xen/5583da64-gnttab-use-per-VCPU-maptrack-free-lists.patch
Charles Arnold 8724a18868 - bnc#935634 - VUL-0: CVE-2015-3259: xen: XSA-137: xl command line
config handling stack overflow
  55a62eb0-xl-correct-handling-of-extra_config-in-main_cpupoolcreate.patch

- bsc#907514 - Bus fatal error & sles12 sudden reboot has been
  observed
- bsc#910258 - SLES12 Xen host crashes with FATAL NMI after
  shutdown of guest with VT-d NIC
- bsc#918984 - Bus fatal error & sles11-SP4 sudden reboot has been
  observed
- bsc#923967 - Partner-L3: Bus fatal error & sles11-SP3 sudden
  reboot has been observed
  552d293b-x86-vMSI-X-honor-all-mask-requests.patch
  552d2966-x86-vMSI-X-add-valid-bits-for-read-acceleration.patch
  5576f143-x86-adjust-PV-I-O-emulation-functions-types.patch
  55795a52-x86-vMSI-X-support-qword-MMIO-access.patch
  5583d9c5-x86-MSI-X-cleanup.patch
  5583da09-x86-MSI-track-host-and-guest-masking-separately.patch
  55b0a218-x86-PCI-CFG-write-intercept.patch
  55b0a255-x86-MSI-X-maskall.patch
  55b0a283-x86-MSI-X-teardown.patch
  55b0a2ab-x86-MSI-X-enable.patch
  55b0a2db-x86-MSI-track-guest-masking.patch
- Upstream patches from Jan 
  552d0f49-x86-traps-identify-the-vcpu-in-context-when-dumping-regs.patch
  559bc633-x86-cpupool-clear-proper-cpu_valid-bit-on-CPU-teardown.patch
  559bc64e-credit1-properly-deal-with-CPUs-not-in-any-pool.patch
  559bc87f-x86-hvmloader-avoid-data-corruption-with-xenstore-rw.patch
  55a66a1e-make-rangeset_report_ranges-report-all-ranges.patch
  55a77e4f-dmar-device-scope-mem-leak-fix.patch

OBS-URL: https://build.opensuse.org/package/show/Virtualization/xen?expand=0&rev=373
2015-08-26 22:28:15 +00:00

285 lines
9.2 KiB
Diff

# Commit dff515dfeac4c1c13422a128c558ac21ddc6c8db
# Date 2015-06-19 11:01:24 +0200
# Author Malcolm Crossley <malcolm.crossley@citrix.com>
# Committer Jan Beulich <jbeulich@suse.com>
gnttab: use per-VCPU maptrack free lists
Performance analysis of aggregate network throughput with many VMs
shows that performance is signficantly limited by contention on the
maptrack lock when obtaining/releasing maptrack handles from the free
list.
Instead of a single free list use a per-VCPU list. This avoids any
contention when obtaining a handle. Handles must be released back to
their original list and since this may occur on a different VCPU there
is some contention on the destination VCPU's free list tail pointer
(but this is much better than a per-domain lock).
Increase the default maximum number of maptrack frames by 4 times
because: a) struct grant_mapping is now 16 bytes (instead of 8); and
b) a guest may not evenly distribute all the grant map operations
across the VCPUs (meaning some VCPUs need more maptrack entries than
others).
Signed-off-by: Malcolm Crossley <malcolm.crossley@citrix.com>
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
--- a/xen/common/domain.c
+++ b/xen/common/domain.c
@@ -126,6 +126,8 @@ struct vcpu *alloc_vcpu(
tasklet_init(&v->continue_hypercall_tasklet, NULL, 0);
+ grant_table_init_vcpu(v);
+
if ( !zalloc_cpumask_var(&v->cpu_hard_affinity) ||
!zalloc_cpumask_var(&v->cpu_hard_affinity_tmp) ||
!zalloc_cpumask_var(&v->cpu_hard_affinity_saved) ||
--- a/xen/common/grant_table.c
+++ b/xen/common/grant_table.c
@@ -37,6 +37,7 @@
#include <xen/iommu.h>
#include <xen/paging.h>
#include <xen/keyhandler.h>
+#include <xen/vmap.h>
#include <xsm/xsm.h>
#include <asm/flushtlb.h>
@@ -57,7 +58,7 @@ integer_param("gnttab_max_frames", max_g
* New options allow to set max_maptrack_frames and
* map_grant_table_frames independently.
