comms: run idle kernel on start-up

- support async flag
- source in packets
- rerouting packets

flake.lock

@@ -11,11 +11,11 @@
         "src-pythonparser": "src-pythonparser"
       },
       "locked": {
-        "lastModified": 1697537883,
+        "lastModified": 1706785107,
-        "narHash": "sha256-GfadmYHFkczltX+rPf08YpAHjYa/31ZmmVD578BcFow=",
+        "narHash": "sha256-Uj72tqigiOCdewSSBBMg6zUpVKhwjAo1HeLJgvyZ3oc=",
         "ref": "refs/heads/master",
-        "rev": "b168f0bb4be1697ff100475c20ee304dcc31fcc2",
+        "rev": "3aaa7e04f26a495e8847e47424bfc16d76d82bf8",
-        "revCount": 8573,
+        "revCount": 8672,
         "type": "git",
         "url": "https://github.com/m-labs/artiq.git"
       },
@@ -37,11 +37,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1693473687,
+        "lastModified": 1701573753,
-        "narHash": "sha256-BdLddCWbvoEyakcGwhph9b5dIU1iA0hCQV7KYgU8nos=",
+        "narHash": "sha256-vhEtXjb9AM6/HnsgfVmhJQeqQ9JqysUm7iWNzTIbexs=",
         "owner": "m-labs",
         "repo": "artiq-comtools",
-        "rev": "f522ef3dbc65961f17b2d3d41e927409d970fd79",
+        "rev": "199bdabf4de49cb7ada8a4ac7133008e0f8434b7",
         "type": "github"
       },
       "original": {
@@ -55,11 +55,11 @@
         "systems": "systems"
       },
       "locked": {
-        "lastModified": 1692799911,
+        "lastModified": 1694529238,
-        "narHash": "sha256-3eihraek4qL744EvQXsK1Ha6C3CR7nnT8X2qWap4RNk=",
+        "narHash": "sha256-zsNZZGTGnMOf9YpHKJqMSsa0dXbfmxeoJ7xHlrt+xmY=",
         "owner": "numtide",
         "repo": "flake-utils",
-        "rev": "f9e7cf818399d17d347f847525c5a5a8032e4e44",
+        "rev": "ff7b65b44d01cf9ba6a71320833626af21126384",
         "type": "github"
       },
       "original": {
@@ -71,11 +71,11 @@
     "mozilla-overlay": {
       "flake": false,
       "locked": {
-        "lastModified": 1695805681,
+        "lastModified": 1704373101,
-        "narHash": "sha256-1ElPLD8eFfnuIk0G52HGGpRtQZ4QPCjChRlEOfkZ5ro=",
+        "narHash": "sha256-+gi59LRWRQmwROrmE1E2b3mtocwueCQqZ60CwLG+gbg=",
         "owner": "mozilla",
         "repo": "nixpkgs-mozilla",
-        "rev": "6eabade97bc28d707a8b9d82ad13ef143836736e",
+        "rev": "9b11a87c0cc54e308fa83aac5b4ee1816d5418a2",
         "type": "github"
       },
       "original": {
@@ -87,11 +87,11 @@
     "mozilla-overlay_2": {
       "flake": false,
       "locked": {
-        "lastModified": 1695805681,
+        "lastModified": 1704373101,
-        "narHash": "sha256-1ElPLD8eFfnuIk0G52HGGpRtQZ4QPCjChRlEOfkZ5ro=",
+        "narHash": "sha256-+gi59LRWRQmwROrmE1E2b3mtocwueCQqZ60CwLG+gbg=",
         "owner": "mozilla",
         "repo": "nixpkgs-mozilla",
-        "rev": "6eabade97bc28d707a8b9d82ad13ef143836736e",
+        "rev": "9b11a87c0cc54e308fa83aac5b4ee1816d5418a2",
         "type": "github"
       },
       "original": {
@@ -118,16 +118,16 @@
     },
     "nixpkgs": {
       "locked": {
-        "lastModified": 1697226376,
+        "lastModified": 1706515015,
-        "narHash": "sha256-cumLLb1QOUtWieUnLGqo+ylNt3+fU8Lcv5Zl+tYbRUE=",
+        "narHash": "sha256-eFfY5A7wlYy3jD/75lx6IJRueg4noE+jowl0a8lIlVo=",
         "owner": "NixOS",
         "repo": "nixpkgs",
-        "rev": "898cb2064b6e98b8c5499f37e81adbdf2925f7c5",
+        "rev": "f4a8d6d5324c327dcc2d863eb7f3cc06ad630df4",
         "type": "github"
       },
       "original": {
         "owner": "NixOS",
-        "ref": "nixos-23.05",
+        "ref": "nixos-23.11",
         "repo": "nixpkgs",
         "type": "github"
       }
@@ -147,11 +147,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1697528004,
+        "lastModified": 1701572254,
-        "narHash": "sha256-FFa2MbhAJEjwY58uOs0swvgymfjubHyWba6Q0X6CbB0=",
+        "narHash": "sha256-ixq8dlpyOytDr+d/OmW8v1Ioy9V2G2ibOlNj8GFDSq4=",
         "owner": "m-labs",
         "repo": "sipyco",
-        "rev": "c0a7ed350ccfb85474217057fc47b3f258ca8d99",
+        "rev": "cceac0df537887135f99aa6b1bdd82853f16b4d6",
         "type": "github"
       },
       "original": {
@@ -163,11 +163,11 @@
     "src-migen": {
       "flake": false,
       "locked": {
-        "lastModified": 1697013661,
+        "lastModified": 1702942348,
-        "narHash": "sha256-qNCqgWyE4vTDmyjE2XMJqW1djuBxT25A36AzQfZqluU=",
+        "narHash": "sha256-gKIfHZxsv+jcgDFRW9mPqmwqbZXuRvXefkZcSFjOGHw=",
         "owner": "m-labs",
         "repo": "migen",
-        "rev": "aadc19df93b7aa9ca761aaebbb98a11e0cf2d7c7",
+        "rev": "50934ad10a87ade47219b796535978b9bdf24023",
         "type": "github"
       },
       "original": {
@@ -179,11 +179,11 @@
     "src-misoc": {
       "flake": false,
       "locked": {
-        "lastModified": 1693709836,
+        "lastModified": 1699352904,
-        "narHash": "sha256-YiCk05RYLzZu1CYkQ2r7XtjwVEqkUGTQn388uOls9tI=",
+        "narHash": "sha256-SglyTmXOPv8jJOjwAjJrj/WhAkItQfUbvKfUqrynwRg=",
         "ref": "refs/heads/master",
-        "rev": "58dc4ee60d165ce9145cf3d904241fc154b6407f",
+        "rev": "a53859f2167c31ab5225b6c09f30cf05527b94f4",
-        "revCount": 2448,
+        "revCount": 2452,
         "submodules": true,
         "type": "git",
         "url": "https://github.com/m-labs/misoc.git"
@@ -234,11 +234,11 @@
         ]
       },
       "locked": {
-        "lastModified": 1697795161,
+        "lastModified": 1702982463,
-        "narHash": "sha256-p89w9eoFJ2VFTDZ5Mrv5vsH0E1Ko9z1C6Ett281hCHg=",
+        "narHash": "sha256-jKR3drE2rsTaYGEgIdv5kUo2LOb1JyIb4tJhVuCXTTc=",
         "ref": "refs/heads/master",
-        "rev": "be672ab662d8134ee11412a651864824f6483d4a",
+        "rev": "4168eb63a7e846863331ae4e656cfd82a867cca8",
-        "revCount": 630,
+        "revCount": 636,
         "type": "git",
         "url": "https://git.m-labs.hk/m-labs/zynq-rs"
       },

flake.nix

@@ -78,6 +78,7 @@
 
       # migen/misoc version checks are broken with pyproject for some reason
       postPatch = ''
+        sed -i "1,4d" pyproject.toml
         substituteInPlace pyproject.toml \
           --replace '"migen@git+https://github.com/m-labs/migen",' ""
         substituteInPlace pyproject.toml \
@@ -122,6 +123,9 @@
         src = ./src;
         cargoLock = { 
           lockFile = src/Cargo.lock;
+          outputHashes = {
+              "tar-no-std-0.1.8" = "sha256-xm17108v4smXOqxdLvHl9CxTCJslmeogjm4Y87IXFuM=";
+            };
         };
 
         nativeBuildInputs = [
@@ -149,6 +153,7 @@
 
         doCheck = false;
         dontFixup = true;
+        auditable = false;
       };
       gateware = pkgs.runCommand "${target}-${variant}-gateware"
         {

src/Cargo.lock

@@ -2,6 +2,12 @@
 # It is not intended for manual editing.
 version = 3
 
+[[package]]
+name = "arrayvec"
+version = "0.7.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "96d30a06541fbafbc7f82ed10c06164cfbd2c401138f6addd8404629c4b16711"
+
 [[package]]
 name = "async-recursion"
 version = "0.3.2"
@@ -474,6 +480,7 @@ dependencies = [
  "log_buffer",
  "num-derive",
  "num-traits",
+ "tar-no-std",
  "unwind",
  "vcell",
  "void",
@@ -538,6 +545,16 @@ dependencies = [
  "unicode-ident",
 ]
 
+[[package]]
+name = "tar-no-std"
+version = "0.1.8"
+source = "git+https://git.m-labs.hk/M-Labs/tar-no-std?rev=2ab6dc5#2ab6dc58e5249c59c4eb03eaf3a119bcdd678d32"
+dependencies = [
+ "arrayvec",
+ "bitflags",
+ "log",
+]
+
 [[package]]
 name = "unicode-ident"
 version = "1.0.5"

src/gateware/kasli_soc.py

@@ -237,12 +237,16 @@ class GenericMaster(SoCCore):
         gtx0 = self.gt_drtio.gtxs[0]
         self.specials += Instance("BUFG", i_I=gtx0.txoutclk, o_O=txout_buf)
 
+        ext_async_rst = Signal()
+
         self.submodules.bootstrap = GTPBootstrapClock(self.platform, clk_freq)
         self.submodules.sys_crg = zynq_clocking.SYSCRG(
             self.platform,
             self.ps7,
             txout_buf,
-            clk_sw=gtx0.tx_init.done)
+            clk_sw=self.gt_drtio.stable_clkin.storage,
+            clk_sw_status=gtx0.tx_init.done,
+            ext_async_rst=ext_async_rst)
         self.csr_devices.append("sys_crg")
         self.crg = self.ps7 # HACK for eem_7series to find the clock
         self.crg.cd_sys = self.sys_crg.cd_sys
@@ -250,6 +254,9 @@ class GenericMaster(SoCCore):
             self.bootstrap.cd_bootstrap.clk, self.sys_crg.cd_sys.clk)
         fix_serdes_timing_path(platform)
 
+        self.comb += ext_async_rst.eq(self.sys_crg.clk_sw_fsm.o_clk_sw & ~gtx0.tx_init.done)
+        self.specials += MultiReg(self.sys_crg.clk_sw_fsm.o_clk_sw & self.sys_crg.mmcm_locked, self.gt_drtio.clk_path_ready, odomain="bootstrap")
+
         self.config["HAS_SI5324"] = None
         self.config["SI5324_SOFT_RESET"] = None
 
@@ -419,12 +426,16 @@ class GenericSatellite(SoCCore):
         gtx0 = self.gt_drtio.gtxs[0]
         self.specials += Instance("BUFG", i_I=gtx0.txoutclk, o_O=txout_buf)
 
+        ext_async_rst = Signal()
+
         self.submodules.bootstrap = GTPBootstrapClock(self.platform, clk_freq)
         self.submodules.sys_crg = zynq_clocking.SYSCRG(
             self.platform, 
             self.ps7,
             txout_buf,
-            clk_sw=gtx0.tx_init.done)
+            clk_sw=self.gt_drtio.stable_clkin.storage,
+            clk_sw_status=gtx0.tx_init.done,
+            ext_async_rst=ext_async_rst)
         platform.add_false_path_constraints(
             self.bootstrap.cd_bootstrap.clk, self.sys_crg.cd_sys.clk)
         self.csr_devices.append("sys_crg")
@@ -433,6 +444,9 @@ class GenericSatellite(SoCCore):
 
         fix_serdes_timing_path(platform)
 
+        self.comb += ext_async_rst.eq(self.sys_crg.clk_sw_fsm.o_clk_sw & ~gtx0.tx_init.done)
+        self.specials += MultiReg(self.sys_crg.clk_sw_fsm.o_clk_sw & self.sys_crg.mmcm_locked, self.gt_drtio.clk_path_ready, odomain="bootstrap")
+
         self.rtio_channels = []
         has_grabber = any(peripheral["type"] == "grabber" for peripheral in description["peripherals"])
         if has_grabber:

src/gateware/zc706.py

@@ -226,6 +226,7 @@ class _MasterBase(SoCCore):
         self.csr_devices.append("gt_drtio")
 
         self.submodules.rtio_tsc = rtio.TSC(glbl_fine_ts_width=3)
+        ext_async_rst = Signal()
         txout_buf = Signal()
         gtx0 = self.gt_drtio.gtxs[0]
         self.specials += Instance("BUFG", i_I=gtx0.txoutclk, o_O=txout_buf)
@@ -234,12 +235,17 @@ class _MasterBase(SoCCore):
             self.platform, 
             self.ps7, 
             txout_buf,
-            clk_sw=gtx0.tx_init.done,
+            clk_sw=self.gt_drtio.stable_clkin.storage,
+            clk_sw_status=gtx0.tx_init.done,
+            ext_async_rst=ext_async_rst,
             freq=clk_freq)
         platform.add_false_path_constraints(
             self.bootstrap.cd_bootstrap.clk, self.sys_crg.cd_sys.clk)
         self.csr_devices.append("sys_crg")
 
+        self.comb += ext_async_rst.eq(self.sys_crg.clk_sw_fsm.o_clk_sw & ~gtx0.tx_init.done)
+        self.specials += MultiReg(self.sys_crg.clk_sw_fsm.o_clk_sw & self.sys_crg.mmcm_locked, self.gt_drtio.clk_path_ready, odomain="bootstrap")
+
         drtio_csr_group = []
         drtioaux_csr_group = []
         drtioaux_memory_group = []
@@ -361,6 +367,7 @@ class _SatelliteBase(SoCCore):
             clk_freq=clk_freq)
         self.csr_devices.append("gt_drtio")
 
+        ext_async_rst = Signal()
         txout_buf = Signal()
         txout_buf.attr.add("keep")
         gtx0 = self.gt_drtio.gtxs[0]
@@ -373,12 +380,17 @@ class _SatelliteBase(SoCCore):
             self.platform, 
             self.ps7, 
             txout_buf,
-            clk_sw=gtx0.tx_init.done,
+            clk_sw=self.gt_drtio.stable_clkin.storage,
+            clk_sw_status=gtx0.tx_init.done,
+            ext_async_rst=ext_async_rst,
             freq=clk_freq)
         platform.add_false_path_constraints(
             self.bootstrap.cd_bootstrap.clk, self.sys_crg.cd_sys.clk)
         self.csr_devices.append("sys_crg")
 
+        self.comb += ext_async_rst.eq(self.sys_crg.clk_sw_fsm.o_clk_sw & ~gtx0.tx_init.done)
+        self.specials += MultiReg(self.sys_crg.clk_sw_fsm.o_clk_sw & self.sys_crg.mmcm_locked, self.gt_drtio.clk_path_ready, odomain="bootstrap")
+
         drtioaux_csr_group = []
         drtioaux_memory_group = []
         drtiorep_csr_group = []

src/gateware/zynq_clocking.py

@@ -65,7 +65,7 @@ class ClockSwitchFSM(Module):
 
 
 class SYSCRG(Module, AutoCSR):
-    def __init__(self, platform, ps7, main_clk, clk_sw=None, freq=125e6):
+    def __init__(self, platform, ps7, main_clk, clk_sw=None, clk_sw_status=None, freq=125e6, ext_async_rst=None, ):
         # assumes bootstrap clock is same freq as main and sys output
         self.clock_domains.cd_sys = ClockDomain()
         self.clock_domains.cd_sys4x = ClockDomain(reset_less=True)
@@ -88,7 +88,7 @@ class SYSCRG(Module, AutoCSR):
         else:
             self.comb += self.clk_sw_fsm.i_clk_sw.eq(clk_sw)
 
-        mmcm_locked = Signal()
+        self.mmcm_locked = Signal()
         mmcm_sys = Signal()
         mmcm_sys4x = Signal()
         mmcm_sys5x = Signal()
@@ -96,7 +96,7 @@ class SYSCRG(Module, AutoCSR):
         mmcm_fb_clk = Signal()
         self.specials += [
             Instance("MMCME2_ADV",
-                p_STARTUP_WAIT="FALSE", o_LOCKED=mmcm_locked,
+                p_STARTUP_WAIT="FALSE", o_LOCKED=self.mmcm_locked,
                 p_BANDWIDTH="HIGH",
                 p_REF_JITTER1=0.001,
                 p_CLKIN1_PERIOD=period, i_CLKIN1=main_clk,
@@ -125,8 +125,17 @@ class SYSCRG(Module, AutoCSR):
             Instance("BUFG", i_I=mmcm_sys, o_O=self.cd_sys.clk),
             Instance("BUFG", i_I=mmcm_sys4x, o_O=self.cd_sys4x.clk),
             Instance("BUFG", i_I=mmcm_clk208, o_O=self.cd_clk200.clk),
-            AsyncResetSynchronizer(self.cd_sys, ~mmcm_locked),
+        ]
-            AsyncResetSynchronizer(self.cd_clk200, ~mmcm_locked),
+
+        if ext_async_rst is not None:
+            self.specials += [
+                AsyncResetSynchronizer(self.cd_sys, ~self.mmcm_locked | ext_async_rst),
+                AsyncResetSynchronizer(self.cd_clk200, ~self.mmcm_locked | ext_async_rst),
+            ]
+        else:
+            self.specials += [
+                AsyncResetSynchronizer(self.cd_sys, ~self.mmcm_locked),
+                AsyncResetSynchronizer(self.cd_clk200, ~self.mmcm_locked),
             ]
 
         reset_counter = Signal(4, reset=15)
@@ -139,4 +148,7 @@ class SYSCRG(Module, AutoCSR):
             )
         self.specials += Instance("IDELAYCTRL", i_REFCLK=ClockSignal("clk200"), i_RST=ic_reset)
 
+        if clk_sw_status is None:
             self.comb += self.current_clock.status.eq(self.clk_sw_fsm.o_clk_sw)
+        else:
+            self.comb += self.current_clock.status.eq(clk_sw_status)

src/libboard_artiq/src/drtioaux_proto.rs

@@ -5,7 +5,7 @@ use io::proto::{ProtoRead, ProtoWrite};
 // used by satellite -> master analyzer, subkernel exceptions
 pub const SAT_PAYLOAD_MAX_SIZE: usize  = /*max size*/512 - /*CRC*/4 - /*packet ID*/1 - /*last*/1 - /*length*/2;
 // used by DDMA, subkernel program data (need to provide extra ID and destination)
-pub const MASTER_PAYLOAD_MAX_SIZE: usize = SAT_PAYLOAD_MAX_SIZE - /*destination*/1 - /*ID*/4;
+pub const MASTER_PAYLOAD_MAX_SIZE: usize = SAT_PAYLOAD_MAX_SIZE - /*source*/1 - /*destination*/1 - /*ID*/4;
 
