gpu_acceleration flag & nvidia hacking

NVIDIA.md

@@ -0,0 +1,6 @@
+install:
+dkms
+linux-headers-$(uname-r)
+
+then run:
+https://us.download.nvidia.com/XFree86/aarch64/580.119.02/NVIDIA-Linux-aarch64-580.119.02.run

build/dpkg-deps/nonfree.depends

@@ -5,7 +5,3 @@
 + firmware-libertas
 + firmware-misc-nonfree
 + firmware-realtek
-+ libnvidia-container-tools
-+ libnvidia-container1
-+ nvidia-container-toolkit
-+ nvidia-driver

core/src/lxc/gpu_config

@@ -0,0 +1,20 @@
+
+# NVIDIA Devices Spe
+lxc.cgroup2.devices.allow: c 235:* rwm
+
+# DRM devices
+lxc.mount.entry: /dev/dri dev/dri none bind,optional,create=di
+
+# Device IDs for /dev/dri, check `ls -l /dev/dri`
+lxc.cgroup2.devices.allow: c 226:* rwm
+
+# Nvidia nodes
+lxc.mount.entry: /dev/nvidia0        dev/nvidia0        none bind,optional,create=file
+lxc.mount.entry: /dev/nvidiactl      dev/nvidiactl      none bind,optional,create=file
+lxc.mount.entry: /dev/nvidia-uvm     dev/nvidia-uvm     none bind,optional,create=file
+lxc.mount.entry: /dev/nvidia-modeset dev/nvidia-modeset none bind,optional,create=file
+
+# Device IDs for /dev/nvidia*, check `ls -l /dev/nvidia*`
+lxc.cgroup2.devices.allow: c 195:* rwm
+lxc.cgroup2.devices.allow: c 500:* rwm
+lxc.cgroup2.devices.allow: c 503:* rwm

core/src/lxc/mod.rs

@@ -174,11 +174,11 @@ impl LxcContainer {
         let machine_id = hex::encode(rand::random::<[u8; 16]>());
         let container_dir = Path::new(LXC_CONTAINER_DIR).join(&*guid);
         tokio::fs::create_dir_all(&container_dir).await?;
-        tokio::fs::write(
-            container_dir.join("config"),
-            format!(include_str!("./config.template"), guid = &*guid),
-        )
-        .await?;
+        let mut config_str = format!(include_str!("./config.template"), guid = &*guid);
+        if config.gpu_acceleration {
+            config_str += include_str!("./gpu_config");
+        }
+        tokio::fs::write(container_dir.join("config"), config_str).await?;
         // TODO: append config
         let rootfs_dir = container_dir.join("rootfs");
         let rootfs = OverlayGuard::mount(
@@ -414,7 +414,10 @@ impl Drop for LxcContainer {
 }
 
 #[derive(Default, Serialize)]
-pub struct LxcConfig {}
+pub struct LxcConfig {
+    pub gpu_acceleration: bool,
+}
+
 pub async fn connect(ctx: &RpcContext, container: &LxcContainer) -> Result<Guid, Error> {
     use axum::extract::ws::Message;
 

core/src/s9pk/v2/compat.rs

@@ -251,6 +251,10 @@ impl TryFrom<ManifestV1> for Manifest {
             git_hash: value.git_hash,
             os_version: value.eos_version,
             sdk_version: None,
+            gpu_acceleration: match value.main {
+                PackageProcedure::Docker(d) => d.gpu_acceleration,
+                PackageProcedure::Script(_) => false,
+            },
         })
     }
 }

core/src/s9pk/v2/manifest.rs

@@ -62,6 +62,8 @@ pub struct Manifest {
     pub dependencies: Dependencies,
     #[serde(default)]
     pub hardware_requirements: HardwareRequirements,
+    #[serde(default)]
+    pub gpu_acceleration: bool,
     pub git_hash: Option<GitHash>,
     #[serde(default = "current_version")]
     #[ts(type = "string")]

core/src/service/persistent_container.rs

@@ -96,7 +96,9 @@ impl PersistentContainer {
                         .join("logs")
                         .join(&s9pk.as_manifest().id),
                 ),
-                LxcConfig::default(),
+                LxcConfig {
+                    gpu_acceleration: s9pk.manifest.gpu_acceleration,
+                },
             )
             .await?;
         let rpc_client = lxc_container.connect_rpc(Some(RPC_CONNECT_TIMEOUT)).await?;