Re: [Qemu-devel] [Qemu-arm] [PATCH 0/2] GICv2 & GICv3: RAZ/WI reserved addresses rather than aborting

[Laszlo Ersek <lersek@redhat.com>]

On 01/09/18 15:24, Peter Maydell wrote:
> Ping for code review?
> 
> thanks
> -- PMM
> 
> On 13 December 2017 at 16:52, Peter Maydell <peter.maydell@linaro.org> wrote:
>> The GICv2 and GICv3 specifications

here's where I had started running from your patch...

>> say that reserved register
>> addresses should RAZ/WI.

...and I remarked only looking over my shoulder that "RAZ/WI" is not a
verb :)

Sorry, no clue about any of this -- where should I read up?	

Ard did ask a question though:

https://www.mail-archive.com/qemu-devel@nongnu.org/msg500055.html

CC'ing Drew, Eric and Wei. In particular (IIUC) Eric has been looking
into a weird "why do we have so many (and so slow) aborts with GICv4
emulation" issue, at launching UEFI.

Thanks
Laszlo

>> This means we need to return MEMTX_OK, not
>> MEMTX_ERROR, because now that we support generating external aborts
>> the latter will cause an abort on new board models.
>>
>> In particular, at least some versions of UEFI try to
>> access a reserved address in the GICv3 redistributor
>> (at SGI_base + 0x184) and fail to boot on the virt board
>> without this.
>>
>> thanks
>> -- PMM
>>
>> Peter Maydell (2):
>>   hw/intc/arm_gicv3: Make reserved register addresses RAZ/WI
>>   hw/intc/arm_gic: reserved register addresses are RAZ/WI
>>
>>  hw/intc/arm_gic.c              |  5 +++--
>>  hw/intc/arm_gicv3_dist.c       | 13 +++++++++++++
>>  hw/intc/arm_gicv3_its_common.c |  8 +++-----
>>  hw/intc/arm_gicv3_redist.c     | 13 +++++++++++++
>>  4 files changed, 32 insertions(+), 7 deletions(-)