Re: [Qemu-devel] [PULL 23/35] x86: ioapic: dump version for "info ioapic"

[Paolo Bonzini]

On 30/01/2017 09:07, Peter Maydell wrote:
> On 20 January 2017 at 13:31, Paolo Bonzini <address@hidden> wrote:
>> From: Peter Xu <address@hidden>
>>
>> Signed-off-by: Peter Xu <address@hidden>
>> Message-Id: <address@hidden>
>> Signed-off-by: Paolo Bonzini <address@hidden>
>> ---
>>  hw/intc/ioapic_common.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/hw/intc/ioapic_common.c b/hw/intc/ioapic_common.c
>> index 1b7ec5e..97c4f9c 100644
>> --- a/hw/intc/ioapic_common.c
>> +++ b/hw/intc/ioapic_common.c
>> @@ -58,7 +58,8 @@ void ioapic_print_redtbl(Monitor *mon, IOAPICCommonState 
>> *s)
>>      uint32_t remote_irr = 0;
>>      int i;
>>
>> -    monitor_printf(mon, "ioapic id=0x%02x sel=0x%02x", s->id, s->ioregsel);
>> +    monitor_printf(mon, "ioapic ver=0x%x id=0x%02x sel=0x%02x",
>> +                   s->version, s->id, s->ioregsel);
>>      if (s->ioregsel) {
>>          monitor_printf(mon, " (redir[%u])\n",
>>                         (s->ioregsel - IOAPIC_REG_REDTBL_BASE) >> 1);
> 
> Coverity points out (CID 1369422) that this is a use of a possibly
> uninitialized field. In kvm_ioapic_dump_state() we do:
> 
>     IOAPICCommonState s;
>     kvm_ioapic_get(&s);
>     ioapic_print_redtbl(mon, &s);
> 
> and kvm_ioapic_get() doesn't initialize s->version, so when we
> come to print it in ioapic_print_redtbl() it's uninitialized.
> 
> The easy fix is to initialize version to something. The
> underlying problem here I think is that we're manufacturing
> a fake IOAPICCommonState rather than finding the one that
> corresponds to the actual IOAPIC device in the system...

Right, we can probably use object_resolve_path to get one.

Thanks,

Paolo