Skip to contentSkip to conversation

iommu/iova: Retry from last rb tree node if iova search fails

PATCH Applied [email protected] 4 participants 9 messages Sep 30, 2020 – Nov 17, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v5 1/2 iommu/iova: Retry from last rb tree node if iova search fails

From: Vijayanand Jitta <[email protected]>

When ever a new iova alloc request comes iova is always searched from the cached node and the nodes which are previous to cached node. So, even if there is free iova space available in the nodes which are next to the cached node iova allocation can still fail because of this approach.

Consider the following sequence of iova alloc and frees on 1GB of iova space

1) alloc - 500MB
2) alloc - 12MB
3) alloc - 499MB
4) free - 12MB which was allocated in step 2
5) alloc - 13MB

After the above sequence we will have 12MB of free iova space and cached node will be pointing to the iova pfn of last alloc of 13MB which will be the lowest iova pfn of that iova space. Now if we get an alloc request of 2MB we just search from cached node and then look for lower iova pfn's for free iova and as they aren't any, iova alloc fails though there is 12MB of free iova space.

To avoid such iova search failures do a retry from the last rb tree node when iova search fails, this will search the entire tree and get an iova if its available.

---
 drivers/iommu/iova.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)
drivers/iommu/iova.c +17 -6
index 30d969a..c3a1a8e 100644
@@ -184,8 +184,9 @@ static int __alloc_and_insert_iova_range(struct iova_domain *iovad,
 	struct rb_node *curr, *prev;
 	struct iova *curr_iova;
 	unsigned long flags;
-	unsigned long new_pfn;
+	unsigned long new_pfn, retry_pfn;
 	unsigned long align_mask = ~0UL;
+	unsigned long high_pfn = limit_pfn, low_pfn = iovad->start_pfn;
 
 	if (size_aligned)
 		align_mask <<= fls_long(size - 1);
@@ -198,15 +199,25 @@ static int __alloc_and_insert_iova_range(struct iova_domain *iovad,
 
 	curr = __get_cached_rbnode(iovad, limit_pfn);
 	curr_iova = rb_entry(curr, struct iova, node);
+	retry_pfn = curr_iova->pfn_hi + 1;
+
+retry:
 	do {
-		limit_pfn = min(limit_pfn, curr_iova->pfn_lo);
-		new_pfn = (limit_pfn - size) & align_mask;
+		high_pfn = min(high_pfn, curr_iova->pfn_lo);
+		new_pfn = (high_pfn - size) & align_mask;
 		prev = curr;
 		curr = rb_prev(curr);
 		curr_iova = rb_entry(curr, struct iova, node);
-	} while (curr && new_pfn <= curr_iova->pfn_hi);
-
-	if (limit_pfn < size || new_pfn < iovad->start_pfn) {
+	} while (curr && new_pfn <= curr_iova->pfn_hi && new_pfn >= low_pfn);
+
+	if (high_pfn < size || new_pfn < low_pfn) {
+		if (low_pfn == iovad->start_pfn && retry_pfn < limit_pfn) {
+			high_pfn = limit_pfn;
+			low_pfn = retry_pfn;
+			curr = &iovad->anchor.node;
+			curr_iova = rb_entry(curr, struct iova, node);
+			goto retry;
+		}
 		iovad->max32_alloc_size = size;
 		goto iova32_full;
 	}
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
2.7.4
#1 lore raw

PATCH v5 2/2 iommu/iova: Free global iova rcache on iova alloc failure

From: Vijayanand Jitta <[email protected]>

When ever an iova alloc request fails we free the iova ranges present in the percpu iova rcaches and then retry but the global iova rcache is not freed as a result we could still see iova alloc failure even after retry as global rcache is holding the iova's which can cause fragmentation. So, free the global iova rcache as well and then go for the retry.

