[Xen-devel] [PATCH] Xen: Spread boot time page scrubbing across all available CPU's (v5)

[Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>]

From: Malcolm Crossley <malcolm.crossley@xxxxxxxxxx>

The page scrubbing is done in 128MB chunks in lockstep across all the
non-SMT CPU's. This allows for the boot CPU to hold the heap_lock whilst each
chunk is being scrubbed and then release the heap_lock when the CPU's are
finished scrubing their individual chunk. This allows for the heap_lock to
not be held continously and for pending softirqs are to be serviced
periodically across the CPU's.

The page scrub memory chunks are allocated to the CPU's in a NUMA aware
fashion to reduce socket interconnect overhead and improve performance.
Specifically in the first phase we scrub at the same time on all the
NUMA nodes that have CPUs - we also weed out the SMT threads so that
we only use cores (that gives a 50% boost). The second phase is for NUMA
nodes that have no CPUs - for that we use the closest NUMA node's CPUs
(non-SMT again) to do the job.

This patch reduces the boot page scrub time on a 128GB 64 core AMD Opteron
6386 machine from 49 seconds to 3 seconds.
On a IvyBridge-EX 8 socket box with 1.5TB it cuts it down from 15 minutes
to 63 seconds.

v2
 - Reduced default chunk size to 128MB
 - Added code to scrub NUMA nodes with no active CPU linked to them
 - Be robust to boot CPU not being linked to a NUMA node

v3:
 - Don't use SMT threads
 - Take care of remainder if the number of CPUs (or memory) is odd
 - Restructure the worker thread
 - s/u64/unsigned long/

v4:
 - Don't use all CPUs on non-CPU NUMA nodes, just closest one
 - Syntax, and docs updates
 - Compile on ARM
 - Fix bug when NUMA node has 0 pages

v5:
 - Properly figure out best NUMA node.
 - Fix comments to be proper style.
Signed-off-by: Malcolm Crossley <malcolm.crossley@xxxxxxxxxx>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
---
 docs/misc/xen-command-line.markdown |  10 ++
 xen/common/page_alloc.c             | 208 ++++++++++++++++++++++++++++++++----
 xen/include/asm-arm/numa.h          |   1 +
 3 files changed, 201 insertions(+), 18 deletions(-)

diff --git a/docs/misc/xen-command-line.markdown 
b/docs/misc/xen-command-line.markdown
index 514c7b2..39a67be 100644
--- a/docs/misc/xen-command-line.markdown
+++ b/docs/misc/xen-command-line.markdown
@@ -198,6 +198,16 @@ Scrub free RAM during boot.  This is a safety feature to 
prevent
 accidentally leaking sensitive VM data into other VMs if Xen crashes
 and reboots.
 
+### bootscrub_chunk_
+> `= <size>`
+
+> Default: `128MB`
+
+Maximum RAM block size chunks to be scrubbed whilst holding the page heap lock
+and not running softirqs. Reduce this if softirqs are not being run frequently
+enough. Setting this to a high value may cause cause boot failure, particularly
+if the NMI watchdog is also enabled.
+
 ### cachesize
 > `= <size>`
 
diff --git a/xen/common/page_alloc.c b/xen/common/page_alloc.c
index 601319c..d7390b6 100644
--- a/xen/common/page_alloc.c
+++ b/xen/common/page_alloc.c
@@ -65,6 +65,13 @@ static bool_t opt_bootscrub __initdata = 1;
 boolean_param("bootscrub", opt_bootscrub);
 
 /*
+ * bootscrub_chunk -> Amount of bytes to scrub lockstep on non-SMT CPUs
+ * on all NUMA nodes.
+ */
+static unsigned long __initdata opt_bootscrub_chunk = MB(128);
+size_param("bootscrub_chunk", opt_bootscrub_chunk);
+
+/*
  * Bit width of the DMA heap -- used to override NUMA-node-first.
  * allocation strategy, which can otherwise exhaust low memory.
  */
@@ -90,6 +97,16 @@ static struct bootmem_region {
 } *__initdata bootmem_region_list;
 static unsigned int __initdata nr_bootmem_regions;
 
+struct scrub_region {
+    unsigned long offset;
+    unsigned long start;
+    unsigned long per_cpu_sz;
+    unsigned long rem;
+    cpumask_t cpus;
+};
+static struct scrub_region __initdata region[MAX_NUMNODES];
+static unsigned long __initdata chunk_size;
+
 static void __init boot_bug(int line)
 {
     panic("Boot BUG at %s:%d", __FILE__, line);
@@ -1256,42 +1273,197 @@ void __init end_boot_allocator(void)
     printk("\n");
 }
 
+static void __init smp_scrub_heap_pages(void *data)
+{
+    unsigned long mfn, start, end;
+    struct page_info *pg;
+    struct scrub_region *r;
+    unsigned int temp_cpu, node, cpu_idx = 0;
+    unsigned int cpu = smp_processor_id();
+
+    if ( data )
+        r = data;
+    else
+    {
+        node = cpu_to_node(cpu);
+        if ( node == NUMA_NO_NODE )
+            return;
+        r = &region[node];
+    }
+
+    /* Determine the current CPU's index into CPU's linked to this node. */
+    for_each_cpu ( temp_cpu, &r->cpus )
+    {
+        if ( cpu == temp_cpu )
+            break;
+        cpu_idx++;
+    }
+
+    /* Calculate the starting mfn for this CPU's memory block. */
+    start = r->start + (r->per_cpu_sz * cpu_idx) + r->offset;
+
+    /* Calculate the end mfn into this CPU's memory block for this iteration. 
*/
+    if ( r->offset + chunk_size >= r->per_cpu_sz )
+    {
+        end = r->start + (r->per_cpu_sz * cpu_idx) + r->per_cpu_sz;
+
+        if ( r->rem && ((cpumask_weight(&r->cpus) - 1 == cpu_idx )) )
+            end += r->rem;
+    }
+    else
+        end = start + chunk_size;
+
+    for ( mfn = start; mfn < end; mfn++ )
+    {
+        pg = mfn_to_page(mfn);
+
+        /* Check the mfn is valid and page is free. */
+        if ( !mfn_valid(mfn) || !page_state_is(pg, free) )
+            continue;
+
+        scrub_one_page(pg);
+    }
+}
+
+static int __init find_non_smt(unsigned int node, cpumask_t *dest)
+{
+    cpumask_t node_cpus;
+    unsigned int i, cpu;
+
+    cpumask_and(&node_cpus, &node_to_cpumask(node), &cpu_online_map);
+    cpumask_clear(dest);
+    for_each_cpu ( i, &node_cpus )
+    {
+        if ( cpumask_intersects(dest, per_cpu(cpu_sibling_mask, i)) )
+            continue;
+        cpu = cpumask_first(per_cpu(cpu_sibling_mask, i));
+        cpumask_set_cpu(cpu, dest);
+    }
+    return cpumask_weight(dest);
+}
+
 /*
- * Scrub all unallocated pages in all heap zones. This function is more
- * convoluted than appears necessary because we do not want to continuously
- * hold the lock while scrubbing very large memory areas.
+ * Scrub all unallocated pages in all heap zones. This function uses all
+ * online cpu's to scrub the memory in parallel.
  */
 void __init scrub_heap_pages(void)
 {
-    unsigned long mfn;
-    struct page_info *pg;
+    cpumask_t node_cpus, all_worker_cpus;
+    unsigned int i, j;
+    unsigned long offset, max_per_cpu_sz = 0;
+    unsigned long start, end;
+    unsigned long rem = 0;
+    int last_distance, best_node;
 
     if ( !opt_bootscrub )
         return;
 
-    printk("Scrubbing Free RAM: ");
+    cpumask_clear(&all_worker_cpus);
+    /* Scrub block size. */
+    chunk_size = opt_bootscrub_chunk >> PAGE_SHIFT;
+    if ( chunk_size == 0 )
+        chunk_size = MB(128);
+
+    /* Round #0 - figure out amounts and which CPUs to use. */
+    for_each_online_node ( i )
+    {
+        if ( !node_spanned_pages(i) )
+            continue;
+        /* Calculate Node memory start and end address. */
+        start = max(node_start_pfn(i), first_valid_mfn);
+        end = min(node_start_pfn(i) + node_spanned_pages(i), max_page);
+        /* CPUs that are online and on this node (if none, that it is OK). */
+        find_non_smt(i, &node_cpus);
+        cpumask_or(&all_worker_cpus, &all_worker_cpus, &node_cpus);
+        if ( cpumask_empty(&node_cpus) )
+        {
+            /* No CPUs on this node. Round #2 will take of it. */
+            rem = 0;
+            region[i].per_cpu_sz = (end - start);
+        } else {
+            rem = (end - start) % cpumask_weight(&node_cpus);
+            region[i].per_cpu_sz = (end - start) / cpumask_weight(&node_cpus);
+            if ( region[i].per_cpu_sz > max_per_cpu_sz )
+                max_per_cpu_sz = region[i].per_cpu_sz;
+        }
+        region[i].start = start;
+        region[i].rem = rem;
+        cpumask_copy(&region[i].cpus, &node_cpus);
+
+    }
+
+    printk("Scrubbing Free RAM on %u nodes using %u CPUs\n", 
num_online_nodes(),
+           cpumask_weight(&all_worker_cpus));
 
-    for ( mfn = first_valid_mfn; mfn < max_page; mfn++ )
+    /* Round: #1 - do NUMA nodes with CPUs. */
+    for ( offset = 0; offset < max_per_cpu_sz; offset += chunk_size )
     {
+        for_each_online_node ( i )
+            region[i].offset = offset;
+
         process_pending_softirqs();
 
-        pg = mfn_to_page(mfn);
+        spin_lock(&heap_lock);
+        on_selected_cpus(&all_worker_cpus, smp_scrub_heap_pages, NULL, 1);
+        spin_unlock(&heap_lock);
 
-        /* Quick lock-free check. */
-        if ( !mfn_valid(mfn) || !page_state_is(pg, free) )
+        printk(".");
+    }
+
+    /*
+     * Round #2: NUMA nodes with no CPUs get scrubbed with CPUs on the node
+     * closest to us and with CPUs.
+     */
+    for_each_online_node ( i )
+    {
+        node_cpus = node_to_cpumask(i);
+
+        if ( !cpumask_empty(&node_cpus) )
             continue;
 
-        /* Every 100MB, print a progress dot. */
-        if ( (mfn % ((100*1024*1024)/PAGE_SIZE)) == 0 )
-            printk(".");
+        last_distance = INT_MAX;
+        best_node = first_node(node_online_map);
+        /* Figure out which NODE CPUs are close. */
+        for_each_online_node ( j )
+        {
+            int distance;
 
-        spin_lock(&heap_lock);
+            if ( i == j )
+                continue;
+            distance = __node_distance(i, j);
+            if ( distance < last_distance )
+            {
+                if ( cpumask_empty(&node_to_cpumask(j)) )
+                    continue;
+                last_distance = distance;
+                best_node = j;
+            }
+        }
+        /*
+         * Use CPUs from best node, and if there are no CPUs on the
+         * first node (the default) use the BSP.
+         */
+        if ( find_non_smt(best_node, &node_cpus) == 0 )
+            cpumask_set_cpu(smp_processor_id(), &node_cpus);
 
-        /* Re-check page status with lock held. */
-        if ( page_state_is(pg, free) )
-            scrub_one_page(pg);
+        /* We already have the node information from round #0. */
+        region[i].rem = region[i].per_cpu_sz % cpumask_weight(&node_cpus);
+        region[i].per_cpu_sz /= cpumask_weight(&node_cpus);
+        max_per_cpu_sz = region[i].per_cpu_sz;
+        cpumask_copy(&region[i].cpus, &node_cpus);
 
-        spin_unlock(&heap_lock);
+        for ( offset = 0; offset < max_per_cpu_sz; offset += chunk_size )
+        {
+            region[i].offset = offset;
+
+            process_pending_softirqs();
+
+            spin_lock(&heap_lock);
+            on_selected_cpus(&node_cpus, smp_scrub_heap_pages, &region[i], 1);
+            spin_unlock(&heap_lock);
+
+            printk(".");
+        }
     }
 
     printk("done.\n");
diff --git a/xen/include/asm-arm/numa.h b/xen/include/asm-arm/numa.h
index cb8f2ba..2c019d7 100644
--- a/xen/include/asm-arm/numa.h
+++ b/xen/include/asm-arm/numa.h
@@ -12,6 +12,7 @@ static inline __attribute__((pure)) int phys_to_nid(paddr_t 
addr)
 
 /* XXX: implement NUMA support */
 #define node_spanned_pages(nid) (total_pages)
+#define node_start_pfn(nid) (frametable_base_mfn)
 #define __node_distance(a, b) (20)
 
 #endif /* __ARCH_ARM_NUMA_H */
-- 
1.9.3


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel