On 03-07-20, 13:23, Will Deacon wrote: > On Tue, Jun 16, 2020 at 01:52:32PM -0700, John Stultz wrote: > > On Tue, Jun 16, 2020 at 12:55 AM Marc Zyngier <maz@xxxxxxxxxx> wrote: > > > On 2020-06-16 07:13, John Stultz wrote: > > > > diff --git a/drivers/iommu/Kconfig b/drivers/iommu/Kconfig > > > > index b510f67dfa49..714893535dd2 100644 > > > > --- a/drivers/iommu/Kconfig > > > > +++ b/drivers/iommu/Kconfig > > > > @@ -381,6 +381,7 @@ config SPAPR_TCE_IOMMU > > > > config ARM_SMMU > > > > tristate "ARM Ltd. System MMU (SMMU) Support" > > > > depends on (ARM64 || ARM || (COMPILE_TEST && !GENERIC_ATOMIC64)) && > > > > MMU > > > > + depends on QCOM_SCM || !QCOM_SCM #if QCOM_SCM=m this can't be =y > > > > > > This looks a bit ugly. Could you explain why we need this at the SMMU > > > level? I'd have expected the dependency to flow the other way around... > > > > Yea, so the arm-smmu-qcom.c file calls directly into the qcom-scm code > > via qcom_scm_qsmmu500_wait_safe_toggle() > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/iommu/arm-smmu-qcom.c?h=v5.8-rc1#n44 > > > > So if ARM_SMMU=y and QCOM_SCM=m we get: > > drivers/iommu/arm-smmu-qcom.o: In function `qcom_smmu500_reset': > > arm-smmu-qcom.c:(.text+0xb4): undefined reference to > > `qcom_scm_qsmmu500_wait_safe_toggle' > > > > Do you have a suggestion for an alternative approach? > > Can you use symbol_get() or something like that? How are module dependencies > handled by other drivers? So drivers deal with this by making rules in Kconfig which will prohibit this case. QCOM_SCM depends on ARM_SMMU with the caveat that if ARM_SMMU is a module, QCOM_SCM cant be inbuilt. This can be done by adding below line in Kconfig for QCOM_SCM: depends on ARM_SMMU || !ARM_SMMU This is quite prevalent is drivers to ensure dependency like this Thanks -- ~Vinod