---
 drivers/iommu/iova.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
drivers/iommu/iova.c +23 -0
index c3a1a8e..faf9b13 100644
@@ -25,6 +25,7 @@ static void init_iova_rcaches(struct iova_domain *iovad);
 static void free_iova_rcaches(struct iova_domain *iovad);
 static void fq_destroy_all_entries(struct iova_domain *iovad);
 static void fq_flush_timeout(struct timer_list *t);
+static void free_global_cached_iovas(struct iova_domain *iovad);
 
 void
 init_iova_domain(struct iova_domain *iovad, unsigned long granule,
@@ -442,6 +443,7 @@ alloc_iova_fast(struct iova_domain *iovad, unsigned long size,
 		flush_rcache = false;
 		for_each_online_cpu(cpu)
 			free_cpu_cached_iovas(cpu, iovad);
+		free_global_cached_iovas(iovad);
 		goto retry;
 	}
 
@@ -1057,5 +1059,26 @@ void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad)
 	}
 }
 
+/*
+ * free all the IOVA ranges of global cache
+ */
+static void free_global_cached_iovas(struct iova_domain *iovad)
+{
+	struct iova_rcache *rcache;
+	unsigned long flags;
+	int i, j;
+
+	for (i = 0; i < IOVA_RANGE_CACHE_MAX_SIZE; ++i) {
+		rcache = &iovad->rcaches[i];
+		spin_lock_irqsave(&rcache->lock, flags);
+		for (j = 0; j < rcache->depot_size; ++j) {
+			iova_magazine_free_pfns(rcache->depot[j], iovad);
+			iova_magazine_free(rcache->depot[j]);
+			rcache->depot[j] = NULL;
+		}
+		rcache->depot_size = 0;
+		spin_unlock_irqrestore(&rcache->lock, flags);
+	}
+}
 MODULE_AUTHOR("Anil S Keshavamurthy <[email protected]>");
 MODULE_LICENSE("GPL");
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
2.7.4
#7 lore raw
In reply to [email protected]
Quoted 71 lines · On 9/30/2020 1:14 PM, [email protected] wrote:

From: Vijayanand Jitta <[email protected]>

When ever a new iova alloc request comes iova is always searched from the cached node and the nodes which are previous to cached node. So, even if there is free iova space available in the nodes which are next to the cached node iova allocation can still fail because of this approach.

Consider the following sequence of iova alloc and frees on 1GB of iova space

1) alloc - 500MB
2) alloc - 12MB
3) alloc - 499MB
4) free - 12MB which was allocated in step 2
5) alloc - 13MB

After the above sequence we will have 12MB of free iova space and cached node will be pointing to the iova pfn of last alloc of 13MB which will be the lowest iova pfn of that iova space. Now if we get an alloc request of 2MB we just search from cached node and then look for lower iova pfn's for free iova and as they aren't any, iova alloc fails though there is 12MB of free iova space.

To avoid such iova search failures do a retry from the last rb tree node when iova search fails, this will search the entire tree and get an iova if its available.

---
 drivers/iommu/iova.c | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)
drivers/iommu/iova.c +10 -6
index 30d969a..c3a1a8e 100644
@@ -184,8 +184,9 @@ static int __alloc_and_insert_iova_range(struct iova_domain *iovad,
 	struct rb_node *curr, *prev;
 	struct iova *curr_iova;
 	unsigned long flags;
-	unsigned long new_pfn;
+	unsigned long new_pfn, retry_pfn;
 	unsigned long align_mask = ~0UL;
+	unsigned long high_pfn = limit_pfn, low_pfn = iovad->start_pfn;
 
 	if (size_aligned)
 		align_mask <<= fls_long(size - 1);
@@ -198,15 +199,25 @@ static int __alloc_and_insert_iova_range(struct iova_domain *iovad,
 
 	curr = __get_cached_rbnode(iovad, limit_pfn);
 	curr_iova = rb_entry(curr, struct iova, node);
+	retry_pfn = curr_iova->pfn_hi + 1;
+
+retry:
 	do {
-		limit_pfn = min(limit_pfn, curr_iova->pfn_lo);
-		new_pfn = (limit_pfn - size) & align_mask;
+		high_pfn = min(high_pfn, curr_iova->pfn_lo);
+		new_pfn = (high_pfn - size) & align_mask;
 		prev = curr;
 		curr = rb_prev(curr);
 		curr_iova = rb_entry(curr, struct iova, node);
-	} while (curr && new_pfn <= curr_iova->pfn_hi);
-
-	if (limit_pfn < size || new_pfn < iovad->start_pfn) {
+	} while (curr && new_pfn <= curr_iova->pfn_hi && new_pfn >= low_pfn);
+
+	if (high_pfn < size || new_pfn < low_pfn) {

drivers/iommu/iova.c static int __alloc_and_insert_iova_range(struct iova_domain *iovad, · line 223

