Messages in this thread | | | Subject | Re: [PATCH v3 2/6] iommu/dma: add support for non-strict mode | From | "Leizhen (ThunderTown)" <> | Date | Thu, 26 Jul 2018 12:15:28 +0800 |
| |
On 2018/7/25 6:01, Robin Murphy wrote: > On 2018-07-12 7:18 AM, Zhen Lei wrote: >> 1. Save the related domain pointer in struct iommu_dma_cookie, make iovad >> capable call domain->ops->flush_iotlb_all to flush TLB. >> 2. Add a new iommu capability: IOMMU_CAP_NON_STRICT, which used to indicate >> that the iommu domain support non-strict mode. >> 3. During the iommu domain initialization phase, call capable() to check >> whether it support non-strcit mode. If so, call init_iova_flush_queue >> to register iovad->flush_cb callback. >> 4. All unmap(contains iova-free) APIs will finally invoke __iommu_dma_unmap >> -->iommu_dma_free_iova. If the domain is non-strict, call queue_iova to >> put off iova freeing. >> >> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com> >> --- >> drivers/iommu/dma-iommu.c | 25 +++++++++++++++++++++++++ >> include/linux/iommu.h | 7 +++++++ >> 2 files changed, 32 insertions(+) >> >> diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c >> index ddcbbdb..9f0c77a 100644 >> --- a/drivers/iommu/dma-iommu.c >> +++ b/drivers/iommu/dma-iommu.c >> @@ -55,6 +55,7 @@ struct iommu_dma_cookie { >> }; >> struct list_head msi_page_list; >> spinlock_t msi_lock; >> + struct iommu_domain *domain_non_strict; >> }; >> static inline size_t cookie_msi_granule(struct iommu_dma_cookie *cookie) >> @@ -257,6 +258,17 @@ static int iova_reserve_iommu_regions(struct device *dev, >> return ret; >> } >> +static void iommu_dma_flush_iotlb_all(struct iova_domain *iovad) >> +{ >> + struct iommu_dma_cookie *cookie; >> + struct iommu_domain *domain; >> + >> + cookie = container_of(iovad, struct iommu_dma_cookie, iovad); >> + domain = cookie->domain_non_strict; >> + >> + domain->ops->flush_iotlb_all(domain); >> +} >> + >> /** >> * iommu_dma_init_domain - Initialise a DMA mapping domain >> * @domain: IOMMU domain previously prepared by iommu_get_dma_cookie() >> @@ -272,6 +284,7 @@ static int iova_reserve_iommu_regions(struct device *dev, >> int iommu_dma_init_domain(struct iommu_domain *domain, dma_addr_t base, >> u64 size, struct device *dev) >> { >> + const struct iommu_ops *ops = domain->ops; >> struct iommu_dma_cookie *cookie = domain->iova_cookie; >> struct iova_domain *iovad = &cookie->iovad; >> unsigned long order, base_pfn, end_pfn; >> @@ -308,6 +321,15 @@ int iommu_dma_init_domain(struct iommu_domain *domain, dma_addr_t base, >> } >> init_iova_domain(iovad, 1UL << order, base_pfn); >> + >> + if ((ops->capable && ops->capable(IOMMU_CAP_NON_STRICT)) && >> + (IOMMU_DOMAIN_STRICT_MODE(domain) == IOMMU_NON_STRICT)) { >> + BUG_ON(!ops->flush_iotlb_all); >> + >> + cookie->domain_non_strict = domain; >> + init_iova_flush_queue(iovad, iommu_dma_flush_iotlb_all, NULL); >> + } >> + >> if (!dev) >> return 0; >> @@ -390,6 +412,9 @@ static void iommu_dma_free_iova(struct iommu_dma_cookie *cookie, >> /* The MSI case is only ever cleaning up its most recent allocation */ >> if (cookie->type == IOMMU_DMA_MSI_COOKIE) >> cookie->msi_iova -= size; >> + else if (cookie->domain_non_strict) >> + queue_iova(iovad, iova_pfn(iovad, iova), >> + size >> iova_shift(iovad), 0); >> else >> free_iova_fast(iovad, iova_pfn(iovad, iova), >> size >> iova_shift(iovad)); >> diff --git a/include/linux/iommu.h b/include/linux/iommu.h >> index 19938ee..82ed979 100644 >> --- a/include/linux/iommu.h >> +++ b/include/linux/iommu.h >> @@ -86,6 +86,12 @@ struct iommu_domain_geometry { >> #define IOMMU_DOMAIN_DMA (__IOMMU_DOMAIN_PAGING | \ >> __IOMMU_DOMAIN_DMA_API) >> +#define IOMMU_STRICT 0 >> +#define IOMMU_NON_STRICT 1 >> +#define IOMMU_STRICT_MODE_MASK 1UL >> +#define IOMMU_DOMAIN_STRICT_MODE(domain) \ >> + (domain->type != IOMMU_DOMAIN_UNMANAGED) >> + >> struct iommu_domain { >> unsigned type; >> const struct iommu_ops *ops; >> @@ -101,6 +107,7 @@ enum iommu_cap { >> transactions */ >> IOMMU_CAP_INTR_REMAP, /* IOMMU supports interrupt isolation */ >> IOMMU_CAP_NOEXEC, /* IOMMU_NOEXEC flag */ >> + IOMMU_CAP_NON_STRICT, /* IOMMU supports non-strict mode */ > > This still isn't a capability of the hardware. Non-strict mode is pure software policy; *every IOMMU ever* supports it. > > If you want to have this kind of per-driver control, make it a domain attribute - that's not only more logical, but should also work out a lot cleaner overall.
OK, I will use quirk, so this capability will be removed in the next version.
> > Robin. > >> }; >> /* >> > > . >
-- Thanks! BestRegards
| |