 #[derive(Debug)]
 pub enum Error {
@@ -19,6 +19,46 @@ impl From<IoError> for Error {
     }
 }
 
+#[derive(PartialEq, Clone, Copy, Debug)]
+#[repr(u8)]
+pub enum PayloadStatus {
+    Middle = 0,
+    First = 1,
+    Last = 2,
+    FirstAndLast = 3,
+}
+
+impl From<u8> for PayloadStatus {
+    fn from(value: u8) -> PayloadStatus {
+        match value {
+            0 => PayloadStatus::Middle,
+            1 => PayloadStatus::First,
+            2 => PayloadStatus::Last,
+            3 => PayloadStatus::FirstAndLast,
+            _ => unreachable!(),
+        }
+    }
+}
+
+impl PayloadStatus {
+    pub fn is_first(self) -> bool {
+        self == PayloadStatus::First || self == PayloadStatus::FirstAndLast
+    }
+
+    pub fn is_last(self) -> bool {
+        self == PayloadStatus::Last || self == PayloadStatus::FirstAndLast
+    }
+
+    pub fn from_status(first: bool, last: bool) -> PayloadStatus {
+        match (first, last) {
+            (true, true) => PayloadStatus::FirstAndLast,
+            (true, false) => PayloadStatus::First,
+            (false, true) => PayloadStatus::Last,
+            (false, false) => PayloadStatus::Middle,
+        }
+    }
+}
+
 #[derive(PartialEq, Debug)]
 pub enum Packet {
     EchoRequest,
@@ -49,6 +89,8 @@ pub enum Packet {
     RoutingSetRank {
         rank: u8,
     },
+    RoutingRetrievePackets,
+    RoutingNoPackets,
     RoutingAck,
 
     MonitorRequest {
@@ -157,31 +199,40 @@ pub enum Packet {
     },
 
     DmaAddTraceRequest {
+        source: u8,
         destination: u8,
         id: u32,
-        last: bool,
+        status: PayloadStatus,
         length: u16,
         trace: [u8; MASTER_PAYLOAD_MAX_SIZE],
     },
     DmaAddTraceReply {
+        source: u8,
+        destination: u8,
+        id: u32,
         succeeded: bool,
     },
     DmaRemoveTraceRequest {
+        source: u8,
         destination: u8,
         id: u32,
     },
     DmaRemoveTraceReply {
+        destination: u8,
         succeeded: bool,
     },
     DmaPlaybackRequest {
+        source: u8,
         destination: u8,
         id: u32,
         timestamp: u64,
     },
     DmaPlaybackReply {
+        destination: u8,
         succeeded: bool,
     },
     DmaPlaybackStatus {
+        source: u8,
         destination: u8,
         id: u32,
         error: u8,
@@ -192,7 +243,7 @@ pub enum Packet {
     SubkernelAddDataRequest {
         destination: u8,
         id: u32,
-        last: bool,
+        status: PayloadStatus,
         length: u16,
         data: [u8; MASTER_PAYLOAD_MAX_SIZE],
     },
@@ -200,22 +251,20 @@ pub enum Packet {
         succeeded: bool,
     },
     SubkernelLoadRunRequest {
+        source: u8,
         destination: u8,
         id: u32,
         run: bool,
     },
     SubkernelLoadRunReply {
-        succeeded: bool,
-    },
-    SubkernelStopRequest {
         destination: u8,
-    },
-    SubkernelStopReply {
         succeeded: bool,
     },
     SubkernelFinished {
+        destination: u8,
         id: u32,
         with_exception: bool,
+        exception_src: u8,
     },
     SubkernelExceptionRequest {
         destination: u8,
@@ -226,9 +275,10 @@ pub enum Packet {
         data: [u8; SAT_PAYLOAD_MAX_SIZE],
     },
     SubkernelMessage {
+        source: u8,
         destination: u8,
         id: u32,
-        last: bool,
+        status: PayloadStatus,
         length: u16,
         data: [u8; MASTER_PAYLOAD_MAX_SIZE],
     },
@@ -275,6 +325,8 @@ impl Packet {
                 rank: reader.read_u8()?,
             },
             0x32 => Packet::RoutingAck,
+            0x33 => Packet::RoutingRetrievePackets,
+            0x34 => Packet::RoutingNoPackets,
 
             0x40 => Packet::MonitorRequest {
                 destination: reader.read_u8()?,
@@ -389,39 +441,49 @@ impl Packet {
             }
 
             0xb0 => {
+                let source = reader.read_u8()?;
                 let destination = reader.read_u8()?;
                 let id = reader.read_u32()?;
-                let last = reader.read_bool()?;
+                let status = reader.read_u8()?;
                 let length = reader.read_u16()?;
                 let mut trace: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
                 reader.read_exact(&mut trace[0..length as usize])?;
                 Packet::DmaAddTraceRequest {
+                    source: source,
                     destination: destination,
                     id: id,
-                    last: last,
+                    status: PayloadStatus::from(status),
                     length: length as u16,
                     trace: trace,
                 }
             }
             0xb1 => Packet::DmaAddTraceReply {
+                source: reader.read_u8()?,
+                destination: reader.read_u8()?,
+                id: reader.read_u32()?,
                 succeeded: reader.read_bool()?,
             },
             0xb2 => Packet::DmaRemoveTraceRequest {
+                source: reader.read_u8()?,
                 destination: reader.read_u8()?,
                 id: reader.read_u32()?,
             },
             0xb3 => Packet::DmaRemoveTraceReply {
+                destination: reader.read_u8()?,
                 succeeded: reader.read_bool()?,
             },
             0xb4 => Packet::DmaPlaybackRequest {
+                source: reader.read_u8()?,
                 destination: reader.read_u8()?,
                 id: reader.read_u32()?,
                 timestamp: reader.read_u64()?,
             },
             0xb5 => Packet::DmaPlaybackReply {
+                destination: reader.read_u8()?,
                 succeeded: reader.read_bool()?,
             },
             0xb6 => Packet::DmaPlaybackStatus {
+                source: reader.read_u8()?,
                 destination: reader.read_u8()?,
                 id: reader.read_u32()?,
                 error: reader.read_u8()?,
@@ -432,14 +494,14 @@ impl Packet {
             0xc0 => {
                 let destination = reader.read_u8()?;
                 let id = reader.read_u32()?;
-                let last = reader.read_bool()?;
+                let status = PayloadStatus::from(reader.read_u8()?);
                 let length = reader.read_u16()?;
                 let mut data: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
                 reader.read_exact(&mut data[0..length as usize])?;
                 Packet::SubkernelAddDataRequest {
                     destination: destination,
                     id: id,
-                    last: last,
+                    status: status,
                     length: length as u16,
                     data: data,
                 }
@@ -448,22 +510,20 @@ impl Packet {
                 succeeded: reader.read_bool()?,
             },
             0xc4 => Packet::SubkernelLoadRunRequest {
+                source: reader.read_u8()?,
                 destination: reader.read_u8()?,
                 id: reader.read_u32()?,
                 run: reader.read_bool()?,
             },
             0xc5 => Packet::SubkernelLoadRunReply {
-                succeeded: reader.read_bool()?,
-            },
-            0xc6 => Packet::SubkernelStopRequest {
                 destination: reader.read_u8()?,
-            },
-            0xc7 => Packet::SubkernelStopReply {
                 succeeded: reader.read_bool()?,
             },
             0xc8 => Packet::SubkernelFinished {
+                destination: reader.read_u8()?,
                 id: reader.read_u32()?,
                 with_exception: reader.read_bool()?,
+                exception_src: reader.read_u8()?,
             },
             0xc9 => Packet::SubkernelExceptionRequest {
                 destination: reader.read_u8()?,
@@ -480,16 +540,18 @@ impl Packet {
                 }
             }
             0xcb => {
+                let source = reader.read_u8()?;
                 let destination = reader.read_u8()?;
                 let id = reader.read_u32()?;
-                let last = reader.read_bool()?;
+                let status = reader.read_u8()?;
                 let length = reader.read_u16()?;
                 let mut data: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
                 reader.read_exact(&mut data[0..length as usize])?;
                 Packet::SubkernelMessage {
+                    source: source,
                     destination: destination,
                     id: id,
-                    last: last,
+                    status: PayloadStatus::from(status),
                     length: length as u16,
                     data: data,
                 }
@@ -540,6 +602,8 @@ impl Packet {
                 writer.write_u8(rank)?;
             }
             Packet::RoutingAck => writer.write_u8(0x32)?,
+            Packet::RoutingRetrievePackets => writer.write_u8(0x33)?,
+            Packet::RoutingNoPackets => writer.write_u8(0x34)?,
 
             Packet::MonitorRequest {
                 destination,
@@ -711,49 +775,69 @@ impl Packet {
             }
 
             Packet::DmaAddTraceRequest {
+                source,
                 destination,
                 id,
-                last,
+                status,
                 trace,
                 length,
             } => {
                 writer.write_u8(0xb0)?;
+                writer.write_u8(source)?;
                 writer.write_u8(destination)?;
                 writer.write_u32(id)?;
-                writer.write_bool(last)?;
+                writer.write_u8(status as u8)?;
                 // trace may be broken down to fit within drtio aux memory limit
                 // will be reconstructed by satellite
                 writer.write_u16(length)?;
                 writer.write_all(&trace[0..length as usize])?;
             }
-            Packet::DmaAddTraceReply { succeeded } => {
+            Packet::DmaAddTraceReply {
+                source,
+                destination,
+                id,
+                succeeded,
+            } => {
                 writer.write_u8(0xb1)?;
+                writer.write_u8(source)?;
+                writer.write_u8(destination)?;
+                writer.write_u32(id)?;
                 writer.write_bool(succeeded)?;
             }
-            Packet::DmaRemoveTraceRequest { destination, id } => {
+            Packet::DmaRemoveTraceRequest {
+                source,
+                destination,
+                id,
+            } => {
                 writer.write_u8(0xb2)?;
+                writer.write_u8(source)?;
                 writer.write_u8(destination)?;
                 writer.write_u32(id)?;
             }
-            Packet::DmaRemoveTraceReply { succeeded } => {
+            Packet::DmaRemoveTraceReply { destination, succeeded } => {
                 writer.write_u8(0xb3)?;
+                writer.write_u8(destination)?;
                 writer.write_bool(succeeded)?;
             }
             Packet::DmaPlaybackRequest {
+                source,
                 destination,
                 id,
                 timestamp,
             } => {
                 writer.write_u8(0xb4)?;
+                writer.write_u8(source)?;
                 writer.write_u8(destination)?;
                 writer.write_u32(id)?;
                 writer.write_u64(timestamp)?;
             }
-            Packet::DmaPlaybackReply { succeeded } => {
+            Packet::DmaPlaybackReply { destination, succeeded } => {
                 writer.write_u8(0xb5)?;
+                writer.write_u8(destination)?;
                 writer.write_bool(succeeded)?;
             }
             Packet::DmaPlaybackStatus {
+                source,
                 destination,
                 id,
                 error,
@@ -761,6 +845,7 @@ impl Packet {
                 timestamp,
             } => {
                 writer.write_u8(0xb6)?;
+                writer.write_u8(source)?;
                 writer.write_u8(destination)?;
                 writer.write_u32(id)?;
                 writer.write_u8(error)?;
@@ -771,14 +856,14 @@ impl Packet {
             Packet::SubkernelAddDataRequest {
                 destination,
                 id,
-                last,
+                status,
                 data,
                 length,
             } => {
                 writer.write_u8(0xc0)?;
                 writer.write_u8(destination)?;
                 writer.write_u32(id)?;
-                writer.write_bool(last)?;
+                writer.write_u8(status as u8)?;
                 writer.write_u16(length)?;
                 writer.write_all(&data[0..length as usize])?;
             }
@@ -786,28 +871,34 @@ impl Packet {
                 writer.write_u8(0xc1)?;
                 writer.write_bool(succeeded)?;
             }
-            Packet::SubkernelLoadRunRequest { destination, id, run } => {
+            Packet::SubkernelLoadRunRequest {
+                source,
+                destination,
+                id,
+                run,
+            } => {
                 writer.write_u8(0xc4)?;
+                writer.write_u8(source)?;
                 writer.write_u8(destination)?;
                 writer.write_u32(id)?;
                 writer.write_bool(run)?;
             }
-            Packet::SubkernelLoadRunReply { succeeded } => {
+            Packet::SubkernelLoadRunReply { destination, succeeded } => {
                 writer.write_u8(0xc5)?;
-                writer.write_bool(succeeded)?;
-            }
-            Packet::SubkernelStopRequest { destination } => {
-                writer.write_u8(0xc6)?;
                 writer.write_u8(destination)?;
-            }
-            Packet::SubkernelStopReply { succeeded } => {
-                writer.write_u8(0xc7)?;
                 writer.write_bool(succeeded)?;
             }
-            Packet::SubkernelFinished { id, with_exception } => {
+            Packet::SubkernelFinished {
+                destination,
+                id,
+                with_exception,
+                exception_src,
+            } => {
                 writer.write_u8(0xc8)?;
+                writer.write_u8(destination)?;
                 writer.write_u32(id)?;
                 writer.write_bool(with_exception)?;
+                writer.write_u8(exception_src)?;
             }
             Packet::SubkernelExceptionRequest { destination } => {
                 writer.write_u8(0xc9)?;
@@ -820,16 +911,18 @@ impl Packet {
                 writer.write_all(&data[0..length as usize])?;
             }
             Packet::SubkernelMessage {
+                source,
                 destination,
                 id,
-                last,
+                status,
                 data,
                 length,
             } => {
                 writer.write_u8(0xcb)?;
+                writer.write_u8(source)?;
                 writer.write_u8(destination)?;
                 writer.write_u32(id)?;
-                writer.write_bool(last)?;
+                writer.write_u8(status as u8)?;
                 writer.write_u16(length)?;
                 writer.write_all(&data[0..length as usize])?;
             }
@@ -840,4 +933,39 @@ impl Packet {
         }
         Ok(())
     }
+
+    pub fn routable_destination(&self) -> Option<u8> {
+        // only for packets that could be re-routed, not only forwarded
+        match self {
+            Packet::DmaAddTraceRequest { destination, .. } => Some(*destination),
+            Packet::DmaAddTraceReply { destination, .. } => Some(*destination),
+            Packet::DmaRemoveTraceRequest { destination, .. } => Some(*destination),
+            Packet::DmaRemoveTraceReply { destination, .. } => Some(*destination),
+            Packet::DmaPlaybackRequest { destination, .. } => Some(*destination),
+            Packet::DmaPlaybackReply { destination, .. } => Some(*destination),
+            Packet::SubkernelLoadRunRequest { destination, .. } => Some(*destination),
+            Packet::SubkernelLoadRunReply { destination, .. } => Some(*destination),
+            Packet::SubkernelMessage { destination, .. } => Some(*destination),
+            Packet::SubkernelMessageAck { destination } => Some(*destination),
+            Packet::DmaPlaybackStatus { destination, .. } => Some(*destination),
+            Packet::SubkernelFinished { destination, .. } => Some(*destination),
+            _ => None,
+        }
+    }
+
+    pub fn expects_response(&self) -> bool {
+        // returns true if the routable packet should elicit a response
+        // e.g. reply, ACK packets end a conversation,
+        // and firmware should not wait for response
+        match self {
+            Packet::DmaAddTraceReply { .. }
+            | Packet::DmaRemoveTraceReply { .. }
+            | Packet::DmaPlaybackReply { .. }
+            | Packet::SubkernelLoadRunReply { .. }
+            | Packet::SubkernelMessageAck { .. }
+            | Packet::DmaPlaybackStatus { .. }
+            | Packet::SubkernelFinished { .. } => false,
+            _ => true,
+        }
+    }
 }

src/libboard_artiq/src/lib.rs

@@ -19,7 +19,7 @@ pub mod drtioaux;
 #[cfg(has_drtio)]
 pub mod drtioaux_async;
 pub mod drtioaux_proto;
-#[cfg(feature = "target_kasli_soc")]
+#[cfg(all(feature = "target_kasli_soc", has_drtio))]
 pub mod io_expander;
 pub mod logger;
 #[cfg(has_drtio)]

src/libksupport/src/kernel/dma.rs

@@ -170,6 +170,7 @@ pub extern "C" fn dma_playback(timestamp: i64, ptr: i32, _uses_ddma: bool) {
         csr::rtio_dma::base_address_write(ptr as u32);
         csr::rtio_dma::time_offset_write(timestamp as u64);
 
+        let old_cri_master = csr::cri_con::selected_read();
         csr::cri_con::selected_write(1);
         csr::rtio_dma::enable_write(1);
         #[cfg(has_drtio)]
@@ -183,7 +184,7 @@ pub extern "C" fn dma_playback(timestamp: i64, ptr: i32, _uses_ddma: bool) {
                 });
         }
         while csr::rtio_dma::enable_read() != 0 {}
-        csr::cri_con::selected_write(0);
+        csr::cri_con::selected_write(old_cri_master);
 
         let error = csr::rtio_dma::error_read();
         if error != 0 {

src/libksupport/src/kernel/mod.rs

@@ -77,6 +77,7 @@ pub enum Message {
     #[cfg(has_drtio)]
     SubkernelLoadRunRequest {
         id: u32,
+        destination: u8,
         run: bool,
     },
     #[cfg(has_drtio)]
@@ -86,7 +87,7 @@ pub enum Message {
     #[cfg(has_drtio)]
     SubkernelAwaitFinishRequest {
         id: u32,
-        timeout: u64,
+        timeout: i64,
     },
     #[cfg(has_drtio)]
     SubkernelAwaitFinishReply {
@@ -95,14 +96,15 @@ pub enum Message {
     #[cfg(has_drtio)]
     SubkernelMsgSend {
         id: u32,
+        destination: Option<u8>,
         data: Vec<u8>,
     },
     #[cfg(has_drtio)]
     SubkernelMsgSent,
     #[cfg(has_drtio)]
     SubkernelMsgRecvRequest {
-        id: u32,
+        id: i32,
-        timeout: u64,
+        timeout: i64,
         tags: Vec<u8>,
     },
     #[cfg(has_drtio)]

src/libksupport/src/kernel/subkernel.rs

@@ -5,12 +5,16 @@ use cslice::CSlice;
 use super::{Message, SubkernelStatus, KERNEL_CHANNEL_0TO1, KERNEL_CHANNEL_1TO0};
 use crate::{artiq_raise, rpc::send_args};
 
-pub extern "C" fn load_run(id: u32, run: bool) {
+pub extern "C" fn load_run(id: u32, destination: u8, run: bool) {
     unsafe {
         KERNEL_CHANNEL_1TO0
             .as_mut()
             .unwrap()
-            .send(Message::SubkernelLoadRunRequest { id: id, run: run });
+            .send(Message::SubkernelLoadRunRequest {
+                id: id,
+                destination: destination,
+                run: run,
+            });
     }
     match unsafe { KERNEL_CHANNEL_0TO1.as_mut().unwrap() }.recv() {
         Message::SubkernelLoadRunReply { succeeded: true } => (),
@@ -21,7 +25,7 @@ pub extern "C" fn load_run(id: u32, run: bool) {
     }
 }
 
