Re: [PATCH v1 00/14] OvmfPkg/Virtio: Add APIs to map system physical to device address

[Laszlo Ersek <lersek@redhat.com>]

On 08/15/17 22:44, Brijesh Singh wrote:
> 
> 
> On 08/15/2017 03:39 PM, Laszlo Ersek wrote:
>> On 08/15/17 22:26, Brijesh Singh wrote:
>>>
>>>
>>> On 08/15/2017 02:48 PM, Laszlo Ersek wrote:
>>>> On 08/15/17 21:32, Brijesh Singh wrote:
>>>>> Hi Laszlo,
>>>>>
>>>>> On 08/15/2017 05:42 AM, Laszlo Ersek wrote:
>>>>> [snip]
>>>>>
>>>>>>>
>>>>>>> I have been following the steps from
>>>>>>> https://wiki.linaro.org/LEG/UEFIforQEMU
>>>>>>>
>>>>>>> qemu-system-aarch64 \
>>>>>>>     -m 1024 \
>>>>>>>     -cpu cortex-a57 \
>>>>>>>     -M virt  \
>>>>>>>     -bios QEMU_EFI.fd \
>>>>>>>     -serial stdio
>>>>>>>
>>>>>>> I tried this steps with and without my patches and it resulted in
>>>>>>> the
>>>>>>> same.
>>>>>>> It seems like I am missing something in the qemu cli, do I need to
>>>>>>> pass
>>>>>>> special dtb file or something similar ?
>>>>>>
>>>>>> The above command line is not right ("-bios"). Please scroll down the
>>>>>> wiki page, to the section heading saying "Using persistent UEFI
>>>>>> variables". There it explains how to pad the images and how to use
>>>>>> two
>>>>>> -pflash options. ... Perhaps even that part of the article is a bit
>>>>>> out-of-date now.
>>>>>>
>>>>>> Basically, today ArmVirtQemu should be used the same way as OVMF,
>>>>>> except
>>>>>> for the padding. The build produces two files:
>>>>>> - QEMU_EFI.fd (fw binary)
>>>>>> - QEMU_VARS.fd (varstore template)
>>>>>>
>>>>>> Each should be padded to 64MiB with zeros at the end (write a small
>>>>>> script for that), then use them with two pflash drives similarly to
>>>>>> OVMF.
>>>>>>
>>>>>
>>>>> Still no luck, you can see my log error [1]. I never get to UEFI
>>>>> shell,
>>>>> I have
>>>>> tried with and without virtio disk.
>>>>>
>>>>> https://gist.github.com/codomania/0aed024702b817761ee55fd30929200a
>>>>>
>>>>> I will continuing googling ...
>>>>
>>>> In order to get as detailed as possible logs, I suggest adding the
>>>> following option to the ArmVirtQemu build command line:
>>>>
>>>>     -D DEBUG_PRINT_ERROR_LEVEL=0x8040004F
>>>>
>>>> The current log looks quite strange to me in places, but I'm not
>>>> sure if
>>>> that's because there are problems in those parts, or because the log
>>>> does not contain DEBUG_VERBOSE entries.
>>>>
>>>
>>>
>>> https://gist.github.com/codomania/8b2fc5424fda259236405c5e257d8f47
>>>
>>> I am using Ubuntu 16.04 for builds and runs
>>>
>>> $ qemu-system-aarch64 --version
>>> QEMU emulator version 2.5.0 (Debian 1:2.5+dfsg-5ubuntu10.14), Copyright
>>> (c) 2003-2008 Fabrice Bellard
>>
>> What is your complete QEMU command line?
>>
> 
> I have been using the following two qemu cli
> 
> # qemu-system-aarch64 -m 2048 -cpu cortex-a57 -M virt \
>      -pflash flash0.img -pflash flash1.img \
>      -nographic
> 
> # qemu-system-aarch64 -m 2048 -cpu cortex-a57 -M virt \
>       -pflash flash0.img -pflash flash1.img \
>      -drive
> if=none,file=/home/brijesh/xenial-server-cloudimg-arm64.img,id=hd0,format=raw
> -device virtio-blk-device,drive=hd0 \
>      -nographic

I can reproduce the boot hang. It looks like an edk2 regression. I'm
currently bisecting the issue.

Thanks
Laszlo