Re: [Qemu-arm] [PATCH] virt: Suppress external aborts on virt-2.10 and earlier

[Philippe Mathieu-Daudé]

On 9/26/18 2:29 PM, Peter Maydell wrote:
> On 26 September 2018 at 11:35, Philippe Mathieu-Daudé <address@hidden> wrote:
>> Hi Peter,
>>
>> On 9/25/18 4:41 PM, Peter Maydell wrote:
>>> In commit c79c0a314c43b78 we enabled emulation of external aborts
>>> when the guest attempts to access a physical address with no
>>> mapped device. In commit 4672cbd7bed88dc6 we suppress this for
>>> most legacy boards to prevent breakage of previously working
>>> guests, but we didn't suppress it in the 'virt' board, with
>>> the rationale "we know that guests won't try to prod devices
>>> that we don't describe in the device tree or ACPI tables". This
>>> is mostly true, but we've had a report of a Linux guest image
>>> that this did break. The problem seems to be that the guest
>>> is (incorrectly) configured with a DEBUG_UART_PHYS value that
>>> tells it there is a uart at 0x10009000 (which is true for
>>> vexpress but not for virt), so in early bootup the kernel
>>> probes this bogus address.
>>>
>>> This is a misconfigured guest, so we don't need to worry
>>> about it too much, but we can arrange that guests that ran
>>> on QEMU v2.10 (before c79c0a314c43b78) will still run on
>>> the "virt-2.10" board model, by suppressing external aborts
>>> only for that version and earlier. This seems a reasonable
>>> compromise.
>>
>> I tried another approach to keep MEMTX_DECODE_ERROR delivered (out of
>> 0x10009000) for 2.10, using the UNIMP device, but the code looks uglier
>> (create too much generic code for a single issue):
> 
> I definitely don't want to do that...
> 
>> I prefer your compromise.
>>
>>>
>>> Cc: address@hidden
>>> Signed-off-by: Peter Maydell <address@hidden>
>>> ---
>>>  hw/arm/virt.c | 1 +
>>>  1 file changed, 1 insertion(+)
>>>
>>> diff --git a/hw/arm/virt.c b/hw/arm/virt.c
>>> index 0b57f87abcb..3ba310a37b6 100644
>>> --- a/hw/arm/virt.c
>>> +++ b/hw/arm/virt.c
>>> @@ -1926,6 +1926,7 @@ static void virt_machine_2_10_options(MachineClass 
>>> *mc)
>>>  {
>>>      virt_machine_2_11_options(mc);
>>>      SET_MACHINE_COMPAT(mc, VIRT_COMPAT_2_10);
>>
>> Can you add a short comment here?
>>
>>        /* See https://bugs.launchpad.net/qemu/+bug/???:
>>         * Some misconfigured Linux guest poke for vexpress uart
>>         * at 0x10009000 */
> 
> I view it as more "virt-2.10 is supposed to behave like
> the 2.10 release virt" rather than a specific fix for
> this guest, I think.

You right, fine. Very useful short explanation btw.