-pub extern "C" fn await_finish(id: u32, timeout: u64) {
+pub extern "C" fn await_finish(id: u32, timeout: i64) {
     unsafe {
         KERNEL_CHANNEL_1TO0
             .as_mut()
@@ -51,7 +55,14 @@ pub extern "C" fn await_finish(id: u32, timeout: u64) {
     }
 }
 
-pub extern "C" fn send_message(id: u32, count: u8, tag: &CSlice<u8>, data: *const *const ()) {
+pub extern "C" fn send_message(
+    id: u32,
+    is_return: bool,
+    destination: u8,
+    count: u8,
+    tag: &CSlice<u8>,
+    data: *const *const (),
+) {
     let mut buffer = Vec::<u8>::new();
     send_args(&mut buffer, 0, tag.as_ref(), data, false).expect("RPC encoding failed");
     // overwrite service tag, include how many tags are in the message
@@ -59,6 +70,7 @@ pub extern "C" fn send_message(id: u32, count: u8, tag: &CSlice<u8>, data: *cons
     unsafe {
         KERNEL_CHANNEL_1TO0.as_mut().unwrap().send(Message::SubkernelMsgSend {
             id: id,
+            destination: if is_return { None } else { Some(destination) },
             data: buffer[3..].to_vec(),
         });
     }
@@ -68,7 +80,7 @@ pub extern "C" fn send_message(id: u32, count: u8, tag: &CSlice<u8>, data: *cons
     }
 }
 
-pub extern "C" fn await_message(id: u32, timeout: u64, tags: &CSlice<u8>, min: u8, max: u8) {
+pub extern "C" fn await_message(id: i32, timeout: i64, tags: &CSlice<u8>, min: u8, max: u8) {
     unsafe {
         KERNEL_CHANNEL_1TO0
             .as_mut()

src/runtime/Cargo.toml.tpl

@@ -41,3 +41,7 @@ libc = { path = "../libc" }
 io = { path = "../libio", features = ["alloc"] }
 ksupport = { path = "../libksupport" }
 libboard_artiq = { path = "../libboard_artiq" }
+
+[dependencies.tar-no-std]
+git = "https://git.m-labs.hk/M-Labs/tar-no-std"
+rev = "2ab6dc5"

src/runtime/src/comms.rs

@@ -3,17 +3,22 @@ use core::{cell::RefCell, fmt, slice, str};
 
 use core_io::Error as IoError;
 use cslice::CSlice;
+use dyld::elf;
 use futures::{future::FutureExt, select_biased};
 #[cfg(has_drtio)]
 use io::Cursor;
 #[cfg(has_drtio)]
 use ksupport::rpc;
 use ksupport::{kernel, resolve_channel_name};
+#[cfg(has_drtio)]
+use libasync::delay;
 use libasync::{smoltcp::{Sockets, TcpStream},
                task};
 use libboard_artiq::drtio_routing;
 #[cfg(feature = "target_kasli_soc")]
 use libboard_zynq::error_led::ErrorLED;
+#[cfg(has_drtio)]
+use libboard_zynq::time::Milliseconds;
 use libboard_zynq::{self as zynq,
                     smoltcp::{self,
                               iface::{EthernetInterfaceBuilder, NeighborCache},
@@ -27,6 +32,8 @@ use libcortex_a9::{mutex::Mutex,
 use log::{error, info, warn};
 use num_derive::{FromPrimitive, ToPrimitive};
 use num_traits::{FromPrimitive, ToPrimitive};
+#[cfg(has_drtio)]
+use tar_no_std::TarArchiveRef;
 
 #[cfg(has_drtio)]
 use crate::pl;
@@ -43,6 +50,8 @@ pub enum Error {
     BufferExhausted,
     #[cfg(has_drtio)]
     SubkernelError(subkernel::Error),
+    #[cfg(has_drtio)]
+    DestinationDown,
 }
 
 pub type Result<T> = core::result::Result<T, Error>;
@@ -57,6 +66,8 @@ impl fmt::Display for Error {
             Error::BufferExhausted => write!(f, "buffer exhausted"),
             #[cfg(has_drtio)]
             Error::SubkernelError(error) => write!(f, "subkernel error: {:?}", error),
+            #[cfg(has_drtio)]
+            Error::DestinationDown => write!(f, "subkernel destination down"),
         }
     }
 }
@@ -390,7 +401,11 @@ async fn handle_run_kernel(
                 control.borrow_mut().tx.async_send(reply).await;
             }
             #[cfg(has_drtio)]
-            kernel::Message::SubkernelLoadRunRequest { id, run } => {
+            kernel::Message::SubkernelLoadRunRequest {
+                id,
+                destination: _,
+                run,
+            } => {
                 let succeeded = match subkernel::load(aux_mutex, routing_table, timer, id, run).await {
                     Ok(()) => true,
                     Err(e) => {
@@ -436,8 +451,9 @@ async fn handle_run_kernel(
                     .await;
             }
             #[cfg(has_drtio)]
-            kernel::Message::SubkernelMsgSend { id, data } => {
+            kernel::Message::SubkernelMsgSend { id, destination, data } => {
-                let res = subkernel::message_send(aux_mutex, routing_table, timer, id, data).await;
+                let res =
+                    subkernel::message_send(aux_mutex, routing_table, timer, id, destination.unwrap(), data).await;
                 match res {
                     Ok(_) => (),
                     Err(e) => {
@@ -452,12 +468,26 @@ async fn handle_run_kernel(
             }
             #[cfg(has_drtio)]
             kernel::Message::SubkernelMsgRecvRequest { id, timeout, tags } => {
-                let message_received = subkernel::message_await(id, timeout, timer).await;
+                let message_received = subkernel::message_await(id as u32, timeout, timer).await;
                 let (status, count) = match message_received {
                     Ok(ref message) => (kernel::SubkernelStatus::NoError, message.count),
                     Err(SubkernelError::Timeout) => (kernel::SubkernelStatus::Timeout, 0),
                     Err(SubkernelError::IncorrectState) => (kernel::SubkernelStatus::IncorrectState, 0),
                     Err(SubkernelError::CommLost) => (kernel::SubkernelStatus::CommLost, 0),
+                    Err(SubkernelError::SubkernelException) => {
+                        error!("Exception in subkernel");
+                        // just retrieve the exception
+                        let status = subkernel::await_finish(aux_mutex, routing_table, timer, id as u32, timeout)
+                            .await
+                            .unwrap();
+                        match stream {
+                            None => (),
+                            Some(stream) => {
+                                write_chunk(stream, &status.exception.unwrap()).await?;
+                            }
+                        }
+                        (kernel::SubkernelStatus::OtherError, 0)
+                    }
                     Err(_) => (kernel::SubkernelStatus::OtherError, 0),
                 };
                 control
@@ -524,6 +554,56 @@ async fn handle_run_kernel(
     Ok(())
 }
 
+async fn handle_flash_kernel(
+    buffer: &Vec<u8>,
+    control: &Rc<RefCell<kernel::Control>>,
+    _up_destinations: &Rc<RefCell<[bool; drtio_routing::DEST_COUNT]>>,
+    _aux_mutex: &Rc<Mutex<bool>>,
+    _routing_table: &drtio_routing::RoutingTable,
+    _timer: GlobalTimer,
+) -> Result<()> {
+    if buffer[0] == elf::ELFMAG0 && buffer[1] == elf::ELFMAG1 && buffer[2] == elf::ELFMAG2 && buffer[3] == elf::ELFMAG3
+    {
+        // assume ELF file, proceed as before
+        load_kernel(buffer, control, None).await
+    } else {
+        #[cfg(has_drtio)]
+        {
+            let archive = TarArchiveRef::new(buffer.as_ref());
+            let entries = archive.entries();
+            let mut main_lib: Vec<u8> = Vec::new();
+            for entry in entries {
+                if entry.filename().as_str() == "main.elf" {
+                    main_lib = entry.data().to_vec();
+                } else {
+                    // subkernel filename must be in format:
+                    // "<subkernel id> <destination>.elf"
+                    let filename = entry.filename();
+                    let mut iter = filename.as_str().split_whitespace();
+                    let sid: u32 = iter.next().unwrap().parse().unwrap();
+                    let dest: u8 = iter.next().unwrap().strip_suffix(".elf").unwrap().parse().unwrap();
+                    let up = _up_destinations.borrow()[dest as usize];
+                    if up {
+                        let subkernel_lib = entry.data().to_vec();
+                        subkernel::add_subkernel(sid, dest, subkernel_lib).await;
+                        match subkernel::upload(_aux_mutex, _routing_table, _timer, sid).await {
+                            Ok(_) => (),
+                            Err(_) => return Err(Error::UnexpectedPattern),
+                        }
+                    } else {
+                        return Err(Error::DestinationDown);
+                    }
+                }
+            }
+            load_kernel(&main_lib, control, None).await
+        }
+        #[cfg(not(has_drtio))]
+        {
+            panic!("multi-kernel libraries are not supported in standalone systems");
+        }
+    }
+}
+
 async fn load_kernel(
     buffer: &Vec<u8>,
     control: &Rc<RefCell<kernel::Control>>,
@@ -638,6 +718,32 @@ async fn handle_connection(
     }
 }
 
+async fn flash_and_run_idle_kernel(
+    buffer: &Vec<u8>,
+    control: &Rc<RefCell<kernel::Control>>,
+    up_destinations: &Rc<RefCell<[bool; drtio_routing::DEST_COUNT]>>,
+    aux_mutex: &Rc<Mutex<bool>>,
+    routing_table: &drtio_routing::RoutingTable,
+    timer: GlobalTimer,
+) {
+    info!("Loading idle kernel");
+    let res = handle_flash_kernel(buffer, control, up_destinations,  aux_mutex, routing_table, timer)
+        .await;
+    match res {
+        #[cfg(has_drtio)]
+        Err(Error::DestinationDown) => {
+            let mut countdown = timer.countdown();
+            delay(&mut countdown, Milliseconds(500)).await;
+        }
+        Err(_) => warn!("error loading idle kernel"),
+        _ => (),
+    }
+    info!("Running idle kernel");
+    let _ = handle_run_kernel(None, control, up_destinations, aux_mutex, routing_table, timer)
+        .await.map_err(|_| warn!("error running idle kernel"));
+    info!("Idle kernel terminated");
+}
+
 pub fn main(timer: GlobalTimer, cfg: Config) {
     let net_addresses = net_settings::get_addresses(&cfg);
     info!("network addresses: {}", net_addresses);
@@ -679,7 +785,6 @@ pub fn main(timer: GlobalTimer, cfg: Config) {
 
     Sockets::init(32);
 
-    // before, mutex was on io, but now that io isn't used...?
     let aux_mutex: Rc<Mutex<bool>> = Rc::new(Mutex::new(false));
     #[cfg(has_drtio)]
     let drtio_routing_table = Rc::new(RefCell::new(drtio_routing::config_routing_table(
@@ -702,9 +807,16 @@ pub fn main(timer: GlobalTimer, cfg: Config) {
     let idle_kernel = Rc::new(cfg.read("idle_kernel").ok());
     if let Ok(buffer) = cfg.read("startup_kernel") {
         info!("Loading startup kernel...");
-        if let Ok(()) = task::block_on(load_kernel(&buffer, &control, None)) {
-            info!("Starting startup kernel...");
         let routing_table = drtio_routing_table.borrow();
+        if let Ok(()) = task::block_on(handle_flash_kernel(
+            &buffer,
+            &control,
+            &up_destinations,
+            &aux_mutex,
+            &routing_table,
+            timer,
+        )) {
+            info!("Starting startup kernel...");
             let _ = task::block_on(handle_run_kernel(
                 None,
                 &control,
@@ -722,8 +834,30 @@ pub fn main(timer: GlobalTimer, cfg: Config) {
     mgmt::start(cfg);
 
     task::spawn(async move {
-        let connection = Rc::new(Semaphore::new(1, 1));
+        let connection = Rc::new(Semaphore::new(0, 1));
         let terminate = Rc::new(Semaphore::new(0, 1));
+        {
+            let control = control.clone();
+            let idle_kernel = idle_kernel.clone();
+            let connection = connection.clone();
+            let terminate = terminate.clone();
+            let up_destinations = up_destinations.clone();
+            let aux_mutex = aux_mutex.clone();
+            let routing_table = drtio_routing_table.clone();
+            task::spawn(async move {
+                let routing_table = routing_table.borrow();
+                select_biased! {
+                    _ = (async {
+                        if let Some(buffer) = &*idle_kernel {
+                            flash_and_run_idle_kernel(&buffer, &control, &up_destinations, &aux_mutex, &routing_table, timer).await;
+                        }
+                    }).fuse() => (),
+                    _ = terminate.async_wait().fuse() => ()
+                }
+                connection.signal();
+            });
+        }
+
         loop {
             let mut stream = TcpStream::accept(1381, 0x10_000, 0x10_000).await.unwrap();
 
@@ -751,13 +885,7 @@ pub fn main(timer: GlobalTimer, cfg: Config) {
                             .await
                             .map_err(|e| warn!("connection terminated: {}", e));
                         if let Some(buffer) = &*idle_kernel {
-                            info!("Loading idle kernel");
+                            flash_and_run_idle_kernel(&buffer, &control, &up_destinations, &aux_mutex, &routing_table, timer).await;
-                            let _ = load_kernel(&buffer, &control, None)
-                                .await.map_err(|_| warn!("error loading idle kernel"));
-                            info!("Running idle kernel");
-                            let _ = handle_run_kernel(None, &control, &up_destinations, &aux_mutex, &routing_table, timer)
-                                .await.map_err(|_| warn!("error running idle kernel"));
-                            info!("Idle kernel terminated");
                         }
                     }).fuse() => (),
                     _ = terminate.async_wait().fuse() => ()

src/runtime/src/main.rs

@@ -8,14 +8,14 @@
 #[macro_use]
 extern crate alloc;
 
-#[cfg(feature = "target_kasli_soc")]
+#[cfg(all(feature = "target_kasli_soc", has_drtio))]
 use core::cell::RefCell;
 
 use ksupport;
 use libasync::task;
 #[cfg(has_drtio_eem)]
 use libboard_artiq::drtio_eem;
-#[cfg(feature = "target_kasli_soc")]
+#[cfg(all(feature = "target_kasli_soc", has_drtio))]
 use libboard_artiq::io_expander;
 use libboard_artiq::{identifier_read, logger, pl};
 use libboard_zynq::{gic, mpcore, timer::GlobalTimer};
@@ -38,7 +38,7 @@ mod rtio_mgt;
 #[cfg(has_drtio)]
 mod subkernel;
 
-#[cfg(feature = "target_kasli_soc")]
+#[cfg(all(feature = "target_kasli_soc", has_drtio))]
 async fn io_expanders_service(
     i2c_bus: RefCell<&mut libboard_zynq::i2c::I2c>,
     io_expander0: RefCell<io_expander::IoExpander>,
@@ -93,15 +93,11 @@ pub fn main_core0() {
     info!("gateware ident: {}", identifier_read(&mut [0; 64]));
 
     ksupport::i2c::init();
-    #[cfg(feature = "target_kasli_soc")]
+    #[cfg(all(feature = "target_kasli_soc", has_drtio))]
-    let i2c_bus = unsafe { (ksupport::i2c::I2C_BUS).as_mut().unwrap() };
-
-    #[cfg(feature = "target_kasli_soc")]
-    let (mut io_expander0, mut io_expander1);
-    #[cfg(feature = "target_kasli_soc")]
     {
-        io_expander0 = io_expander::IoExpander::new(i2c_bus, 0).unwrap();
+        let i2c_bus = unsafe { (ksupport::i2c::I2C_BUS).as_mut().unwrap() };
-        io_expander1 = io_expander::IoExpander::new(i2c_bus, 1).unwrap();
+        let mut io_expander0 = io_expander::IoExpander::new(i2c_bus, 0).unwrap();
+        let mut io_expander1 = io_expander::IoExpander::new(i2c_bus, 1).unwrap();
         io_expander0
             .init(i2c_bus)
             .expect("I2C I/O expander #0 initialization failed");
@@ -115,6 +111,11 @@ pub fn main_core0() {
         io_expander1.set(1, 1, false);
         io_expander0.service(i2c_bus).unwrap();
         io_expander1.service(i2c_bus).unwrap();
+        task::spawn(io_expanders_service(
+            RefCell::new(i2c_bus),
+            RefCell::new(io_expander0),
+            RefCell::new(io_expander1),
+        ));
     }
 
     let cfg = match Config::new() {
@@ -135,11 +136,5 @@ pub fn main_core0() {
 
     task::spawn(ksupport::report_async_rtio_errors());
 
-    #[cfg(feature = "target_kasli_soc")]
-    task::spawn(io_expanders_service(
-        RefCell::new(i2c_bus),
-        RefCell::new(io_expander0),
-        RefCell::new(io_expander1),
-    ));
     comms::main(timer, cfg);
 }

src/runtime/src/rtio_dma.rs

@@ -142,9 +142,9 @@ pub mod remote_dma {
             }
         }
 
-        pub async fn playback_done(&mut self, destination: u8, error: u8, channel: u32, timestamp: u64) {
+        pub async fn playback_done(&mut self, source: u8, error: u8, channel: u32, timestamp: u64) {
             let mut traces_locked = self.traces.async_lock().await;
-            let mut trace = traces_locked.get_mut(&destination).unwrap();
+            let mut trace = traces_locked.get_mut(&source).unwrap();
             trace.state = RemoteState::PlaybackEnded {
                 error: error,
                 channel: channel,

src/runtime/src/rtio_mgt.rs

@@ -13,8 +13,10 @@ pub mod drtio {
     use ksupport::{resolve_channel_name, ASYNC_ERROR_BUSY, ASYNC_ERROR_COLLISION, ASYNC_ERROR_SEQUENCE_ERROR,
                    SEEN_ASYNC_ERRORS};
     use libasync::{delay, task};
-    use libboard_artiq::{drtioaux::Error, drtioaux_async, drtioaux_async::Packet,
+    use libboard_artiq::{drtioaux::Error,
-                         drtioaux_proto::MASTER_PAYLOAD_MAX_SIZE};
+                         drtioaux_async,
+                         drtioaux_async::Packet,
+                         drtioaux_proto::{PayloadStatus, MASTER_PAYLOAD_MAX_SIZE}};
     use libboard_zynq::time::Milliseconds;
     use log::{error, info, warn};
 
@@ -41,39 +43,102 @@ pub mod drtio {
         unsafe { (csr::DRTIO[linkno].rx_up_read)() == 1 }
     }
 
