On 07/13/2011 01:41 AM, Alex Williamson wrote:
> static int rtl8139_post_load(void *opaque, int version_id)
> {
> @@ -3283,7 +3242,7 @@ static void rtl8139_pre_save(void *opaque)
> rtl8139_set_next_tctr_time(s, current_time);
> s->TCTR = muldiv64(current_time - s->TCTR_base, PCI_FREQUENCY,
> get_ticks_per_sec());
> - s->rtl8139_mmio_io_addr_dummy = s->rtl8139_mmio_io_addr;
> + s->rtl8139_mmio_io_addr_dummy = 0;
This makes the dummy value fairly useless for preserving new->old
migration.
Drop it altogether and bump the version or add a subsection
to prevent migration to old versions that consume this?
That means we can't migrate to 0.14, even though 0.14 is safe.
How about we fix 0.13.blah? And make the rule that we only support
backwards migration to fully patched releases. There's no problem
requiring an updated target; only an updated source is an issue.
--
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.
