[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH v3 14/16] i2c:smbus_eeprom: Add vmstate handling

From: Corey Minyard
Subject: Re: [Qemu-devel] [PATCH v3 14/16] i2c:smbus_eeprom: Add vmstate handling to the smbus eeprom
Date: Wed, 19 Dec 2018 12:52:09 -0600
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.2.1

On 11/29/18 7:29 AM, Dr. David Alan Gilbert wrote:
* address@hidden (address@hidden) wrote:
From: Corey Minyard <address@hidden>

Transfer the state of the EEPROM on a migration.  This way the
data remains consistent on migration.

This required moving the actual data to a separate array and
using the data provided in the init function as a separate
initialization array, since a pointer property has to be a
void * and the array needs to be uint8_t[].
So I think I'm OK with this, but I'd like other peoples comments as
well; see comments.

Well, no comments so far.

I have a question on this.  If this particular device were modified to be able
to support different size eeproms, would this need a new version of the
vmstate structure?  I know the vmstate structure itself would have to
change to have an array size, but would it have to change in such a
way that the array size would need to be transmitted?

I ask because someday, someone might need an eeprom of a different
size, and it would be nice if it could be done now in a way that avoided
creating a new version.



Signed-off-by: Corey Minyard <address@hidden>
Cc: Paolo Bonzini <address@hidden>
Cc: Michael S. Tsirkin <address@hidden>
Cc: Dr. David Alan Gilbert <address@hidden>
Cc: Peter Maydell <address@hidden>
  hw/i2c/smbus_eeprom.c | 34 ++++++++++++++++++++++++++++++++--
  1 file changed, 32 insertions(+), 2 deletions(-)

diff --git a/hw/i2c/smbus_eeprom.c b/hw/i2c/smbus_eeprom.c
index 8e9b734c09..942057dc10 100644
--- a/hw/i2c/smbus_eeprom.c
+++ b/hw/i2c/smbus_eeprom.c
@@ -24,6 +24,7 @@
#include "qemu/osdep.h"
  #include "hw/hw.h"
+#include "hw/boards.h"
  #include "hw/i2c/i2c.h"
  #include "hw/i2c/smbus_slave.h"
  #include "hw/i2c/smbus_eeprom.h"
@@ -39,8 +40,10 @@
typedef struct SMBusEEPROMDevice {
      SMBusDevice smbusdev;
-    void *data;
+    uint8_t data[SMBUS_EEPROM_SIZE];
+    void *init_data;
      uint8_t offset;
+    bool accessed;
  } SMBusEEPROMDevice;
static uint8_t eeprom_receive_byte(SMBusDevice *dev)
@@ -49,6 +52,7 @@ static uint8_t eeprom_receive_byte(SMBusDevice *dev)
      uint8_t *data = eeprom->data;
      uint8_t val = data[eeprom->offset++];
+ eeprom->accessed = true;
  #ifdef DEBUG
      printf("eeprom_receive_byte: addr=0x%02x val=0x%02x\n",
             dev->i2c.address, val);
@@ -61,6 +65,7 @@ static int eeprom_write_data(SMBusDevice *dev, uint8_t *buf, 
uint8_t len)
      SMBusEEPROMDevice *eeprom = SMBUS_EEPROM(dev);
      uint8_t *data = eeprom->data;
+ eeprom->accessed = true;
  #ifdef DEBUG
      printf("eeprom_write_byte: addr=0x%02x cmd=0x%02x val=0x%02x\n",
             dev->i2c.address, cmd, buf[0]);
@@ -78,15 +83,39 @@ static int eeprom_write_data(SMBusDevice *dev, uint8_t 
*buf, uint8_t len)
      return 0;
+static bool smbus_eeprom_vmstate_needed(void *opaque)
+    MachineClass *mc = MACHINE_GET_CLASS(qdev_get_machine());
+    SMBusEEPROMDevice *eeprom = opaque;
+    return (eeprom->accessed || smbus_vmstate_needed(&eeprom->smbusdev)) &&
+        !mc->smbus_no_migration_support;
That's a little complx, but OK.

+static const VMStateDescription vmstate_smbus_eeprom = {
+    .name = "smbus-eeprom",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .needed = smbus_eeprom_vmstate_needed,
+    .fields      = (VMStateField[]) {
+        VMSTATE_SMBUS_DEVICE(smbusdev, SMBusEEPROMDevice),
+        VMSTATE_UINT8(offset, SMBusEEPROMDevice),
+        VMSTATE_BOOL(accessed, SMBusEEPROMDevice),
OK, good - including 'accessed' means that if we migrate a->b->c and
it's not changed while it's on b, then 'c' still gets the updated

+    }
  static void smbus_eeprom_realize(DeviceState *dev, Error **errp)
      SMBusEEPROMDevice *eeprom = SMBUS_EEPROM(dev);
+ memcpy(eeprom->data, eeprom->init_data, SMBUS_EEPROM_SIZE);
      eeprom->offset = 0;
static Property smbus_eeprom_properties[] = {
-    DEFINE_PROP_PTR("data", SMBusEEPROMDevice, data),
+    DEFINE_PROP_PTR("data", SMBusEEPROMDevice, init_data),
@@ -99,6 +128,7 @@ static void smbus_eeprom_class_initfn(ObjectClass *klass, void *data)
      sc->receive_byte = eeprom_receive_byte;
      sc->write_data = eeprom_write_data;
      dc->props = smbus_eeprom_properties;
+    dc->vmsd = &vmstate_smbus_eeprom;
      /* Reason: pointer property "data" */
      dc->user_creatable = false;
so from migration side:

Reviewed-by: Dr. David Alan Gilbert <address@hidden>


Dr. David Alan Gilbert / address@hidden / Manchester, UK

reply via email to

[Prev in Thread] Current Thread [Next in Thread]