*/
-#define DEFAULT_MAX_MAPTRACK_FRAMES 256
+#define DEFAULT_MAX_MAPTRACK_FRAMES 1024
static unsigned int __read_mostly max_maptrack_frames;
integer_param("gnttab_max_maptrack_frames", max_maptrack_frames);
@@ -279,62 +280,103 @@ double_gt_unlock(struct grant_table *lgt
static inline int
__get_maptrack_handle(
- struct grant_table *t)
+ struct grant_table *t,
+ struct vcpu *v)
{
- unsigned int h;
- if ( unlikely((h = t->maptrack_head) == MAPTRACK_TAIL) )
+ unsigned int head, next;
+
+ /* No maptrack pages allocated for this VCPU yet? */
+ head = v->maptrack_head;
+ if ( unlikely(head == MAPTRACK_TAIL) )
return -1;
- t->maptrack_head = maptrack_entry(t, h).ref;
- return h;
+
+ /*
+ * Always keep one entry in the free list to make it easier to add
+ * free entries to the tail.
+ */
+ next = read_atomic(&maptrack_entry(t, head).ref);
+ if ( unlikely(next == MAPTRACK_TAIL) )
+ return -1;
+
+ v->maptrack_head = next;
+
+ return head;
}
static inline void
put_maptrack_handle(
struct grant_table *t, int handle)
{
- spin_lock(&t->maptrack_lock);
- maptrack_entry(t, handle).ref = t->maptrack_head;
- t->maptrack_head = handle;
- spin_unlock(&t->maptrack_lock);
+ struct domain *currd = current->domain;
+ struct vcpu *v;
+ unsigned int prev_tail, cur_tail;
+
+ /* 1. Set entry to be a tail. */
+ maptrack_entry(t, handle).ref = MAPTRACK_TAIL;
+
+ /* 2. Add entry to the tail of the list on the original VCPU. */
+ v = currd->vcpu[maptrack_entry(t, handle).vcpu];
+
+ cur_tail = read_atomic(&v->maptrack_tail);
+ do {
+ prev_tail = cur_tail;
+ cur_tail = cmpxchg(&v->maptrack_tail, prev_tail, handle);
+ } while ( cur_tail != prev_tail );
+
+ /* 3. Update the old tail entry to point to the new entry. */
+ write_atomic(&maptrack_entry(t, prev_tail).ref, handle);
}
static inline int
get_maptrack_handle(
struct grant_table *lgt)
{
+ struct vcpu *curr = current;
int i;
grant_handle_t handle;
struct grant_mapping *new_mt;
- unsigned int new_mt_limit, nr_frames;
+
+ handle = __get_maptrack_handle(lgt, curr);
+ if ( likely(handle != -1) )
+ return handle;
spin_lock(&lgt->maptrack_lock);
- while ( unlikely((handle = __get_maptrack_handle(lgt)) == -1) )
+ if ( nr_maptrack_frames(lgt) >= max_maptrack_frames )
{
- nr_frames = nr_maptrack_frames(lgt);
- if ( nr_frames >= max_maptrack_frames )
- break;
+ spin_unlock(&lgt->maptrack_lock);
+ return -1;
+ }
- new_mt = alloc_xenheap_page();
- if ( !new_mt )
- break;
+ new_mt = alloc_xenheap_page();
+ if ( !new_mt )
+ {
+ spin_unlock(&lgt->maptrack_lock);
+ return -1;
+ }
+ clear_page(new_mt);
- clear_page(new_mt);
+ /*
+ * Use the first new entry and add the remaining entries to the
+ * head of the free list.