Quoted 10 lines
drivers/iommu/iova.c +7 -0
+		if (low_pfn == iovad->start_pfn && retry_pfn < limit_pfn) {
+			high_pfn = limit_pfn;
+			low_pfn = retry_pfn;
+			curr = &iovad->anchor.node;
+			curr_iova = rb_entry(curr, struct iova, node);
+			goto retry;
+		}
 		iovad->max32_alloc_size = size;
 		goto iova32_full;
 	}

Gentle ping.

Thanks,
Vijay

Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a
member of Code Aurora Forum, hosted by The Linux Foundation
#2 lore raw

Re: [PATCH v5 2/2] iommu/iova: Free global iova rcache on iova alloc failure

In reply to [email protected]
Quoted 53 lines · On 9/30/2020 1:14 PM, [email protected] wrote:

From: Vijayanand Jitta <[email protected]>

When ever an iova alloc request fails we free the iova ranges present in the percpu iova rcaches and then retry but the global iova rcache is not freed as a result we could still see iova alloc failure even after retry as global rcache is holding the iova's which can cause fragmentation. So, free the global iova rcache as well and then go for the retry.

---
 drivers/iommu/iova.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)
drivers/iommu/iova.c +15 -0
index c3a1a8e..faf9b13 100644
@@ -25,6 +25,7 @@ static void init_iova_rcaches(struct iova_domain *iovad);
 static void free_iova_rcaches(struct iova_domain *iovad);
 static void fq_destroy_all_entries(struct iova_domain *iovad);
 static void fq_flush_timeout(struct timer_list *t);
+static void free_global_cached_iovas(struct iova_domain *iovad);
 
 void
 init_iova_domain(struct iova_domain *iovad, unsigned long granule,
@@ -442,6 +443,7 @@ alloc_iova_fast(struct iova_domain *iovad, unsigned long size,
 		flush_rcache = false;
 		for_each_online_cpu(cpu)
 			free_cpu_cached_iovas(cpu, iovad);
+		free_global_cached_iovas(iovad);
 		goto retry;
 	}
 
@@ -1057,5 +1059,26 @@ void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad)
 	}
 }
 
+/*
+ * free all the IOVA ranges of global cache
+ */
+static void free_global_cached_iovas(struct iova_domain *iovad)
+{
+	struct iova_rcache *rcache;
+	unsigned long flags;
+	int i, j;
+
+	for (i = 0; i < IOVA_RANGE_CACHE_MAX_SIZE; ++i) {
+		rcache = &iovad->rcaches[i];
+		spin_lock_irqsave(&rcache->lock, flags);
+		for (j = 0; j < rcache->depot_size; ++j) {

drivers/iommu/iova.c void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad) · line 1084

Quoted 10 lines
drivers/iommu/iova.c +8 -0
+			iova_magazine_free_pfns(rcache->depot[j], iovad);
+			iova_magazine_free(rcache->depot[j]);
+			rcache->depot[j] = NULL;
+		}
+		rcache->depot_size = 0;
+		spin_unlock_irqrestore(&rcache->lock, flags);
+	}
+}
 MODULE_AUTHOR("Anil S Keshavamurthy <[email protected]>");
 MODULE_LICENSE("GPL");

Gentle ping.

Thanks,
Vijay

Signature · 6 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a
member of Code Aurora Forum, hosted by The Linux Foundation
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
maintainer Reviewed-by on v5 2/2
#3 lore raw

Re: [PATCH v5 2/2] iommu/iova: Free global iova rcache on iova alloc failure

In reply to [email protected]
Quoted 9 lines · On 2020-09-30 08:44, [email protected] wrote:

From: Vijayanand Jitta <[email protected]>

When ever an iova alloc request fails we free the iova ranges present in the percpu iova rcaches and then retry but the global iova rcache is not freed as a result we could still see iova alloc failure even after retry as global rcache is holding the iova's which can cause fragmentation. So, free the global iova rcache as well and then go for the retry.

