Re: [RFC PATCH 0/5] ARM: reduce the memory consumed when mapping UEFI flash images

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> On 11/16/20 2:48 PM, Markus Armbruster wrote:
>> Philippe Mathieu-Daudé <philmd@redhat.com> writes:
>> 
>>> Hi David,
>>>
>>> On 11/16/20 11:42 AM, David Edmondson wrote:
>>>> Currently ARM UEFI images are typically built as 2MB/768kB flash
>>>> images for code and variables respectively. These images are both then
>>>> padded out to 64MB before being loaded by QEMU.
>>>>
>>>> Because the images are 64MB each, QEMU allocates 128MB of memory to
>>>> read them, and then proceeds to read all 128MB from disk (dirtying the
>>>> memory). Of this 128MB less than 3MB is useful - the rest is zero
>>>> padding.
>>>
>>> 2 years ago I commented the same problem, and suggested to access the
>>> underlying storage by "block", as this is a "block storage".
>>>
>>> Back then the response was "do not try to fix something that works".
>>> This is why we choose the big hammer "do not accept image size not
>>> matching device size" way.
>>>
>>> While your series seems to help, it only postpone the same
>>> implementation problem. If what you want is use the least memory
>>> required, I still believe accessing the device by block is the
>>> best approach.
>> 
>> "Do not try to fix something that works" is hard to disagree with.
>> However, at least some users seem to disagree with "this works".  Enough
>> to overcome the resistance to change?
>
> Yeah, at least 3 big users so far:
>
> - Huawei
> https://www.mail-archive.com/qemu-devel@nongnu.org/msg607292.html
> - Tencent
> https://www.mail-archive.com/qemu-devel@nongnu.org/msg742066.html
> - Oracle
> (this series).
>
> Then Huawei tried the MicroVM approach:
> https://www.mail-archive.com/qemu-devel@nongnu.org/msg680103.html
>
> I simply wanted to save David time by remembering this other approach,
> since Peter already commented on David's one (see Huawei link).

IIRC the two questions that came up were:

  - what would reading memory not covered by a file look like (0's or
    something more like real HW, 7f?).

  - what happens when the guest writes beyond the bounds of a backing
    file?

I'm guessing for these cloudy type applications no one cares about
persistence of EFI variables? Maybe we just need a formulation for the
second pflash which is explicit about writes being ephemeral while also
being accepted?

>
> Regards,
>
> Phil.


-- 
Alex Bennée