From e60aa7b53845a261dd419652f12ab9f89e668843 Mon Sep 17 00:00:00 2001 From: Robin Murphy Date: Thu, 21 Sep 2017 16:52:44 +0100 Subject: iommu/iova: Extend rbtree node caching The cached node mechanism provides a significant performance benefit for allocations using a 32-bit DMA mask, but in the case of non-PCI devices or where the 32-bit space is full, the loss of this benefit can be significant - on large systems there can be many thousands of entries in the tree, such that walking all the way down to find free space every time becomes increasingly awful. Maintain a similar cached node for the whole IOVA space as a superset of the 32-bit space so that performance can remain much more consistent. Inspired by work by Zhen Lei . Tested-by: Ard Biesheuvel Tested-by: Zhen Lei Tested-by: Nate Watterson Signed-off-by: Robin Murphy Signed-off-by: Joerg Roedel --- include/linux/iova.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/iova.h b/include/linux/iova.h index d179b9bf7814..69ea3e258ff2 100644 --- a/include/linux/iova.h +++ b/include/linux/iova.h @@ -70,7 +70,8 @@ struct iova_fq { struct iova_domain { spinlock_t iova_rbtree_lock; /* Lock to protect update of rbtree */ struct rb_root rbroot; /* iova domain rbtree root */ - struct rb_node *cached32_node; /* Save last alloced node */ + struct rb_node *cached_node; /* Save last alloced node */ + struct rb_node *cached32_node; /* Save last 32-bit alloced node */ unsigned long granule; /* pfn granularity for this domain */ unsigned long start_pfn; /* Lower limit for this domain */ unsigned long dma_32bit_pfn; -- cgit v1.2.3 From aa3ac9469c1850ed00741955b975c3a19029763a Mon Sep 17 00:00:00 2001 From: Zhen Lei Date: Thu, 21 Sep 2017 16:52:45 +0100 Subject: iommu/iova: Make dma_32bit_pfn implicit Now that the cached node optimisation can apply to all allocations, the couple of users which were playing tricks with dma_32bit_pfn in order to benefit from it can stop doing so. Conversely, there is also no need for all the other users to explicitly calculate a 'real' 32-bit PFN, when init_iova_domain() can happily do that itself from the page granularity. CC: Thierry Reding CC: Jonathan Hunter CC: David Airlie CC: Sudeep Dutt CC: Ashutosh Dixit Signed-off-by: Zhen Lei Tested-by: Ard Biesheuvel Tested-by: Zhen Lei Tested-by: Nate Watterson [rm: use iova_shift(), rewrote commit message] Signed-off-by: Robin Murphy Signed-off-by: Joerg Roedel --- include/linux/iova.h | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'include/linux') diff --git a/include/linux/iova.h b/include/linux/iova.h index 69ea3e258ff2..953cfd20f152 100644 --- a/include/linux/iova.h +++ b/include/linux/iova.h @@ -154,7 +154,7 @@ struct iova *reserve_iova(struct iova_domain *iovad, unsigned long pfn_lo, unsigned long pfn_hi); void copy_reserved_iova(struct iova_domain *from, struct iova_domain *to); void init_iova_domain(struct iova_domain *iovad, unsigned long granule, - unsigned long start_pfn, unsigned long pfn_32bit); + unsigned long start_pfn); int init_iova_flush_queue(struct iova_domain *iovad, iova_flush_cb flush_cb, iova_entry_dtor entry_dtor); struct iova *find_iova(struct iova_domain *iovad, unsigned long pfn); @@ -230,8 +230,7 @@ static inline void copy_reserved_iova(struct iova_domain *from, static inline void init_iova_domain(struct iova_domain *iovad, unsigned long granule, - unsigned long start_pfn, - unsigned long pfn_32bit) + unsigned long start_pfn) { } -- cgit v1.2.3 From bb68b2fbfbd643d4407541f9c7a16a2c9b3a57c7 Mon Sep 17 00:00:00 2001 From: Robin Murphy Date: Thu, 21 Sep 2017 16:52:46 +0100 Subject: iommu/iova: Add rbtree anchor node Add a permanent dummy IOVA reservation to the rbtree, such that we can always access the top of the address space instantly. The immediate benefit is that we remove the overhead of the rb_last() traversal when not using the cached node, but it also paves the way for further simplifications. Signed-off-by: Robin Murphy Signed-off-by: Joerg Roedel --- include/linux/iova.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/iova.h b/include/linux/iova.h index 953cfd20f152..c696ee81054e 100644 --- a/include/linux/iova.h +++ b/include/linux/iova.h @@ -75,6 +75,7 @@ struct iova_domain { unsigned long granule; /* pfn granularity for this domain */ unsigned long start_pfn; /* Lower limit for this domain */ unsigned long dma_32bit_pfn; + struct iova anchor; /* rbtree lookup anchor */ struct iova_rcache rcaches[IOVA_RANGE_CACHE_MAX_SIZE]; /* IOVA range caches */ iova_flush_cb flush_cb; /* Call-Back function to flush IOMMU -- cgit v1.2.3 From ec154bf56b276a0bb36079a5d22a267b5f417801 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Fri, 6 Oct 2017 15:00:53 +0200 Subject: iommu/vt-d: Don't register bus-notifier under dmar_global_lock The notifier function will take the dmar_global_lock too, so lockdep complains about inverse locking order when the notifier is registered under the dmar_global_lock. Reported-by: Jan Kiszka Fixes: 59ce0515cdaf ('iommu/vt-d: Update DRHD/RMRR/ATSR device scope caches when PCI hotplug happens') Signed-off-by: Joerg Roedel --- include/linux/dmar.