[Xen-devel] [PATCH v11 2/3] x86/mem_sharing: reset a fork

[Tamas K Lengyel <tamas.lengyel@xxxxxxxxx>]

Implement hypercall that allows a fork to shed all memory that got allocated
for it during its execution and re-load its vCPU context from the parent VM.
This allows the forked VM to reset into the same state the parent VM is in a
faster way then creating a new fork would be. Measurements show about a 2x
speedup during normal fuzzing operations. Performance may vary depending how
much memory got allocated for the forked VM. If it has been completely
deduplicated from the parent VM then creating a new fork would likely be more
performant.

Signed-off-by: Tamas K Lengyel <tamas.lengyel@xxxxxxxxx>
---
v11: use copy_settings function to reset special pages
---
 xen/arch/x86/mm/mem_sharing.c | 115 ++++++++++++++++++++++++++++++++++
 xen/include/public/memory.h   |   4 ++
 2 files changed, 119 insertions(+)

diff --git a/xen/arch/x86/mm/mem_sharing.c b/xen/arch/x86/mm/mem_sharing.c
index 4a840d09dd..a458096b01 100644
--- a/xen/arch/x86/mm/mem_sharing.c
+++ b/xen/arch/x86/mm/mem_sharing.c
@@ -1775,6 +1775,91 @@ static int fork(struct domain *cd, struct domain *d)
     return rc;
 }
 
+/*
+ * The fork reset operation is intended to be used on short-lived forks only.
+ */
+static int fork_reset(struct domain *d, struct domain *cd,
+                      struct mem_sharing_op_fork_reset *fr)
+{
+    int rc = 0;
+    struct p2m_domain* p2m = p2m_get_hostp2m(cd);
+    struct page_info *page, *tmp;
+    unsigned long list_position = 0, preempt_count = 0, restart = fr->opaque;
+
+    domain_pause(cd);
+
+    page_list_for_each_safe(page, tmp, &cd->page_list)
+    {
+        p2m_type_t p2mt;
+        p2m_access_t p2ma;
+        gfn_t gfn;
+        mfn_t mfn;
+        bool shared = false;
+
+        list_position++;
+
+        /* Resume were we left of before preemption */
+        if ( restart && list_position < restart )
+            continue;
+
+        mfn = page_to_mfn(page);
+        if ( mfn_valid(mfn) )
+        {
+
+            gfn = mfn_to_gfn(cd, mfn);
+            mfn = __get_gfn_type_access(p2m, gfn_x(gfn), &p2mt, &p2ma,
+                                        0, NULL, false);
+
+            if ( p2m_is_ram(p2mt) && !p2m_is_shared(p2mt) )
+            {
+                /* take an extra reference, must work for a shared page */
+                if( !get_page(page, cd) )
+                {
+                    ASSERT_UNREACHABLE();
+                    return -EINVAL;
+                }
+
+                shared = true;
+                preempt_count += 0x10;
+
+                /*
+                 * Must succeed, it's a shared page that exists and
+                 * thus its size is guaranteed to be 4k so we are not splitting
+                 * large pages.
+                 */
+                rc = p2m->set_entry(p2m, gfn, INVALID_MFN, PAGE_ORDER_4K,
+                                    p2m_invalid, p2m_access_rwx, -1);
+                ASSERT(!rc);
+
+                put_page_alloc_ref(page);
+                put_page(page);
+            }
+        }
+
+        if ( !shared )
+            preempt_count++;
+
+        /* Preempt every 2MiB (shared) or 32MiB (unshared) - arbitrary. */
+        if ( preempt_count >= 0x2000 )
+        {
+            if ( hypercall_preempt_check() )
+            {
+                rc = -ERESTART;
+                break;
+            }
+            preempt_count = 0;
+        }
+    }
+
+    if ( rc )
+        fr->opaque = list_position;
+    else
+        rc = copy_settings(cd, d);
+
+    domain_unpause(cd);
+    return rc;
+}
+
 int mem_sharing_memop(XEN_GUEST_HANDLE_PARAM(xen_mem_sharing_op_t) arg)
 {
     int rc;
@@ -2066,6 +2151,36 @@ int 
mem_sharing_memop(XEN_GUEST_HANDLE_PARAM(xen_mem_sharing_op_t) arg)
         break;
     }
 
+    case XENMEM_sharing_op_fork_reset:
+    {
+        struct domain *pd;
+
+        rc = -ENOSYS;
+        if ( !mem_sharing_is_fork(d) )
+            goto out;
+
+        rc = rcu_lock_live_remote_domain_by_id(d->parent->domain_id, &pd);
+        if ( rc )
+            goto out;
+
+        rc = fork_reset(pd, d, &mso.u.fork_reset);
+
+        rcu_unlock_domain(pd);
+
+        if ( rc > 0 )
+        {
+            if ( __copy_to_guest(arg, &mso, 1) )
+                rc = -EFAULT;
+            else
+                rc = hypercall_create_continuation(__HYPERVISOR_memory_op,
+                                                   "lh", XENMEM_sharing_op,
+                                                   arg);
+        }
+        else
+            mso.u.fork_reset.opaque = 0;
+        break;
+    }
+
     default:
         rc = -ENOSYS;
         break;
diff --git a/xen/include/public/memory.h b/xen/include/public/memory.h
index c1dbad060e..7ca07c01dd 100644
--- a/xen/include/public/memory.h
+++ b/xen/include/public/memory.h
@@ -483,6 +483,7 @@ DEFINE_XEN_GUEST_HANDLE(xen_mem_access_op_t);
 #define XENMEM_sharing_op_audit             7
 #define XENMEM_sharing_op_range_share       8
 #define XENMEM_sharing_op_fork              9
+#define XENMEM_sharing_op_fork_reset        10
 
 #define XENMEM_SHARING_OP_S_HANDLE_INVALID  (-10)
 #define XENMEM_SHARING_OP_C_HANDLE_INVALID  (-9)
@@ -537,6 +538,9 @@ struct xen_mem_sharing_op {
             domid_t parent_domain;        /* IN: parent's domain id */
             uint16_t _pad[3];             /* Must be set to 0 */
         } fork;
+        struct mem_sharing_op_fork_reset {   /* OP_FORK_RESET */
+            uint64_aligned_t opaque;         /* Must be set to 0 */
+        } fork_reset;
     } u;
 };
 typedef struct xen_mem_sharing_op xen_mem_sharing_op_t;
-- 
2.20.1


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel