On 05/01/2021 11:06, Philippe Mathieu-Daudé wrote:

> ping?
> 
> On 12/5/20 4:09 PM, Philippe Mathieu-Daudé wrote:
>> Per the "NCR89C105 Chip Specification" referenced in the header:
>>
>>                    Chip-level Address Map
>>
>>    ------------------------------------------------------------------
>>    | 1D0 0000 ->   | Counter/Timers                        | W,D    |
>>    |   1DF FFFF    |                                       |        |
>>    ...
>>
>>    The address map indicated the allowed accesses at each address.
>>    [...] W indicates a word access, and D indicates a double-word
>>    access.
>>
>> The SLAVIO timer controller is implemented expecting 32-bit accesses.
>> Commit a3d12d073e1 restricted the memory accesses to 32-bit, while
>> the device allows 64-bit accesses.
>>
>> This was not an issue until commit 5d971f9e67 which reverted
>> ("memory: accept mismatching sizes in memory_region_access_valid").
>>
>> Fix by renaming .valid MemoryRegionOps as .impl, and add the valid
>> access range (W -> 4, D -> 8).
>>
>> Since commit 21786c7e598 ("memory: Log invalid memory accesses")
>> this class of bug can be quickly debugged displaying 'guest_errors'
>> accesses, as:
>>
>>    $ qemu-system-sparc -M SS-20 -m 256 -bios ss20_v2.25_rom -serial stdio -d guest_errors
>>
>>    Power-ON Reset
>>    Invalid access at addr 0x0, size 8, region 'timer-1', reason: invalid size (min:4 max:4)
>>
>>    $ qemu-system-sparc -M SS-20 -m 256 -bios ss20_v2.25_rom -monitor stdio -S
>>    (qemu) info mtree
>>    address-space: memory
>>      0000000000000000-ffffffffffffffff (prio 0, i/o): system
>>        ...
>>        0000000ff1300000-0000000ff130000f (prio 0, i/o): timer-1
>>               ^^^^^^^^^                                 ^^^^^^^
>>                     \ memory region base address and name /
>>
>>    (qemu) info qtree
>>    bus: main-system-bus
>>      dev: slavio_timer, id ""              <-- device type name
>>        gpio-out "sysbus-irq" 17
>>        num_cpus = 1 (0x1)
>>        mmio 0000000ff1310000/0000000000000014
>>        mmio 0000000ff1300000/0000000000000010 <--- base address
>>        mmio 0000000ff1301000/0000000000000010
>>        mmio 0000000ff1302000/0000000000000010
>>        ...
>>
>> Reported-by: Yap KV <email address hidden>
>> Buglink: https://bugs.launchpad.net/bugs/1906905
>> Fixes: a3d12d073e1 ("slavio_timer: convert to memory API")
>> Signed-off-by: Philippe Mathieu-Daudé <email address hidden>
>> ---
>> Cc: Benoit Canet <email address hidden>
>> Cc: <email address hidden>
>> Signed-off-by: Philippe Mathieu-Daudé <email address hidden>
>> ---
>>   hw/timer/slavio_timer.c | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/hw/timer/slavio_timer.c b/hw/timer/slavio_timer.c
>> index 5b2d20cb6a5..03e33fc5926 100644
>> --- a/hw/timer/slavio_timer.c
>> +++ b/hw/timer/slavio_timer.c
>> @@ -331,6 +331,10 @@ static const MemoryRegionOps slavio_timer_mem_ops = {
>>       .write = slavio_timer_mem_writel,
>>       .endianness = DEVICE_NATIVE_ENDIAN,
>>       .valid = {
>> +        .min_access_size = 4,
>> +        .max_access_size = 8,
>> +    },
>> +    .impl = {
>>           .min_access_size = 4,
>>           .max_access_size = 4,
>>       },

Looks like I did queue this to my local qemu-sparc branch, but was waiting to see if 
there was feedback on the other outstanding SPARC patches before sending a PR. The 
patches are all fairly minor, however I would prefer an Ack for the outstanding LEON 
patchset first. I'll go chase that one up now.


ATB,

Mark.