-    async fn process_async_packets(aux_mutex: &Mutex<bool>, linkno: u8, packet: Packet) -> Option<Packet> {
+    async fn link_has_async_ready(linkno: u8) -> bool {
-        // returns None if an async packet has been consumed
+        let linkno = linkno as usize;
+        let async_ready;
+        unsafe {
+            async_ready = (csr::DRTIO[linkno].async_messages_ready_read)() == 1;
+            (csr::DRTIO[linkno].async_messages_ready_write)(1);
+        }
+        async_ready
+    }
+
+    async fn process_async_packets(
+        aux_mutex: &Mutex<bool>,
+        linkno: u8,
+        routing_table: &drtio_routing::RoutingTable,
+        timer: GlobalTimer,
+    ) {
+        if link_has_async_ready(linkno).await {
+            loop {
+                let reply = aux_transact(aux_mutex, linkno, &Packet::RoutingRetrievePackets, timer).await;
+                if let Ok(packet) = reply {
                     match packet {
                         Packet::DmaPlaybackStatus {
                             id,
-                destination,
+                            source,
+                            destination: 0,
                             error,
                             channel,
                             timestamp,
                         } => {
-                remote_dma::playback_done(id, destination, error, channel, timestamp).await;
+                            remote_dma::playback_done(id, source, error, channel, timestamp).await;
-                None
                         }
-            Packet::SubkernelFinished { id, with_exception } => {
+                        Packet::SubkernelFinished {
-                subkernel::subkernel_finished(id, with_exception).await;
+                            id,
-                None
+                            destination: 0,
+                            with_exception,
+                            exception_src,
+                        } => {
+                            subkernel::subkernel_finished(id, with_exception, exception_src).await;
                         }
                         Packet::SubkernelMessage {
                             id,
-                destination: from,
+                            source,
-                last,
+                            destination: 0,
+                            status,
                             length,
                             data,
                         } => {
-                subkernel::message_handle_incoming(id, last, length as usize, &data).await;
+                            subkernel::message_handle_incoming(id, status, length as usize, &data).await;
                             // acknowledge receiving part of the message
                             let _lock = aux_mutex.async_lock().await;
-                drtioaux_async::send(linkno, &Packet::SubkernelMessageAck { destination: from })
+                            drtioaux_async::send(linkno, &Packet::SubkernelMessageAck { destination: source })
                                 .await
                                 .unwrap();
-                None
+                            let mut countdown = timer.countdown();
+                            // give the satellites some time to process
+                            delay(&mut countdown, Milliseconds(10)).await;
+                        }
+                        // routable packets
+                        Packet::DmaAddTraceRequest { destination, .. }
+                        | Packet::DmaAddTraceReply { destination, .. }
+                        | Packet::DmaRemoveTraceRequest { destination, .. }
+                        | Packet::DmaRemoveTraceReply { destination, .. }
+                        | Packet::DmaPlaybackRequest { destination, .. }
+                        | Packet::DmaPlaybackReply { destination, .. }
+                        | Packet::SubkernelLoadRunRequest { destination, .. }
+                        | Packet::SubkernelLoadRunReply { destination, .. }
+                        | Packet::SubkernelMessage { destination, .. }
+                        | Packet::SubkernelMessageAck { destination, .. }
+                        | Packet::DmaPlaybackStatus { destination, .. }
+                        | Packet::SubkernelFinished { destination, .. } => {
+                            let dest_link = routing_table.0[destination as usize][0] - 1;
+                            if dest_link == linkno {
+                                warn!(
+                                    "[LINK#{}] Re-routed packet would return to the same link, dropping: {:?}",
+                                    linkno, packet
+                                );
+                            } else if destination == 0 {
+                                warn!("[LINK#{}] Received invalid routable packet: {:?}", linkno, packet)
+                            } else {
+                                drtioaux_async::send(dest_link, &packet).await.unwrap();
+                            }
+                        }
+
+                        Packet::RoutingNoPackets => break,
+
+                        other => warn!("[LINK#{}] Received an unroutable packet: {:?}", linkno, other),
+                    }
+                } else {
+                    warn!(
+                        "[LINK#{}] Error handling async packets ({})",
+                        linkno,
+                        reply.unwrap_err()
+                    );
+                    return;
+                }
             }
-            other => Some(other),
         }
     }
 
@@ -208,11 +273,7 @@ pub mod drtio {
     async fn process_unsolicited_aux(aux_mutex: &Rc<Mutex<bool>>, linkno: u8) {
         let _lock = aux_mutex.async_lock().await;
         match drtioaux_async::recv(linkno).await {
-            Ok(Some(packet)) => {
+            Ok(Some(packet)) => warn!("[LINK#{}] unsolicited aux packet: {:?}", linkno, packet),
-                if let Some(packet) = process_async_packets(aux_mutex, linkno, packet).await {
-                    warn!("[LINK#{}] unsolicited aux packet: {:?}", linkno, packet);
-                }
-            }
             Ok(None) => (),
             Err(_) => warn!("[LINK#{}] aux packet error", linkno),
         }
@@ -281,7 +342,6 @@ pub mod drtio {
                 let linkno = hop - 1;
                 if destination_up(up_destinations, destination).await {
                     if up_links[linkno as usize] {
-                        loop {
                         let reply = aux_transact(
                             aux_mutex,
                             linkno,
@@ -294,13 +354,7 @@ pub mod drtio {
                         match reply {
                             Ok(Packet::DestinationDownReply) => {
                                 destination_set_up(routing_table, up_destinations, destination, false).await;
-                                    remote_dma::destination_changed(
+                                remote_dma::destination_changed(aux_mutex, routing_table, timer, destination, false)
-                                        aux_mutex,
-                                        routing_table,
-                                        timer,
-                                        destination,
-                                        false,
-                                    )
                                     .await;
                                 subkernel::destination_changed(aux_mutex, routing_table, timer, destination, false)
                                     .await;
@@ -333,16 +387,9 @@ pub mod drtio {
                                 );
                                 unsafe { SEEN_ASYNC_ERRORS |= ASYNC_ERROR_BUSY };
                             }
-                                Ok(packet) => match process_async_packets(aux_mutex, linkno, packet).await {
+                            Ok(packet) => error!("[DEST#{}] received unexpected aux packet: {:?}", destination, packet),
-                                    Some(packet) => {
-                                        error!("[DEST#{}] received unexpected aux packet: {:?}", destination, packet)
-                                    }
-                                    None => continue,
-                                },
                             Err(e) => error!("[DEST#{}] communication failed ({})", destination, e),
                         }
-                            break;
-                        }
                     } else {
                         destination_set_up(routing_table, up_destinations, destination, false).await;
                         remote_dma::destination_changed(aux_mutex, routing_table, timer, destination, false).await;
@@ -391,6 +438,7 @@ pub mod drtio {
                 if up_links[linkno as usize] {
                     /* link was previously up */
                     if link_rx_up(linkno).await {
+                        process_async_packets(aux_mutex, linkno, routing_table, timer).await;
                         process_unsolicited_aux(aux_mutex, linkno).await;
                         process_local_errors(linkno).await;
                     } else {
@@ -463,7 +511,7 @@ pub mod drtio {
         reply_handler_f: HandlerF,
     ) -> Result<(), &'static str>
     where
-        PacketF: Fn(&[u8; MASTER_PAYLOAD_MAX_SIZE], bool, usize) -> Packet,
+        PacketF: Fn(&[u8; MASTER_PAYLOAD_MAX_SIZE], PayloadStatus, usize) -> Packet,
         HandlerF: Fn(&Packet) -> Result<(), &'static str>,
     {
         let mut i = 0;
@@ -474,10 +522,12 @@ pub mod drtio {
             } else {
                 data.len() - i
             } as usize;
+            let first = i == 0;
             let last = i + len == data.len();
             slice[..len].clone_from_slice(&data[i..i + len]);
             i += len;
-            let packet = packet_f(&slice, last, len);
+            let status = PayloadStatus::from_status(first, last);
+            let packet = packet_f(&slice, status, len);
             let reply = aux_transact(aux_mutex, linkno, &packet, timer).await?;
             reply_handler_f(&reply)?;
         }
@@ -498,16 +548,25 @@ pub mod drtio {
             aux_mutex,
             timer,
             trace,
-            |slice, last, len| Packet::DmaAddTraceRequest {
+            |slice, status, len| Packet::DmaAddTraceRequest {
                 id: id,
+                source: 0,
                 destination: destination,
-                last: last,
+                status: status,
                 length: len as u16,
                 trace: *slice,
             },
             |reply| match reply {
-                Packet::DmaAddTraceReply { succeeded: true } => Ok(()),
+                Packet::DmaAddTraceReply {
-                Packet::DmaAddTraceReply { succeeded: false } => Err("error adding trace on satellite"),
+                    destination: 0,
+                    succeeded: true,
+                    ..
+                } => Ok(()),
+                Packet::DmaAddTraceReply {
+                    destination: 0,
+                    succeeded: false,
+                    ..
+                } => Err("error adding trace on satellite"),
                 _ => Err("adding DMA trace failed, unexpected aux packet"),
             },
         )
@@ -527,14 +586,21 @@ pub mod drtio {
             linkno,
             &Packet::DmaRemoveTraceRequest {
                 id: id,
+                source: 0,
                 destination: destination,
             },
             timer,
         )
         .await;
         match reply {
-            Ok(Packet::DmaRemoveTraceReply { succeeded: true }) => Ok(()),
+            Ok(Packet::DmaRemoveTraceReply {
-            Ok(Packet::DmaRemoveTraceReply { succeeded: false }) => Err("satellite DMA erase error"),
+                destination: 0,
+                succeeded: true,
+            }) => Ok(()),
+            Ok(Packet::DmaRemoveTraceReply {
+                destination: 0,
+                succeeded: false,
+            }) => Err("satellite DMA erase error"),
             Ok(_) => Err("adding trace failed, unexpected aux packet"),
             Err(_) => Err("erasing trace failed, aux error"),
         }
@@ -554,6 +620,7 @@ pub mod drtio {
             linkno,
             &Packet::DmaPlaybackRequest {
                 id: id,
+                source: 0,
                 destination: destination,
                 timestamp: timestamp,
             },
@@ -561,8 +628,14 @@ pub mod drtio {
         )
         .await;
         match reply {
-            Ok(Packet::DmaPlaybackReply { succeeded: true }) => Ok(()),
+            Ok(Packet::DmaPlaybackReply {
-            Ok(Packet::DmaPlaybackReply { succeeded: false }) => Err("error on DMA playback request"),
+                destination: 0,
+                succeeded: true,
+            }) => Ok(()),
+            Ok(Packet::DmaPlaybackReply {
+                destination: 0,
+                succeeded: false,
+            }) => Err("error on DMA playback request"),
             Ok(_) => Err("received unexpected aux packet during DMA playback"),
             Err(_) => Err("aux error on DMA playback"),
         }
@@ -655,10 +728,10 @@ pub mod drtio {
             aux_mutex,
             timer,
             data,
-            |slice, last, len| Packet::SubkernelAddDataRequest {
+            |slice, status, len| Packet::SubkernelAddDataRequest {
                 id: id,
                 destination: destination,
-                last: last,
+                status: status,
                 length: len as u16,
                 data: *slice,
             },
@@ -685,6 +758,7 @@ pub mod drtio {
             linkno,
             &Packet::SubkernelLoadRunRequest {
                 id: id,
+                source: 0,
                 destination: destination,
                 run: run,
             },
@@ -692,8 +766,14 @@ pub mod drtio {
         )
         .await?;
         match reply {
-            Packet::SubkernelLoadRunReply { succeeded: true } => return Ok(()),
+            Packet::SubkernelLoadRunReply {
-            Packet::SubkernelLoadRunReply { succeeded: false } => return Err("error on subkernel run request"),
+                destination: 0,
+                succeeded: true,
+            } => return Ok(()),
+            Packet::SubkernelLoadRunReply {
+                destination: 0,
+                succeeded: false,
+            } => return Err("error on subkernel run request"),
             _ => return Err("received unexpected aux packet during subkernel run"),
         }
     }
@@ -742,10 +822,11 @@ pub mod drtio {
             aux_mutex,
             timer,
             message,
-            |slice, last, len| Packet::SubkernelMessage {
+            |slice, status, len| Packet::SubkernelMessage {
+                source: 0,
                 destination: destination,
                 id: id,
-                last: last,
+                status: status,
                 length: len as u16,
                 data: *slice,
             },

src/runtime/src/subkernel.rs

@@ -1,10 +1,11 @@
 use alloc::{collections::BTreeMap, rc::Rc, vec::Vec};
 
 use libasync::task;
-use libboard_artiq::{drtio_routing::RoutingTable, drtioaux_proto::MASTER_PAYLOAD_MAX_SIZE};
+use libboard_artiq::{drtio_routing::RoutingTable,
+                     drtioaux_proto::{PayloadStatus, MASTER_PAYLOAD_MAX_SIZE}};
 use libboard_zynq::{time::Milliseconds, timer::GlobalTimer};
 use libcortex_a9::mutex::Mutex;
-use log::error;
+use log::{error, warn};
 
 use crate::rtio_mgt::drtio;
 
@@ -12,7 +13,7 @@ use crate::rtio_mgt::drtio;
 pub enum FinishStatus {
     Ok,
     CommLost,
-    Exception,
+    Exception(u8), // exception source
 }
 
 #[derive(Debug, PartialEq, Clone, Copy)]
@@ -28,6 +29,7 @@ pub enum Error {
     Timeout,
     IncorrectState,
     SubkernelNotFound,
+    SubkernelException,
     CommLost,
     DrtioError(&'static str),
 }
@@ -119,17 +121,19 @@ pub async fn clear_subkernels() {
     CURRENT_MESSAGES.async_lock().await.clear();
 }
 
-pub async fn subkernel_finished(id: u32, with_exception: bool) {
+pub async fn subkernel_finished(id: u32, with_exception: bool, exception_src: u8) {
     // called upon receiving DRTIO SubkernelRunDone
     // may be None if session ends and is cleared
     if let Some(subkernel) = SUBKERNELS.async_lock().await.get_mut(&id) {
+        if subkernel.state == SubkernelState::Running {
             subkernel.state = SubkernelState::Finished {
                 status: match with_exception {
-                true => FinishStatus::Exception,
+                    true => FinishStatus::Exception(exception_src),
                     false => FinishStatus::Ok,
                 },
             }
         }
+    }
 }
 
 pub async fn destination_changed(
@@ -165,26 +169,35 @@ pub async fn await_finish(
     routing_table: &RoutingTable,
     timer: GlobalTimer,
     id: u32,
-    timeout: u64,
+    timeout: i64,
 ) -> Result<SubkernelFinished, Error> {
     match SUBKERNELS.async_lock().await.get(&id).unwrap().state {
         SubkernelState::Running | SubkernelState::Finished { .. } => (),
         _ => return Err(Error::IncorrectState),
     }
-    let max_time = timer.get_time() + Milliseconds(timeout);
+    if timeout > 0 {
+        let max_time = timer.get_time() + Milliseconds(timeout as u64);
         while timer.get_time() < max_time {
-        {
             match SUBKERNELS.async_lock().await.get(&id).unwrap().state {
                 SubkernelState::Finished { .. } => break,
                 _ => (),
             };
-        }
             task::r#yield().await;
         }
         if timer.get_time() >= max_time {
             error!("Remote subkernel finish await timed out");
             return Err(Error::Timeout);
         }
+    } else {
+        // no timeout, wait forever
+        loop {
+            match SUBKERNELS.async_lock().await.get(&id).unwrap().state {
+                SubkernelState::Finished { .. } => break,
+                _ => (),
+            };
+            task::r#yield().await;
+        }
+    }
     if let Some(subkernel) = SUBKERNELS.async_lock().await.get_mut(&id) {
         match subkernel.state {
             SubkernelState::Finished { status } => {
@@ -192,11 +205,8 @@ pub async fn await_finish(
                 Ok(SubkernelFinished {
                     id: id,
                     status: status,
-                    exception: if status == FinishStatus::Exception {
+                    exception: if let FinishStatus::Exception(dest) = status {
-                        Some(
+                        Some(drtio::subkernel_retrieve_exception(aux_mutex, routing_table, timer, dest).await?)
-                            drtio::subkernel_retrieve_exception(aux_mutex, routing_table, timer, subkernel.destination)
-                                .await?,
-                        )
                     } else {
                         None
                     },
@@ -220,13 +230,28 @@ static MESSAGE_QUEUE: Mutex<Vec<Message>> = Mutex::new(Vec::new());
 // currently under construction message(s) (can be from multiple sources)
 static CURRENT_MESSAGES: Mutex<BTreeMap<u32, Message>> = Mutex::new(BTreeMap::new());
 
-pub async fn message_handle_incoming(id: u32, last: bool, length: usize, data: &[u8; MASTER_PAYLOAD_MAX_SIZE]) {
+pub async fn message_handle_incoming(
+    id: u32,
+    status: PayloadStatus,
+    length: usize,
+    data: &[u8; MASTER_PAYLOAD_MAX_SIZE],
+) {
     // called when receiving a message from satellite
-    if SUBKERNELS.async_lock().await.get(&id).is_none() {
+    {
-        // do not add messages for non-existing or deleted subkernels
+        let subkernel_lock = SUBKERNELS.async_lock().await;
+        let subkernel = subkernel_lock.get(&id);
+        if subkernel.is_some() && subkernel.unwrap().state != SubkernelState::Running {
+            // do not add messages for non-running or deleted subkernels
+            warn!("received a message for a non-running subkernel #{}", id);
             return;
         }
+    }
     let mut current_messages = CURRENT_MESSAGES.async_lock().await;
+
+    if status.is_first() {
+        current_messages.remove(&id);
+    }
+
     match current_messages.get_mut(&id) {
         Some(message) => message.data.extend(&data[..length]),
         None => {
@@ -240,7 +265,7 @@ pub async fn message_handle_incoming(id: u32, last: bool, length: usize, data: &
             );
         }
     };
-    if last {
+    if status.is_last() {
         // when done, remove from working queue
         MESSAGE_QUEUE
             .async_lock()
@@ -249,7 +274,9 @@ pub async fn message_handle_incoming(id: u32, last: bool, length: usize, data: &
     }
 }
 
-pub async fn message_await(id: u32, timeout: u64, timer: GlobalTimer) -> Result<Message, Error> {
+pub async fn message_await(id: u32, timeout: i64, timer: GlobalTimer) -> Result<Message, Error> {
+    let is_subkernel = SUBKERNELS.async_lock().await.get(&id).is_some();
+    if is_subkernel {
         match SUBKERNELS.async_lock().await.get(&id).unwrap().state {
             SubkernelState::Finished {
                 status: FinishStatus::CommLost,
@@ -257,8 +284,9 @@ pub async fn message_await(id: u32, timeout: u64, timer: GlobalTimer) -> Result<
             SubkernelState::Running | SubkernelState::Finished { .. } => (),
             _ => return Err(Error::IncorrectState),
         }
-    let max_time = timer.get_time() + Milliseconds(timeout);
+    }
-    while timer.get_time() < max_time {
+    let max_time = timer.get_time() + Milliseconds(timeout as u64);
+    while timeout < 0 || (timeout > 0 && timer.get_time() < max_time) {
         {
             let mut message_queue = MESSAGE_QUEUE.async_lock().await;
             for i in 0..message_queue.len() {
@@ -269,6 +297,17 @@ pub async fn message_await(id: u32, timeout: u64, timer: GlobalTimer) -> Result<
                 }
             }
         }
+        if is_subkernel {
+            match SUBKERNELS.async_lock().await.get(&id).unwrap().state {
+                SubkernelState::Finished {
+                    status: FinishStatus::CommLost,
+                } => return Err(Error::CommLost),
+                SubkernelState::Finished {
+                    status: FinishStatus::Exception(_),
+                } => return Err(Error::SubkernelException),
+                _ => (),
+            }
+        }
         task::r#yield().await;
     }
     Err(Error::Timeout)
@@ -279,9 +318,8 @@ pub async fn message_send<'a>(
     routing_table: &RoutingTable,
     timer: GlobalTimer,
     id: u32,
+    destination: u8,
     message: Vec<u8>,
 ) -> Result<(), Error> {
-    let destination = SUBKERNELS.async_lock().await.get(&id).unwrap().destination;
-    // rpc data prepared by the kernel core already
     Ok(drtio::subkernel_send_message(aux_mutex, routing_table, timer, id, destination, &message).await?)
 }

src/satman/src/dma.rs

@@ -1,7 +1,13 @@
-use alloc::{collections::btree_map::BTreeMap, vec::Vec};
+use alloc::{collections::btree_map::BTreeMap, string::String, vec::Vec};
+use core::mem;
 
-use libboard_artiq::pl::csr;
+use ksupport::kernel::DmaRecorder;
+use libboard_artiq::{drtio_routing::RoutingTable,
+                     drtioaux_proto::{Packet, PayloadStatus, MASTER_PAYLOAD_MAX_SIZE},
+                     pl::csr};
 use libcortex_a9::cache::dcci_slice;
+use routing::{Router, Sliceable};
+use subkernel::Manager as KernelManager;
 
 const ALIGNMENT: usize = 64;
 
@@ -12,16 +18,20 @@ enum ManagerState {
 }
 
 pub struct RtioStatus {
+    pub source: u8,
     pub id: u32,
     pub error: u8,
     pub channel: u32,
     pub timestamp: u64,
 }
 
+#[derive(Debug)]
 pub enum Error {
     IdNotFound,
     PlaybackInProgress,
     EntryNotComplete,
+    MasterDmaFound,
+    UploadFail,
 }
 
 #[derive(Debug)]
@@ -29,13 +39,228 @@ struct Entry {
     trace: Vec<u8>,
     padding_len: usize,
     complete: bool,
+    duration: i64, // relevant for local DMA
+}
+
+impl Entry {
+    pub fn from_vec(data: Vec<u8>, duration: i64) -> Entry {
+        let mut entry = Entry {
+            trace: data,
+            padding_len: 0,
+            complete: true,
+            duration: duration,
+        };
+        entry.realign();
+        entry
+    }
+
+    pub fn id(&self) -> u32 {
+        self.trace[self.padding_len..].as_ptr() as u32
+    }
+
+    pub fn realign(&mut self) {
+        self.trace.push(0);
+        let data_len = self.trace.len();
+
+        self.trace.reserve(ALIGNMENT - 1);
+        let padding = ALIGNMENT - self.trace.as_ptr() as usize % ALIGNMENT;
+        let padding = if padding == ALIGNMENT { 0 } else { padding };
+        for _ in 0..padding {
+            // Vec guarantees that this will not reallocate
+            self.trace.push(0)
+        }
+        for i in 1..data_len + 1 {
+            self.trace[data_len + padding - i] = self.trace[data_len - i]
+        }
+        self.complete = true;
+        self.padding_len = padding;
+
+        dcci_slice(&self.trace);
+    }
+}
+
+#[derive(Debug)]
+enum RemoteTraceState {
+    Unsent,
+    Sending(usize),
+    Ready,
+    Running(usize),
+}
+
+#[derive(Debug)]
+struct RemoteTraces {
+    remote_traces: BTreeMap<u8, Sliceable>,
+    state: RemoteTraceState,
+}
+
+impl RemoteTraces {
+    pub fn new(traces: BTreeMap<u8, Sliceable>) -> RemoteTraces {
+        RemoteTraces {
+            remote_traces: traces,
+            state: RemoteTraceState::Unsent,
+        }
+    }
+
+    // on subkernel request
+    pub fn upload_traces(
+        &mut self,
+        id: u32,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) -> usize {
+        let len = self.remote_traces.len();
+        if len > 0 {
+            self.state = RemoteTraceState::Sending(self.remote_traces.len());
+            for (dest, trace) in self.remote_traces.iter_mut() {
+                // queue up the first packet for all destinations, rest will be sent after first ACK
+                let mut data_slice: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
+                let meta = trace.get_slice_master(&mut data_slice);
+                router.route(
+                    Packet::DmaAddTraceRequest {
+                        source: self_destination,
+                        destination: *dest,
+                        id: id,
+                        status: meta.status,
+                        length: meta.len,
+                        trace: data_slice,
+                    },
+                    routing_table,
+                    rank,
+                    self_destination,
+                );
+            }
+        }
+        len
+    }
+
+    // on incoming Packet::DmaAddTraceReply
+    pub fn ack_upload(
+        &mut self,
+        kernel_manager: &mut KernelManager,
+        source: u8,
+        id: u32,
+        succeeded: bool,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) {
+        if let RemoteTraceState::Sending(count) = self.state {
+            if let Some(trace) = self.remote_traces.get_mut(&source) {
+                if trace.at_end() {
+                    if count - 1 == 0 {
+                        self.state = RemoteTraceState::Ready;
+                        if let Some((id, timestamp)) = kernel_manager.ddma_remote_uploaded(succeeded) {
+                            self.playback(id, timestamp, router, rank, self_destination, routing_table);
+                        }
+                    } else {
+                        self.state = RemoteTraceState::Sending(count - 1);
+                    }
+                } else {
+                    // send next slice
+                    let mut data_slice: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
+                    let meta = trace.get_slice_master(&mut data_slice);
+                    router.route(
+                        Packet::DmaAddTraceRequest {
+                            source: self_destination,
+                            destination: meta.destination,
+                            id: id,
+                            status: meta.status,
+                            length: meta.len,
+                            trace: data_slice,
+                        },
+                        routing_table,
+                        rank,
+                        self_destination,
+                    );
+                }
+            }
+        }
+    }
+
+    // on subkernel request
+    pub fn playback(
+        &mut self,
+        id: u32,
+        timestamp: u64,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) {
+        // route all the playback requests
+        // remote traces (local trace runs on core1 unlike mainline firmware)
+        self.state = RemoteTraceState::Running(self.remote_traces.len());
+        for (dest, _) in self.remote_traces.iter() {
+            router.route(
+                Packet::DmaPlaybackRequest {
+                    source: self_destination,
+                    destination: *dest,
+                    id: id,
+                    timestamp: timestamp,
+                },
+                routing_table,
+                rank,
+                self_destination,
+            );
+            // response will be ignored (succeeded = false handled by the main thread)
+        }
+    }
+
+    // on incoming Packet::DmaPlaybackDone
+    pub fn remote_finished(&mut self, kernel_manager: &mut KernelManager, error: u8, channel: u32, timestamp: u64) {
+        if let RemoteTraceState::Running(count) = self.state {
+            if error != 0 || count - 1 == 0 {
+                // notify the kernel about a DDMA error or finish
+                kernel_manager.ddma_finished(error, channel, timestamp);
+                self.state = RemoteTraceState::Ready;
+                // further messages will be ignored (if there was an error)
+            } else {
+                // no error and not the last one awaited
+                self.state = RemoteTraceState::Running(count - 1);
+            }
+        }
+    }
+
+    pub fn erase(
+        &mut self,
+        id: u32,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) {
+        for (dest, _) in self.remote_traces.iter() {
+            router.route(
+                Packet::DmaRemoveTraceRequest {
+                    source: self_destination,
+                    destination: *dest,
+                    id: id,
+                },
+                routing_table,
+                rank,
+                self_destination,
+            );
+            // response will be ignored as this object will stop existing too
+        }
+    }
+
+    pub fn has_remote_traces(&self) -> bool {
+        self.remote_traces.len() > 0
+    }
 }
 
 #[derive(Debug)]
 pub struct Manager {
-    entries: BTreeMap<u32, Entry>,
+    entries: BTreeMap<(u8, u32), Entry>,
     state: ManagerState,
-    currentid: u32,
+    current_id: u32,
+    current_source: u8,
+
+    remote_entries: BTreeMap<u32, RemoteTraces>,
+    name_map: BTreeMap<String, u32>,
 }
 
 impl Manager {
@@ -45,79 +270,238 @@ impl Manager {
         unsafe { while csr::rtio_dma::enable_read() != 0 {} }
         Manager {
             entries: BTreeMap::new(),
-            currentid: 0,
+            current_id: 0,
+            current_source: 0,
             state: ManagerState::Idle,
+            remote_entries: BTreeMap::new(),
+            name_map: BTreeMap::new(),
         }
     }
 
-    pub fn add(&mut self, id: u32, last: bool, trace: &[u8], trace_len: usize) -> Result<(), Error> {
+    pub fn add(
-        let entry = match self.entries.get_mut(&id) {
+        &mut self,
+        source: u8,
+        id: u32,
+        status: PayloadStatus,
+        trace: &[u8],
+        trace_len: usize,
+    ) -> Result<(), Error> {
+        let entry = match self.entries.get_mut(&(source, id)) {
             Some(entry) => {
-                if entry.complete {
+                if entry.complete || status.is_first() {
                     // replace entry
-                    self.entries.remove(&id);
+                    self.entries.remove(&(source, id));
                     self.entries.insert(
-                        id,
+                        (source, id),
                         Entry {
                             trace: Vec::new(),
                             padding_len: 0,
                             complete: false,
+                            duration: 0,
                         },
                     );
-                    self.entries.get_mut(&id).unwrap()
+                    self.entries.get_mut(&(source, id)).unwrap()
                 } else {
                     entry
                 }
             }
             None => {
                 self.entries.insert(
-                    id,
+                    (source, id),
                     Entry {
                         trace: Vec::new(),
                         padding_len: 0,
                         complete: false,
+                        duration: 0,
                     },
                 );
-                self.entries.get_mut(&id).unwrap()
+                self.entries.get_mut(&(source, id)).unwrap()
             }
         };
         entry.trace.extend(&trace[0..trace_len]);
 
-        if last {
+        if status.is_last() {
-            entry.trace.push(0);
+            entry.realign();
-            let data_len = entry.trace.len();
-
-            // Realign.
-            entry.trace.reserve(ALIGNMENT - 1);
-            let padding = ALIGNMENT - entry.trace.as_ptr() as usize % ALIGNMENT;
-            let padding = if padding == ALIGNMENT { 0 } else { padding };
-            for _ in 0..padding {
-                // Vec guarantees that this will not reallocate
-                entry.trace.push(0)
-            }
-            for i in 1..data_len + 1 {
-                entry.trace[data_len + padding - i] = entry.trace[data_len - i]
-            }
-            entry.complete = true;
-            entry.padding_len = padding;
-            dcci_slice(&entry.trace);
         }
         Ok(())
     }
 
-    pub fn erase(&mut self, id: u32) -> Result<(), Error> {
+    // api for DRTIO
-        match self.entries.remove(&id) {
+    pub fn erase(&mut self, source: u8, id: u32) -> Result<(), Error> {
+        match self.entries.remove(&(source, id)) {
             Some(_) => Ok(()),
             None => Err(Error::IdNotFound),
         }
     }
 
-    pub fn playback(&mut self, id: u32, timestamp: u64) -> Result<(), Error> {
+    // API for subkernel
+    pub fn erase_name(
+        &mut self,
+        name: &str,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) {
+        if let Some(id) = self.name_map.get(name) {
+            if let Some(traces) = self.remote_entries.get_mut(&id) {
+                traces.erase(*id, router, rank, self_destination, routing_table);
+                self.remote_entries.remove(&id);
+            }
+            self.entries.remove(&(self_destination, *id));
+            self.name_map.remove(name);
+        }
+    }
+
+    pub fn remote_finished(
+        &mut self,
+        kernel_manager: &mut KernelManager,
+        id: u32,
+        error: u8,
+        channel: u32,
+        timestamp: u64,
+    ) {
+        if let Some(entry) = self.remote_entries.get_mut(&id) {
+            entry.remote_finished(kernel_manager, error, channel, timestamp);
+        }
+    }
+
+    pub fn ack_upload(
+        &mut self,
+        kernel_manager: &mut KernelManager,
+        source: u8,
+        id: u32,
+        succeeded: bool,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) {
+        if let Some(entry) = self.remote_entries.get_mut(&id) {
+            entry.ack_upload(
+                kernel_manager,
+                source,
+                id,
+                succeeded,
+                router,
+                rank,
+                self_destination,
+                routing_table,
+            );
+        }
+    }
+
+    // API for subkernel
+    pub fn upload_traces(
+        &mut self,
+        id: u32,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) -> Result<usize, Error> {
+        let remote_traces = self.remote_entries.get_mut(&id);
+        let mut len = 0;
+        if let Some(traces) = remote_traces {
+            len = traces.upload_traces(id, router, rank, self_destination, routing_table);
+        }
+        Ok(len)
+    }
+
+    // API for subkernel
+    pub fn playback_remote(
+        &mut self,
+        id: u32,
+        timestamp: u64,
+        router: &mut Router,
+        rank: u8,
+        self_destination: u8,
+        routing_table: &RoutingTable,
+    ) -> Result<(), Error> {
+        if let Some(traces) = self.remote_entries.get_mut(&id) {
+            traces.playback(id, timestamp, router, rank, self_destination, routing_table);
+            Ok(())
+        } else {
+            Err(Error::IdNotFound)
+        }
+    }
+
+    // API for subkernel
+    pub fn cleanup(&mut self, router: &mut Router, rank: u8, self_destination: u8, routing_table: &RoutingTable) {
+        // after subkernel ends, remove all self-generated traces
+        for (_, id) in self.name_map.iter_mut() {
+            if let Some(traces) = self.remote_entries.get_mut(&id) {
+                traces.erase(*id, router, rank, self_destination, routing_table);
+                self.remote_entries.remove(&id);
+            }
+            self.entries.remove(&(self_destination, *id));
+        }
+        self.name_map.clear();
+    }
+
+    // API for subkernel
+    pub fn retrieve(&self, self_destination: u8, name: &String) -> Option<(i32, i64, bool)> {
+        let id = self.name_map.get(name)?;
+        let duration = self.entries.get(&(self_destination, *id))?.duration;
+        let uses_ddma = self.has_remote_traces(*id);
+        Some((*id as i32, duration, uses_ddma))
+    }
+
+    pub fn has_remote_traces(&self, id: u32) -> bool {
+        match self.remote_entries.get(&id) {
+            Some(traces) => traces.has_remote_traces(),
+            _ => false,
+        }
+    }
+
+    pub fn put_record(&mut self, mut recorder: DmaRecorder, self_destination: u8) -> Result<u32, Error> {
+        let mut remote_traces: BTreeMap<u8, Sliceable> = BTreeMap::new();
+
+        let mut local_trace: Vec<u8> = Vec::new();
+        // analyze each entry and put in proper buckets, as the kernel core
+        // sends whole chunks, to limit comms/kernel CPU communication,
+        // and as only comms core has access to varios DMA buffers.
+        let mut ptr = 0;
+        recorder.buffer.push(0);
+        while recorder.buffer[ptr] != 0 {
+            // ptr + 3 = tgt >> 24 (destination)
+            let len = recorder.buffer[ptr] as usize;
+            let destination = recorder.buffer[ptr + 3];
+            if destination == 0 {
+                return Err(Error::MasterDmaFound);
+            } else if destination == self_destination {
+                local_trace.extend(&recorder.buffer[ptr..ptr + len]);
+            } else {
+                if let Some(remote_trace) = remote_traces.get_mut(&destination) {
+                    remote_trace.extend(&recorder.buffer[ptr..ptr + len]);
+                } else {
+                    remote_traces.insert(
+                        destination,
+                        Sliceable::new(destination, recorder.buffer[ptr..ptr + len].to_vec()),
+                    );
+                }
+            }
+            // and jump to the next event
+            ptr += len;
+        }
+        let local_entry = Entry::from_vec(local_trace, recorder.duration);
+
+        let id = local_entry.id();
+        self.entries.insert((self_destination, id), local_entry);
+        self.remote_entries.insert(id, RemoteTraces::new(remote_traces));
+        let mut name = String::new();
+        mem::swap(&mut recorder.name, &mut name);
+        self.name_map.insert(name, id);
+
+        Ok(id)
+    }
+
+    pub fn playback(&mut self, source: u8, id: u32, timestamp: u64) -> Result<(), Error> {
         if self.state != ManagerState::Idle {
             return Err(Error::PlaybackInProgress);
         }
 
-        let entry = match self.entries.get(&id) {
+        let entry = match self.entries.get(&(source, id)) {
             Some(entry) => entry,
             None => {
                 return Err(Error::IdNotFound);
@@ -130,7 +514,8 @@ impl Manager {
         assert!(ptr as u32 % 64 == 0);
 
         self.state = ManagerState::Playback;
-        self.currentid = id;
+        self.current_id = id;
+        self.current_source = source;
 
         unsafe {
             csr::rtio_dma::base_address_write(ptr as u32);
@@ -162,7 +547,8 @@ impl Manager {
                     csr::rtio_dma::error_write(1);
                 }
                 return Some(RtioStatus {
-                    id: self.currentid,
+                    source: self.current_source,
+                    id: self.current_id,
                     error: error,
                     channel: channel,
                     timestamp: timestamp,

src/satman/src/main.rs

@@ -39,11 +39,13 @@ use libboard_zynq::{i2c::I2c, print, println, time::Milliseconds, timer::GlobalT
 use libcortex_a9::{l2c::enable_l2_cache, regs::MPIDR};
 use libregister::RegisterR;
 use libsupport_zynq::ram;
+use routing::Router;
 use subkernel::Manager as KernelManager;
 
 mod analyzer;
 mod dma;
 mod repeater;
+mod routing;
 mod subkernel;
 
 fn drtiosat_reset(reset: bool) {
@@ -72,6 +74,12 @@ fn drtiosat_tsc_loaded() -> bool {
     }
 }
 