This looks reasonable to me - it's mildly annoying that we end up with so many similar-looking functions, but the necessary differences are right down in the middle of the loops so nothing can reasonably be factored out :(

Quoted 53 lines
---
  drivers/iommu/iova.c | 23 +++++++++++++++++++++++
  1 file changed, 23 insertions(+)
drivers/iommu/iova.c +23 -0
index c3a1a8e..faf9b13 100644
@@ -25,6 +25,7 @@ static void init_iova_rcaches(struct iova_domain *iovad);
  static void free_iova_rcaches(struct iova_domain *iovad);
  static void fq_destroy_all_entries(struct iova_domain *iovad);
  static void fq_flush_timeout(struct timer_list *t);
+static void free_global_cached_iovas(struct iova_domain *iovad);
 
  void
  init_iova_domain(struct iova_domain *iovad, unsigned long granule,
@@ -442,6 +443,7 @@ alloc_iova_fast(struct iova_domain *iovad, unsigned long size,
  		flush_rcache = false;
  		for_each_online_cpu(cpu)
  			free_cpu_cached_iovas(cpu, iovad);
+		free_global_cached_iovas(iovad);
  		goto retry;
  	}
 
@@ -1057,5 +1059,26 @@ void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad)
  	}
  }
 
+/*
+ * free all the IOVA ranges of global cache
+ */
+static void free_global_cached_iovas(struct iova_domain *iovad)
+{
+	struct iova_rcache *rcache;
+	unsigned long flags;
+	int i, j;
+
+	for (i = 0; i < IOVA_RANGE_CACHE_MAX_SIZE; ++i) {
+		rcache = &iovad->rcaches[i];
+		spin_lock_irqsave(&rcache->lock, flags);
+		for (j = 0; j < rcache->depot_size; ++j) {
+			iova_magazine_free_pfns(rcache->depot[j], iovad);
+			iova_magazine_free(rcache->depot[j]);
+			rcache->depot[j] = NULL;
+		}
+		rcache->depot_size = 0;
+		spin_unlock_irqrestore(&rcache->lock, flags);
+	}
+}
  MODULE_AUTHOR("Anil S Keshavamurthy <[email protected]>");
  MODULE_LICENSE("GPL");
#4 lore raw

Re: [PATCH v5 2/2] iommu/iova: Free global iova rcache on iova alloc failure

Quoted 11 lines · Robin Murphy
Quoted 9 lines · On 2020-09-30 08:44, [email protected] wrote:

From: Vijayanand Jitta <[email protected]>

When ever an iova alloc request fails we free the iova ranges present in the percpu iova rcaches and then retry but the global iova rcache is not freed as a result we could still see iova alloc failure even after retry as global rcache is holding the iova's which can cause fragmentation. So, free the global iova rcache as well and then go for the retry.

If we do clear all the CPU rcaches, it would nice to have something immediately available to replenish, i.e. use the global rcache, instead of flushing it, if that is not required...

Quoted 1 line

This looks reasonable to me - it's mildly annoying that we end up with so many similar-looking functions,

Well I did add a function to clear all CPU rcaches here, if you would like to check:

https://lore.kernel.org/linux-iommu/[email protected]/

Quoted 9 lines

but the necessary differences are right down in the middle of the loops so nothing can reasonably be factored out :(

Quoted 5 lines

---
  drivers/iommu/iova.c | 23 +++++++++++++++++++++++

  1 file changed, 23 insertions(+)
Quoted 10 lines

drivers/iommu/iova.c static void init_iova_rcaches(struct iova_domain · line 29

Quoted 10 lines
drivers/iommu/iova.c +1 -0
index c3a1a8e..faf9b13 100644
@@ -25,6 +25,7 @@ static void init_iova_rcaches(struct iova_domain 
 *iovad);
   static void free_iova_rcaches(struct iova_domain *iovad);
   static void fq_destroy_all_entries(struct iova_domain *iovad);
   static void fq_flush_timeout(struct timer_list *t);
+static void free_global_cached_iovas(struct iova_domain *iovad);

a thought: It would be great if the file could be rearranged at some point where we don't require so many forward declarations.

