Skip to contentSkip to conversation

serial: Fix double migration data

PATCH Applied Dr. David Alan Gilbert (git) 3 participants 5 messages Mar 30, 2020 – Mar 31, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 serial: Fix double migration data

From: "Dr. David Alan Gilbert" <[email protected]>

After c9808d60281 we have both an object representing the serial-isa device and a separate object representing the underlying common serial uart. Both of these have vmsd's associated with them and thus the migration stream ends up with two copies of the migration data - the
serial-isa includes the vmstate of the core serial. Besides
being wrong, it breaks backwards migration compatibility.

Fix this by removing the dc->vmsd from the core device, so it only gets migrated by any parent devices including it.
Add a vmstate_serial_mm so that any device that uses serial_mm_init rather than creating a device still gets migrated.
(That doesn't fix backwards migration for serial_mm_init users, but does seem to work forwards for ppce500).

Fixes c9808d60281 ('serial: realize the serial device')
---
 hw/char/serial.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)
hw/char/serial.c +11 -1
index 2ab8b69e03..c822a9ae6c 100644
@@ -1043,7 +1043,6 @@ static void serial_class_init(ObjectClass *klass, void* data)
     dc->user_creatable = false;
     dc->realize = serial_realize;
     dc->unrealize = serial_unrealize;
-    dc->vmsd = &vmstate_serial;
     device_class_set_props(dc, serial_properties);
 }
 
@@ -1113,6 +1112,16 @@ static void serial_mm_realize(DeviceState *dev, Error **errp)
     sysbus_init_irq(SYS_BUS_DEVICE(smm), &smm->serial.irq);
 }
 
+static const VMStateDescription vmstate_serial_mm = {
+    .name = "serial",
+    .version_id = 3,
+    .minimum_version_id = 2,
+    .fields = (VMStateField[]) {
+        VMSTATE_STRUCT(serial, SerialMM, 0, vmstate_serial, SerialState),
+        VMSTATE_END_OF_LIST()
+    }
+};
+
 SerialMM *serial_mm_init(MemoryRegion *address_space,
                          hwaddr base, int regshift,
                          qemu_irq irq, int baudbase,
@@ -1162,6 +1171,7 @@ static void serial_mm_class_init(ObjectClass *oc, void *data)
 
     device_class_set_props(dc, serial_mm_properties);
     dc->realize = serial_mm_realize;
+    dc->vmsd = &vmstate_serial_mm;
 }
 
 static const TypeInfo serial_mm_info = {
Signature · 1 line
2.25.1
#1 lore raw

Hi

Quoted 40 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

After c9808d60281 we have both an object representing the serial-isa device and a separate object representing the underlying common serial uart. Both of these have vmsd's associated with them and thus the migration stream ends up with two copies of the migration data - the
serial-isa includes the vmstate of the core serial. Besides
being wrong, it breaks backwards migration compatibility.

Fix this by removing the dc->vmsd from the core device, so it only gets migrated by any parent devices including it.
Add a vmstate_serial_mm so that any device that uses serial_mm_init rather than creating a device still gets migrated.
(That doesn't fix backwards migration for serial_mm_init users, but does seem to work forwards for ppce500).

Fixes c9808d60281 ('serial: realize the serial device')
---
 hw/char/serial.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)
hw/char/serial.c +0 -1
index 2ab8b69e03..c822a9ae6c 100644
@@ -1043,7 +1043,6 @@ static void serial_class_init(ObjectClass *klass, void* data)
     dc->user_creatable = false;
     dc->realize = serial_realize;
     dc->unrealize = serial_unrealize;
-    dc->vmsd = &vmstate_serial;
     device_class_set_props(dc, serial_properties);
 }
 
@@ -1113,6 +1112,16 @@ static void serial_mm_realize(DeviceState *dev, Error **errp)
     sysbus_init_irq(SYS_BUS_DEVICE(smm), &smm->serial.irq);
 }
 

hw/char/serial.c static void serial_mm_realize(DeviceState *dev, Error **errp) · line 1124

Quoted 10 lines
hw/char/serial.c +10 -0
+static const VMStateDescription vmstate_serial_mm = {
+    .name = "serial",
+    .version_id = 3,
+    .minimum_version_id = 2,
+    .fields = (VMStateField[]) {
+        VMSTATE_STRUCT(serial, SerialMM, 0, vmstate_serial, SerialState),
+        VMSTATE_END_OF_LIST()
+    }
+};
+

Why do you make it a sub-state?