+fn drtiosat_async_ready() {
+    unsafe {
+        csr::drtiosat::async_messages_ready_write(1);
+    }
+}
+
 #[cfg(has_drtio_routing)]
 macro_rules! forward {
     ($routing_table:expr, $destination:expr, $rank:expr, $repeaters:expr, $packet:expr, $timer:expr) => {{
@@ -79,7 +87,11 @@ macro_rules! forward {
         if hop != 0 {
             let repno = (hop - 1) as usize;
             if repno < $repeaters.len() {
+                if $packet.expects_response() {
                     return $repeaters[repno].aux_forward($packet, $timer);
+                } else {
+                    return $repeaters[repno].aux_send($packet);
+                }
             } else {
                 return Err(drtioaux::Error::RoutingError);
             }
@@ -95,13 +107,15 @@ macro_rules! forward {
 fn process_aux_packet(
     _repeaters: &mut [repeater::Repeater],
     _routing_table: &mut drtio_routing::RoutingTable,
-    _rank: &mut u8,
+    rank: &mut u8,
+    self_destination: &mut u8,
     packet: drtioaux::Packet,
     timer: &mut GlobalTimer,
     i2c: &mut I2c,
     dma_manager: &mut DmaManager,
     analyzer: &mut Analyzer,
     kernel_manager: &mut KernelManager,
+    router: &mut Router,
 ) -> Result<(), drtioaux::Error> {
     // In the code below, *_chan_sel_write takes an u8 if there are fewer than 256 channels,
     // and u16 otherwise; hence the `as _` conversion.
@@ -122,54 +136,12 @@ fn process_aux_packet(
 
         drtioaux::Packet::DestinationStatusRequest { destination } => {
             #[cfg(has_drtio_routing)]
-            let hop = _routing_table.0[destination as usize][*_rank as usize];
+            let hop = _routing_table.0[destination as usize][*rank as usize];
             #[cfg(not(has_drtio_routing))]
             let hop = 0;
 
             if hop == 0 {
-                if let Some(status) = dma_manager.check_state() {
+                *self_destination = destination;
-                    info!(
-                        "playback done, error: {}, channel: {}, timestamp: {}",
-                        status.error, status.channel, status.timestamp
-                    );
-                    drtioaux::send(
-                        0,
-                        &drtioaux::Packet::DmaPlaybackStatus {
-                            destination: destination,
-                            id: status.id,
-                            error: status.error,
-                            channel: status.channel,
-                            timestamp: status.timestamp,
-                        },
-                    )?;
-                } else if let Some(subkernel_finished) = kernel_manager.get_last_finished() {
-                    info!(
-                        "subkernel {} finished, with exception: {}",
-                        subkernel_finished.id, subkernel_finished.with_exception
-                    );
-                    drtioaux::send(
-                        0,
-                        &drtioaux::Packet::SubkernelFinished {
-                            id: subkernel_finished.id,
-                            with_exception: subkernel_finished.with_exception,
-                        },
-                    )?;
-                } else if kernel_manager.message_is_ready() {
-                    let mut data_slice: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
-                    match kernel_manager.message_get_slice(&mut data_slice) {
-                        Some(meta) => drtioaux::send(
-                            0,
-                            &drtioaux::Packet::SubkernelMessage {
-                                destination: destination,
-                                id: kernel_manager.get_current_id().unwrap(),
-                                last: meta.last,
-                                length: meta.len as u16,
-                                data: data_slice,
-                            },
-                        )?,
-                        None => warn!("subkernel message is ready but no message is present"),
-                    }
-                } else {
                 let errors;
                 unsafe {
                     errors = csr::drtiosat::rtio_error_read();
@@ -199,7 +171,6 @@ fn process_aux_packet(
                     drtioaux::send(0, &drtioaux::Packet::DestinationOkReply)?;
                 }
             }
-            }
 
             #[cfg(has_drtio_routing)]
             {
@@ -242,11 +213,11 @@ fn process_aux_packet(
             drtioaux::send(0, &drtioaux::Packet::RoutingAck)
         }
         #[cfg(has_drtio_routing)]
-        drtioaux::Packet::RoutingSetRank { rank } => {
+        drtioaux::Packet::RoutingSetRank { rank: new_rank } => {
-            *_rank = rank;
+            *rank = new_rank;
-            drtio_routing::interconnect_enable_all(_routing_table, rank);
+            drtio_routing::interconnect_enable_all(_routing_table, new_rank);
 
-            let rep_rank = rank + 1;
+            let rep_rank = new_rank + 1;
             for rep in _repeaters.iter() {
                 if let Err(e) = rep.set_rank(rep_rank, timer) {
                     error!("failed to set rank ({:?})", e);
@@ -267,12 +238,20 @@ fn process_aux_packet(
         #[cfg(not(has_drtio_routing))]
         drtioaux::Packet::RoutingSetRank { rank: _ } => drtioaux::send(0, &drtioaux::Packet::RoutingAck),
 
+        drtioaux::Packet::RoutingRetrievePackets => {
+            let packet = router
+                .get_upstream_packet()
+                .or(Some(drtioaux::Packet::RoutingNoPackets))
+                .unwrap();
+            drtioaux::send(0, &packet)
+        }
+
         drtioaux::Packet::MonitorRequest {
             destination: _destination,
             channel,
             probe,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let value;
             #[cfg(has_rtio_moninj)]
             unsafe {
@@ -294,7 +273,7 @@ fn process_aux_packet(
             overrd,
             value,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             #[cfg(has_rtio_moninj)]
             unsafe {
                 csr::rtio_moninj::inj_chan_sel_write(channel as _);
@@ -308,7 +287,7 @@ fn process_aux_packet(
             channel,
             overrd,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let value;
             #[cfg(has_rtio_moninj)]
             unsafe {
@@ -327,7 +306,7 @@ fn process_aux_packet(
             destination: _destination,
             busno: _busno,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let succeeded = i2c.start().is_ok();
             drtioaux::send(0, &drtioaux::Packet::I2cBasicReply { succeeded: succeeded })
         }
@@ -335,7 +314,7 @@ fn process_aux_packet(
             destination: _destination,
             busno: _busno,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let succeeded = i2c.restart().is_ok();
             drtioaux::send(0, &drtioaux::Packet::I2cBasicReply { succeeded: succeeded })
         }
@@ -343,7 +322,7 @@ fn process_aux_packet(
             destination: _destination,
             busno: _busno,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let succeeded = i2c.stop().is_ok();
             drtioaux::send(0, &drtioaux::Packet::I2cBasicReply { succeeded: succeeded })
         }
@@ -352,7 +331,7 @@ fn process_aux_packet(
             busno: _busno,
             data,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             match i2c.write(data) {
                 Ok(ack) => drtioaux::send(
                     0,
@@ -375,7 +354,7 @@ fn process_aux_packet(
             busno: _busno,
             ack,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             match i2c.read(ack) {
                 Ok(data) => drtioaux::send(
                     0,
@@ -399,7 +378,7 @@ fn process_aux_packet(
             address,
             mask,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let ch = match mask {
                 //decode from mainline, PCA9548-centric API
                 0x00 => None,
@@ -425,7 +404,7 @@ fn process_aux_packet(
             div: _div,
             cs: _cs,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             // todo: reimplement when/if SPI is available
             //let succeeded = spi::set_config(busno, flags, length, div, cs).is_ok();
             drtioaux::send(0, &drtioaux::Packet::SpiBasicReply { succeeded: false })
@@ -435,7 +414,7 @@ fn process_aux_packet(
             busno: _busno,
             data: _data,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             // todo: reimplement when/if SPI is available
             //let succeeded = spi::write(busno, data).is_ok();
             drtioaux::send(0, &drtioaux::Packet::SpiBasicReply { succeeded: false })
@@ -444,7 +423,7 @@ fn process_aux_packet(
             destination: _destination,
             busno: _busno,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             // todo: reimplement when/if SPI is available
             // match spi::read(busno) {
             //     Ok(data) => drtioaux::send(0,
@@ -464,7 +443,7 @@ fn process_aux_packet(
         drtioaux::Packet::AnalyzerHeaderRequest {
             destination: _destination,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let header = analyzer.get_header();
             drtioaux::send(
                 0,
@@ -478,7 +457,7 @@ fn process_aux_packet(
         drtioaux::Packet::AnalyzerDataRequest {
             destination: _destination,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let mut data_slice: [u8; SAT_PAYLOAD_MAX_SIZE] = [0; SAT_PAYLOAD_MAX_SIZE];
             let meta = analyzer.get_data(&mut data_slice);
             drtioaux::send(
@@ -492,55 +471,135 @@ fn process_aux_packet(
         }
 
         drtioaux::Packet::DmaAddTraceRequest {
-            destination: _destination,
+            source,
+            destination,
             id,
-            last,
+            status,
             length,
             trace,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, destination, *rank, _repeaters, &packet, timer);
-            let succeeded = dma_manager.add(id, last, &trace, length as usize).is_ok();
+            *self_destination = destination;
-            drtioaux::send(0, &drtioaux::Packet::DmaAddTraceReply { succeeded: succeeded })
+            let succeeded = dma_manager.add(source, id, status, &trace, length as usize).is_ok();
+            router.send(
+                drtioaux::Packet::DmaAddTraceReply {
+                    source: *self_destination,
+                    destination: source,
+                    id: id,
+                    succeeded: succeeded,
+                },
+                _routing_table,
+                *rank,
+                *self_destination,
+            )
+        }
+        drtioaux::Packet::DmaAddTraceReply {
+            source,
+            destination: _destination,
+            id,
+            succeeded,
+        } => {
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
+            dma_manager.ack_upload(
+                kernel_manager,
+                source,
+                id,
+                succeeded,
+                router,
+                *rank,
+                *self_destination,
+                _routing_table,
+            );
+            Ok(())
         }
         drtioaux::Packet::DmaRemoveTraceRequest {
+            source,
             destination: _destination,
             id,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
-            let succeeded = dma_manager.erase(id).is_ok();
+            let succeeded = dma_manager.erase(source, id).is_ok();
-            drtioaux::send(0, &drtioaux::Packet::DmaRemoveTraceReply { succeeded: succeeded })
+            router.send(
+                drtioaux::Packet::DmaRemoveTraceReply {
+                    destination: source,
+                    succeeded: succeeded,
+                },
+                _routing_table,
+                *rank,
+                *self_destination,
+            )
+        }
+        drtioaux::Packet::DmaRemoveTraceReply {
+            destination: _destination,
+            succeeded: _,
+        } => {
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
+            Ok(())
         }
         drtioaux::Packet::DmaPlaybackRequest {
+            source,
             destination: _destination,
             id,
             timestamp,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let succeeded = if !kernel_manager.running() {
-                dma_manager.playback(id, timestamp).is_ok()
+                dma_manager.playback(source, id, timestamp).is_ok()
             } else {
                 false
             };
-            drtioaux::send(0, &drtioaux::Packet::DmaPlaybackReply { succeeded: succeeded })
+            router.send(
+                drtioaux::Packet::DmaPlaybackReply {
+                    destination: source,
+                    succeeded: succeeded,
+                },
+                _routing_table,
+                *rank,
+                *self_destination,
+            )
+        }
+        drtioaux::Packet::DmaPlaybackReply {
+            destination: _destination,
+            succeeded,
+        } => {
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
+            if !succeeded {
+                kernel_manager.ddma_nack();
+            }
+            Ok(())
+        }
+        drtioaux::Packet::DmaPlaybackStatus {
+            source: _,
+            destination: _destination,
+            id,
+            error,
+            channel,
+            timestamp,
+        } => {
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
+            dma_manager.remote_finished(kernel_manager, id, error, channel, timestamp);
+            Ok(())
         }
 
         drtioaux::Packet::SubkernelAddDataRequest {
-            destination: _destination,
+            destination,
             id,
-            last,
+            status,
             length,
             data,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, destination, *rank, _repeaters, &packet, timer);
-            let succeeded = kernel_manager.add(id, last, &data, length as usize).is_ok();
+            *self_destination = destination;
+            let succeeded = kernel_manager.add(id, status, &data, length as usize).is_ok();
             drtioaux::send(0, &drtioaux::Packet::SubkernelAddDataReply { succeeded: succeeded })
         }
         drtioaux::Packet::SubkernelLoadRunRequest {
+            source,
             destination: _destination,
             id,
             run,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let mut succeeded = kernel_manager.load(id).is_ok();
             // allow preloading a kernel with delayed run
             if run {
@@ -548,59 +607,91 @@ fn process_aux_packet(
                     // cannot run kernel while DDMA is running
                     succeeded = false;
                 } else {
-                    succeeded |= kernel_manager.run(id).is_ok();
+                    succeeded |= kernel_manager.run(source, id).is_ok();
                 }
             }
-            drtioaux::send(0, &drtioaux::Packet::SubkernelLoadRunReply { succeeded: succeeded })
+            router.send(
+                drtioaux::Packet::SubkernelLoadRunReply {
+                    destination: source,
+                    succeeded: succeeded,
+                },
+                _routing_table,
+                *rank,
+                *self_destination,
+            )
+        }
+        drtioaux::Packet::SubkernelLoadRunReply {
+            destination: _destination,
+            succeeded,
+        } => {
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
+            // received if local subkernel started another, remote subkernel
+            kernel_manager.subkernel_load_run_reply(succeeded);
+            Ok(())
+        }
+        drtioaux::Packet::SubkernelFinished {
+            destination: _destination,
+            id,
+            with_exception,
+            exception_src,
+        } => {
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
+            kernel_manager.remote_subkernel_finished(id, with_exception, exception_src);
+            Ok(())
         }
         drtioaux::Packet::SubkernelExceptionRequest {
             destination: _destination,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             let mut data_slice: [u8; SAT_PAYLOAD_MAX_SIZE] = [0; SAT_PAYLOAD_MAX_SIZE];
             let meta = kernel_manager.exception_get_slice(&mut data_slice);
             drtioaux::send(
                 0,
                 &drtioaux::Packet::SubkernelException {
-                    last: meta.last,
+                    last: meta.status.is_last(),
                     length: meta.len,
                     data: data_slice,
                 },
             )
         }
         drtioaux::Packet::SubkernelMessage {
-            destination,
+            source,
-            id: _id,
+            destination: _destination,
-            last,
+            id,
+            status,
             length,
             data,
         } => {
-            forward!(_routing_table, destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
-            kernel_manager.message_handle_incoming(last, length as usize, &data);
+            kernel_manager.message_handle_incoming(status, id, length as usize, &data);
-            drtioaux::send(
+            router.send(
-                0,
+                drtioaux::Packet::SubkernelMessageAck { destination: source },
-                &drtioaux::Packet::SubkernelMessageAck {
+                _routing_table,
-                    destination: destination,
+                *rank,
-                },
+                *self_destination,
             )
         }
         drtioaux::Packet::SubkernelMessageAck {
             destination: _destination,
         } => {
-            forward!(_routing_table, _destination, *_rank, _repeaters, &packet, timer);
+            forward!(_routing_table, _destination, *rank, _repeaters, &packet, timer);
             if kernel_manager.message_ack_slice() {
                 let mut data_slice: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
                 if let Some(meta) = kernel_manager.message_get_slice(&mut data_slice) {
-                    drtioaux::send(
+                    // route and not send immediately as ACKs are not a beginning of a transaction
-                        0,
+                    router.route(
-                        &drtioaux::Packet::SubkernelMessage {
+                        drtioaux::Packet::SubkernelMessage {
-                            destination: *_rank,
+                            source: *self_destination,
+                            destination: meta.destination,
                             id: kernel_manager.get_current_id().unwrap(),
-                            last: meta.last,
+                            status: meta.status,
                             length: meta.len as u16,
                             data: data_slice,
                         },
-                    )?;
+                        _routing_table,
+                        *rank,
+                        *self_destination,
+                    );
                 } else {
                     error!("Error receiving message slice");
                 }
@@ -608,8 +699,8 @@ fn process_aux_packet(
             Ok(())
         }
 
-        _ => {
+        p => {
-            warn!("received unexpected aux packet");
+            warn!("received unexpected aux packet: {:?}", p);
             Ok(())
         }
     }
@@ -619,32 +710,35 @@ fn process_aux_packets(
     repeaters: &mut [repeater::Repeater],
     routing_table: &mut drtio_routing::RoutingTable,
     rank: &mut u8,
+    self_destination: &mut u8,
     timer: &mut GlobalTimer,
     i2c: &mut I2c,
     dma_manager: &mut DmaManager,
     analyzer: &mut Analyzer,
     kernel_manager: &mut KernelManager,
+    router: &mut Router,
 ) {
     let result = drtioaux::recv(0).and_then(|packet| {
-        if let Some(packet) = packet {
+        if let Some(packet) = packet.or_else(|| router.get_local_packet()) {
             process_aux_packet(
                 repeaters,
                 routing_table,
                 rank,
+                self_destination,
                 packet,
                 timer,
                 i2c,
                 dma_manager,
                 analyzer,
                 kernel_manager,
+                router,
             )
         } else {
             Ok(())
         }
     });
-    match result {
+    if let Err(e) = result {
-        Ok(()) => (),
+        warn!("aux packet error ({:?})", e);
-        Err(e) => warn!("aux packet error ({:?})", e),
     }
 }
 
@@ -800,17 +894,20 @@ pub extern "C" fn main_core0() -> i32 {
     }
     let mut routing_table = drtio_routing::RoutingTable::default_empty();
     let mut rank = 1;
+    let mut destination = 1;
 
     let mut hardware_tick_ts = 0;
 
     let mut control = ksupport::kernel::Control::start();
 
     loop {
+        let mut router = Router::new();
+
         while !drtiosat_link_rx_up() {
             drtiosat_process_errors();
             #[allow(unused_mut)]
             for mut rep in repeaters.iter_mut() {
-                rep.service(&routing_table, rank, &mut timer);
+                rep.service(&routing_table, rank, destination, &mut router, &mut timer);
             }
             #[cfg(feature = "target_kasli_soc")]
             {
@@ -849,15 +946,17 @@ pub extern "C" fn main_core0() -> i32 {
                 &mut repeaters,
                 &mut routing_table,
                 &mut rank,
+                &mut destination,
                 &mut timer,
                 &mut i2c,
                 &mut dma_manager,
                 &mut analyzer,
                 &mut kernel_manager,
+                &mut router,
             );
             #[allow(unused_mut)]
             for mut rep in repeaters.iter_mut() {
-                rep.service(&routing_table, rank, &mut timer);
+                rep.service(&routing_table, rank, destination, &mut router, &mut timer);
             }
             #[cfg(feature = "target_kasli_soc")]
             {
@@ -880,7 +979,45 @@ pub extern "C" fn main_core0() -> i32 {
                     error!("aux packet error: {:?}", e);
                 }
             }
-            kernel_manager.process_kern_requests(rank, timer);
+            if let Some(status) = dma_manager.check_state() {
+                info!(
+                    "playback done, error: {}, channel: {}, timestamp: {}",
+                    status.error, status.channel, status.timestamp
+                );
+                router.route(
+                    drtioaux::Packet::DmaPlaybackStatus {
+                        source: destination,
+                        destination: status.source,
+                        id: status.id,
+                        error: status.error,
+                        channel: status.channel,
+                        timestamp: status.timestamp,
+                    },
+                    &routing_table,
+                    rank,
+                    destination,
+                );
+            }
+
+            kernel_manager.process_kern_requests(
+                &mut router,
+                &routing_table,
+                rank,
+                destination,
+                &mut dma_manager,
+                &timer,
+            );
+
+            #[cfg(has_drtio_routing)]
+            if let Some((repno, packet)) = router.get_downstream_packet() {
+                if let Err(e) = repeaters[repno].aux_send(&packet) {
+                    warn!("[REP#{}] Error when sending packet to satellite ({:?})", repno, e)
+                }
+            }
+
+            if router.any_upstream_waiting() {
+                drtiosat_async_ready();
+            }
         }
 
         drtiosat_reset_phy(true);

src/satman/src/repeater.rs

@@ -6,6 +6,7 @@ use libboard_artiq::{drtio_routing, drtioaux};
 #[cfg(has_drtio_routing)]
 use libboard_zynq::time::Milliseconds;
 use libboard_zynq::timer::GlobalTimer;
+use routing::Router;
 