+ */
+ handle = lgt->maptrack_limit;
- new_mt_limit = lgt->maptrack_limit + MAPTRACK_PER_PAGE;
+ for ( i = 0; i < MAPTRACK_PER_PAGE; i++ )
+ {
+ new_mt[i].ref = handle + i + 1;
+ new_mt[i].vcpu = curr->vcpu_id;
+ }
+ new_mt[i - 1].ref = curr->maptrack_head;
- for ( i = 1; i < MAPTRACK_PER_PAGE; i++ )
- new_mt[i - 1].ref = lgt->maptrack_limit + i;
- new_mt[i - 1].ref = lgt->maptrack_head;
- lgt->maptrack_head = lgt->maptrack_limit;
+ /* Set tail directly if this is the first page for this VCPU. */
+ if ( curr->maptrack_tail == MAPTRACK_TAIL )
+ curr->maptrack_tail = handle + MAPTRACK_PER_PAGE - 1;
- lgt->maptrack[nr_frames] = new_mt;
- smp_wmb();
- lgt->maptrack_limit = new_mt_limit;
+ curr->maptrack_head = handle + 1;
- gdprintk(XENLOG_INFO, "Increased maptrack size to %u frames\n",
- nr_frames + 1);
- }
+ lgt->maptrack[nr_maptrack_frames(lgt)] = new_mt;
+ lgt->maptrack_limit += MAPTRACK_PER_PAGE;
spin_unlock(&lgt->maptrack_lock);
@@ -2919,16 +2961,9 @@ grant_table_create(
}
/* Tracking of mapped foreign frames table */
- if ( (t->maptrack = xzalloc_array(struct grant_mapping *,
- max_maptrack_frames)) == NULL )
+ t->maptrack = vzalloc(max_maptrack_frames * sizeof(*t->maptrack));
+ if ( t->maptrack == NULL )
goto no_mem_2;
- if ( (t->maptrack[0] = alloc_xenheap_page()) == NULL )
- goto no_mem_3;
- clear_page(t->maptrack[0]);
- t->maptrack_limit = MAPTRACK_PER_PAGE;
- for ( i = 1; i < MAPTRACK_PER_PAGE; i++ )
- t->maptrack[0][i - 1].ref = i;
- t->maptrack[0][i - 1].ref = MAPTRACK_TAIL;
/* Shared grant table. */
if ( (t->shared_raw = xzalloc_array(void *, max_grant_frames)) == NULL )
@@ -2960,8 +2995,7 @@ grant_table_create(
free_xenheap_page(t->shared_raw[i]);
xfree(t->shared_raw);
no_mem_3:
- free_xenheap_page(t->maptrack[0]);
- xfree(t->maptrack);
+ vfree(t->maptrack);
no_mem_2:
for ( i = 0;
i < num_act_frames_from_sha_frames(INITIAL_NR_GRANT_FRAMES); i++ )
@@ -3096,7 +3130,7 @@ grant_table_destroy(
for ( i = 0; i < nr_maptrack_frames(t); i++ )
free_xenheap_page(t->maptrack[i]);
- xfree(t->maptrack);
+ vfree(t->maptrack);
for ( i = 0; i < nr_active_grant_frames(t); i++ )
free_xenheap_page(t->active[i]);
@@ -3110,6 +3144,12 @@ grant_table_destroy(
d->grant_table = NULL;
}
+void grant_table_init_vcpu(struct vcpu *v)
+{
+ v->maptrack_head = MAPTRACK_TAIL;
+ v->maptrack_tail = MAPTRACK_TAIL;
+}
+
static void gnttab_usage_print(struct domain *rd)
{
int first = 1;
--- a/xen/include/xen/grant_table.h
+++ b/xen/include/xen/grant_table.h
@@ -60,6 +60,8 @@ struct grant_mapping {
u32 ref; /* grant ref */
u16 flags; /* 0-4: GNTMAP_* ; 5-15: unused */
domid_t domid; /* granting domain */
+ u32 vcpu; /* vcpu which created the grant mapping */
+ u32 pad; /* round size to a power of 2 */
};
/* Per-domain grant information. */
@@ -83,9 +85,8 @@ struct grant_table {
grant_status_t **status;
/* Active grant table. */
struct active_grant_entry **active;
- /* Mapping tracking table. */
+ /* Mapping tracking table per vcpu. */
struct grant_mapping **maptrack;
- unsigned int maptrack_head;
unsigned int maptrack_limit;
/* Lock protecting the maptrack page list, head, and limit */
spinlock_t maptrack_lock;
@@ -99,6 +100,7 @@ int grant_table_create(
struct domain *d);
void grant_table_destroy(
struct domain *d);
+void grant_table_init_vcpu(struct vcpu *v);
/* Domain death release of granted mappings of other domains' memory. */
void
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -219,6 +219,10 @@ struct vcpu
/* VCPU paused by system controller. */
int controller_pause_count;
+ /* Maptrack */
+ unsigned int maptrack_head;
+ unsigned int maptrack_tail;
+
/* IRQ-safe virq_lock protects against delivering VIRQ to stale evtchn. */
evtchn_port_t virq_to_evtchn[NR_VIRQS];
spinlock_t virq_lock;