[Qemu-devel] [PATCH v2 2/2] numa, spapr: equally distribute memory on nodes

[Laurent Vivier]

When there are more nodes than memory available to put the minimum
allowed memory by node, all the memory is put on the last node.

This is because we put (ram_size / nb_numa_nodes) &
~((1 << mc->numa_mem_align_shift) - 1); on each node, and in this
case the value is 0. This is particularly true with pseries,
as the memory must be aligned to 256MB.

To avoid this problem, this patch uses an error diffusion algorithm [1]
to distribute equally the memory on nodes.

Example:

qemu-system-ppc64 -S -nographic  -nodefaults -monitor stdio -m 1G -smp 8 \
                  -numa node -numa node -numa node \
                  -numa node -numa node -numa node

Before:

(qemu) info numa
6 nodes
node 0 cpus: 0 6
node 0 size: 0 MB
node 1 cpus: 1 7
node 1 size: 0 MB
node 2 cpus: 2
node 2 size: 0 MB
node 3 cpus: 3
node 3 size: 0 MB
node 4 cpus: 4
node 4 size: 0 MB
node 5 cpus: 5
node 5 size: 1024 MB

After:
(qemu) info numa
6 nodes
node 0 cpus: 0 6
node 0 size: 0 MB
node 1 cpus: 1 7
node 1 size: 256 MB
node 2 cpus: 2
node 2 size: 0 MB
node 3 cpus: 3
node 3 size: 256 MB
node 4 cpus: 4
node 4 size: 256 MB
node 5 cpus: 5
node 5 size: 256 MB

[1] https://en.wikipedia.org/wiki/Error_diffusion

Signed-off-by: Laurent Vivier <address@hidden>
---
 hw/ppc/spapr.c         | 21 ++++++++++++++++++++-
 include/hw/ppc/spapr.h |  3 ++-
 2 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
index 80d12d0..be498e2 100644
--- a/hw/ppc/spapr.c
+++ b/hw/ppc/spapr.c
@@ -3106,6 +3106,23 @@ static void spapr_pic_print_info(InterruptStatsProvider 
*obj,
     ics_pic_print_info(spapr->ics, mon);
 }
 
+static void spapr_numa_auto_assign_ram(uint64_t *nodes, int nb_nodes,
+                                       ram_addr_t size)
+{
+    int i;
+    uint64_t usedmem = 0, node_mem;
+    uint64_t granularity = size / nb_nodes;
+    uint64_t propagate = 0;
+
+    for (i = 0; i < nb_nodes - 1; i++) {
+        node_mem = (granularity + propagate) & ~(SPAPR_MEMORY_BLOCK_SIZE - 1);
+        propagate = granularity + propagate - node_mem;
+        nodes[i] = node_mem;
+        usedmem += node_mem;
+    }
+    nodes[i] = ram_size - usedmem;
+}
+
 static void spapr_machine_class_init(ObjectClass *oc, void *data)
 {
     MachineClass *mc = MACHINE_CLASS(oc);
@@ -3162,7 +3179,8 @@ static void spapr_machine_class_init(ObjectClass *oc, 
void *data)
      * SPAPR_MEMORY_BLOCK_SIZE (256M) since that's the granularity
      * in which LMBs are represented and hot-added
      */
-    mc->numa_mem_align_shift = 28;
+    mc->numa_mem_align_shift = SPAPR_MEMORY_BLOCK_SIZE_SHIFT;
+    mc->numa_auto_assign_ram = spapr_numa_auto_assign_ram;
 }
 
 static const TypeInfo spapr_machine_info = {
@@ -3242,6 +3260,7 @@ static void spapr_machine_2_9_class_options(MachineClass 
*mc)
 {
     spapr_machine_2_10_class_options(mc);
     SET_MACHINE_COMPAT(mc, SPAPR_COMPAT_2_9);
+    mc->numa_auto_assign_ram = NULL;
 }
 
 DEFINE_SPAPR_MACHINE(2_9, "2.9", false);
diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
index 5802f88..8f4a588 100644
--- a/include/hw/ppc/spapr.h
+++ b/include/hw/ppc/spapr.h
@@ -653,7 +653,8 @@ int spapr_rtc_import_offset(sPAPRRTCState *rtc, int64_t 
legacy_offset);
 
 int spapr_rng_populate_dt(void *fdt);
 
-#define SPAPR_MEMORY_BLOCK_SIZE (1 << 28) /* 256MB */
+#define SPAPR_MEMORY_BLOCK_SIZE_SHIFT 28 /* 256MB */
+#define SPAPR_MEMORY_BLOCK_SIZE (1 << SPAPR_MEMORY_BLOCK_SIZE_SHIFT)
 
 /*
  * This defines the maximum number of DIMM slots we can have for sPAPR
-- 
2.9.3