 #[cfg(has_drtio_routing)]
 fn rep_link_rx_up(repno: u8) -> bool {
@@ -53,7 +54,14 @@ impl Repeater {
         self.state == RepeaterState::Up
     }
 
-    pub fn service(&mut self, routing_table: &drtio_routing::RoutingTable, rank: u8, timer: &mut GlobalTimer) {
+    pub fn service(
+        &mut self,
+        routing_table: &drtio_routing::RoutingTable,
+        rank: u8,
+        destination: u8,
+        router: &mut Router,
+        timer: &mut GlobalTimer,
+    ) {
         self.process_local_errors();
 
         match self.state {
@@ -116,6 +124,11 @@ impl Repeater {
                     info!("[REP#{}] link is down", self.repno);
                     self.state = RepeaterState::Down;
                 }
+                if self.async_messages_ready() {
+                    if let Err(e) = self.handle_async(routing_table, rank, destination, router, timer) {
+                        warn!("[REP#{}] Error handling async messages ({:?})", self.repno, e);
+                    }
+                }
             }
             RepeaterState::Failed => {
                 if !rep_link_rx_up(self.repno) {
@@ -173,6 +186,34 @@ impl Repeater {
         }
     }
 
+    fn async_messages_ready(&self) -> bool {
+        let async_rdy;
+        unsafe {
+            async_rdy = (csr::DRTIOREP[self.repno as usize].async_messages_ready_read)();
+            (csr::DRTIOREP[self.repno as usize].async_messages_ready_write)(0);
+        }
+        async_rdy == 1
+    }
+
+    fn handle_async(
+        &self,
+        routing_table: &drtio_routing::RoutingTable,
+        rank: u8,
+        self_destination: u8,
+        router: &mut Router,
+        timer: &mut GlobalTimer,
+    ) -> Result<(), drtioaux::Error> {
+        loop {
+            drtioaux::send(self.auxno, &drtioaux::Packet::RoutingRetrievePackets).unwrap();
+            let reply = self.recv_aux_timeout(200, timer)?;
+            match reply {
+                drtioaux::Packet::RoutingNoPackets => break,
+                packet => router.route(packet, routing_table, rank, self_destination),
+            }
+        }
+        Ok(())
+    }
+
     fn recv_aux_timeout(&self, timeout: u32, timer: &mut GlobalTimer) -> Result<drtioaux::Packet, drtioaux::Error> {
         let max_time = timer.get_time() + Milliseconds(timeout.into());
         loop {
@@ -191,15 +232,19 @@ impl Repeater {
     }
 
     pub fn aux_forward(&self, request: &drtioaux::Packet, timer: &mut GlobalTimer) -> Result<(), drtioaux::Error> {
-        if self.state != RepeaterState::Up {
+        self.aux_send(request)?;
-            return Err(drtioaux::Error::LinkDown);
-        }
-        drtioaux::send(self.auxno, request).unwrap();
         let reply = self.recv_aux_timeout(200, timer)?;
         drtioaux::send(0, &reply).unwrap();
         Ok(())
     }
 
+    pub fn aux_send(&self, request: &drtioaux::Packet) -> Result<(), drtioaux::Error> {
+        if self.state != RepeaterState::Up {
+            return Err(drtioaux::Error::LinkDown);
+        }
+        drtioaux::send(self.auxno, request)
+    }
+
     pub fn sync_tsc(&self, timer: &mut GlobalTimer) -> Result<(), drtioaux::Error> {
         if self.state != RepeaterState::Up {
             return Ok(());
@@ -302,7 +347,15 @@ impl Repeater {
         Repeater::default()
     }
 
-    pub fn service(&self, _routing_table: &drtio_routing::RoutingTable, _rank: u8, _timer: &mut GlobalTimer) {}
+    pub fn service(
+        &self,
+        _routing_table: &drtio_routing::RoutingTable,
+        _rank: u8,
+        _destination: u8,
+        _router: &mut Router,
+        _timer: &mut GlobalTimer,
+    ) {
+    }
 
     pub fn sync_tsc(&self, _timer: &mut GlobalTimer) -> Result<(), drtioaux::Error> {
         Ok(())

src/satman/src/routing.rs

@@ -0,0 +1,190 @@
+use alloc::{collections::vec_deque::VecDeque, vec::Vec};
+use core::cmp::min;
+
+#[cfg(has_drtio_routing)]
+use libboard_artiq::pl::csr;
+use libboard_artiq::{drtio_routing, drtioaux,
+                     drtioaux_proto::{PayloadStatus, MASTER_PAYLOAD_MAX_SIZE, SAT_PAYLOAD_MAX_SIZE}};
+
+pub struct SliceMeta {
+    pub destination: u8,
+    pub len: u16,
+    pub status: PayloadStatus,
+}
+
+/* represents data that has to be sent to Master */
+#[derive(Debug)]
+pub struct Sliceable {
+    it: usize,
+    data: Vec<u8>,
+    destination: u8,
+}
+
+macro_rules! get_slice_fn {
+    ($name:tt, $size:expr) => {
+        pub fn $name(&mut self, data_slice: &mut [u8; $size]) -> SliceMeta {
+            let first = self.it == 0;
+            let len = min($size, self.data.len() - self.it);
+            let last = self.it + len == self.data.len();
+            let status = PayloadStatus::from_status(first, last);
+
+            data_slice[..len].clone_from_slice(&self.data[self.it..self.it + len]);
+            self.it += len;
+
+            SliceMeta {
+                destination: self.destination,
+                len: len as u16,
+                status: status,
+            }
+        }
+    };
+}
+
+impl Sliceable {
+    pub fn new(destination: u8, data: Vec<u8>) -> Sliceable {
+        Sliceable {
+            it: 0,
+            data: data,
+            destination: destination,
+        }
+    }
+
+    pub fn at_end(&self) -> bool {
+        self.it == self.data.len()
+    }
+
+    pub fn extend(&mut self, data: &[u8]) {
+        self.data.extend(data);
+    }
+
+    get_slice_fn!(get_slice_sat, SAT_PAYLOAD_MAX_SIZE);
+    get_slice_fn!(get_slice_master, MASTER_PAYLOAD_MAX_SIZE);
+}
+
+// Packets from downstream (further satellites) are received and routed appropriately.
+// they're passed as soon as possible downstream (within the subtree), or sent upstream,
+// which is notified about pending packets.
+// for rank 1 (connected to master) satellites, these packets are passed as an answer to DestinationStatusRequest;
+// for higher ranks, after getting a notification, it will transact with downstream to get the pending packets.
+
+// forward! macro is not deprecated, as routable packets are only these that can originate
+// from both master and satellite, e.g. DDMA and Subkernel.
+
+pub struct Router {
+    upstream_queue: VecDeque<drtioaux::Packet>,
+    local_queue: VecDeque<drtioaux::Packet>,
+    #[cfg(has_drtio_routing)]
+    downstream_queue: VecDeque<(usize, drtioaux::Packet)>,
+    upstream_notified: bool,
+}
+
+impl Router {
+    pub fn new() -> Router {
+        Router {
+            upstream_queue: VecDeque::new(),
+            local_queue: VecDeque::new(),
+            #[cfg(has_drtio_routing)]
+            downstream_queue: VecDeque::new(),
+            upstream_notified: false,
+        }
+    }
+
+    // Called by local sources (DDMA, kernel) and by repeaters on receiving async data;
+    // messages are always buffered for both upstream and downstream
+    pub fn route(
+        &mut self,
+        packet: drtioaux::Packet,
+        _routing_table: &drtio_routing::RoutingTable,
+        _rank: u8,
+        self_destination: u8,
+    ) {
+        let destination = packet.routable_destination();
+        #[cfg(has_drtio_routing)]
+        {
+            if let Some(destination) = destination {
+                let hop = _routing_table.0[destination as usize][_rank as usize] as usize;
+                if destination == self_destination {
+                    self.local_queue.push_back(packet);
+                } else if hop > 0 && hop < csr::DRTIOREP.len() {
+                    let repno = (hop - 1) as usize;
+                    self.downstream_queue.push_back((repno, packet));
+                } else {
+                    self.upstream_queue.push_back(packet);
+                }
+            } else {
+                error!("Received an unroutable packet: {:?}", packet);
+            }
+        }
+        #[cfg(not(has_drtio_routing))]
+        {
+            if destination == Some(self_destination) {
+                self.local_queue.push_back(packet);
+            } else {
+                self.upstream_queue.push_back(packet);
+            }
+        }
+    }
+
+    // Sends a packet to a required destination, routing if necessary
+    pub fn send(
+        &mut self,
+        packet: drtioaux::Packet,
+        _routing_table: &drtio_routing::RoutingTable,
+        _rank: u8,
+        _destination: u8,
+    ) -> Result<(), drtioaux::Error> {
+        #[cfg(has_drtio_routing)]
+        {
+            let destination = packet.routable_destination();
+            if let Some(destination) = destination {
+                let hop = _routing_table.0[destination as usize][_rank as usize] as usize;
+                if destination == 0 {
+                    // response is needed immediately if master required it
+                    drtioaux::send(0, &packet)?;
+                } else if !(hop > 0 && hop < csr::DRTIOREP.len()) {
+                    // higher rank can wait
+                    self.upstream_queue.push_back(packet);
+                } else {
+                    let repno = (hop - 1) as usize;
+                    // transaction will occur at closest possible opportunity
+                    self.downstream_queue.push_back((repno, packet));
+                }
+                Ok(())
+            } else {
+                // packet not supported in routing, fallback - sent directly
+                drtioaux::send(0, &packet)
+            }
+        }
+        #[cfg(not(has_drtio_routing))]
+        {
+            drtioaux::send(0, &packet)
+        }
+    }
+
+    pub fn any_upstream_waiting(&mut self) -> bool {
+        let empty = self.upstream_queue.is_empty();
+        if !empty && !self.upstream_notified {
+            self.upstream_notified = true; // so upstream will not get spammed with notifications
+            true
+        } else {
+            false
+        }
+    }
+
+    pub fn get_upstream_packet(&mut self) -> Option<drtioaux::Packet> {
+        let packet = self.upstream_queue.pop_front();
+        if packet.is_none() {
+            self.upstream_notified = false;
+        }
+        packet
+    }
+
+    #[cfg(has_drtio_routing)]
+    pub fn get_downstream_packet(&mut self) -> Option<(usize, drtioaux::Packet)> {
+        self.downstream_queue.pop_front()
+    }
+
+    pub fn get_local_packet(&mut self) -> Option<drtioaux::Packet> {
+        self.local_queue.pop_front()
+    }
+}

src/satman/src/subkernel.rs

@@ -1,26 +1,52 @@
-use alloc::{collections::{BTreeMap, VecDeque},
+use alloc::{collections::BTreeMap,
             format,
             string::{String, ToString},
             vec::Vec};
-use core::{cmp::min, option::NoneError, slice, str};
+use core::{option::NoneError, slice, str};
 
 use core_io::{Error as IoError, Write};
 use cslice::AsCSlice;
+use dma::{Error as DmaError, Manager as DmaManager};
 use io::{Cursor, ProtoWrite};
 use ksupport::{eh_artiq, kernel, rpc};
-use libboard_artiq::{drtioaux_proto::{MASTER_PAYLOAD_MAX_SIZE, SAT_PAYLOAD_MAX_SIZE},
+use libboard_artiq::{drtio_routing::RoutingTable,
+                     drtioaux,
+                     drtioaux_proto::{PayloadStatus, MASTER_PAYLOAD_MAX_SIZE, SAT_PAYLOAD_MAX_SIZE},
                      pl::csr};
 use libboard_zynq::{time::Milliseconds, timer::GlobalTimer};
 use libcortex_a9::sync_channel::Receiver;
 use log::warn;
+use routing::{Router, SliceMeta, Sliceable};
 
 #[derive(Debug, Clone, PartialEq)]
 enum KernelState {
     Absent,
     Loaded,
     Running,
-    MsgAwait(Milliseconds, Vec<u8>),
+    MsgAwait {
+        max_time: Option<Milliseconds>,
+        id: u32,
+        tags: Vec<u8>,
+    },
     MsgSending,
+    SubkernelAwaitLoad,
+    SubkernelAwaitFinish {
+        max_time: Option<Milliseconds>,
+        id: u32,
+    },
+    DmaUploading,
+    DmaPendingPlayback {
+        id: u32,
+        timestamp: u64,
+    },
+    DmaPendingAwait {
+        id: u32,
+        timestamp: u64,
+        max_time: Milliseconds,
+    },
+    DmaAwait {
+        max_time: Milliseconds,
+    },
 }
 
 #[derive(Debug)]
@@ -31,7 +57,9 @@ pub enum Error {
     NoMessage,
     AwaitingMessage,
     SubkernelIoError,
+    DrtioError,
     KernelException(Sliceable),
+    DmaError(DmaError),
 }
 
 impl From<NoneError> for Error {
@@ -46,33 +74,38 @@ impl From<IoError> for Error {
     }
 }
 
+impl From<DmaError> for Error {
+    fn from(value: DmaError) -> Error {
+        Error::DmaError(value)
+    }
+}
+
 impl From<()> for Error {
     fn from(_: ()) -> Error {
         Error::NoMessage
     }
 }
 
+impl From<drtioaux::Error> for Error {
+    fn from(_value: drtioaux::Error) -> Error {
+        Error::DrtioError
+    }
+}
+
 macro_rules! unexpected {
     ($($arg:tt)*) => (return Err(Error::Unexpected(format!($($arg)*))));
 }
 
-/* represents data that has to be sent to Master */
-#[derive(Debug)]
-pub struct Sliceable {
-    it: usize,
-    data: Vec<u8>,
-}
-
 /* represents interkernel messages */
 struct Message {
     count: u8,
+    id: u32,
     data: Vec<u8>,
 }
 
 #[derive(PartialEq)]
 enum OutMessageState {
     NoMessage,
-    MessageReady,
     MessageBeingSent,
     MessageSent,
     MessageAcknowledged,
@@ -82,7 +115,7 @@ enum OutMessageState {
 struct MessageManager {
     out_message: Option<Sliceable>,
     out_state: OutMessageState,
-    in_queue: VecDeque<Message>,
+    in_queue: Vec<Message>,
     in_buffer: Option<Message>,
 }
 
@@ -92,6 +125,8 @@ struct Session {
     kernel_state: KernelState,
     last_exception: Option<Sliceable>,
     messages: MessageManager,
+    source: u8, // which destination requested running the kernel
+    subkernels_finished: Vec<u32>,
 }
 
 impl Session {
@@ -101,13 +136,15 @@ impl Session {
             kernel_state: KernelState::Absent,
             last_exception: None,
             messages: MessageManager::new(),
+            source: 0,
+            subkernels_finished: Vec::new(),
         }
     }
 
     fn running(&self) -> bool {
         match self.kernel_state {
             KernelState::Absent | KernelState::Loaded => false,
-            KernelState::Running | KernelState::MsgAwait { .. } | KernelState::MsgSending => true,
+            _ => true,
         }
     }
 }
@@ -129,40 +166,8 @@ pub struct Manager<'a> {
 pub struct SubkernelFinished {
     pub id: u32,
     pub with_exception: bool,
-}
+    pub exception_source: u8,
-
+    pub source: u8,
-pub struct SliceMeta {
-    pub len: u16,
-    pub last: bool,
-}
-
-macro_rules! get_slice_fn {
-    ($name:tt, $size:expr) => {
-        pub fn $name(&mut self, data_slice: &mut [u8; $size]) -> SliceMeta {
-            if self.data.len() == 0 {
-                return SliceMeta { len: 0, last: true };
-            }
-            let len = min($size, self.data.len() - self.it);
-            let last = self.it + len == self.data.len();
-
-            data_slice[..len].clone_from_slice(&self.data[self.it..self.it + len]);
-            self.it += len;
-
-            SliceMeta {
-                len: len as u16,
-                last: last,
-            }
-        }
-    };
-}
-
-impl Sliceable {
-    pub fn new(data: Vec<u8>) -> Sliceable {
-        Sliceable { it: 0, data: data }
-    }
-
-    get_slice_fn!(get_slice_sat, SAT_PAYLOAD_MAX_SIZE);
-    get_slice_fn!(get_slice_master, MASTER_PAYLOAD_MAX_SIZE);
 }
 
 impl MessageManager {
@@ -170,36 +175,35 @@ impl MessageManager {
         MessageManager {
             out_message: None,
             out_state: OutMessageState::NoMessage,
-            in_queue: VecDeque::new(),
+            in_queue: Vec::new(),
             in_buffer: None,
         }
     }
 
-    pub fn handle_incoming(&mut self, last: bool, length: usize, data: &[u8; MASTER_PAYLOAD_MAX_SIZE]) {
+    pub fn handle_incoming(
+        &mut self,
+        status: PayloadStatus,
+        id: u32,
+        length: usize,
+        data: &[u8; MASTER_PAYLOAD_MAX_SIZE],
+    ) {
         // called when receiving a message from master
+        if status.is_first() {
+            self.in_buffer = None;
+        }
         match self.in_buffer.as_mut() {
             Some(message) => message.data.extend(&data[..length]),
             None => {
                 self.in_buffer = Some(Message {
                     count: data[0],
+                    id: id,
                     data: data[1..length].to_vec(),
                 });
             }
         };
-        if last {
+        if status.is_last() {
             // when done, remove from working queue
-            self.in_queue.push_back(self.in_buffer.take().unwrap());
+            self.in_queue.push(self.in_buffer.take().unwrap());
-        }
-    }
-
-    pub fn is_outgoing_ready(&mut self) -> bool {
-        // called by main loop, to see if there's anything to send, will send it afterwards
-        match self.out_state {
-            OutMessageState::MessageReady => {
-                self.out_state = OutMessageState::MessageBeingSent;
-                true
-            }
-            _ => false,
         }
     }
 
@@ -218,7 +222,7 @@ impl MessageManager {
             return None;
         }
         let meta = self.out_message.as_mut()?.get_slice_master(data_slice);
-        if meta.last {
+        if meta.status.is_last() {
             // clear the message slot
             self.out_message = None;
             // notify kernel with a flag that message is sent
@@ -242,15 +246,44 @@ impl MessageManager {
         }
     }
 
-    pub fn accept_outgoing(&mut self, message: Vec<u8>) -> Result<(), Error> {
+    pub fn accept_outgoing(
-        // service tag skipped in kernel
+        &mut self,
-        self.out_message = Some(Sliceable::new(message));
+        id: u32,
-        self.out_state = OutMessageState::MessageReady;
+        self_destination: u8,
+        destination: u8,
+        message: Vec<u8>,
+        routing_table: &RoutingTable,
+        rank: u8,
+        router: &mut Router,
+    ) -> Result<(), Error> {
+        self.out_message = Some(Sliceable::new(destination, message));
+
+        let mut data_slice: [u8; MASTER_PAYLOAD_MAX_SIZE] = [0; MASTER_PAYLOAD_MAX_SIZE];
+        self.out_state = OutMessageState::MessageBeingSent;
+        let meta = self.get_outgoing_slice(&mut data_slice).unwrap();
+        router.route(
+            drtioaux::Packet::SubkernelMessage {
+                source: self_destination,
+                destination: destination,
+                id: id,
+                status: meta.status,
+                length: meta.len as u16,
+                data: data_slice,
+            },
+            routing_table,
+            rank,
+            self_destination,
+        );
         Ok(())
     }
 
-    pub fn get_incoming(&mut self) -> Option<Message> {
+    pub fn get_incoming(&mut self, id: u32) -> Option<Message> {
-        self.in_queue.pop_front()
+        for i in 0..self.in_queue.len() {
+            if self.in_queue[i].id == id {
+                return Some(self.in_queue.remove(i));
+            }
+        }
+        None
     }
 }
 
@@ -265,10 +298,10 @@ impl<'a> Manager<'_> {
         }
     }
 
-    pub fn add(&mut self, id: u32, last: bool, data: &[u8], data_len: usize) -> Result<(), Error> {
+    pub fn add(&mut self, id: u32, status: PayloadStatus, data: &[u8], data_len: usize) -> Result<(), Error> {
         let kernel = match self.kernels.get_mut(&id) {
             Some(kernel) => {
-                if kernel.complete {
+                if kernel.complete || status.is_first() {
                     // replace entry
                     self.kernels.remove(&id);
                     self.kernels.insert(
@@ -296,7 +329,7 @@ impl<'a> Manager<'_> {
         };
         kernel.library.extend(&data[0..data_len]);
 
-        kernel.complete = last;
+        kernel.complete = status.is_last();
         Ok(())
     }
 
@@ -311,12 +344,12 @@ impl<'a> Manager<'_> {
         }
     }
 
-    pub fn run(&mut self, id: u32) -> Result<(), Error> {
+    pub fn run(&mut self, source: u8, id: u32) -> Result<(), Error> {
-        info!("starting subkernel #{}", id);
         if self.session.kernel_state != KernelState::Loaded || self.session.id != id {
             self.load(id)?;
         }
         self.session.kernel_state = KernelState::Running;
+        self.session.source = source;
         unsafe {
             csr::cri_con::selected_write(2);
         }
@@ -325,11 +358,17 @@ impl<'a> Manager<'_> {
         Ok(())
     }
 
-    pub fn message_handle_incoming(&mut self, last: bool, length: usize, slice: &[u8; MASTER_PAYLOAD_MAX_SIZE]) {
+    pub fn message_handle_incoming(
+        &mut self,
+        status: PayloadStatus,
+        id: u32,
+        length: usize,
+        slice: &[u8; MASTER_PAYLOAD_MAX_SIZE],
+    ) {
         if !self.running() {
             return;
         }
-        self.session.messages.handle_incoming(last, length, slice);
+        self.session.messages.handle_incoming(status, id, length, slice);
     }
 
     pub fn message_get_slice(&mut self, slice: &mut [u8; MASTER_PAYLOAD_MAX_SIZE]) -> Option<SliceMeta> {
@@ -347,10 +386,6 @@ impl<'a> Manager<'_> {
         self.session.messages.ack_slice()
     }
 
-    pub fn message_is_ready(&mut self) -> bool {
-        self.session.messages.is_outgoing_ready()
-    }
-
     pub fn load(&mut self, id: u32) -> Result<(), Error> {
         if self.session.id == id && self.session.kernel_state == KernelState::Loaded {
             return Ok(());
@@ -378,14 +413,14 @@ impl<'a> Manager<'_> {
     pub fn exception_get_slice(&mut self, data_slice: &mut [u8; SAT_PAYLOAD_MAX_SIZE]) -> SliceMeta {
         match self.session.last_exception.as_mut() {
             Some(exception) => exception.get_slice_sat(data_slice),
-            None => SliceMeta { len: 0, last: true },
+            None => SliceMeta {
+                destination: 0,
+                len: 0,
+                status: PayloadStatus::FirstAndLast,
+            },
         }
     }
 
-    pub fn get_last_finished(&mut self) -> Option<SubkernelFinished> {
-        self.last_finished.take()
-    }
-
     fn kernel_stop(&mut self) {
         self.session.kernel_state = KernelState::Absent;
         unsafe {
@@ -415,13 +450,92 @@ impl<'a> Manager<'_> {
             &[],
             0,
         ) {
-            Ok(_) => self.session.last_exception = Some(Sliceable::new(writer.into_inner())),
+            Ok(_) => self.session.last_exception = Some(Sliceable::new(0, writer.into_inner())),
             Err(_) => error!("Error writing exception data"),
         }
         self.kernel_stop();
     }
 
-    pub fn process_kern_requests(&mut self, rank: u8, timer: GlobalTimer) {
+    pub fn ddma_finished(&mut self, error: u8, channel: u32, timestamp: u64) {
+        if let KernelState::DmaAwait { .. } = self.session.kernel_state {
+            self.control.tx.send(kernel::Message::DmaAwaitRemoteReply {
+                timeout: false,
+                error: error,
+                channel: channel,
+                timestamp: timestamp,
+            });
+            self.session.kernel_state = KernelState::Running;
+        }
+    }
+
+    pub fn ddma_nack(&mut self) {
+        // for simplicity treat it as a timeout...
+        if let KernelState::DmaAwait { .. } = self.session.kernel_state {
+            self.control.tx.send(kernel::Message::DmaAwaitRemoteReply {
+                timeout: true,
+                error: 0,
+                channel: 0,
+                timestamp: 0,
+            });
+            self.session.kernel_state = KernelState::Running;
+        }
+    }
+
+    pub fn ddma_remote_uploaded(&mut self, succeeded: bool) -> Option<(u32, u64)> {
+        // returns a tuple of id, timestamp in case a playback needs to be started immediately
+        if !succeeded {
+            self.kernel_stop();
+            self.runtime_exception(Error::DmaError(DmaError::UploadFail));
+        }
+        let res = match self.session.kernel_state {
+            KernelState::DmaPendingPlayback { id, timestamp } => {
+                self.session.kernel_state = KernelState::Running;
+                Some((id, timestamp))
+            }
+            KernelState::DmaPendingAwait {
+                id,
+                timestamp,
+                max_time,
+            } => {
+                self.session.kernel_state = KernelState::DmaAwait { max_time: max_time };
+                Some((id, timestamp))
+            }
+            KernelState::DmaUploading => {
+                self.session.kernel_state = KernelState::Running;
+                None
+            }
+            _ => None,
+        };
+        res
+    }
+
+    pub fn process_kern_requests(
+        &mut self,
+        router: &mut Router,
+        routing_table: &RoutingTable,
+        rank: u8,
+        destination: u8,
+        dma_manager: &mut DmaManager,
+        timer: &GlobalTimer,
+    ) {
+        if let Some(subkernel_finished) = self.last_finished.take() {
+            info!(
+                "subkernel {} finished, with exception: {}",
+                subkernel_finished.id, subkernel_finished.with_exception
+            );
+            router.route(
+                drtioaux::Packet::SubkernelFinished {
+                    destination: subkernel_finished.source,
+                    id: subkernel_finished.id,
+                    with_exception: subkernel_finished.with_exception,
+                    exception_src: subkernel_finished.exception_source,
+                },
+                &routing_table,
+                rank,
+                destination,
+            );
+        }
+
         if !self.running() {
             return;
         }
@@ -434,6 +548,8 @@ impl<'a> Manager<'_> {
                 self.last_finished = Some(SubkernelFinished {
                     id: self.session.id,
                     with_exception: true,
+                    exception_source: destination,
+                    source: self.session.source,
                 });
             }
             Err(e) => {
@@ -442,15 +558,19 @@ impl<'a> Manager<'_> {
                 self.last_finished = Some(SubkernelFinished {
                     id: self.session.id,
                     with_exception: true,
+                    exception_source: destination,
+                    source: self.session.source,
                 });
             }
         }
 
-        match self.process_kern_message(rank, timer) {
+        match self.process_kern_message(router, routing_table, rank, destination, dma_manager, timer) {
             Ok(true) => {
                 self.last_finished = Some(SubkernelFinished {
                     id: self.session.id,
                     with_exception: false,
+                    exception_source: 0,
+                    source: self.session.source,
                 });
             }
             Ok(false) | Err(Error::NoMessage) => (),
@@ -459,6 +579,8 @@ impl<'a> Manager<'_> {
                 self.last_finished = Some(SubkernelFinished {
                     id: self.session.id,
                     with_exception: true,
+                    exception_source: destination,
+                    source: self.session.source,
                 });
             }
             Err(e) => {
@@ -467,16 +589,52 @@ impl<'a> Manager<'_> {
                 self.last_finished = Some(SubkernelFinished {
                     id: self.session.id,
                     with_exception: true,
+                    exception_source: destination,
+                    source: self.session.source,
                 });
             }
         }
     }
 
-    fn process_kern_message(&mut self, rank: u8, timer: GlobalTimer) -> Result<bool, Error> {
+    pub fn subkernel_load_run_reply(&mut self, succeeded: bool) {
+        if self.session.kernel_state == KernelState::SubkernelAwaitLoad {
+            self.control
+                .tx
+                .send(kernel::Message::SubkernelLoadRunReply { succeeded: succeeded });
+            self.session.kernel_state = KernelState::Running;
+        } else {
+            warn!("received unsolicited SubkernelLoadRunReply");
+        }
+    }
+
+    pub fn remote_subkernel_finished(&mut self, id: u32, with_exception: bool, exception_source: u8) {
+        if with_exception {
+            self.kernel_stop();
+            self.last_finished = Some(SubkernelFinished {
+                source: self.session.source,
+                id: self.session.id,
+                with_exception: true,
+                exception_source: exception_source,
+            })
+        } else {
+            self.session.subkernels_finished.push(id);
+        }
+    }
+
+    fn process_kern_message(
+        &mut self,
+        router: &mut Router,
+        routing_table: &RoutingTable,
+        rank: u8,
+        self_destination: u8,
+        dma_manager: &mut DmaManager,
+        timer: &GlobalTimer,
+    ) -> Result<bool, Error> {
         let reply = self.control.rx.try_recv()?;
         match reply {
             kernel::Message::KernelFinished(_async_errors) => {
                 self.kernel_stop();
+                dma_manager.cleanup(router, rank, self_destination, routing_table);
                 return Ok(true);
             }
             kernel::Message::KernelException(exceptions, stack_pointers, backtrace, async_errors) => {
@@ -493,7 +651,7 @@ impl<'a> Manager<'_> {
                     Err(_) => error!("Error writing exception data"),
                 }
                 self.kernel_stop();
-                return Err(Error::KernelException(Sliceable::new(writer.into_inner())));
+                return Err(Error::KernelException(Sliceable::new(0, writer.into_inner())));
             }
             kernel::Message::CachePutRequest(key, value) => {
                 self.cache.insert(key, value);
@@ -503,18 +661,117 @@ impl<'a> Manager<'_> {
                 let value = self.cache.get(&key).unwrap_or(&DEFAULT).clone();
                 self.control.tx.send(kernel::Message::CacheGetReply(value));
             }
-            kernel::Message::SubkernelMsgSend { id: _, data } => {
+
-                self.session.messages.accept_outgoing(data)?;
+            kernel::Message::DmaPutRequest(recorder) => {
+                // ddma is always used on satellites
+                if let Ok(id) = dma_manager.put_record(recorder, self_destination) {
+                    dma_manager.upload_traces(id, router, rank, self_destination, routing_table)?;
+                    self.session.kernel_state = KernelState::DmaUploading;
+                } else {
+                    unexpected!("DMAError: found an unsupported call to RTIO devices on master")
+                }
+            }
+            kernel::Message::DmaEraseRequest(name) => {
+                dma_manager.erase_name(&name, router, rank, self_destination, routing_table);
+            }
+            kernel::Message::DmaGetRequest(name) => {
+                let dma_meta = dma_manager.retrieve(self_destination, &name);
+                self.control.tx.send(kernel::Message::DmaGetReply(dma_meta));
+            }
+            kernel::Message::DmaStartRemoteRequest { id, timestamp } => {
+                if self.session.kernel_state != KernelState::DmaUploading {
+                    dma_manager.playback_remote(
+                        id as u32,
+                        timestamp as u64,
+                        router,
+                        rank,
+                        self_destination,
+                        routing_table,
+                    )?;
+                } else {
+                    self.session.kernel_state = KernelState::DmaPendingPlayback {
+                        id: id as u32,
+                        timestamp: timestamp as u64,
+                    };
+                }
+            }
+            kernel::Message::DmaAwaitRemoteRequest(_id) => {
+                let max_time = timer.get_time() + Milliseconds(10000);
+                self.session.kernel_state = match self.session.kernel_state {
+                    // if we are still waiting for the traces to be uploaded, extend the state by timeout
+                    KernelState::DmaPendingPlayback { id, timestamp } => KernelState::DmaPendingAwait {
+                        id: id,
+                        timestamp: timestamp,
+                        max_time: max_time,
+                    },
+                    _ => KernelState::DmaAwait { max_time: max_time },
+                };
+            }
+
+            kernel::Message::SubkernelMsgSend {
+                id: _id,
+                destination: msg_dest,
+                data,
+            } => {
+                let msg_dest = msg_dest.or(Some(self.session.source)).unwrap();
+                self.session.messages.accept_outgoing(
+                    self.session.id,
+                    self_destination,
+                    msg_dest,
+                    data,
+                    routing_table,
+                    rank,
+                    router,
+                )?;
                 self.session.kernel_state = KernelState::MsgSending;
             }
-            kernel::Message::SubkernelMsgRecvRequest { id: _, timeout, tags } => {
+            kernel::Message::SubkernelMsgRecvRequest { id, timeout, tags } => {
-                let max_time = timer.get_time() + Milliseconds(timeout);
+                let id = if id == -1 { self.session.id } else { id as u32 };
-                self.session.kernel_state = KernelState::MsgAwait(max_time, tags);
+                let max_time = if timeout > 0 {
+                    Some(timer.get_time() + Milliseconds(timeout as u64))
+                } else {
+                    None
+                };
+                self.session.kernel_state = KernelState::MsgAwait {
+                    max_time: max_time,
+                    id: id,
+                    tags: tags,
+                };
+            }
+            kernel::Message::SubkernelLoadRunRequest {
+                id,
+                destination: sk_destination,
+                run,
+            } => {
+                self.session.kernel_state = KernelState::SubkernelAwaitLoad;
+                router.route(
+                    drtioaux::Packet::SubkernelLoadRunRequest {
+                        source: self_destination,
+                        destination: sk_destination,
+                        id: id,
+                        run: run,
+                    },
+                    routing_table,
+                    rank,
+                    self_destination,
+                );
+            }
+
+            kernel::Message::SubkernelAwaitFinishRequest { id, timeout } => {
+                let max_time = if timeout > 0 {
+                    Some(timer.get_time() + Milliseconds(timeout as u64))
+                } else {
+                    None
+                };
+                self.session.kernel_state = KernelState::SubkernelAwaitFinish {
+                    max_time: max_time,
+                    id: id,
+                };
             }
             kernel::Message::UpDestinationsRequest(destination) => {
-                self.control
+                self.control.tx.send(kernel::Message::UpDestinationsReply(
-                    .tx
+                    destination == (self_destination as i32),
-                    .send(kernel::Message::UpDestinationsReply(destination == (rank as i32)));
+                ));
             }
             _ => {
                 unexpected!("unexpected message from core1 while kernel was running: {:?}", reply);
@@ -523,10 +780,11 @@ impl<'a> Manager<'_> {
         Ok(false)
     }
 
-    fn process_external_messages(&mut self, timer: GlobalTimer) -> Result<(), Error> {
+    fn process_external_messages(&mut self, timer: &GlobalTimer) -> Result<(), Error> {
         match &self.session.kernel_state {
-            KernelState::MsgAwait(timeout, tags) => {
+            KernelState::MsgAwait { max_time, id, tags } => {
-                if timer.get_time() > *timeout {
+                if let Some(max_time) = *max_time {
+                    if timer.get_time() > max_time {
                         self.control.tx.send(kernel::Message::SubkernelMsgRecvReply {
                             status: kernel::SubkernelStatus::Timeout,
                             count: 0,
@@ -534,7 +792,8 @@ impl<'a> Manager<'_> {
                         self.session.kernel_state = KernelState::Running;
                         return Ok(());
                     }
-                if let Some(message) = self.session.messages.get_incoming() {
+                }
+                if let Some(message) = self.session.messages.get_incoming(*id) {
                     self.control.tx.send(kernel::Message::SubkernelMsgRecvReply {
                         status: kernel::SubkernelStatus::NoError,
                         count: message.count,
@@ -555,11 +814,47 @@ impl<'a> Manager<'_> {
                     Err(Error::AwaitingMessage)
                 }
             }
+            KernelState::SubkernelAwaitFinish { max_time, id } => {
+                if let Some(max_time) = *max_time {
+                    if timer.get_time() > max_time {
+                        self.control.tx.send(kernel::Message::SubkernelAwaitFinishReply {
+                            status: kernel::SubkernelStatus::Timeout,
+                        });
+                        self.session.kernel_state = KernelState::Running;
+                        return Ok(());
+                    }
+                }
+                let mut i = 0;
+                for status in &self.session.subkernels_finished {
+                    if *status == *id {
+                        self.control.tx.send(kernel::Message::SubkernelAwaitFinishReply {
+                            status: kernel::SubkernelStatus::NoError,
+                        });
+                        self.session.kernel_state = KernelState::Running;
+                        self.session.subkernels_finished.swap_remove(i);
+                        break;
+                    }
+                    i += 1;
+                }
+                Ok(())
+            }
+            KernelState::DmaAwait { max_time } | KernelState::DmaPendingAwait { max_time, .. } => {
+                if timer.get_time() > *max_time {
+                    self.control.tx.send(kernel::Message::DmaAwaitRemoteReply {
+                        timeout: true,
+                        error: 0,
+                        channel: 0,
+                        timestamp: 0,
+                    });
+                    self.session.kernel_state = KernelState::Running;
+                }
+                Ok(())
+            }
             _ => Ok(()),
         }
     }
 
-    fn pass_message_to_kernel(&mut self, message: &Message, tags: Vec<u8>, timer: GlobalTimer) -> Result<(), Error> {
+    fn pass_message_to_kernel(&mut self, message: &Message, tags: Vec<u8>, timer: &GlobalTimer) -> Result<(), Error> {
         let mut reader = Cursor::new(&message.data);
         let mut current_tags: &[u8] = &tags;
         let mut i = message.count;
@@ -582,7 +877,7 @@ impl<'a> Manager<'_> {
                             let mut writer = Cursor::new(buf);
                             match write_exception(&mut writer, exceptions, stack_pointers, backtrace, async_errors) {
                                 Ok(()) => {
-                                    exception = Some(Sliceable::new(writer.into_inner()));
+                                    exception = Some(Sliceable::new(0, writer.into_inner()));
                                 }
                                 Err(_) => {
                                     unexpected = Some("Error writing exception data".to_string());
@@ -676,7 +971,7 @@ where
 
 fn recv_w_timeout(
     rx: &mut Receiver<'_, kernel::Message>,
-    timer: GlobalTimer,
+    timer: &GlobalTimer,
     timeout: u64,
 ) -> Result<kernel::Message, Error> {
     let max_time = timer.get_time() + Milliseconds(timeout);