Skip to contentSkip to conversation

pc: support user provided NIC naming/indexing

RFC In review Igor Mammedov 2 participants 14 messages Dec 22, 2020 – Jan 26, 2021

lore
Share & tools
lore
#0 lore raw

cover pc: support user provided NIC naming/indexing

Series implements support for 'onboard' naming scheme for network interfaces (1), which is based on PCI firmware spec and lets user to explicitly specify index that will be used by guest to name network interface, ex:

    -device e1000,acpi-index=33

should make guest rename NIC name to 'eno33' where 'eno' is default prefix for this scheme.

Hope is that it will allow to simplify launching VMs from template disk images with different set VM configurations without need to reconfigure guest network intrfaces or risk of loosing network connectivity.

For more detailed description/examples see patches [3-4/5]

1)
https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/

Git repo for testing:
https://github.com/imammedo/qemu/branches acpi-index-rfc

Igor Mammedov (5):
acpi: add aml_to_decimalstring() and aml_call6() helpers
tests: acpi: temporary whitelist DSDT changes
pci: introduce apci-index property for PCI device
pci: acpi: add _DSM method to PCI devices
tests: acpi: update expected data files

 include/hw/acpi/aml-build.h                 |   3 +
 include/hw/acpi/pci.h                       |   1 +
 include/hw/acpi/pcihp.h                     |   7 +-
 include/hw/pci/pci.h                        |   1 +
 tests/qtest/bios-tables-test-allowed-diff.h |  21 +++++
 hw/acpi/aml-build.c                         |  28 +++++++
 hw/acpi/pci.c                               |  84 ++++++++++++++++++++
 hw/acpi/pcihp.c                             |  25 +++++-
 hw/i386/acpi-build.c                        |  31 +++++++-
 hw/pci/pci.c                                |   1 +
 tests/data/acpi/pc/DSDT                     | Bin 5065 -> 6023 bytes
 tests/data/acpi/pc/DSDT.acpihmat            | Bin 6390 -> 7348 bytes
 tests/data/acpi/pc/DSDT.bridge              | Bin 6924 -> 8689 bytes
 tests/data/acpi/pc/DSDT.cphp                | Bin 5529 -> 6487 bytes
 tests/data/acpi/pc/DSDT.dimmpxm             | Bin 6719 -> 7677 bytes
 tests/data/acpi/pc/DSDT.hpbridge            | Bin 5026 -> 5990 bytes
 tests/data/acpi/pc/DSDT.hpbrroot            | Bin 3084 -> 3177 bytes
 tests/data/acpi/pc/DSDT.ipmikcs             | Bin 5137 -> 6095 bytes
 tests/data/acpi/pc/DSDT.memhp               | Bin 6424 -> 7382 bytes
 tests/data/acpi/pc/DSDT.numamem             | Bin 5071 -> 6029 bytes
 tests/data/acpi/pc/DSDT.roothp              | Bin 5261 -> 6324 bytes
 tests/data/acpi/q35/DSDT                    | Bin 7801 -> 7863 bytes
 tests/data/acpi/q35/DSDT.acpihmat           | Bin 9126 -> 9188 bytes
 tests/data/acpi/q35/DSDT.bridge             | Bin 7819 -> 7911 bytes
 tests/data/acpi/q35/DSDT.cphp               | Bin 8265 -> 8327 bytes
 tests/data/acpi/q35/DSDT.dimmpxm            | Bin 9455 -> 9517 bytes
 tests/data/acpi/q35/DSDT.ipmibt             | Bin 7876 -> 7938 bytes
 tests/data/acpi/q35/DSDT.memhp              | Bin 9160 -> 9222 bytes
 tests/data/acpi/q35/DSDT.mmio64             | Bin 8932 -> 9024 bytes
 tests/data/acpi/q35/DSDT.numamem            | Bin 7807 -> 7869 bytes
 tests/data/acpi/q35/DSDT.tis                | Bin 8407 -> 8468 bytes
 31 files changed, 197 insertions(+), 5 deletions(-)
Signature · 1 line
2.27.0
#1 lore raw

RFC 1/5 acpi: add aml_to_decimalstring() and aml_call6() helpers

it will be used by follow up patches

---
 include/hw/acpi/aml-build.h |  3 +++
 hw/acpi/aml-build.c         | 28 ++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+)
include/hw/acpi/aml-build.h +3 -0
index e727bea1bc..695b7e6323 100644
@@ -301,6 +301,7 @@ Aml *aml_arg(int pos);
 Aml *aml_to_integer(Aml *arg);
 Aml *aml_to_hexstring(Aml *src, Aml *dst);
 Aml *aml_to_buffer(Aml *src, Aml *dst);
+Aml *aml_to_decimalstring(Aml *src, Aml *dst);
 Aml *aml_store(Aml *val, Aml *target);
 Aml *aml_and(Aml *arg1, Aml *arg2, Aml *dst);
 Aml *aml_or(Aml *arg1, Aml *arg2, Aml *dst);
@@ -323,6 +324,8 @@ Aml *aml_call3(const char *method, Aml *arg1, Aml *arg2, Aml *arg3);
 Aml *aml_call4(const char *method, Aml *arg1, Aml *arg2, Aml *arg3, Aml *arg4);
 Aml *aml_call5(const char *method, Aml *arg1, Aml *arg2, Aml *arg3, Aml *arg4,
                Aml *arg5);
+Aml *aml_call6(const char *method, Aml *arg1, Aml *arg2, Aml *arg3, Aml *arg4,
+               Aml *arg5, Aml *arg6);
 Aml *aml_gpio_int(AmlConsumerAndProducer con_and_pro,
                   AmlLevelAndEdge edge_level,
                   AmlActiveHighAndLow active_level, AmlShared shared,
hw/acpi/aml-build.c +28 -0
index f976aa667b..89702be788 100644
@@ -633,6 +633,19 @@ Aml *aml_to_buffer(Aml *src, Aml *dst)
     return var;
 }
 
+/* ACPI 2.0a: 17.2.4.4 Type 2 Opcodes Encoding: DefToDecimalString */
+Aml *aml_to_decimalstring(Aml *src, Aml *dst)
+{
+    Aml *var = aml_opcode(0x97 /* ToDecimalStringOp */);
+    aml_append(var, src);
+    if (dst) {
+        aml_append(var, dst);
+    } else {
+        build_append_byte(var->buf, 0x00 /* NullNameOp */);
+    }
+    return var;
+}
+
 /* ACPI 1.0b: 16.2.5.4 Type 2 Opcodes Encoding: DefStore */
 Aml *aml_store(Aml *val, Aml *target)
 {
@@ -834,6 +847,21 @@ Aml *aml_call5(const char *method, Aml *arg1, Aml *arg2, Aml *arg3, Aml *arg4,
     return var;
 }
 
+/* helper to call method with 5 arguments */
+Aml *aml_call6(const char *method, Aml *arg1, Aml *arg2, Aml *arg3, Aml *arg4,
+               Aml *arg5, Aml *arg6)
+{
+    Aml *var = aml_alloc();
+    build_append_namestring(var->buf, "%s", method);
+    aml_append(var, arg1);
+    aml_append(var, arg2);
+    aml_append(var, arg3);
+    aml_append(var, arg4);
+    aml_append(var, arg5);
+    aml_append(var, arg6);
+    return var;
+}
+
 /*
  * ACPI 5.0: 6.4.3.8.1 GPIO Connection Descriptor
  * Type 1, Large Item Name 0xC
Signature · 1 line
2.27.0
#2 lore raw

RFC 2/5 tests: acpi: temporary whitelist DSDT changes

In reply to Igor Mammedov
---
 tests/qtest/bios-tables-test-allowed-diff.h | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)
tests/qtest/bios-tables-test-allowed-diff.h +21 -0
index dfb8523c8b..5c695cdf37 100644
@@ -1 +1,22 @@
 /* List of comma-separated changed AML files to ignore */
+"tests/data/acpi/pc/DSDT",
+"tests/data/acpi/pc/DSDT.acpihmat",
+"tests/data/acpi/pc/DSDT.bridge",
+"tests/data/acpi/pc/DSDT.cphp",
+"tests/data/acpi/pc/DSDT.dimmpxm",
+"tests/data/acpi/pc/DSDT.hpbridge",
+"tests/data/acpi/pc/DSDT.hpbrroot",
+"tests/data/acpi/pc/DSDT.ipmikcs",
+"tests/data/acpi/pc/DSDT.memhp",
+"tests/data/acpi/pc/DSDT.numamem",
+"tests/data/acpi/pc/DSDT.roothp",
+"tests/data/acpi/q35/DSDT",
+"tests/data/acpi/q35/DSDT.acpihmat",
+"tests/data/acpi/q35/DSDT.bridge",
+"tests/data/acpi/q35/DSDT.cphp",
+"tests/data/acpi/q35/DSDT.dimmpxm",
+"tests/data/acpi/q35/DSDT.ipmibt",
+"tests/data/acpi/q35/DSDT.memhp",
+"tests/data/acpi/q35/DSDT.mmio64",
+"tests/data/acpi/q35/DSDT.numamem",
+"tests/data/acpi/q35/DSDT.tis",
Signature · 1 line
2.27.0
#3 lore raw

RFC 3/5 pci: introduce apci-index property for PCI device

In reply to Igor Mammedov

In x86/ACPI world, since systemd v197, linux distros are using predictable network interface naming since systemd v197. Which on QEMU based VMs results into path based naming scheme, that names network interfaces based on PCI topology.

With this one has to plug NIC in exacly the same bus/slot, which was used when disk image was first provisioned/configured or one risks to loose network configuration due to NIC being renamed to actually used topology.
That also restricts freedom reshape PCI configuration of VM without need to reconfigure used guest image.

systemd also offers "onboard" naming scheme which is preffered over PCI slot/topology one, provided that firmware implements:
"
PCI Firmware Specification 3.1
4.6.7. DSM for Naming a PCI or PCI Express Device Under
Operating Systems
"
that allows to assign user defined index to PCI device, which systemd will use to name NIC. For example, using
-device e1000,acpi-index=100
guest will rename NIC to 'eno100', where 'eno' is default prefix for "onboard" naming scheme. This doesn't reqiure any advance configuration on guest side.

Hope is that 'acpi-index' will be easier to consume by mangment layer, compared to forcing specic PCI topology and/or having several disk image templates for different topologies and will help to simplify process of spawning VM from the same template without need to reconfigure guest network configuration.

this patch adds, 'acpi-index'* property and wires up (abuses) unused pci hotplug registers to pass index value to AML code at runtime.
Following patch will add corresponding _DSM code and wire it up to PCI devices described in ACPI.

