bool iommu_dma_strict __read_mostly = true;
>> +static bool iommu_dma_strict __read_mostly =
>> +ÂÂÂÂÂÂÂÂÂÂÂ IS_ENABLED(CONFIG_IOMMU_DEFAULT_STRICT);
>>
>> Âstruct iommu_group {
>> ÂÂÂÂ struct kobject kobj;
>>
> 
> 
> 
> .
>