# qemu-system-ppc -M ppce500 -monitor stdio -serial pty
in 4.2 and 5.0:
    "serial (8)": {
        "divider": "0x00d9",
        "rbr": "0x00",
        "ier": "0x00",
        "iir": "0xc1",
        "lcr": "0x03",
        "mcr": "0x03",
        "lsr": "0x60",
        "msr": "0xb0",
        "scr": "0x00",
        "fcr_vmstate": "0x01"
    },

With this patch:

    "serial (8)": {
        "serial": {
            "divider": "0x00d9",
            "rbr": "0x00",
            "ier": "0x00",
            "iir": "0xc1",
            "lcr": "0x03",
            "mcr": "0x03",
            "lsr": "0x60",
            "msr": "0xb0",
            "scr": "0x00",
            "fcr_vmstate": "0x01"
        }
    },
Quoted 12 lines
hw/char/serial.c +1 -1
 SerialMM *serial_mm_init(MemoryRegion *address_space,
                          hwaddr base, int regshift,
                          qemu_irq irq, int baudbase,
@@ -1162,6 +1171,7 @@ static void serial_mm_class_init(ObjectClass *oc, void *data)
 
     device_class_set_props(dc, serial_mm_properties);
     dc->realize = serial_mm_realize;
+    dc->vmsd = &vmstate_serial_mm;
 }
 
 static const TypeInfo serial_mm_info = {
--
Quoted 3 lines

2.25.1

I understand removing the serial state from SerialClass solves the double state issue for ISA. But at the same time, I think we should aim to migrate ISASerial state to SerialClass state. I can take a look if you want.

Signature · 1 line
Marc-André Lureau
#2 lore raw
Quoted 44 lines · * Marc-André Lureau ([email protected]) wrote:

Hi

Quoted 40 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

After c9808d60281 we have both an object representing the serial-isa device and a separate object representing the underlying common serial uart. Both of these have vmsd's associated with them and thus the migration stream ends up with two copies of the migration data - the
serial-isa includes the vmstate of the core serial. Besides
being wrong, it breaks backwards migration compatibility.

Fix this by removing the dc->vmsd from the core device, so it only gets migrated by any parent devices including it.
Add a vmstate_serial_mm so that any device that uses serial_mm_init rather than creating a device still gets migrated.
(That doesn't fix backwards migration for serial_mm_init users, but does seem to work forwards for ppce500).

Fixes c9808d60281 ('serial: realize the serial device')
---
 hw/char/serial.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)
hw/char/serial.c +0 -1
index 2ab8b69e03..c822a9ae6c 100644
@@ -1043,7 +1043,6 @@ static void serial_class_init(ObjectClass *klass, void* data)
     dc->user_creatable = false;
     dc->realize = serial_realize;
     dc->unrealize = serial_unrealize;
-    dc->vmsd = &vmstate_serial;
     device_class_set_props(dc, serial_properties);
 }
 
@@ -1113,6 +1112,16 @@ static void serial_mm_realize(DeviceState *dev, Error **errp)
     sysbus_init_irq(SYS_BUS_DEVICE(smm), &smm->serial.irq);
 }
 
Quoted 12 lines

hw/char/serial.c static void serial_mm_realize(DeviceState *dev, Error **errp) · line 1124

Quoted 10 lines
hw/char/serial.c +10 -0
+static const VMStateDescription vmstate_serial_mm = {
+    .name = "serial",
+    .version_id = 3,
+    .minimum_version_id = 2,
+    .fields = (VMStateField[]) {
+        VMSTATE_STRUCT(serial, SerialMM, 0, vmstate_serial, SerialState),
+        VMSTATE_END_OF_LIST()
+    }
+};
+

Why do you make it a sub-state?

Because it's consistent with serial-isa and it's simple.

Quoted 41 lines
# qemu-system-ppc -M ppce500 -monitor stdio -serial pty
in 4.2 and 5.0:
    "serial (8)": {
        "divider": "0x00d9",
        "rbr": "0x00",
        "ier": "0x00",
        "iir": "0xc1",
        "lcr": "0x03",
        "mcr": "0x03",
        "lsr": "0x60",
        "msr": "0xb0",
        "scr": "0x00",
        "fcr_vmstate": "0x01"
    },

With this patch:

    "serial (8)": {
        "serial": {
            "divider": "0x00d9",
            "rbr": "0x00",
            "ier": "0x00",
            "iir": "0xc1",
            "lcr": "0x03",
            "mcr": "0x03",
            "lsr": "0x60",
            "msr": "0xb0",
            "scr": "0x00",
            "fcr_vmstate": "0x01"
        }
    },
Quoted 10 lines
hw/char/serial.c +1 -0
 SerialMM *serial_mm_init(MemoryRegion *address_space,
                          hwaddr base, int regshift,
                          qemu_irq irq, int baudbase,
@@ -1162,6 +1171,7 @@ static void serial_mm_class_init(ObjectClass *oc, void *data)
 
     device_class_set_props(dc, serial_mm_properties);
     dc->realize = serial_mm_realize;
+    dc->vmsd = &vmstate_serial_mm;
 }
 
Quoted 10 lines
Quoted 5 lines
hw/char/serial.c +0 -1
 static const TypeInfo serial_mm_info = {
--

2.25.1

I understand removing the serial state from SerialClass solves the double state issue for ISA. But at the same time, I think we should aim to migrate ISASerial state to SerialClass state. I can take a look if you want.