*) name comes from linux kernel terminology

---
CC: [email protected]

 include/hw/acpi/pcihp.h |  7 ++++++-
 include/hw/pci/pci.h    |  1 +
 hw/acpi/pci.c           |  6 ++++++
 hw/acpi/pcihp.c         | 25 ++++++++++++++++++++++++-
 hw/i386/acpi-build.c    | 10 ++++++++++
 hw/pci/pci.c            |  1 +
 6 files changed, 48 insertions(+), 2 deletions(-)
include/hw/acpi/pcihp.h +6 -1
index dfd375820f..72d1773ca1 100644
@@ -46,6 +46,7 @@ typedef struct AcpiPciHpPciStatus {
 typedef struct AcpiPciHpState {
     AcpiPciHpPciStatus acpi_pcihp_pci_status[ACPI_PCIHP_MAX_HOTPLUG_BUS];
     uint32_t hotplug_select;
+    uint32_t acpi_index;
     PCIBus *root;
     MemoryRegion io;
     bool legacy_piix;
@@ -71,6 +72,8 @@ void acpi_pcihp_reset(AcpiPciHpState *s, bool acpihp_root_off);
 
 extern const VMStateDescription vmstate_acpi_pcihp_pci_status;
 
+bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id);
+
 #define VMSTATE_PCI_HOTPLUG(pcihp, state, test_pcihp) \
         VMSTATE_UINT32_TEST(pcihp.hotplug_select, state, \
                             test_pcihp), \
@@ -78,6 +81,8 @@ extern const VMStateDescription vmstate_acpi_pcihp_pci_status;
                                   ACPI_PCIHP_MAX_HOTPLUG_BUS, \
                                   test_pcihp, 1, \
                                   vmstate_acpi_pcihp_pci_status, \
-                                  AcpiPciHpPciStatus)
+                                  AcpiPciHpPciStatus), \
+        VMSTATE_UINT32_TEST(pcihp.acpi_index, state, \
+                            vmstate_acpi_pcihp_use_acpi_index)
 
 #endif
include/hw/pci/pci.h +1 -0
index 259f9c992d..e592532558 100644
@@ -357,6 +357,7 @@ struct PCIDevice {
 
     /* ID of standby device in net_failover pair */
     char *failover_pair_id;
+    uint32_t acpi_index;
 };
 
 void pci_register_bar(PCIDevice *pci_dev, int region_num,
hw/acpi/pci.c +6 -0
index 9510597a19..07d5101d83 100644
@@ -27,6 +27,7 @@
 #include "hw/acpi/aml-build.h"
 #include "hw/acpi/pci.h"
 #include "hw/pci/pcie_host.h"
+#include "hw/acpi/pcihp.h"
 
 void build_mcfg(GArray *table_data, BIOSLinker *linker, AcpiMcfgInfo *info)
 {
@@ -59,3 +60,8 @@ void build_mcfg(GArray *table_data, BIOSLinker *linker, AcpiMcfgInfo *info)
                  "MCFG", table_data->len - mcfg_start, 1, NULL, NULL);
 }
 
+bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id)
+{
+     AcpiPciHpState *s = opaque;
+     return s->acpi_index;
+}
hw/acpi/pcihp.c +24 -1
index 9dc4d3e2db..9634567e3a 100644
@@ -347,7 +347,8 @@ static uint64_t pci_read(void *opaque, hwaddr addr, unsigned int size)
         trace_acpi_pci_down_read(val);
         break;
     case PCI_EJ_BASE:
-        /* No feature defined yet */
+        val = s->acpi_index;
+        s->acpi_index = 0;
         trace_acpi_pci_features_read(val);
         break;
     case PCI_RMV_BASE:
@@ -367,8 +368,30 @@ static uint64_t pci_read(void *opaque, hwaddr addr, unsigned int size)
 static void pci_write(void *opaque, hwaddr addr, uint64_t data,
                       unsigned int size)
 {
+    int slot;
+    PCIBus *bus;
+    BusChild *kid, *next;
     AcpiPciHpState *s = opaque;
+
+    s->acpi_index = 0;
     switch (addr) {
+    case PCI_UP_BASE:
+        slot = ctz32(data);
+
+        if (s->hotplug_select >= ACPI_PCIHP_MAX_HOTPLUG_BUS) {
+            break;
+        }
+
+        bus = acpi_pcihp_find_hotplug_bus(s, s->hotplug_select);
+        QTAILQ_FOREACH_SAFE(kid, &bus->qbus.children, sibling, next) {
+            Object *o = OBJECT(kid->child);
+            PCIDevice *dev = PCI_DEVICE(o);
+            if (PCI_SLOT(dev->devfn) == slot) {
+                s->acpi_index = object_property_get_uint(o, "acpi-index", NULL);
+                break;
+            }
+        }
+        break;
     case PCI_EJ_BASE:
         if (s->hotplug_select >= ACPI_PCIHP_MAX_HOTPLUG_BUS) {
             break;
hw/i386/acpi-build.c +10 -0
index f18b71dea9..27d2958e25 100644
@@ -1132,6 +1132,16 @@ static void build_piix4_pci_hotplug(Aml *table)
     aml_append(method, aml_return(aml_int(0)));
     aml_append(scope, method);
 
+    method = aml_method("AIDX", 2, AML_NOTSERIALIZED);
+    aml_append(method, aml_acquire(aml_name("BLCK"), 0xFFFF));
+    aml_append(method, aml_store(aml_arg(0), aml_name("BNUM")));
+    aml_append(method,
+        aml_store(aml_shiftleft(aml_int(1), aml_arg(1)), aml_name("PCIU")));
+    aml_append(method, aml_store(aml_name("B0EJ"), aml_local(0)));
+    aml_append(method, aml_release(aml_name("BLCK")));
+    aml_append(method, aml_return(aml_local(0)));
+    aml_append(scope, method);
+
     aml_append(table, scope);
 }
 
hw/pci/pci.c +1 -0
index d4349ea577..617f48ff3b 100644
@@ -76,6 +76,7 @@ static Property pci_props[] = {
                     QEMU_PCIE_EXTCAP_INIT_BITNR, true),
     DEFINE_PROP_STRING("failover_pair_id", PCIDevice,
                        failover_pair_id),
+    DEFINE_PROP_UINT32("acpi-index",  PCIDevice, acpi_index, -1),
     DEFINE_PROP_END_OF_LIST()
 };
 
Signature · 1 line
2.27.0
#4 lore raw

RFC 4/5 pci: acpi: add _DSM method to PCI devices

In reply to Igor Mammedov

Implement _DSM according to:
PCI Firmware Specification 3.1
4.6.7. DSM for Naming a PCI or PCI Express Device Under
Operating Systems
and wire it up to cold and hot-plugged PCI devices. Feature depends on ACPI hotplug being enabled (as that provides PCI devices descriptions in ACPI and MMIO registers that are reused to fetch acpi-index).

acpi-index should work for
- cold plugged NICs:

      $QEMU -evice e1000,acpi-index=100
         => 'eno100'
  - hot-plugged
      (monitor) device_add e1000,acpi-index=200,id=remove_me
         => 'eno200'
  - replugged
      (monitor) device_del remove_me
      (monitor) device_add e1000,acpi-index=1
         => 'eno1'

Windows also sees index under "PCI Label Id" field in properties dialog but otherwise it doesn't seem to have any effect.

---
 include/hw/acpi/pci.h |  1 +
 hw/acpi/pci.c         | 78 +++++++++++++++++++++++++++++++++++++++++++
 hw/i386/acpi-build.c  | 21 ++++++++++--
 3 files changed, 97 insertions(+), 3 deletions(-)
include/hw/acpi/pci.h +1 -0
index bf2a3ed0ba..5e1eb2a96a 100644
@@ -34,4 +34,5 @@ typedef struct AcpiMcfgInfo {
 } AcpiMcfgInfo;
 
 void build_mcfg(GArray *table_data, BIOSLinker *linker, AcpiMcfgInfo *info);
+Aml *aml_pci_device_dsm(void);
 #endif
hw/acpi/pci.c +78 -0

81 lines not on this page · raw message

hw/i386/acpi-build.c +18 -3
index 27d2958e25..447ad39c35 100644
@@ -385,6 +385,13 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
                     aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
                 );
                 aml_append(dev, method);
+                method = aml_method("_DSM", 4, AML_SERIALIZED);
+                aml_append(method,
+                    aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1),
+                                         aml_arg(2), aml_arg(3),
+                                         aml_name("BSEL"), aml_name("_SUN")))
+                );
+                aml_append(dev, method);
                 aml_append(parent_scope, dev);
 
                 build_append_pcihp_notify_entry(notify_method, slot);
@@ -412,6 +419,14 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
         dev = aml_device("S%.02X", PCI_DEVFN(slot, 0));
         aml_append(dev, aml_name_decl("_ADR", aml_int(slot << 16)));
 
+        aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
+        method = aml_method("_DSM", 4, AML_SERIALIZED);
+        aml_append(method,
+           aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1), aml_arg(2),
+                                aml_arg(3), aml_name("BSEL"), aml_name("_SUN")))
+        );
+        aml_append(dev, method);
+
         if (pc->class_id == PCI_CLASS_DISPLAY_VGA) {
             /* add VGA specific AML methods */
             int s3d;
@@ -434,9 +449,7 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
             aml_append(method, aml_return(aml_int(s3d)));
             aml_append(dev, method);
         } else if (hotplug_enabled_dev) {
-            /* add _SUN/_EJ0 to make slot hotpluggable  */
-            aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
-
+            /* add _EJ0 to make slot hotpluggable  */
             method = aml_method("_EJ0", 1, AML_NOTSERIALIZED);
             aml_append(method,
                 aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
@@ -1142,6 +1155,8 @@ static void build_piix4_pci_hotplug(Aml *table)
     aml_append(method, aml_return(aml_local(0)));
     aml_append(scope, method);
 
+    aml_append(scope, aml_pci_device_dsm());
+
     aml_append(table, scope);
 }
 
Signature · 1 line
2.27.0
#5 lore raw

Re: [RFC 4/5] pci: acpi: add _DSM method to PCI devices

Quoted 110 lines · Igor Mammedov

Implement _DSM according to:
PCI Firmware Specification 3.1
4.6.7. DSM for Naming a PCI or PCI Express Device Under
Operating Systems
and wire it up to cold and hot-plugged PCI devices. Feature depends on ACPI hotplug being enabled (as that provides PCI devices descriptions in ACPI and MMIO registers that are reused to fetch acpi-index).

acpi-index should work for
- cold plugged NICs:

      $QEMU -evice e1000,acpi-index=100
         => 'eno100'
  - hot-plugged
      (monitor) device_add e1000,acpi-index=200,id=remove_me
         => 'eno200'
  - replugged
      (monitor) device_del remove_me
      (monitor) device_add e1000,acpi-index=1
         => 'eno1'

Windows also sees index under "PCI Label Id" field in properties dialog but otherwise it doesn't seem to have any effect.

