[pve-devel] [RFC v2 qemu-server 1/3] fix #2255: add support for nvme emulation

[Oguz Bektas]

now we can add nvme drives;

nvme0: local-lvm:vm-103-disk-0,size=32G

or

qm set VMID --nvme0 local-lvm:32

max number is 8 for now, as most real hardware has 1-3 nvme slots and
can have a few more with pcie. most cases won't go over 8, if there's an
actual usecase at some point this can always be increased without
breaking anything (although the same isn't valid for decreasing it).

Signed-off-by: Oguz Bektas <o.bektas at proxmox.com>
---

v1->v2:
* serial can be configured by user
* add missing "id=drive-nvme[X]" in the device string, without this we
can't unplug at all

 PVE/QemuServer.pm       | 23 ++++++++++++++++++++---
 PVE/QemuServer/Drive.pm | 21 +++++++++++++++++++++
 2 files changed, 41 insertions(+), 3 deletions(-)

diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index dcf05df..3623fdb 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -406,7 +406,7 @@ EODESC
 	optional => 1,
 	type => 'string', format => 'pve-qm-bootdisk',
 	description => "Enable booting from specified disk.",
-	pattern => '(ide|sata|scsi|virtio)\d+',
+	pattern => '(ide|sata|scsi|virtio|nvme)\d+',
     },
     smp => {
 	optional => 1,
@@ -1424,7 +1424,10 @@ sub print_drivedevice_full {
 	    $device .= ",rotation_rate=1";
 	}
 	$device .= ",wwn=$drive->{wwn}" if $drive->{wwn};
-
+    } elsif ($drive->{interface} eq 'nvme') {
+	my $path = $drive->{file};
+	$drive->{serial} //= "$drive->{interface}$drive->{index}"; # serial is mandatory for nvme
+	$device = "nvme,drive=drive-$drive->{interface}$drive->{index},id=$drive->{interface}$drive->{index}";
     } elsif ($drive->{interface} eq 'ide' || $drive->{interface} eq 'sata') {
 	my $maxdev = ($drive->{interface} eq 'sata') ? $PVE::QemuServer::Drive::MAX_SATA_DISKS : 2;
 	my $controller = int($drive->{index} / $maxdev);
@@ -2157,7 +2160,7 @@ sub parse_vm_config {
 	    } else {
 		$key = 'ide2' if $key eq 'cdrom';
 		my $fmt = $confdesc->{$key}->{format};
-		if ($fmt && $fmt =~ /^pve-qm-(?:ide|scsi|virtio|sata)$/) {
+		if ($fmt && $fmt =~ /^pve-qm-(?:ide|scsi|virtio|sata|nvme)$/) {
 		    my $v = parse_drive($key, $value);
 		    if (my $volid = filename_to_volume_id($vmid, $v->{file}, $v->{media})) {
 			$v->{file} = $volid;
@@ -3784,7 +3787,17 @@ sub vm_deviceplug {
 	    warn $@ if $@;
 	    die $err;
         }
+    } elsif ($deviceid =~ m/^(nvme)(\d+)$/) {
+
+	qemu_driveadd($storecfg, $vmid, $device);
 
+	my $devicefull = print_drivedevice_full($storecfg, $conf, $vmid, $device, $arch, $machine_type);
+	eval { qemu_deviceadd($vmid, $devicefull); };
+	if (my $err = $@) {
+	    eval { qemu_drivedel($vmid, $deviceid); };
+	    warn $@ if $@;
+	    die $err;
+        }
     } elsif ($deviceid =~ m/^(net)(\d+)$/) {
 
 	return undef if !qemu_netdevadd($vmid, $conf, $arch, $device, $deviceid);
@@ -3862,6 +3875,10 @@ sub vm_deviceunplug {
         qemu_drivedel($vmid, $deviceid);
 	qemu_deletescsihw($conf, $vmid, $deviceid);
 
+    } elsif ($deviceid =~ m/^(nvme)(\d+)$/) {
+	qemu_devicedel($vmid, $deviceid);
+	qemu_drivedel($vmid, $deviceid);
+
     } elsif ($deviceid =~ m/^(net)(\d+)$/) {
 
         qemu_devicedel($vmid, $deviceid);
diff --git a/PVE/QemuServer/Drive.pm b/PVE/QemuServer/Drive.pm
index f84333f..b8a553a 100644
--- a/PVE/QemuServer/Drive.pm
+++ b/PVE/QemuServer/Drive.pm
@@ -27,6 +27,7 @@ PVE::JSONSchema::register_standard_option('pve-qm-image-format', {
 
 my $MAX_IDE_DISKS = 4;
 my $MAX_SCSI_DISKS = 31;
+my $MAX_NVME_DISKS = 8;
 my $MAX_VIRTIO_DISKS = 16;
 our $MAX_SATA_DISKS = 6;
 our $MAX_UNUSED_DISKS = 256;
@@ -275,6 +276,20 @@ my $scsidesc = {
 };
 PVE::JSONSchema::register_standard_option("pve-qm-scsi", $scsidesc);
 
+my $nvme_fmt = {
+    %drivedesc_base,
+    %ssd_fmt,
+    %wwn_fmt,
+};
+
+my $nvmedesc = {
+    optional => 1,
+    type => 'string', format => $nvme_fmt,
+    description => "Use volume as NVME disk (n is 0 to " . ($MAX_NVME_DISKS -1) . ").",
+};
+
+PVE::JSONSchema::register_standard_option("pve-qm-nvme", $nvmedesc);
+
 my $sata_fmt = {
     %drivedesc_base,
     %ssd_fmt,
@@ -364,6 +379,11 @@ for (my $i = 0; $i < $MAX_SCSI_DISKS; $i++)  {
     $drivedesc_hash->{"scsi$i"} = $scsidesc;
 }
 
+for (my $i = 0; $i < $MAX_NVME_DISKS; $i++)  {
+    $drivedesc_hash->{"nvme$i"} = $nvmedesc;
+}
+
+
 for (my $i = 0; $i < $MAX_VIRTIO_DISKS; $i++)  {
     $drivedesc_hash->{"virtio$i"} = $virtiodesc;
 }
@@ -380,6 +400,7 @@ sub valid_drive_names {
             (map { "scsi$_" } (0 .. ($MAX_SCSI_DISKS - 1))),
             (map { "virtio$_" } (0 .. ($MAX_VIRTIO_DISKS - 1))),
             (map { "sata$_" } (0 .. ($MAX_SATA_DISKS - 1))),
+            (map { "nvme$_" } (0 .. ($MAX_NVME_DISKS - 1))),
             'efidisk0');
 }
 
-- 
2.20.1