From mboxrd@z Thu Jan 1 00:00:00 1970 From: Robin Murphy Subject: Re: [PATCH 1/3] iommu: Add fast hook for getting DMA domains Date: Fri, 17 Aug 2018 12:11:57 +0100 Message-ID: <29a126d2-4bbe-1750-5265-d3dbfcb658ae@arm.com> References: <4d50587e62c16f7b5cfc45dee808d774655155cf.1534250425.git.robin.murphy@arm.com> Mime-Version: 1.0 Content-Type: text/plain; charset="windows-1252"; Format="flowed" Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: Content-Language: en-GB List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: iommu-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org Errors-To: iommu-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org To: John Garry , joro-zLv9SwRftAIdnm+yROfE0A@public.gmane.org, will.deacon-5wv7dgnIgG8@public.gmane.org, catalin.marinas-5wv7dgnIgG8@public.gmane.org Cc: linuxarm-hv44wF8Li93QT0dZR+AlfA@public.gmane.org, iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org, guohanjun-hv44wF8Li93QT0dZR+AlfA@public.gmane.org, linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org List-Id: iommu@lists.linux-foundation.org On 17/08/18 10:36, John Garry wrote: > On 14/08/2018 14:04, Robin Murphy wrote: >> While iommu_get_domain_for_dev() is the robust way for arbitrary IOMMU >> API callers to retrieve the domain pointer, for DMA ops domains it >> doesn't scale well for large systems and multi-queue devices, since the >> momentary refcount adjustment will lead to exclusive cacheline contention >> when multiple CPUs are operating in parallel on different mappings for >> the same device. >> >> In the case of DMA ops domains, however, this refcounting is actually >> unnecessary, since they already imply that the group exists and is >> managed by platform code and IOMMU internals (by virtue of >> iommu_group_get_for_dev()) such that a reference will already be held >> for the lifetime of the device. Thus we can avoid the bottleneck by >> providing a fast lookup specifically for the DMA code to retrieve the >> default domain it already knows it has set up - a simple read-only >> dereference plays much nicer with cache-coherency protocols. >> >> Signed-off-by: Robin Murphy >> --- >> =A0drivers/iommu/iommu.c | 9 +++++++++ >> =A0include/linux/iommu.h | 1 + >> =A02 files changed, 10 insertions(+) >> >> diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c >> index 63b37563db7e..63c586875df5 100644 >> --- a/drivers/iommu/iommu.c >> +++ b/drivers/iommu/iommu.c >> @@ -1379,6 +1379,15 @@ struct iommu_domain = >> *iommu_get_domain_for_dev(struct device *dev) >> =A0} >> =A0EXPORT_SYMBOL_GPL(iommu_get_domain_for_dev); >> >> +/* >> + * For IOMMU_DOMAIN_DMA implementations which already provide their own >> + * guarantees that the group and its default domain are valid and = >> correct. >> + */ >> +struct iommu_domain *iommu_get_dma_domain(struct device *dev) >> +{ >> +=A0=A0=A0 return dev->iommu_group->default_domain; >> +} >> + >> =A0/* >> =A0 * IOMMU groups are really the natrual working unit of the IOMMU, but >> =A0 * the IOMMU API works on domains and devices.=A0 Bridge that gap by >> diff --git a/include/linux/iommu.h b/include/linux/iommu.h >> index 19938ee6eb31..16f2172698e5 100644 >> --- a/include/linux/iommu.h >> +++ b/include/linux/iommu.h >> @@ -297,6 +297,7 @@ extern int iommu_attach_device(struct iommu_domain = >> *domain, >> =A0extern void iommu_detach_device(struct iommu_domain *domain, >> =A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 struct device *dev); >> =A0extern struct iommu_domain *iommu_get_domain_for_dev(struct device = >> *dev); >> +extern struct iommu_domain *iommu_get_dma_domain(struct device *dev); > = > Hi Robin, > = > I was wondering whether it's standard to provide a stubbed version of = > this function for !CONFIG_IOMMU_API? Nope, that's deliberate - this is one of those hooks where any = legitimate caller will already be dependent on IOMMU_API itself. Robin. > = > Cheers, > John > = >> =A0extern int iommu_map(struct iommu_domain *domain, unsigned long iova, >> =A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 phys_addr_t paddr, size_t size, = int prot); >> =A0extern size_t iommu_unmap(struct iommu_domain *domain, unsigned long = >> iova, >> > = > =