---
 include/hw/acpi/pci.h |  1 +
 hw/acpi/pci.c         | 78 +++++++++++++++++++++++++++++++++++++++++++
 hw/i386/acpi-build.c  | 21 ++++++++++--
 3 files changed, 97 insertions(+), 3 deletions(-)
include/hw/acpi/pci.h +1 -0
index bf2a3ed0ba..5e1eb2a96a 100644
@@ -34,4 +34,5 @@ typedef struct AcpiMcfgInfo {
 } AcpiMcfgInfo;
 
 void build_mcfg(GArray *table_data, BIOSLinker *linker, AcpiMcfgInfo *info);
+Aml *aml_pci_device_dsm(void);
 #endif
hw/acpi/pci.c +61 -0
index 07d5101d83..6d49d515d3 100644
@@ -65,3 +65,81 @@ bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id)
      AcpiPciHpState *s = opaque;
      return s->acpi_index;
 }
+
+Aml *aml_pci_device_dsm(void)
+{
+    Aml *method, *UUID, *ifctx, *ifctx1, *ifctx2, *ifctx3, *elsectx;
+    Aml *acpi_index = aml_local(0);
+    Aml *zero = aml_int(0);
+    Aml *bnum = aml_arg(4);
+    Aml *sun = aml_arg(5);
+
+    method = aml_method("PDSM", 6, AML_SERIALIZED);
+
+    /*
+     * PCI Firmware Specification 3.1
+     * 4.6.  _DSM Definitions for PCI
+     */
+    UUID = aml_touuid("E5C937D0-3553-4D7A-9117-EA4D19C3434D");
+    ifctx = aml_if(aml_equal(aml_arg(0), UUID));
+    {
+        aml_append(ifctx, aml_store(aml_call2("AIDX", bnum, sun), acpi_index));
+        ifctx1 = aml_if(aml_equal(aml_arg(2), zero));
+        {
+            uint8_t byte_list[1];
+
+            ifctx2 = aml_if(aml_equal(aml_arg(1), aml_int(2)));
+            {
+                /*
+                 * advertise function 7 if device has acpi-index
+                 */
+                ifctx3 = aml_if(aml_lnot(aml_equal(acpi_index, zero)));
+                {
+                    byte_list[0] =
+                        1 /* have supported functions */ |
+                        1 << 7 /* support for function 7 */
+                    ;
+                    aml_append(ifctx3, aml_return(aml_buffer(1, byte_list)));
+                }
+                aml_append(ifctx2, ifctx3);
+             }
+             aml_append(ifctx1, ifctx2);
+
+             byte_list[0] = 0; /* nothing supported */
+             aml_append(ifctx1, aml_return(aml_buffer(1, byte_list)));
+         }
+         aml_append(ifctx, ifctx1);
+         elsectx = aml_else();
+         /*
+          * PCI Firmware Specification 3.1
+          * 4.6.7. _DSM for Naming a PCI or PCI Express Device Under
+          *        Operating Systems
+          */
+         ifctx1 = aml_if(aml_equal(aml_arg(2), aml_int(7)));
+         {
+             Aml *pkg = aml_package(2);
+             Aml *label = aml_local(2);
+             Aml *ret = aml_local(1);
+
+             aml_append(ifctx1, aml_concatenate(aml_string("PCI Device "),
+                 aml_to_decimalstring(acpi_index, NULL), label));
+
+             aml_append(pkg, zero);
+             aml_append(pkg, aml_string("placeholder"));

hw/acpi/pci.c bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id) · line 138

Quoted 10 lines
hw/acpi/pci.c +10 -0
+             aml_append(ifctx1, aml_store(pkg, ret));
+             /*
+              * update apci-index to actual value
+              */
+             aml_append(ifctx1, aml_store(acpi_index, aml_index(ret, zero)));
+             /*
+              * update device label to actual value
+              */
+             aml_append(ifctx1, aml_store(label, aml_index(ret, aml_int(1))));
+             aml_append(ifctx1, aml_return(ret));

This code needs more comments to explain what it's doing. E.g. what is device label?
Spec seems to say the string is optional. Is it relevant somehow?

spec also says string must be unique. Do we need to enforce that?

Quoted 62 lines
hw/acpi/pci.c +7 -0
+         }
+         aml_append(elsectx, ifctx1);
+         aml_append(ifctx, elsectx);
+    }
+    aml_append(method, ifctx);
+    return method;
+}
hw/i386/acpi-build.c +18 -3
index 27d2958e25..447ad39c35 100644
@@ -385,6 +385,13 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
                     aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
                 );
                 aml_append(dev, method);
+                method = aml_method("_DSM", 4, AML_SERIALIZED);
+                aml_append(method,
+                    aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1),
+                                         aml_arg(2), aml_arg(3),
+                                         aml_name("BSEL"), aml_name("_SUN")))
+                );
+                aml_append(dev, method);
                 aml_append(parent_scope, dev);
 
                 build_append_pcihp_notify_entry(notify_method, slot);
@@ -412,6 +419,14 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
         dev = aml_device("S%.02X", PCI_DEVFN(slot, 0));
         aml_append(dev, aml_name_decl("_ADR", aml_int(slot << 16)));
 
+        aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
+        method = aml_method("_DSM", 4, AML_SERIALIZED);
+        aml_append(method,
+           aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1), aml_arg(2),
+                                aml_arg(3), aml_name("BSEL"), aml_name("_SUN")))
+        );
+        aml_append(dev, method);
+
         if (pc->class_id == PCI_CLASS_DISPLAY_VGA) {
             /* add VGA specific AML methods */
             int s3d;
@@ -434,9 +449,7 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
             aml_append(method, aml_return(aml_int(s3d)));
             aml_append(dev, method);
         } else if (hotplug_enabled_dev) {
-            /* add _SUN/_EJ0 to make slot hotpluggable  */
-            aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
-
+            /* add _EJ0 to make slot hotpluggable  */
             method = aml_method("_EJ0", 1, AML_NOTSERIALIZED);
             aml_append(method,
                 aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
@@ -1142,6 +1155,8 @@ static void build_piix4_pci_hotplug(Aml *table)
     aml_append(method, aml_return(aml_local(0)));
     aml_append(scope, method);
 
+    aml_append(scope, aml_pci_device_dsm());
+
     aml_append(table, scope);
 }
 
Signature · 1 line
2.27.0
#6 lore raw

Re: [RFC 4/5] pci: acpi: add _DSM method to PCI devices

Quoted 117 lines · On Wed, 13 Jan 2021 07:13:04 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 116 lines · Igor Mammedov

Implement _DSM according to:
PCI Firmware Specification 3.1
4.6.7. DSM for Naming a PCI or PCI Express Device Under
Operating Systems
and wire it up to cold and hot-plugged PCI devices. Feature depends on ACPI hotplug being enabled (as that provides PCI devices descriptions in ACPI and MMIO registers that are reused to fetch acpi-index).

acpi-index should work for
- cold plugged NICs:

      $QEMU -evice e1000,acpi-index=100  
         => 'eno100'  
  - hot-plugged
      (monitor) device_add e1000,acpi-index=200,id=remove_me  
         => 'eno200'  
  - replugged
      (monitor) device_del remove_me
      (monitor) device_add e1000,acpi-index=1  
         => 'eno1'  

Windows also sees index under "PCI Label Id" field in properties dialog but otherwise it doesn't seem to have any effect.

---
 include/hw/acpi/pci.h |  1 +
 hw/acpi/pci.c         | 78 +++++++++++++++++++++++++++++++++++++++++++
 hw/i386/acpi-build.c  | 21 ++++++++++--
 3 files changed, 97 insertions(+), 3 deletions(-)
include/hw/acpi/pci.h +1 -0
index bf2a3ed0ba..5e1eb2a96a 100644
@@ -34,4 +34,5 @@ typedef struct AcpiMcfgInfo {
 } AcpiMcfgInfo;
 
 void build_mcfg(GArray *table_data, BIOSLinker *linker, AcpiMcfgInfo *info);
+Aml *aml_pci_device_dsm(void);
 #endif
hw/acpi/pci.c +67 -0
index 07d5101d83..6d49d515d3 100644
@@ -65,3 +65,81 @@ bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id)
      AcpiPciHpState *s = opaque;
      return s->acpi_index;
 }
+
+Aml *aml_pci_device_dsm(void)
+{
+    Aml *method, *UUID, *ifctx, *ifctx1, *ifctx2, *ifctx3, *elsectx;
+    Aml *acpi_index = aml_local(0);
+    Aml *zero = aml_int(0);
+    Aml *bnum = aml_arg(4);
+    Aml *sun = aml_arg(5);
+
+    method = aml_method("PDSM", 6, AML_SERIALIZED);
+
+    /*
+     * PCI Firmware Specification 3.1
+     * 4.6.  _DSM Definitions for PCI
+     */
+    UUID = aml_touuid("E5C937D0-3553-4D7A-9117-EA4D19C3434D");
+    ifctx = aml_if(aml_equal(aml_arg(0), UUID));
+    {
+        aml_append(ifctx, aml_store(aml_call2("AIDX", bnum, sun), acpi_index));
+        ifctx1 = aml_if(aml_equal(aml_arg(2), zero));
+        {
+            uint8_t byte_list[1];
+
+            ifctx2 = aml_if(aml_equal(aml_arg(1), aml_int(2)));
+            {
+                /*
+                 * advertise function 7 if device has acpi-index
+                 */
+                ifctx3 = aml_if(aml_lnot(aml_equal(acpi_index, zero)));
+                {
+                    byte_list[0] =
+                        1 /* have supported functions */ |
+                        1 << 7 /* support for function 7 */
+                    ;
+                    aml_append(ifctx3, aml_return(aml_buffer(1, byte_list)));
+                }
+                aml_append(ifctx2, ifctx3);
+             }
+             aml_append(ifctx1, ifctx2);
+
+             byte_list[0] = 0; /* nothing supported */
+             aml_append(ifctx1, aml_return(aml_buffer(1, byte_list)));
+         }
+         aml_append(ifctx, ifctx1);
+         elsectx = aml_else();
+         /*
+          * PCI Firmware Specification 3.1
+          * 4.6.7. _DSM for Naming a PCI or PCI Express Device Under
+          *        Operating Systems
+          */
+         ifctx1 = aml_if(aml_equal(aml_arg(2), aml_int(7)));
+         {
+             Aml *pkg = aml_package(2);
+             Aml *label = aml_local(2);
+             Aml *ret = aml_local(1);
+
+             aml_append(ifctx1, aml_concatenate(aml_string("PCI Device "),
+                 aml_to_decimalstring(acpi_index, NULL), label));
+
+             aml_append(pkg, zero);
+             aml_append(pkg, aml_string("placeholder"));
+             aml_append(ifctx1, aml_store(pkg, ret));
+             /*
+              * update apci-index to actual value
+              */
+             aml_append(ifctx1, aml_store(acpi_index, aml_index(ret, zero)));
+             /*
Quoted 8 lines

hw/acpi/pci.c bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id) · line 138

Quoted 4 lines
hw/acpi/pci.c +4 -0
+              * update device label to actual value
+              */
+             aml_append(ifctx1, aml_store(label, aml_index(ret, aml_int(1))));
+             aml_append(ifctx1, aml_return(ret));  

This code needs more comments to explain what it's doing. E.g. what is device label?
Spec seems to say the string is optional. Is it relevant somehow?

potentially label may be used (with custom rules) to for interface naming or as human readable PCI device description.
We probably can drop it, or provide and additional PCIDevice option to let user specify it on CLI along with acpi-index.