I don't think there's anything wrong with having a separate layer here; the physical reality of what we have is a UART (Serial) that is on the ISA bus where the ISA bus interfacing doesn't require any extra state to be migrated.

Dave

Quoted 5 lines
Signature · 2 lines
-- 
Marc-André Lureau

--
Dr. David Alan Gilbert / [email protected] / Manchester, UK

Reviewed-by
#3 lore raw

Hi

Quoted 59 lines · Dr. David Alan Gilbert
Quoted 44 lines · * Marc-André Lureau ([email protected]) wrote:

Hi

Quoted 40 lines · Dr. David Alan Gilbert (git)

From: "Dr. David Alan Gilbert" <[email protected]>

After c9808d60281 we have both an object representing the serial-isa device and a separate object representing the underlying common serial uart. Both of these have vmsd's associated with them and thus the migration stream ends up with two copies of the migration data - the
serial-isa includes the vmstate of the core serial. Besides
being wrong, it breaks backwards migration compatibility.

Fix this by removing the dc->vmsd from the core device, so it only gets migrated by any parent devices including it.
Add a vmstate_serial_mm so that any device that uses serial_mm_init rather than creating a device still gets migrated.
(That doesn't fix backwards migration for serial_mm_init users, but does seem to work forwards for ppce500).

Fixes c9808d60281 ('serial: realize the serial device')
---
 hw/char/serial.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)
hw/char/serial.c +0 -1
index 2ab8b69e03..c822a9ae6c 100644
@@ -1043,7 +1043,6 @@ static void serial_class_init(ObjectClass *klass, void* data)
     dc->user_creatable = false;
     dc->realize = serial_realize;
     dc->unrealize = serial_unrealize;
-    dc->vmsd = &vmstate_serial;
     device_class_set_props(dc, serial_properties);
 }
 
@@ -1113,6 +1112,16 @@ static void serial_mm_realize(DeviceState *dev, Error **errp)
     sysbus_init_irq(SYS_BUS_DEVICE(smm), &smm->serial.irq);
 }
 
Quoted 12 lines

hw/char/serial.c static void serial_mm_realize(DeviceState *dev, Error **errp) · line 1124

Quoted 10 lines
hw/char/serial.c +10 -0
+static const VMStateDescription vmstate_serial_mm = {
+    .name = "serial",
+    .version_id = 3,
+    .minimum_version_id = 2,
+    .fields = (VMStateField[]) {
+        VMSTATE_STRUCT(serial, SerialMM, 0, vmstate_serial, SerialState),
+        VMSTATE_END_OF_LIST()
+    }
+};
+

Why do you make it a sub-state?

Quoted 1 line

Because it's consistent with serial-isa and it's simple.

ok, lgtm then

Quoted 62 lines
Quoted 42 lines
# qemu-system-ppc -M ppce500 -monitor stdio -serial pty
in 4.2 and 5.0:
    "serial (8)": {
        "divider": "0x00d9",
        "rbr": "0x00",
        "ier": "0x00",
        "iir": "0xc1",
        "lcr": "0x03",
        "mcr": "0x03",
        "lsr": "0x60",
        "msr": "0xb0",
        "scr": "0x00",
        "fcr_vmstate": "0x01"
    },

With this patch:

    "serial (8)": {
        "serial": {
            "divider": "0x00d9",
            "rbr": "0x00",
            "ier": "0x00",
            "iir": "0xc1",
            "lcr": "0x03",
            "mcr": "0x03",
            "lsr": "0x60",
            "msr": "0xb0",
            "scr": "0x00",
            "fcr_vmstate": "0x01"
        }
    },
Quoted 10 lines
hw/char/serial.c +1 -0
 SerialMM *serial_mm_init(MemoryRegion *address_space,
                          hwaddr base, int regshift,
                          qemu_irq irq, int baudbase,
@@ -1162,6 +1171,7 @@ static void serial_mm_class_init(ObjectClass *oc, void *data)
 
     device_class_set_props(dc, serial_mm_properties);
     dc->realize = serial_mm_realize;
+    dc->vmsd = &vmstate_serial_mm;
 }
 
Quoted 10 lines
Quoted 5 lines
hw/char/serial.c +0 -1
 static const TypeInfo serial_mm_info = {
--

2.25.1

I understand removing the serial state from SerialClass solves the double state issue for ISA. But at the same time, I think we should aim to migrate ISASerial state to SerialClass state. I can take a look if you want.

I don't think there's anything wrong with having a separate layer here; the physical reality of what we have is a UART (Serial) that is on the ISA bus where the ISA bus interfacing doesn't require any extra state to be migrated.

Dave

Quoted 2 lines
Quoted 6 lines
Quoted 3 lines

--
Marc-André Lureau

--
Dr. David Alan Gilbert / [email protected] / Manchester, UK

thanks!

Signature · 1 line
Marc-André Lureau