Quoted 22 lines
Quoted 22 lines

  void
  init_iova_domain(struct iova_domain *iovad, unsigned long granule,

drivers/iommu/iova.c +9 -0
@@ -442,6 +443,7 @@ alloc_iova_fast(struct iova_domain *iovad, 
 unsigned long size,
           flush_rcache = false;
           for_each_online_cpu(cpu)
               free_cpu_cached_iovas(cpu, iovad);
+        free_global_cached_iovas(iovad);
           goto retry;
       }
@@ -1057,5 +1059,26 @@ void free_cpu_cached_iovas(unsigned int cpu, 
 struct iova_domain *iovad)
       }
   }
+/*
+ * free all the IOVA ranges of global cache
+ */
+static void free_global_cached_iovas(struct iova_domain *iovad)
+{
+    struct iova_rcache *rcache;
+    unsigned long flags;
+    int i, j;
Quoted 8 lines

drivers/iommu/iova.c void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad) · line 1077

Quoted 8 lines
drivers/iommu/iova.c +8 -0
+
+    for (i = 0; i < IOVA_RANGE_CACHE_MAX_SIZE; ++i) {
+        rcache = &iovad->rcaches[i];
+        spin_lock_irqsave(&rcache->lock, flags);
+        for (j = 0; j < rcache->depot_size; ++j) {
+            iova_magazine_free_pfns(rcache->depot[j], iovad);
+            iova_magazine_free(rcache->depot[j]);
+            rcache->depot[j] = NULL;

I don't think that NULLify is strictly necessary

Quoted 8 lines

drivers/iommu/iova.c void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad) · line 1084

Quoted 8 lines
drivers/iommu/iova.c +5 -0
+        }
+        rcache->depot_size = 0;
+        spin_unlock_irqrestore(&rcache->lock, flags);
+    }
+}
   MODULE_AUTHOR("Anil S Keshavamurthy <[email protected]>");
   MODULE_LICENSE("GPL");
Quoted 5 lines

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
.

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu

maintainer
#5 lore raw

Re: [PATCH v5 2/2] iommu/iova: Free global iova rcache on iova alloc failure

Quoted 13 lines · John Garry
Quoted 11 lines · Robin Murphy
Quoted 9 lines · On 2020-09-30 08:44, [email protected] wrote:

From: Vijayanand Jitta <[email protected]>

When ever an iova alloc request fails we free the iova ranges present in the percpu iova rcaches and then retry but the global iova rcache is not freed as a result we could still see iova alloc failure even after retry as global rcache is holding the iova's which can cause fragmentation. So, free the global iova rcache as well and then go for the retry.

Quoted 1 line

If we do clear all the CPU rcaches, it would nice to have something immediately available to replenish, i.e. use the global rcache, instead of flushing it, if that is not required...

If we've reached the point of clearing *any* caches, though, I think any hope of maintaining performance is already long gone. We've walked the rbtree for the entire address space and found that it's still too full to allocate from; we're teetering on the brink of hard failure and this is a last-ditch attempt to claw back as much as possible in the hope that it gives us a usable space.

TBH I'm not entirely sure what allocation pattern was expected by the original code such that purging only some of the caches made sense, nor what kind of pattern leads to lots of smaller IOVAs being allocated, freed, and never reused to the point of blocking larger allocations, but either way the reasoning does at least seem to hold up in abstract.

Quoted 5 lines
Quoted 1 line

This looks reasonable to me - it's mildly annoying that we end up with so many similar-looking functions,

Well I did add a function to clear all CPU rcaches here, if you would like to check:

https://lore.kernel.org/linux-iommu/[email protected]/

I was thinking more of the way free_iova_rcaches(), free_cpu_cached_iovas(), and free_global_cached_iovas() all look pretty much the same shape at a glance.