Quoted 2 lines

spec also says string must be unique. Do we need to enforce that?

I think we should enforce it (Though I intentionally dropped it from RFC)

Quoted 63 lines
Quoted 62 lines
hw/acpi/pci.c +7 -0
+         }
+         aml_append(elsectx, ifctx1);
+         aml_append(ifctx, elsectx);
+    }
+    aml_append(method, ifctx);
+    return method;
+}
hw/i386/acpi-build.c +18 -3
index 27d2958e25..447ad39c35 100644
@@ -385,6 +385,13 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
                     aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
                 );
                 aml_append(dev, method);
+                method = aml_method("_DSM", 4, AML_SERIALIZED);
+                aml_append(method,
+                    aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1),
+                                         aml_arg(2), aml_arg(3),
+                                         aml_name("BSEL"), aml_name("_SUN")))
+                );
+                aml_append(dev, method);
                 aml_append(parent_scope, dev);
 
                 build_append_pcihp_notify_entry(notify_method, slot);
@@ -412,6 +419,14 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
         dev = aml_device("S%.02X", PCI_DEVFN(slot, 0));
         aml_append(dev, aml_name_decl("_ADR", aml_int(slot << 16)));
 
+        aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
+        method = aml_method("_DSM", 4, AML_SERIALIZED);
+        aml_append(method,
+           aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1), aml_arg(2),
+                                aml_arg(3), aml_name("BSEL"), aml_name("_SUN")))
+        );
+        aml_append(dev, method);
+
         if (pc->class_id == PCI_CLASS_DISPLAY_VGA) {
             /* add VGA specific AML methods */
             int s3d;
@@ -434,9 +449,7 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
             aml_append(method, aml_return(aml_int(s3d)));
             aml_append(dev, method);
         } else if (hotplug_enabled_dev) {
-            /* add _SUN/_EJ0 to make slot hotpluggable  */
-            aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
-
+            /* add _EJ0 to make slot hotpluggable  */
             method = aml_method("_EJ0", 1, AML_NOTSERIALIZED);
             aml_append(method,
                 aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
@@ -1142,6 +1155,8 @@ static void build_piix4_pci_hotplug(Aml *table)
     aml_append(method, aml_return(aml_local(0)));
     aml_append(scope, method);
 
+    aml_append(scope, aml_pci_device_dsm());
+
     aml_append(table, scope);
 }
 
Signature · 1 line
2.27.0  
#7 lore raw

Re: [RFC 4/5] pci: acpi: add _DSM method to PCI devices

In reply to Igor Mammedov
Quoted 46 lines · Igor Mammedov

Implement _DSM according to:
PCI Firmware Specification 3.1
4.6.7. DSM for Naming a PCI or PCI Express Device Under
Operating Systems
and wire it up to cold and hot-plugged PCI devices. Feature depends on ACPI hotplug being enabled (as that provides PCI devices descriptions in ACPI and MMIO registers that are reused to fetch acpi-index).

acpi-index should work for
- cold plugged NICs:

      $QEMU -evice e1000,acpi-index=100
         => 'eno100'
  - hot-plugged
      (monitor) device_add e1000,acpi-index=200,id=remove_me
         => 'eno200'
  - replugged
      (monitor) device_del remove_me
      (monitor) device_add e1000,acpi-index=1
         => 'eno1'

Windows also sees index under "PCI Label Id" field in properties dialog but otherwise it doesn't seem to have any effect.

---
 include/hw/acpi/pci.h |  1 +
 hw/acpi/pci.c         | 78 +++++++++++++++++++++++++++++++++++++++++++
 hw/i386/acpi-build.c  | 21 ++++++++++--
 3 files changed, 97 insertions(+), 3 deletions(-)
include/hw/acpi/pci.h +1 -0
index bf2a3ed0ba..5e1eb2a96a 100644
@@ -34,4 +34,5 @@ typedef struct AcpiMcfgInfo {
 } AcpiMcfgInfo;
 
 void build_mcfg(GArray *table_data, BIOSLinker *linker, AcpiMcfgInfo *info);
+Aml *aml_pci_device_dsm(void);
 #endif
hw/acpi/pci.c +0 -0
index 07d5101d83..6d49d515d3 100644

hw/acpi/pci.c bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id) · line 71

Quoted 7 lines
hw/acpi/pci.c +4 -0
@@ -65,3 +65,81 @@ bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id)
      AcpiPciHpState *s = opaque;
      return s->acpi_index;
 }
+
+Aml *aml_pci_device_dsm(void)
+{
+    Aml *method, *UUID, *ifctx, *ifctx1, *ifctx2, *ifctx3, *elsectx;

s/UUID/uuid/ I think ...

And can we move ifctx things to the correct scope?

Quoted 129 lines
hw/acpi/pci.c +74 -0
+    Aml *acpi_index = aml_local(0);
+    Aml *zero = aml_int(0);
+    Aml *bnum = aml_arg(4);
+    Aml *sun = aml_arg(5);
+
+    method = aml_method("PDSM", 6, AML_SERIALIZED);
+
+    /*
+     * PCI Firmware Specification 3.1
+     * 4.6.  _DSM Definitions for PCI
+     */
+    UUID = aml_touuid("E5C937D0-3553-4D7A-9117-EA4D19C3434D");
+    ifctx = aml_if(aml_equal(aml_arg(0), UUID));
+    {
+        aml_append(ifctx, aml_store(aml_call2("AIDX", bnum, sun), acpi_index));
+        ifctx1 = aml_if(aml_equal(aml_arg(2), zero));
+        {
+            uint8_t byte_list[1];
+
+            ifctx2 = aml_if(aml_equal(aml_arg(1), aml_int(2)));
+            {
+                /*
+                 * advertise function 7 if device has acpi-index
+                 */
+                ifctx3 = aml_if(aml_lnot(aml_equal(acpi_index, zero)));
+                {
+                    byte_list[0] =
+                        1 /* have supported functions */ |
+                        1 << 7 /* support for function 7 */
+                    ;
+                    aml_append(ifctx3, aml_return(aml_buffer(1, byte_list)));
+                }
+                aml_append(ifctx2, ifctx3);
+             }
+             aml_append(ifctx1, ifctx2);
+
+             byte_list[0] = 0; /* nothing supported */
+             aml_append(ifctx1, aml_return(aml_buffer(1, byte_list)));
+         }
+         aml_append(ifctx, ifctx1);
+         elsectx = aml_else();
+         /*
+          * PCI Firmware Specification 3.1
+          * 4.6.7. _DSM for Naming a PCI or PCI Express Device Under
+          *        Operating Systems
+          */
+         ifctx1 = aml_if(aml_equal(aml_arg(2), aml_int(7)));
+         {
+             Aml *pkg = aml_package(2);
+             Aml *label = aml_local(2);
+             Aml *ret = aml_local(1);
+
+             aml_append(ifctx1, aml_concatenate(aml_string("PCI Device "),
+                 aml_to_decimalstring(acpi_index, NULL), label));
+
+             aml_append(pkg, zero);
+             aml_append(pkg, aml_string("placeholder"));
+             aml_append(ifctx1, aml_store(pkg, ret));
+             /*
+              * update apci-index to actual value
+              */
+             aml_append(ifctx1, aml_store(acpi_index, aml_index(ret, zero)));
+             /*
+              * update device label to actual value
+              */
+             aml_append(ifctx1, aml_store(label, aml_index(ret, aml_int(1))));
+             aml_append(ifctx1, aml_return(ret));
+         }
+         aml_append(elsectx, ifctx1);
+         aml_append(ifctx, elsectx);
+    }
+    aml_append(method, ifctx);
+    return method;
+}
hw/i386/acpi-build.c +18 -3
index 27d2958e25..447ad39c35 100644
@@ -385,6 +385,13 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
                     aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
                 );
                 aml_append(dev, method);
+                method = aml_method("_DSM", 4, AML_SERIALIZED);
+                aml_append(method,
+                    aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1),
+                                         aml_arg(2), aml_arg(3),
+                                         aml_name("BSEL"), aml_name("_SUN")))
+                );
+                aml_append(dev, method);
                 aml_append(parent_scope, dev);
 
                 build_append_pcihp_notify_entry(notify_method, slot);
@@ -412,6 +419,14 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
         dev = aml_device("S%.02X", PCI_DEVFN(slot, 0));
         aml_append(dev, aml_name_decl("_ADR", aml_int(slot << 16)));
 
+        aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
+        method = aml_method("_DSM", 4, AML_SERIALIZED);
+        aml_append(method,
+           aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1), aml_arg(2),
+                                aml_arg(3), aml_name("BSEL"), aml_name("_SUN")))
+        );
+        aml_append(dev, method);
+
         if (pc->class_id == PCI_CLASS_DISPLAY_VGA) {
             /* add VGA specific AML methods */
             int s3d;
@@ -434,9 +449,7 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
             aml_append(method, aml_return(aml_int(s3d)));
             aml_append(dev, method);
         } else if (hotplug_enabled_dev) {
-            /* add _SUN/_EJ0 to make slot hotpluggable  */
-            aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
-
+            /* add _EJ0 to make slot hotpluggable  */
             method = aml_method("_EJ0", 1, AML_NOTSERIALIZED);
             aml_append(method,
                 aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
@@ -1142,6 +1155,8 @@ static void build_piix4_pci_hotplug(Aml *table)
     aml_append(method, aml_return(aml_local(0)));
     aml_append(scope, method);
 
+    aml_append(scope, aml_pci_device_dsm());
+
     aml_append(table, scope);
 }
 
Signature · 1 line
2.27.0
#8 lore raw

Re: [RFC 4/5] pci: acpi: add _DSM method to PCI devices

Quoted 57 lines · On Tue, 26 Jan 2021 06:16:50 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 46 lines · Igor Mammedov

Implement _DSM according to:
PCI Firmware Specification 3.1
4.6.7. DSM for Naming a PCI or PCI Express Device Under
Operating Systems
and wire it up to cold and hot-plugged PCI devices. Feature depends on ACPI hotplug being enabled (as that provides PCI devices descriptions in ACPI and MMIO registers that are reused to fetch acpi-index).

acpi-index should work for
- cold plugged NICs:

      $QEMU -evice e1000,acpi-index=100  
         => 'eno100'  
  - hot-plugged
      (monitor) device_add e1000,acpi-index=200,id=remove_me  
         => 'eno200'  
  - replugged
      (monitor) device_del remove_me
      (monitor) device_add e1000,acpi-index=1  
         => 'eno1'  

Windows also sees index under "PCI Label Id" field in properties dialog but otherwise it doesn't seem to have any effect.

