[pve-devel] [PATCH qemu-server] memory: hugepages: map numa node IDs to host and guest correctly

Alexandre DERUMIER aderumier at odiso.com
Tue Aug 16 08:01:45 CEST 2016


Hi Wolfgang, sorry to be late, I was offline last week (some more holidays)

I'll try to test it this week.


----- Mail original -----
De: "Wolfgang Bumiller" <w.bumiller at proxmox.com>
À: "pve-devel" <pve-devel at pve.proxmox.com>
Cc: "aderumier" <aderumier at odiso.com>
Envoyé: Vendredi 29 Juillet 2016 13:40:54
Objet: [PATCH qemu-server] memory: hugepages: map numa node IDs to host and guest correctly

foreach_dimm() provides a guest numa node index, when used 
in conjunction with the guest-to-host numa node topology 
mapping one has to make sure that the correct host-side 
indices are used. 

This covers situations where the user defines a numaX with 
hostnodes=Y with Y != X. 

For instance: 
cores: 2 
hotplug: disk,network,cpu,memory 
hugepages: 2 
memory: 2048 
numa: 1 
numa1: memory=512,hostnodes=0,cpus=0-1,policy=bind 
numa2: memory=512,hostnodes=0,cpus=2-3,policy=bind 

Both numa IDs 1 and 2 passed by foreach_dimm() have to be 
mapped to host node 0. 

Note that this also reverses the foreach_reverse_dimm() numa 
node numbering as the current code, while walking sizes 
backwards, walked the numa IDs inside each size forward, 
which makes more sense. (Memory hot-unplug is still working 
with this.) 
--- 
Note: this is to be applied on top of Alexandre's patch #2: 
hugepages: use hostnodes value as numanode for topology 

@Alexandre: Please review if you get the chance. 

PVE/QemuServer/Memory.pm | 44 +++++++++++++++++++++++++++++++++++++++++--- 
1 file changed, 41 insertions(+), 3 deletions(-) 

diff --git a/PVE/QemuServer/Memory.pm b/PVE/QemuServer/Memory.pm 
index 37f285a..087584b 100644 
--- a/PVE/QemuServer/Memory.pm 
+++ b/PVE/QemuServer/Memory.pm 
@@ -10,6 +10,33 @@ my $MAX_NUMA = 8; 
my $MAX_MEM = 4194304; 
my $STATICMEM = 1024; 

+sub get_numa_node_list { 
+ my ($conf) = @_; 
+ my @numa_map; 
+ for (my $i = 0; $i < $MAX_NUMA; $i++) { 
+ my $entry = $conf->{"numa$i"} or next; 
+ my $numa = PVE::QemuServer::parse_numa($entry) or next; 
+ push @numa_map, $i; 
+ } 
+ return @numa_map if @numa_map; 
+ my $sockets = $conf->{sockets} || 1; 
+ return (0..($sockets-1)); 
+} 
+ 
+# only valid when numa nodes map to a single host node 
+sub get_numa_guest_to_host_map { 
+ my ($conf) = @_; 
+ my $map = {}; 
+ for (my $i = 0; $i < $MAX_NUMA; $i++) { 
+ my $entry = $conf->{"numa$i"} or next; 
+ my $numa = PVE::QemuServer::parse_numa($entry) or next; 
+ $map->{$i} = print_numa_hostnodes($numa->{hostnodes}); 
+ } 
+ return $map if %$map; 
+ my $sockets = $conf->{sockets} || 1; 
+ return map { $_ => $_ } (0..($sockets-1)); 
+} 
+ 
sub foreach_dimm{ 
my ($conf, $vmid, $memory, $sockets, $func) = @_; 

@@ -27,11 +54,13 @@ sub foreach_dimm{ 

return if $current_size == $memory; 

+ my @numa_map = get_numa_node_list($conf); 
+ 
for (my $j = 0; $j < 8; $j++) { 
for (my $i = 0; $i < 32; $i++) { 
my $name = "dimm${dimm_id}"; 
$dimm_id++; 
- my $numanode = $i % $sockets; 
+ my $numanode = $numa_map[$i % @numa_map]; 
$current_size += $dimm_size; 
&$func($conf, $vmid, $name, $dimm_size, $numanode, $current_size, $memory); 
return $current_size if $current_size >= $memory; 
@@ -57,11 +86,13 @@ sub foreach_reverse_dimm { 

return if $current_size == $memory; 

+ my @numa_map = get_numa_node_list($conf); 
+ 
for (my $j = 0; $j < 8; $j++) { 
for (my $i = 0; $i < 32; $i++) { 
my $name = "dimm${dimm_id}"; 
$dimm_id--; 
- my $numanode = $i % $sockets; 
+ my $numanode = $numa_map[(31-$i) % @numa_map]; 
$current_size -= $dimm_size; 
&$func($conf, $vmid, $name, $dimm_size, $numanode, $current_size, $memory); 
return $current_size if $current_size <= $memory; 
@@ -90,6 +121,8 @@ sub qemu_memory_hotplug { 

if($value > $memory) { 

+ my $numa_hostmap = get_numa_guest_to_host_map($conf) if $conf->{hugepages}; 
+ 
foreach_dimm($conf, $vmid, $value, $sockets, sub { 
my ($conf, $vmid, $name, $dimm_size, $numanode, $current_size, $memory) = @_; 

@@ -99,7 +132,8 @@ sub qemu_memory_hotplug { 

my $hugepages_size = hugepages_size($conf, $dimm_size); 
my $path = hugepages_mount_path($hugepages_size); 
- my $hugepages_topology->{$hugepages_size}->{$numanode} = hugepages_nr($dimm_size, $hugepages_size); 
+ my $host_numanode = $numa_hostmap->{$numanode}; 
+ my $hugepages_topology->{$hugepages_size}->{$host_numanode} = hugepages_nr($dimm_size, $hugepages_size); 

my $code = sub { 
my $hugepages_host_topology = hugepages_host_topology(); 
@@ -436,9 +470,13 @@ sub hugepages_topology { 
} 

if ($hotplug_features->{memory}) { 
+ my $numa_hostmap = get_numa_guest_to_host_map($conf); 
+ 
foreach_dimm($conf, undef, $memory, $sockets, sub { 
my ($conf, undef, $name, $dimm_size, $numanode, $current_size, $memory) = @_; 

+ $numanode = $numa_hostmap->{$numanode}; 
+ 
my $hugepages_size = hugepages_size($conf, $dimm_size); 
$hugepages_topology->{$hugepages_size}->{$numanode} += hugepages_nr($dimm_size, $hugepages_size); 
}); 
-- 
2.1.4 




More information about the pve-devel mailing list