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/dmar.h b/include/linux/dmar.h index e8ffba1052d3..e2433bc50210 100644 --- a/include/linux/dmar.h +++ b/include/linux/dmar.h @@ -112,6 +112,7 @@ static inline bool dmar_rcu_check(void) extern int dmar_table_init(void); extern int dmar_dev_scope_init(void); +extern void dmar_register_bus_notifier(void); extern int dmar_parse_dev_scope(void *start, void *end, int *cnt, struct dmar_dev_scope **devices, u16 segment); extern void *dmar_alloc_dev_scope(void *start, void *end, int *cnt); -- cgit v1.2.3 From 538d5b333216c3daa7a5821307164f10af73ec8c Mon Sep 17 00:00:00 2001 From: Tomasz Nowicki Date: Wed, 20 Sep 2017 10:52:02 +0200 Subject: iommu/iova: Make rcache flush optional on IOVA allocation failure Since IOVA allocation failure is not unusual case we need to flush CPUs' rcache in hope we will succeed in next round. However, it is useful to decide whether we need rcache flush step because of two reasons: - Not scalability. On large system with ~100 CPUs iterating and flushing rcache for each CPU becomes serious bottleneck so we may want to defer it. - free_cpu_cached_iovas() does not care about max PFN we are interested in. Thus we may flush our rcaches and still get no new IOVA like in the commonly used scenario: if (dma_limit > DMA_BIT_MASK(32) && dev_is_pci(dev)) iova = alloc_iova_fast(iovad, iova_len, DMA_BIT_MASK(32) >> shift); if (!iova) iova = alloc_iova_fast(iovad, iova_len, dma_limit >> shift); 1. First alloc_iova_fast() call is limited to DMA_BIT_MASK(32) to get PCI devices a SAC address 2. alloc_iova() fails due to full 32-bit space 3. rcaches contain PFNs out of 32-bit space so free_cpu_cached_iovas() throws entries away for nothing and alloc_iova() fails again 4. Next alloc_iova_fast() call cannot take advantage of rcache since we have just defeated caches. In this case we pick the slowest option to proceed. This patch reworks flushed_rcache local flag to be additional function argument instead and control rcache flush step. Also, it updates all users to do the flush as the last chance. Signed-off-by: Tomasz Nowicki Reviewed-by: Robin Murphy Tested-by: Nate Watterson Signed-off-by: Joerg Roedel --- include/linux/iova.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/iova.h b/include/linux/iova.h index c696ee81054e..928442dda565 100644 --- a/include/linux/iova.h +++ b/include/linux/iova.h @@ -150,7 +150,7 @@ void queue_iova(struct iova_domain *iovad, unsigned long pfn, unsigned long pages, unsigned long data); unsigned long alloc_iova_fast(struct iova_domain *iovad, unsigned long size, - unsigned long limit_pfn); + unsigned long limit_pfn, bool flush_rcache); struct iova *reserve_iova(struct iova_domain *iovad, unsigned long pfn_lo, unsigned long pfn_hi); void copy_reserved_iova(struct iova_domain *from, struct iova_domain *to); @@ -212,7 +212,8 @@ static inline void queue_iova(struct iova_domain *iovad, static inline unsigned long alloc_iova_fast(struct iova_domain *iovad, unsigned long size, - unsigned long limit_pfn) + unsigned long limit_pfn, + bool flush_rcache) { return 0; } -- cgit v1.2.3 From 973b546451fdf11e518cc96d1b137af893a38db5 Mon Sep 17 00:00:00 2001 From: Lu Baolu Date: Fri, 3 Nov 2017 10:51:33 -0600 Subject: iommu/vt-d: Clear Page Request Overflow fault bit Currently Page Request Overflow bit in IOMMU Fault Status register is not cleared. Not clearing this bit would mean that any future page-request is going to be automatically dropped by IOMMU. Suggested-by: Ashok Raj Signed-off-by: Lu Baolu Signed-off-by: Alex Williamson --- include/linux/intel-iommu.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux') diff --git a/include/linux/intel-iommu.h b/include/linux/intel-iommu.h index 485a5b48f038..f3274d9f46a2 100644 --- a/include/linux/intel-iommu.h +++ b/include/linux/intel-iommu.h @@ -212,6 +212,7 @@ #define DMA_FSTS_IQE (1 << 4) #define DMA_FSTS_ICE (1 << 5) #define DMA_FSTS_ITE (1 << 6) +#define DMA_FSTS_PRO (1 << 7) #define dma_fsts_fault_record_index(s) (((s) >> 8) & 0xff) /* FRCD_REG, 32 bits access */ -- cgit v1.2.3