---
 include/hw/acpi/pci.h |  1 +
 hw/acpi/pci.c         | 78 +++++++++++++++++++++++++++++++++++++++++++
 hw/i386/acpi-build.c  | 21 ++++++++++--
 3 files changed, 97 insertions(+), 3 deletions(-)
include/hw/acpi/pci.h +1 -0
index bf2a3ed0ba..5e1eb2a96a 100644
@@ -34,4 +34,5 @@ typedef struct AcpiMcfgInfo {
 } AcpiMcfgInfo;
 
 void build_mcfg(GArray *table_data, BIOSLinker *linker, AcpiMcfgInfo *info);
+Aml *aml_pci_device_dsm(void);
 #endif
hw/acpi/pci.c +0 -0
index 07d5101d83..6d49d515d3 100644

hw/acpi/pci.c bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id) · line 71

Quoted 7 lines
hw/acpi/pci.c +4 -0
@@ -65,3 +65,81 @@ bool vmstate_acpi_pcihp_use_acpi_index(void *opaque, int version_id)
      AcpiPciHpState *s = opaque;
      return s->acpi_index;
 }
+
+Aml *aml_pci_device_dsm(void)
+{
+    Aml *method, *UUID, *ifctx, *ifctx1, *ifctx2, *ifctx3, *elsectx;  

s/UUID/uuid/ I think ...

Quoted 1 line

And can we move ifctx things to the correct scope?

I'll try to do it.

Quoted 131 lines
Quoted 130 lines
hw/acpi/pci.c +74 -0
+    Aml *acpi_index = aml_local(0);
+    Aml *zero = aml_int(0);
+    Aml *bnum = aml_arg(4);
+    Aml *sun = aml_arg(5);
+
+    method = aml_method("PDSM", 6, AML_SERIALIZED);
+
+    /*
+     * PCI Firmware Specification 3.1
+     * 4.6.  _DSM Definitions for PCI
+     */
+    UUID = aml_touuid("E5C937D0-3553-4D7A-9117-EA4D19C3434D");
+    ifctx = aml_if(aml_equal(aml_arg(0), UUID));
+    {
+        aml_append(ifctx, aml_store(aml_call2("AIDX", bnum, sun), acpi_index));
+        ifctx1 = aml_if(aml_equal(aml_arg(2), zero));
+        {
+            uint8_t byte_list[1];
+
+            ifctx2 = aml_if(aml_equal(aml_arg(1), aml_int(2)));
+            {
+                /*
+                 * advertise function 7 if device has acpi-index
+                 */
+                ifctx3 = aml_if(aml_lnot(aml_equal(acpi_index, zero)));
+                {
+                    byte_list[0] =
+                        1 /* have supported functions */ |
+                        1 << 7 /* support for function 7 */
+                    ;
+                    aml_append(ifctx3, aml_return(aml_buffer(1, byte_list)));
+                }
+                aml_append(ifctx2, ifctx3);
+             }
+             aml_append(ifctx1, ifctx2);
+
+             byte_list[0] = 0; /* nothing supported */
+             aml_append(ifctx1, aml_return(aml_buffer(1, byte_list)));
+         }
+         aml_append(ifctx, ifctx1);
+         elsectx = aml_else();
+         /*
+          * PCI Firmware Specification 3.1
+          * 4.6.7. _DSM for Naming a PCI or PCI Express Device Under
+          *        Operating Systems
+          */
+         ifctx1 = aml_if(aml_equal(aml_arg(2), aml_int(7)));
+         {
+             Aml *pkg = aml_package(2);
+             Aml *label = aml_local(2);
+             Aml *ret = aml_local(1);
+
+             aml_append(ifctx1, aml_concatenate(aml_string("PCI Device "),
+                 aml_to_decimalstring(acpi_index, NULL), label));
+
+             aml_append(pkg, zero);
+             aml_append(pkg, aml_string("placeholder"));
+             aml_append(ifctx1, aml_store(pkg, ret));
+             /*
+              * update apci-index to actual value
+              */
+             aml_append(ifctx1, aml_store(acpi_index, aml_index(ret, zero)));
+             /*
+              * update device label to actual value
+              */
+             aml_append(ifctx1, aml_store(label, aml_index(ret, aml_int(1))));
+             aml_append(ifctx1, aml_return(ret));
+         }
+         aml_append(elsectx, ifctx1);
+         aml_append(ifctx, elsectx);
+    }
+    aml_append(method, ifctx);
+    return method;
+}
hw/i386/acpi-build.c +18 -3
index 27d2958e25..447ad39c35 100644
@@ -385,6 +385,13 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
                     aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
                 );
                 aml_append(dev, method);
+                method = aml_method("_DSM", 4, AML_SERIALIZED);
+                aml_append(method,
+                    aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1),
+                                         aml_arg(2), aml_arg(3),
+                                         aml_name("BSEL"), aml_name("_SUN")))
+                );
+                aml_append(dev, method);
                 aml_append(parent_scope, dev);
 
                 build_append_pcihp_notify_entry(notify_method, slot);
@@ -412,6 +419,14 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
         dev = aml_device("S%.02X", PCI_DEVFN(slot, 0));
         aml_append(dev, aml_name_decl("_ADR", aml_int(slot << 16)));
 
+        aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
+        method = aml_method("_DSM", 4, AML_SERIALIZED);
+        aml_append(method,
+           aml_return(aml_call6("PDSM", aml_arg(0), aml_arg(1), aml_arg(2),
+                                aml_arg(3), aml_name("BSEL"), aml_name("_SUN")))
+        );
+        aml_append(dev, method);
+
         if (pc->class_id == PCI_CLASS_DISPLAY_VGA) {
             /* add VGA specific AML methods */
             int s3d;
@@ -434,9 +449,7 @@ static void build_append_pci_bus_devices(Aml *parent_scope, PCIBus *bus,
             aml_append(method, aml_return(aml_int(s3d)));
             aml_append(dev, method);
         } else if (hotplug_enabled_dev) {
-            /* add _SUN/_EJ0 to make slot hotpluggable  */
-            aml_append(dev, aml_name_decl("_SUN", aml_int(slot)));
-
+            /* add _EJ0 to make slot hotpluggable  */
             method = aml_method("_EJ0", 1, AML_NOTSERIALIZED);
             aml_append(method,
                 aml_call2("PCEJ", aml_name("BSEL"), aml_name("_SUN"))
@@ -1142,6 +1155,8 @@ static void build_piix4_pci_hotplug(Aml *table)
     aml_append(method, aml_return(aml_local(0)));
     aml_append(scope, method);
 
+    aml_append(scope, aml_pci_device_dsm());
+
     aml_append(table, scope);
 }
 
Signature · 1 line
2.27.0  
#9 lore raw

RFC 5/5 tests: acpi: update expected data files

In reply to Igor Mammedov

Expected change is addition of
AIDX and PDSM methods at PCI0 scope and
_DSM methods (+ _SUN field where it was missing) to each PCI device slot

@@ -277,6 +277,54 @@ DefinitionBlock ("", "DSDT", 1, "BOCHS ", "BXPCDSDT", 0x00000001)

             Release (BLCK)
             Return (Zero)
         }
+
+        Method (AIDX, 2, NotSerialized)
+        {
+            Acquire (BLCK, 0xFFFF)
+            BNUM = Arg0
+            PCIU = (One << Arg1)
+            Local0 = B0EJ /* \_SB_.PCI0.B0EJ */
+            Release (BLCK)
+            Return (Local0)

+ }
+
+ Method (PDSM, 6, Serialized)
+ {
+ If ((Arg0 == ToUUID ("e5c937d0-3553-4d7a-9117-ea4d19c3434d") /* Device Labeling Interface */))
+ {

+                Local0 = AIDX (Arg4, Arg5)
+                If ((Arg2 == Zero))
+                {

+ If ((Arg1 == 0x02))
+ {
+ If ((Local0 != Zero))
+ {
+ Return (Buffer (One)
+ {
+ 0x81 // .
+ })
+ }
+ }
+
+ Return (Buffer (One)
+ {
+ 0x00 // .
+ })

+                }
+                ElseIf ((Arg2 == 0x07))
+                {

+ Concatenate ("PCI Device ", ToDecimalString (Local0), Local2)
+ Local1 = Package (0x02)
+ {
+ Zero,
+ "placeholder"
+ }

+                    Local1 [Zero] = Local0
+                    Local1 [One] = Local2
+                    Return (Local1)
+                }
+            }
+        }
     }

     Scope (_SB)
@@ -883,11 +931,22 @@ DefinitionBlock ("", "DSDT", 1, "BOCHS ", "BXPCDSDT", 0x00000001)
             Device (S00)
             {
                 Name (_ADR, Zero)  // _ADR: Address
+                Name (_SUN, Zero)  // _SUN: Slot User Number
+                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
+                {
+                    Return (PDSM (Arg0, Arg1, Arg2, Arg3, BSEL, _SUN))
+                }
             }
[...]
@@ -912,6 +971,11 @@ DefinitionBlock ("", "DSDT", 1, "BOCHS ", "BXPCDSDT", 0x00000001)
                 {
                     PCEJ (BSEL, _SUN)
                 }

+
+ Method (_DSM, 4, Serialized) // _DSM: Device-Specific Method
+ {
+ Return (PDSM (Arg0, Arg1, Arg2, Arg3, BSEL, _SUN))
+ }

             }

[...]

---
 tests/data/acpi/pc/DSDT           | Bin 5065 -> 6023 bytes
 tests/data/acpi/pc/DSDT.acpihmat  | Bin 6390 -> 7348 bytes
 tests/data/acpi/pc/DSDT.bridge    | Bin 6924 -> 8689 bytes
 tests/data/acpi/pc/DSDT.cphp      | Bin 5529 -> 6487 bytes
 tests/data/acpi/pc/DSDT.dimmpxm   | Bin 6719 -> 7677 bytes
 tests/data/acpi/pc/DSDT.hpbridge  | Bin 5026 -> 5990 bytes
 tests/data/acpi/pc/DSDT.hpbrroot  | Bin 3084 -> 3177 bytes
 tests/data/acpi/pc/DSDT.ipmikcs   | Bin 5137 -> 6095 bytes
 tests/data/acpi/pc/DSDT.memhp     | Bin 6424 -> 7382 bytes
 tests/data/acpi/pc/DSDT.numamem   | Bin 5071 -> 6029 bytes
 tests/data/acpi/pc/DSDT.roothp    | Bin 5261 -> 6324 bytes
 tests/data/acpi/q35/DSDT          | Bin 7801 -> 7863 bytes
 tests/data/acpi/q35/DSDT.acpihmat | Bin 9126 -> 9188 bytes
 tests/data/acpi/q35/DSDT.bridge   | Bin 7819 -> 7911 bytes
 tests/data/acpi/q35/DSDT.cphp     | Bin 8265 -> 8327 bytes
 tests/data/acpi/q35/DSDT.dimmpxm  | Bin 9455 -> 9517 bytes
 tests/data/acpi/q35/DSDT.ipmibt   | Bin 7876 -> 7938 bytes
 tests/data/acpi/q35/DSDT.memhp    | Bin 9160 -> 9222 bytes
 tests/data/acpi/q35/DSDT.mmio64   | Bin 8932 -> 9024 bytes
 tests/data/acpi/q35/DSDT.numamem  | Bin 7807 -> 7869 bytes
 tests/data/acpi/q35/DSDT.tis      | Bin 8407 -> 8468 bytes
 21 files changed, 0 insertions(+), 0 deletions(-)
tests/data/acpi/pc/DSDT binary +0 -0
index f6173df1d598767a79aa34ad7585ad7d45c5d4f3..b9afc0a458d77b7346fca9cd2d8797a4d4ed8302 100644
GIT binary patch
tests/data/acpi/pc/DSDT.acpihmat binary +0 -0
index 67f3f7249eaaa9404ebf0f2d0a324b8c8e3bd445..806bcbee30dc38117fbb37f433bfbcec171ce34b 100644
GIT binary patch
tests/data/acpi/pc/DSDT.bridge binary +0 -0
index 643390f4c4138b37fc481656d3f555d0eeedcb02..fd0c98e8adacada8252ed74f887773feb275bfc8 100644
GIT binary patch
tests/data/acpi/pc/DSDT.cphp binary +0 -0
index 1ddcf7d8812f5d8d4d38fe7e7b35fd5885806046..8f453a59f297be7a2afaa8b8d9dbd9a45dc20fdd 100644
GIT binary patch
tests/data/acpi/pc/DSDT.dimmpxm binary +0 -0
index c44385cc01879324738ffb7f997b8cdd762cbf97..0dc9455c519a0a9116ba1f34026a9bdbabf6575e 100644
GIT binary patch
tests/data/acpi/pc/DSDT.hpbridge binary +0 -0
index 4ecf1eb13bf49499f729b53a6d0114672a76e28d..78f97cbda1c17f3e09cc4332e87d473fe99646b2 100644
GIT binary patch
tests/data/acpi/pc/DSDT.hpbrroot binary +0 -0
index a3046226ec1dcb234b726029b3790dfedb3b9221..4f930ed9644a4018da96ef3e5dbedcb89ef19ee0 100644
GIT binary patch
tests/data/acpi/pc/DSDT.ipmikcs binary +0 -0
index f1638c5d079a9442c09390426a913010df6efd8d..3231292e1ca1be6c6a5ac0297a886ab23a52ebb3 100644
GIT binary patch
tests/data/acpi/pc/DSDT.memhp binary +0 -0
index 4c19e45e66918c61674785c99e4474e58866f125..df83eaf0cf87e29ae7a6bf6f6a2dac26015e8689 100644
GIT binary patch
tests/data/acpi/pc/DSDT.numamem binary +0 -0
index 40cfd933259af05ac2aee07fca32f22122255211..dbb6b182fbd3b45a6ca38d8492892867f9000a4e 100644
GIT binary patch
tests/data/acpi/pc/DSDT.roothp binary +0 -0
index 078fc8031b479cc77b6527a2b7b4bd576b6e6028..fccb574c38ef112aa5e415ddf9a049d12b1c142c 100644
GIT binary patch
tests/data/acpi/q35/DSDT binary +0 -0
index d25cd7072932886d6967f4023faac1e1fa6e836c..9506b60117147fd8ebe30fc3bf9f124773e6a3c5 100644
GIT binary patch
tests/data/acpi/q35/DSDT.acpihmat binary +0 -0
index 722e06af83abcde203a2b96a8ec81fd3bab9fc98..ae51a89dc056a13f50bc515e05207e875c369a51 100644
GIT binary patch
tests/data/acpi/q35/DSDT.bridge binary +0 -0
index 06bac139d668ddfc7914e258b471a303c9dbd192..0cf9f067d25ae5d596f236cda3b1a495a527249b 100644
GIT binary patch
tests/data/acpi/q35/DSDT.cphp binary +0 -0
index 2b933ac482e6883efccbd7d6c96089602f2c0b4d..3d17ac851fe4a988006f49deccdffd426581b6a7 100644
GIT binary patch
tests/data/acpi/q35/DSDT.dimmpxm binary +0 -0
index bd8f8305b028ef20f9b6d1a0c69ac428d027e3d1..acde6cb0a22f031c0f8743e5cceb3b707e3aab61 100644
GIT binary patch
tests/data/acpi/q35/DSDT.ipmibt binary +0 -0
index a8f868e23c25688ab1c0371016c071f23e9d732f..77dac3feb29b6769b0ccb6b29a16538a4ac31879 100644
GIT binary patch
tests/data/acpi/q35/DSDT.memhp binary +0 -0
index 9a802e4c67022386442976d5cb997ea3fc57b58f..5e3c1a4b1a34a63edcf9394455dbbaa86a24b044 100644
GIT binary patch
tests/data/acpi/q35/DSDT.mmio64 binary +0 -0
index 948c2dc7264c31932b490ca00691a7c4d9aefdb0..09faed2cd326d591b6c33203dd9c0504808a0952 100644
GIT binary patch
tests/data/acpi/q35/DSDT.numamem binary +0 -0
index 44ec1b0af400da6d298284aa959aa38add7e6dd5..642a7b39cc8bb174daa098c4a377df8cd5938289 100644
GIT binary patch
tests/data/acpi/q35/DSDT.tis binary +0 -0
index 30da3ec27958881801dacc954a343321ba26a2ae..bef6c030b8fa08abf538bfb4ac244d1567c83c2e 100644
GIT binary patch
Signature · 1 line
2.27.0
#10 lore raw
In reply to Igor Mammedov
Quoted 5 lines · Igor Mammedov

Series implements support for 'onboard' naming scheme for network interfaces (1), which is based on PCI firmware spec and lets user to explicitly specify index that will be used by guest to name network interface, ex:

Quoted 8 lines
    -device e1000,acpi-index=33

should make guest rename NIC name to 'eno33' where 'eno' is default prefix for this scheme.

Hope is that it will allow to simplify launching VMs from template disk images with different set VM configurations without need to reconfigure guest network intrfaces or risk of loosing network connectivity.

Questions:
the spec says:
Assignment of specific device names to multi-function devices installed in expansion slots, and/or PCI or PCI Express devices that are hot-added to expansion slots in operating system- environment would be handled in operating system-specific manner, and is not specified via this specification.

Accordingly, link below says:
" Names incorporating Firmware/BIOS provided index numbers for on-board devices (example: eno1)"

to what extend does guest assume the index is for on-board devices? it seems things work for fedora but how confident are we that this will keep working.

Further, code seems to only look at the slot level. According to this, and according to the spec, this does not work with multifunction devices, does it?

The link you supplied lists another option:
"Names incorporating Firmware/BIOS provided PCI Express hotplug slot index numbers (example: ens1)" these are under management control already ...

Also if we ask users to supply the property on the slot then it seems that the property can be baked into ACPI when it's created instead of being loaded from host - we can avoid adding new registers, this seems preferable. Could someone from management side chime in on whether that is sufficient?

More questions:

does all this affect windows guests at all?

where does the "acpi index" terminology come from? the pci firmware spec talks about "instance number", right?

Quoted 50 lines

For more detailed description/examples see patches [3-4/5]

1)
https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/

Git repo for testing:
https://github.com/imammedo/qemu/branches acpi-index-rfc

Igor Mammedov (5):
acpi: add aml_to_decimalstring() and aml_call6() helpers
tests: acpi: temporary whitelist DSDT changes
pci: introduce apci-index property for PCI device
pci: acpi: add _DSM method to PCI devices
tests: acpi: update expected data files

 include/hw/acpi/aml-build.h                 |   3 +
 include/hw/acpi/pci.h                       |   1 +
 include/hw/acpi/pcihp.h                     |   7 +-
 include/hw/pci/pci.h                        |   1 +
 tests/qtest/bios-tables-test-allowed-diff.h |  21 +++++
 hw/acpi/aml-build.c                         |  28 +++++++
 hw/acpi/pci.c                               |  84 ++++++++++++++++++++
 hw/acpi/pcihp.c                             |  25 +++++-
 hw/i386/acpi-build.c                        |  31 +++++++-
 hw/pci/pci.c                                |   1 +
 tests/data/acpi/pc/DSDT                     | Bin 5065 -> 6023 bytes
 tests/data/acpi/pc/DSDT.acpihmat            | Bin 6390 -> 7348 bytes
 tests/data/acpi/pc/DSDT.bridge              | Bin 6924 -> 8689 bytes
 tests/data/acpi/pc/DSDT.cphp                | Bin 5529 -> 6487 bytes
 tests/data/acpi/pc/DSDT.dimmpxm             | Bin 6719 -> 7677 bytes
 tests/data/acpi/pc/DSDT.hpbridge            | Bin 5026 -> 5990 bytes
 tests/data/acpi/pc/DSDT.hpbrroot            | Bin 3084 -> 3177 bytes
 tests/data/acpi/pc/DSDT.ipmikcs             | Bin 5137 -> 6095 bytes
 tests/data/acpi/pc/DSDT.memhp               | Bin 6424 -> 7382 bytes
 tests/data/acpi/pc/DSDT.numamem             | Bin 5071 -> 6029 bytes
 tests/data/acpi/pc/DSDT.roothp              | Bin 5261 -> 6324 bytes
 tests/data/acpi/q35/DSDT                    | Bin 7801 -> 7863 bytes
 tests/data/acpi/q35/DSDT.acpihmat           | Bin 9126 -> 9188 bytes
 tests/data/acpi/q35/DSDT.bridge             | Bin 7819 -> 7911 bytes
 tests/data/acpi/q35/DSDT.cphp               | Bin 8265 -> 8327 bytes
 tests/data/acpi/q35/DSDT.dimmpxm            | Bin 9455 -> 9517 bytes
 tests/data/acpi/q35/DSDT.ipmibt             | Bin 7876 -> 7938 bytes
 tests/data/acpi/q35/DSDT.memhp              | Bin 9160 -> 9222 bytes
 tests/data/acpi/q35/DSDT.mmio64             | Bin 8932 -> 9024 bytes
 tests/data/acpi/q35/DSDT.numamem            | Bin 7807 -> 7869 bytes
 tests/data/acpi/q35/DSDT.tis                | Bin 8407 -> 8468 bytes
 31 files changed, 197 insertions(+), 5 deletions(-)
Signature · 1 line
2.27.0
#11 lore raw
Quoted 22 lines · On Wed, 13 Jan 2021 07:09:56 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 5 lines · Igor Mammedov

Series implements support for 'onboard' naming scheme for network interfaces (1), which is based on PCI firmware spec and lets user to explicitly specify index that will be used by guest to name network interface, ex:

Quoted 8 lines
    -device e1000,acpi-index=33

should make guest rename NIC name to 'eno33' where 'eno' is default prefix for this scheme.

Hope is that it will allow to simplify launching VMs from template disk images with different set VM configurations without need to reconfigure guest network intrfaces or risk of loosing network connectivity.

Questions:
the spec says:
Assignment of specific device names to multi-function devices installed in expansion slots, and/or PCI or PCI Express devices that are hot-added to expansion slots in operating system- environment would be handled in operating system-specific manner, and is not specified via this specification.

Quoted 6 lines

Accordingly, link below says:
" Names incorporating Firmware/BIOS provided index numbers for on-board devices (example: eno1)"

to what extend does guest assume the index is for on-board devices? it seems things work for fedora but how confident are we that this will keep working.

code itself is not limiting it to onboard devices in any way. (I can only speculate here, reason for calling it onboard is that on real hardware ACPI is mostly static tables and firmware provides an option to set index for only built-in NICs). Technically there is no reason to call in onboad though.

I'd believe it should work with any distribution that uses recent enough systemd/udev (released starting from 2003).

Quoted 3 lines

Further, code seems to only look at the slot level. According to this, and according to the spec, this does not work with multifunction devices, does it?

we probably should disable it for multifunction devices, any suggestions how to detect those in QEMU?

Quoted 3 lines

The link you supplied lists another option:
"Names incorporating Firmware/BIOS provided PCI Express hotplug slot index numbers (example: ens1)" these are under management control already ...

with it interface name continues to depend on PCI topology (and theoretically limited to PCI expess). That's becomes harder to consume as complexity grows (i.e. mgmt needs to keep NIC in the same place for which guest image was configured for). acpi-index doesn't impose such limitation.

In case of 1 NIC, it could be moved anywhere within PCI hierarchy and guest doesn't have to be reconfigured to account for new interface name (i.e without loosing network connectivity - that's the actual issue coming from upper layers that made me look into acpi-index approach).

In another words acpi index is easier to consume for users above libvirt and frees mgmt hands in a way it could distribute PCI devices. Even better would be if guest image could carry index as metadata

Quoted 5 lines

Also if we ask users to supply the property on the slot then it seems that the property can be baked into ACPI when it's created instead of being loaded from host - we can avoid adding new registers, this seems preferable. Could someone from management side chime in on whether that is sufficient?

I did consider it (it would be simpler, but not much), however unless we disable PCI hotplug for affected slots it won't work. (unplug device from such slot and plug another in that place will still return boot time index. That's why I ended up with hotplug variant.

I chose abusing existing PCI hotplug registers for it, but we can use a new range if that's preferred.

Quoted 3 lines

More questions:

does all this affect windows guests at all?

I don't know (spec shows examples that reminded me about NIC naming which Windows use(s|d)). But I won't bet on it.
If I recall correctly, for e1000 NIC, it didn't made any difference in naming (pre-existing guest image).

Quoted 2 lines

where does the "acpi index" terminology come from? the pci firmware spec talks about "instance number", right?

it comes from linux kernel (that's how it's named in sysfs) and systemd/udev uses it. So I tried to avoid making up another one.

Quoted 52 lines
Quoted 50 lines

For more detailed description/examples see patches [3-4/5]

1)
https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/

Git repo for testing:
https://github.com/imammedo/qemu/branches acpi-index-rfc

Igor Mammedov (5):
acpi: add aml_to_decimalstring() and aml_call6() helpers
tests: acpi: temporary whitelist DSDT changes
pci: introduce apci-index property for PCI device
pci: acpi: add _DSM method to PCI devices
tests: acpi: update expected data files

 include/hw/acpi/aml-build.h                 |   3 +
 include/hw/acpi/pci.h                       |   1 +
 include/hw/acpi/pcihp.h                     |   7 +-
 include/hw/pci/pci.h                        |   1 +
 tests/qtest/bios-tables-test-allowed-diff.h |  21 +++++
 hw/acpi/aml-build.c                         |  28 +++++++
 hw/acpi/pci.c                               |  84 ++++++++++++++++++++
 hw/acpi/pcihp.c                             |  25 +++++-
 hw/i386/acpi-build.c                        |  31 +++++++-
 hw/pci/pci.c                                |   1 +
 tests/data/acpi/pc/DSDT                     | Bin 5065 -> 6023 bytes
 tests/data/acpi/pc/DSDT.acpihmat            | Bin 6390 -> 7348 bytes
 tests/data/acpi/pc/DSDT.bridge              | Bin 6924 -> 8689 bytes
 tests/data/acpi/pc/DSDT.cphp                | Bin 5529 -> 6487 bytes
 tests/data/acpi/pc/DSDT.dimmpxm             | Bin 6719 -> 7677 bytes
 tests/data/acpi/pc/DSDT.hpbridge            | Bin 5026 -> 5990 bytes
 tests/data/acpi/pc/DSDT.hpbrroot            | Bin 3084 -> 3177 bytes
 tests/data/acpi/pc/DSDT.ipmikcs             | Bin 5137 -> 6095 bytes
 tests/data/acpi/pc/DSDT.memhp               | Bin 6424 -> 7382 bytes
 tests/data/acpi/pc/DSDT.numamem             | Bin 5071 -> 6029 bytes
 tests/data/acpi/pc/DSDT.roothp              | Bin 5261 -> 6324 bytes
 tests/data/acpi/q35/DSDT                    | Bin 7801 -> 7863 bytes
 tests/data/acpi/q35/DSDT.acpihmat           | Bin 9126 -> 9188 bytes
 tests/data/acpi/q35/DSDT.bridge             | Bin 7819 -> 7911 bytes
 tests/data/acpi/q35/DSDT.cphp               | Bin 8265 -> 8327 bytes
 tests/data/acpi/q35/DSDT.dimmpxm            | Bin 9455 -> 9517 bytes
 tests/data/acpi/q35/DSDT.ipmibt             | Bin 7876 -> 7938 bytes
 tests/data/acpi/q35/DSDT.memhp              | Bin 9160 -> 9222 bytes
 tests/data/acpi/q35/DSDT.mmio64             | Bin 8932 -> 9024 bytes
 tests/data/acpi/q35/DSDT.numamem            | Bin 7807 -> 7869 bytes
 tests/data/acpi/q35/DSDT.tis                | Bin 8407 -> 8468 bytes
 31 files changed, 197 insertions(+), 5 deletions(-)
Signature · 1 line
2.27.0  
#12 lore raw
Quoted 32 lines · Igor Mammedov
Quoted 22 lines · On Wed, 13 Jan 2021 07:09:56 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 5 lines · Igor Mammedov

Series implements support for 'onboard' naming scheme for network interfaces (1), which is based on PCI firmware spec and lets user to explicitly specify index that will be used by guest to name network interface, ex:

Quoted 8 lines
    -device e1000,acpi-index=33

should make guest rename NIC name to 'eno33' where 'eno' is default prefix for this scheme.

Hope is that it will allow to simplify launching VMs from template disk images with different set VM configurations without need to reconfigure guest network intrfaces or risk of loosing network connectivity.

Questions:
the spec says:
Assignment of specific device names to multi-function devices installed in expansion slots, and/or PCI or PCI Express devices that are hot-added to expansion slots in operating system- environment would be handled in operating system-specific manner, and is not specified via this specification.

Quoted 6 lines

Accordingly, link below says:
" Names incorporating Firmware/BIOS provided index numbers for on-board devices (example: eno1)"

to what extend does guest assume the index is for on-board devices? it seems things work for fedora but how confident are we that this will keep working.

Quoted 5 lines

code itself is not limiting it to onboard devices in any way. (I can only speculate here, reason for calling it onboard is that on real hardware ACPI is mostly static tables and firmware provides an option to set index for only built-in NICs). Technically there is no reason to call in onboad though.

Well there's an smbios thing that I believe says onboard in the spec. Come to think of it that one actually applies to multifunction devices too.

Quoted 10 lines

I'd believe it should work with any distribution that uses recent enough systemd/udev (released starting from 2003).

Quoted 3 lines

Further, code seems to only look at the slot level. According to this, and according to the spec, this does not work with multifunction devices, does it?

we probably should disable it for multifunction devices, any suggestions how to detect those in QEMU?

QEMU_PCI_CAP_MULTIFUNCTION_BITNR ?

Enforcing this will help but can we make this self-documenting somehow?

Quoted 10 lines
Quoted 4 lines

The link you supplied lists another option:
"Names incorporating Firmware/BIOS provided PCI Express hotplug slot index numbers (example: ens1)" these are under management control already ...

with it interface name continues to depend on PCI topology (and theoretically limited to PCI expess). That's becomes harder to consume as complexity grows (i.e. mgmt needs to keep NIC in the same place for which guest image was configured for). acpi-index doesn't impose such limitation.

Quoted 4 lines

In case of 1 NIC, it could be moved anywhere within PCI hierarchy and guest doesn't have to be reconfigured to account for new interface name (i.e without loosing network connectivity - that's the actual issue coming from upper layers that made me look into acpi-index approach).

Could you describe the issue in a bit more detail in the commit log pls?

Quoted 10 lines

In another words acpi index is easier to consume for users above libvirt and frees mgmt hands in a way it could distribute PCI devices. Even better would be if guest image could carry index as metadata

Quoted 5 lines

Also if we ask users to supply the property on the slot then it seems that the property can be baked into ACPI when it's created instead of being loaded from host - we can avoid adding new registers, this seems preferable. Could someone from management side chime in on whether that is sufficient?

Quoted 3 lines

I did consider it (it would be simpler, but not much), however unless we disable PCI hotplug for affected slots it won't work. (unplug device from such slot and plug another in that place will still return boot time index.

Well it's interesting that you mention it. Let's say you remove a device then plug in a different index immediately ... Guest gets a wrong name temporarily, right?
Should we worry about that?

Quoted 4 lines

That's why I ended up with hotplug variant.

I chose abusing existing PCI hotplug registers for it, but we can use a new range if that's preferred.

Would it work for multi-function devices though?
If yes that's a big advantage ...

Quoted 9 lines
Quoted 3 lines

More questions:

does all this affect windows guests at all?

I don't know (spec shows examples that reminded me about NIC naming which Windows use(s|d)). But I won't bet on it.
If I recall correctly, for e1000 NIC, it didn't made any difference in naming (pre-existing guest image).

Quoted 5 lines
Quoted 2 lines

where does the "acpi index" terminology come from? the pci firmware spec talks about "instance number", right?

it comes from linux kernel (that's how it's named in sysfs) and systemd/udev uses it. So I tried to avoid making up another one.

OTOH that's guest specific ... and does not seem to imply the limitations unless you look at the code ...