Quoted 46 lines
Quoted 9 lines

but the necessary differences are right down in the middle of the loops so nothing can reasonably be factored out :(

Quoted 5 lines

---
  drivers/iommu/iova.c | 23 +++++++++++++++++++++++

  1 file changed, 23 insertions(+)
Quoted 10 lines

drivers/iommu/iova.c static void init_iova_rcaches(struct iova_domain · line 29

Quoted 10 lines
drivers/iommu/iova.c +1 -0
index c3a1a8e..faf9b13 100644
@@ -25,6 +25,7 @@ static void init_iova_rcaches(struct iova_domain 
 *iovad);
   static void free_iova_rcaches(struct iova_domain *iovad);
   static void fq_destroy_all_entries(struct iova_domain *iovad);
   static void fq_flush_timeout(struct timer_list *t);
+static void free_global_cached_iovas(struct iova_domain *iovad);

a thought: It would be great if the file could be rearranged at some point where we don't require so many forward declarations.

Quoted 24 lines
Quoted 24 lines

  void
  init_iova_domain(struct iova_domain *iovad, unsigned long granule,

drivers/iommu/iova.c +11 -0
@@ -442,6 +443,7 @@ alloc_iova_fast(struct iova_domain *iovad, 
 unsigned long size,
           flush_rcache = false;
           for_each_online_cpu(cpu)
               free_cpu_cached_iovas(cpu, iovad);
+        free_global_cached_iovas(iovad);
           goto retry;
       }
@@ -1057,5 +1059,26 @@ void free_cpu_cached_iovas(unsigned int cpu, 
 struct iova_domain *iovad)
       }
   }
+/*
+ * free all the IOVA ranges of global cache
+ */
+static void free_global_cached_iovas(struct iova_domain *iovad)
+{
+    struct iova_rcache *rcache;
+    unsigned long flags;
+    int i, j;
+
+    for (i = 0; i < IOVA_RANGE_CACHE_MAX_SIZE; ++i) {
Quoted 8 lines
Quoted 6 lines

drivers/iommu/iova.c void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad) · line 1077

Quoted 6 lines
drivers/iommu/iova.c +6 -0
+        rcache = &iovad->rcaches[i];
+        spin_lock_irqsave(&rcache->lock, flags);
+        for (j = 0; j < rcache->depot_size; ++j) {
+            iova_magazine_free_pfns(rcache->depot[j], iovad);
+            iova_magazine_free(rcache->depot[j]);
+            rcache->depot[j] = NULL;

I don't think that NULLify is strictly necessary

True, we don't explicitly clear depot entries in __iova_rcache_get() for normal operation, so there's not much point in doing so here.

Robin.

Quoted 14 lines
Quoted 13 lines

drivers/iommu/iova.c void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad) · line 1084

Quoted 8 lines
drivers/iommu/iova.c +5 -0
+        }
+        rcache->depot_size = 0;
+        spin_unlock_irqrestore(&rcache->lock, flags);
+    }
+}
   MODULE_AUTHOR("Anil S Keshavamurthy <[email protected]>");
   MODULE_LICENSE("GPL");

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
.

Acked-by
#6 lore raw

Re: [PATCH v5 2/2] iommu/iova: Free global iova rcache on iova alloc failure

Quoted 10 lines · Robin Murphy
Quoted 7 lines
Quoted 5 lines
Quoted 4 lines

alloc failure even after retry as global
rcache is holding the iova's which can cause fragmentation. So, free the global iova rcache as well and then go for the retry.

If we do clear all the CPU rcaches, it would nice to have something immediately available to replenish, i.e. use the global rcache, instead of flushing it, if that is not required...

If we've reached the point of clearing *any* caches, though, I think any hope of maintaining performance is already long gone. We've walked the rbtree for the entire address space and found that it's still too full to allocate from; we're teetering on the brink of hard failure and this is a last-ditch attempt to claw back as much as possible in the hope that it gives us a usable space. > TBH I'm not entirely sure what allocation pattern was expected by the original code such that purging only some of the caches made sense,

I'd say that the assumption is that once the CPU rcaches are flushed, then we should have space again. No need to go any further.

Quoted 1 line

nor what kind of pattern leads to lots of smaller IOVAs being allocated, freed, and never reused to the point of blocking larger allocations, but either way the reasoning does at least seem to hold up in abstract.

Ok, but I'd like to see that hard failure (if you get my meaning). Flushing the depot rcache may be papering over some other bug.