Quoted 53 lines
Quoted 53 lines
Quoted 51 lines

For more detailed description/examples see patches [3-4/5]

1)
https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/

Git repo for testing:
https://github.com/imammedo/qemu/branches acpi-index-rfc

Igor Mammedov (5):
acpi: add aml_to_decimalstring() and aml_call6() helpers
tests: acpi: temporary whitelist DSDT changes
pci: introduce apci-index property for PCI device
pci: acpi: add _DSM method to PCI devices
tests: acpi: update expected data files

 include/hw/acpi/aml-build.h                 |   3 +
 include/hw/acpi/pci.h                       |   1 +
 include/hw/acpi/pcihp.h                     |   7 +-
 include/hw/pci/pci.h                        |   1 +
 tests/qtest/bios-tables-test-allowed-diff.h |  21 +++++
 hw/acpi/aml-build.c                         |  28 +++++++
 hw/acpi/pci.c                               |  84 ++++++++++++++++++++
 hw/acpi/pcihp.c                             |  25 +++++-
 hw/i386/acpi-build.c                        |  31 +++++++-
 hw/pci/pci.c                                |   1 +
 tests/data/acpi/pc/DSDT                     | Bin 5065 -> 6023 bytes
 tests/data/acpi/pc/DSDT.acpihmat            | Bin 6390 -> 7348 bytes
 tests/data/acpi/pc/DSDT.bridge              | Bin 6924 -> 8689 bytes
 tests/data/acpi/pc/DSDT.cphp                | Bin 5529 -> 6487 bytes
 tests/data/acpi/pc/DSDT.dimmpxm             | Bin 6719 -> 7677 bytes
 tests/data/acpi/pc/DSDT.hpbridge            | Bin 5026 -> 5990 bytes
 tests/data/acpi/pc/DSDT.hpbrroot            | Bin 3084 -> 3177 bytes
 tests/data/acpi/pc/DSDT.ipmikcs             | Bin 5137 -> 6095 bytes
 tests/data/acpi/pc/DSDT.memhp               | Bin 6424 -> 7382 bytes
 tests/data/acpi/pc/DSDT.numamem             | Bin 5071 -> 6029 bytes
 tests/data/acpi/pc/DSDT.roothp              | Bin 5261 -> 6324 bytes
 tests/data/acpi/q35/DSDT                    | Bin 7801 -> 7863 bytes
 tests/data/acpi/q35/DSDT.acpihmat           | Bin 9126 -> 9188 bytes
 tests/data/acpi/q35/DSDT.bridge             | Bin 7819 -> 7911 bytes
 tests/data/acpi/q35/DSDT.cphp               | Bin 8265 -> 8327 bytes
 tests/data/acpi/q35/DSDT.dimmpxm            | Bin 9455 -> 9517 bytes
 tests/data/acpi/q35/DSDT.ipmibt             | Bin 7876 -> 7938 bytes
 tests/data/acpi/q35/DSDT.memhp              | Bin 9160 -> 9222 bytes
 tests/data/acpi/q35/DSDT.mmio64             | Bin 8932 -> 9024 bytes
 tests/data/acpi/q35/DSDT.numamem            | Bin 7807 -> 7869 bytes
 tests/data/acpi/q35/DSDT.tis                | Bin 8407 -> 8468 bytes
 31 files changed, 197 insertions(+), 5 deletions(-)
Signature · 1 line
2.27.0  
#13 lore raw
Quoted 39 lines · On Sun, 17 Jan 2021 05:59:18 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 32 lines · Igor Mammedov
Quoted 22 lines · On Wed, 13 Jan 2021 07:09:56 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 5 lines · Igor Mammedov

Series implements support for 'onboard' naming scheme for network interfaces (1), which is based on PCI firmware spec and lets user to explicitly specify index that will be used by guest to name network interface, ex:

Quoted 8 lines
    -device e1000,acpi-index=33

should make guest rename NIC name to 'eno33' where 'eno' is default prefix for this scheme.

Hope is that it will allow to simplify launching VMs from template disk images with different set VM configurations without need to reconfigure guest network intrfaces or risk of loosing network connectivity.

Questions:
the spec says:
Assignment of specific device names to multi-function devices installed in expansion slots, and/or PCI or PCI Express devices that are hot-added to expansion slots in operating system- environment would be handled in operating system-specific manner, and is not specified via this specification.

Quoted 6 lines

Accordingly, link below says:
" Names incorporating Firmware/BIOS provided index numbers for on-board devices (example: eno1)"

to what extend does guest assume the index is for on-board devices? it seems things work for fedora but how confident are we that this will keep working.

Quoted 5 lines

code itself is not limiting it to onboard devices in any way. (I can only speculate here, reason for calling it onboard is that on real hardware ACPI is mostly static tables and firmware provides an option to set index for only built-in NICs). Technically there is no reason to call in onboad though.

Quoted 3 lines

Well there's an smbios thing that I believe says onboard in the spec. Come to think of it that one actually applies to multifunction devices too.

Can you point to the specific place in the spec, pls?

Quoted 10 lines
Quoted 10 lines

I'd believe it should work with any distribution that uses recent enough systemd/udev (released starting from 2003).

Quoted 3 lines

Further, code seems to only look at the slot level. According to this, and according to the spec, this does not work with multifunction devices, does it?

Quoted 6 lines
Quoted 2 lines

we probably should disable it for multifunction devices, any suggestions how to detect those in QEMU?

QEMU_PCI_CAP_MULTIFUNCTION_BITNR ?

Enforcing this will help but can we make this self-documenting somehow?

what do you mean saying "to make it self-documenting somehow"?

Quoted 9 lines
Quoted 9 lines
Quoted 3 lines

The link you supplied lists another option:
"Names incorporating Firmware/BIOS provided PCI Express hotplug slot index numbers (example: ens1)" these are under management control already ...

with it interface name continues to depend on PCI topology (and theoretically limited to PCI expess). That's becomes harder to consume as complexity grows (i.e. mgmt needs to keep NIC in the same place for which guest image was configured for). acpi-index doesn't impose such limitation.

Quoted 7 lines
Quoted 4 lines

In case of 1 NIC, it could be moved anywhere within PCI hierarchy and guest doesn't have to be reconfigured to account for new interface name (i.e without loosing network connectivity - that's the actual issue coming from upper layers that made me look into acpi-index approach).

Could you describe the issue in a bit more detail in the commit log pls?

sure

Quoted 10 lines
Quoted 10 lines

In another words acpi index is easier to consume for users above libvirt and frees mgmt hands in a way it could distribute PCI devices. Even better would be if guest image could carry index as metadata

Quoted 5 lines

Also if we ask users to supply the property on the slot then it seems that the property can be baked into ACPI when it's created instead of being loaded from host - we can avoid adding new registers, this seems preferable. Could someone from management side chime in on whether that is sufficient?

Quoted 8 lines
Quoted 3 lines

I did consider it (it would be simpler, but not much), however unless we disable PCI hotplug for affected slots it won't work. (unplug device from such slot and plug another in that place will still return boot time index.

Well it's interesting that you mention it. Let's say you remove a device then plug in a different index immediately ... Guest gets a wrong name temporarily, right?
Should we worry about that?

guest will get not a wrong name but rather name that user asked for. (i.e. NIC with old index could be moved to other slot and still keep its name)

Quoted 7 lines
Quoted 4 lines

That's why I ended up with hotplug variant.

I chose abusing existing PCI hotplug registers for it, but we can use a new range if that's preferred.

Would it work for multi-function devices though?
If yes that's a big advantage ...

Does ACPI hotplug work with multi-function devices (I thought not)? If not, I'd prefer explicitly disable new feature
for multi-function devices (the code is heavily based on ACPI hotplug infrastructure we already have).
(
in other words, I don't see immediate benefit in grabbing new registers )

Quoted 9 lines
Quoted 9 lines
Quoted 3 lines

More questions:

does all this affect windows guests at all?

I don't know (spec shows examples that reminded me about NIC naming which Windows use(s|d)). But I won't bet on it.
If I recall correctly, for e1000 NIC, it didn't made any difference in naming (pre-existing guest image).

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

where does the "acpi index" terminology come from? the pci firmware spec talks about "instance number", right?

it comes from linux kernel (that's how it's named in sysfs) and systemd/udev uses it. So I tried to avoid making up another one.

OTOH that's guest specific ... and does not seem to imply the limitations unless you look at the code ...

"instance number" looks to me too generic (it's fine in context of specific chapter in spec). While linux kernel's variant looks to me as as better option that fits the bill.

If you don't object, I'd prefer 'acpi-index'.

Quoted 54 lines
Quoted 53 lines
Quoted 53 lines
Quoted 51 lines

For more detailed description/examples see patches [3-4/5]

1)
https://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames/

Git repo for testing:
https://github.com/imammedo/qemu/branches acpi-index-rfc

Igor Mammedov (5):
acpi: add aml_to_decimalstring() and aml_call6() helpers
tests: acpi: temporary whitelist DSDT changes
pci: introduce apci-index property for PCI device
pci: acpi: add _DSM method to PCI devices
tests: acpi: update expected data files

 include/hw/acpi/aml-build.h                 |   3 +
 include/hw/acpi/pci.h                       |   1 +
 include/hw/acpi/pcihp.h                     |   7 +-
 include/hw/pci/pci.h                        |   1 +
 tests/qtest/bios-tables-test-allowed-diff.h |  21 +++++
 hw/acpi/aml-build.c                         |  28 +++++++
 hw/acpi/pci.c                               |  84 ++++++++++++++++++++
 hw/acpi/pcihp.c                             |  25 +++++-
 hw/i386/acpi-build.c                        |  31 +++++++-
 hw/pci/pci.c                                |   1 +
 tests/data/acpi/pc/DSDT                     | Bin 5065 -> 6023 bytes
 tests/data/acpi/pc/DSDT.acpihmat            | Bin 6390 -> 7348 bytes
 tests/data/acpi/pc/DSDT.bridge              | Bin 6924 -> 8689 bytes
 tests/data/acpi/pc/DSDT.cphp                | Bin 5529 -> 6487 bytes
 tests/data/acpi/pc/DSDT.dimmpxm             | Bin 6719 -> 7677 bytes
 tests/data/acpi/pc/DSDT.hpbridge            | Bin 5026 -> 5990 bytes
 tests/data/acpi/pc/DSDT.hpbrroot            | Bin 3084 -> 3177 bytes
 tests/data/acpi/pc/DSDT.ipmikcs             | Bin 5137 -> 6095 bytes
 tests/data/acpi/pc/DSDT.memhp               | Bin 6424 -> 7382 bytes
 tests/data/acpi/pc/DSDT.numamem             | Bin 5071 -> 6029 bytes
 tests/data/acpi/pc/DSDT.roothp              | Bin 5261 -> 6324 bytes
 tests/data/acpi/q35/DSDT                    | Bin 7801 -> 7863 bytes
 tests/data/acpi/q35/DSDT.acpihmat           | Bin 9126 -> 9188 bytes
 tests/data/acpi/q35/DSDT.bridge             | Bin 7819 -> 7911 bytes
 tests/data/acpi/q35/DSDT.cphp               | Bin 8265 -> 8327 bytes
 tests/data/acpi/q35/DSDT.dimmpxm            | Bin 9455 -> 9517 bytes
 tests/data/acpi/q35/DSDT.ipmibt             | Bin 7876 -> 7938 bytes
 tests/data/acpi/q35/DSDT.memhp              | Bin 9160 -> 9222 bytes
 tests/data/acpi/q35/DSDT.mmio64             | Bin 8932 -> 9024 bytes
 tests/data/acpi/q35/DSDT.numamem            | Bin 7807 -> 7869 bytes
 tests/data/acpi/q35/DSDT.tis                | Bin 8407 -> 8468 bytes
 31 files changed, 197 insertions(+), 5 deletions(-)
Signature · 1 line
2.27.0