Either way, I don't feel to strongly, so if you're happy then I won't try to block, so [apart from comment, below]:

Quoted 65 lines
Quoted 6 lines
Quoted 2 lines

This looks reasonable to me - it's mildly annoying that we end up with so many similar-looking functions,

Well I did add a function to clear all CPU rcaches here, if you would like to check:

https://lore.kernel.org/linux-iommu/[email protected]/

I was thinking more of the way free_iova_rcaches(), free_cpu_cached_iovas(), and free_global_cached_iovas() all look pretty much the same shape at a glance.

Quoted 46 lines
Quoted 9 lines

but the necessary differences are right down in the middle of the loops so nothing can reasonably be factored out :(

Quoted 5 lines

---
  drivers/iommu/iova.c | 23 +++++++++++++++++++++++

  1 file changed, 23 insertions(+)
Quoted 10 lines

drivers/iommu/iova.c static void init_iova_rcaches(struct iova_domain · line 29

Quoted 10 lines
drivers/iommu/iova.c +1 -0
index c3a1a8e..faf9b13 100644
@@ -25,6 +25,7 @@ static void init_iova_rcaches(struct iova_domain 
 *iovad);
   static void free_iova_rcaches(struct iova_domain *iovad);
   static void fq_destroy_all_entries(struct iova_domain *iovad);
   static void fq_flush_timeout(struct timer_list *t);
+static void free_global_cached_iovas(struct iova_domain *iovad);

a thought: It would be great if the file could be rearranged at some point where we don't require so many forward declarations.

Quoted 24 lines
Quoted 24 lines

  void
  init_iova_domain(struct iova_domain *iovad, unsigned long granule,

drivers/iommu/iova.c +11 -0
@@ -442,6 +443,7 @@ alloc_iova_fast(struct iova_domain *iovad, 
 unsigned long size,
           flush_rcache = false;
           for_each_online_cpu(cpu)
               free_cpu_cached_iovas(cpu, iovad);
+        free_global_cached_iovas(iovad);
           goto retry;
       }
@@ -1057,5 +1059,26 @@ void free_cpu_cached_iovas(unsigned int cpu, 
 struct iova_domain *iovad)
       }
   }
+/*
+ * free all the IOVA ranges of global cache
+ */
+static void free_global_cached_iovas(struct iova_domain *iovad)
+{
+    struct iova_rcache *rcache;
+    unsigned long flags;
+    int i, j;
+
+    for (i = 0; i < IOVA_RANGE_CACHE_MAX_SIZE; ++i) {
Quoted 8 lines
Quoted 6 lines

drivers/iommu/iova.c void free_cpu_cached_iovas(unsigned int cpu, struct iova_domain *iovad) · line 1077

Quoted 6 lines
drivers/iommu/iova.c +6 -0
+        rcache = &iovad->rcaches[i];
+        spin_lock_irqsave(&rcache->lock, flags);
+        for (j = 0; j < rcache->depot_size; ++j) {
+            iova_magazine_free_pfns(rcache->depot[j], iovad);
+            iova_magazine_free(rcache->depot[j]);
+            rcache->depot[j] = NULL;

I don't think that NULLify is strictly necessary

Quoted 1 line

True, we don't explicitly clear depot entries in __iova_rcache_get() for normal operation, so there's not much point in doing so here.

Right, so for consistency, I think that it would be nice not to NULLify, for consistency.

Quoted 2 lines

Robin.

maintainer Applied
#8 lore raw
In reply to [email protected]
Quoted 10 lines · On Wed, 30 Sep 2020 13:14:23 +0530, [email protected] wrote:

When ever a new iova alloc request comes iova is always searched from the cached node and the nodes which are previous to cached node. So, even if there is free iova space available in the nodes which are next to the cached node iova allocation can still fail because of this approach.

Consider the following sequence of iova alloc and frees on 1GB of iova space

[...]

Applied to arm64 (for-next/iommu/iova), thanks!

[1/2] iommu/iova: Retry from last rb tree node if iova search fails

      https://git.kernel.org/arm64/c/4e89dce72521
[2/2] iommu/iova: Free global iova rcache on iova alloc failure
      https://git.kernel.org/arm64/c/6fa3525b455a

Cheers,

Signature · 9 lines
Will

https://fixes.arm64.dev
https://next.arm64.dev
https://will.arm64.dev
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu