+ Troy

Hi Cédric,

> Subject: Re: [PULL 02/16] tests/functional: Add SDK tests with Linux 5.15
> 
> Hello,
> 
> + Jamin,
> 
> On 2/16/26 16:08, Cédric Le Goater wrote:
> > On 2/16/26 15:59, Daniel P. Berrangé wrote:
> >> On Mon, Feb 16, 2026 at 11:17:38AM +0000, Peter Maydell wrote:
> >>> On Mon, 16 Feb 2026 at 11:13, Peter Maydell <[email protected]>
> wrote:
> >>>>
> >>>> On Thu, 12 Feb 2026 at 21:42, Cédric Le Goater <[email protected]>
> wrote:
> >>>>>
> >>>>> Add functional tests for AST2500 and AST2600 machines using the
> >>>>> OpenBMC SDK v11.00 with Linux kernel 5.15. These tests complement
> >>>>> the existing SDK tests and verify that QEMU correctly boots older
> >>>>> kernel versions on these platforms.
> >>>>
> >>>>
> >>>>> +    ASSET_SDK_V1100_AST2600_515 = Asset(
> >>>>> +
> >>>>>
> +'https://github.com/AspeedTech-BMC/openbmc/releases/download/v11.
> >>>>> +00/ast2600-default-515-obmc.tar.gz',
> >>>>> +
> >>>>>
> +'ece1a934095378929780f03e7d092e562f4b33b2841b80ad7c3d12a85744c0f6
> >>>>> +')
> >>>>
> >>>> This asset file is over a gigabyte in size, it takes forever to
> >>>> download, and our functional test asset download framework has no
> >>>> mechanism for giving any kind of progress indication for downloads
> >>>> (or even for telling you the URL it is downloading until after the
> >>>> download is complete).
> >>>>
> >>>> Can we have a size limit for test assets please, at least until the
> >>>> framework can support large assets a bit better ?
> >>>
> >>> Followup: we have six test assets that are over 500MB in size, and
> >>> they are all aspeed tests. Please can you have a look at slimming
> >>> these down?
> >>
> >> I see 2 big assets from aspeed only
> >>
> >>    64d8926a7d01b649168be96c986603b5690f06391286c438a3a772c8c7
> 039e93
> >>    ece1a934095378929780f03e7d092e562f4b33b2841b80ad7c3d12a857
> 44c0f6
> >>
> >>
> >> what are the names of the other files in $HOME/.cache/qemu/download
> >> you have that are over 500 MB ?
> >>
> >> With regards,
> >> Daniel
> >
> > These are the files :
> >
> >
> 5732255d4617d98b76bbbc116d331d6ac89906fa212969eb8213fdc4aea86451
>  166M
> > ast2500-default-515-obmc.tar.gz
> >
> 2768e2526bdd3f928df31341801cb6de3b9a40ca942b538a3d0800cd26fbdbfe
>  166M
> > ast2500-default-obmc.tar.gz
> >
> ece1a934095378929780f03e7d092e562f4b33b2841b80ad7c3d12a85744c0f6
>  1.1G
> > ast2600-default-515-obmc.tar.gz
> >
> 64d8926a7d01b649168be96c986603b5690f06391286c438a3a772c8c7039e93
>   1.2G
> > ast2600-default-obmc.tar.gz
> >
> 4f8778be176ece1b57d33c4aee13bb989be114c3e4703150eaeb6f996bd5587f
>  582M
> > ast2700-a1-dcscm-obmc.tar.gz
> >
> d5ceed511cd0dfefbb102fff2d731159e0472948a28066dc0d90bcd54be76525
> 520M
> > ast2700-a1-obmc.tar.gz
> >
> 0e93f7976139da71fab9df7952a58bdd80650e23e7abf5853b0eb6695deb02d0
>  581M
> > ast2700-dcscm-obmc.tar.gz
> >
> e2b8f043fe8063dd3b6ded93422e38bd41914dc9c3202199507652df024de4dc
>  519M
> > ast2700-default-obmc.tar.gz
> 
> The image size for v09.08 was around 250 MB for the ast2500,
> ast2600 and ast2700 SoCs. It would be best to keep it like this for testing.
> 
> Would it be possible in v12.00 ?
> 


Thank you for your suggestion and for reporting this issue.

To allow QEMU to accept the image tarball size, I will create new tarballs 
specifically for QEMU functional testing.

The "ast2600-default-image.tar.gz" tarball will be based on 
"ast2600-default-obmc.tar.gz". 
It will contain only the necessary images from ast2600-default-obmc.tar.gz for 
QEMU functional testing, such as image-bmc. 
Including only image-bmc in the tarball will significantly reduce its size. In 
SDK v11.01, the size should be around 64MB.

I plan to create the following new tarballs for QEMU functional testing:
1. ast2600-default-image.tar.gz 
2. ast2600-default-66-image.tar.gz
3. ast2700-default-image.tar.gz
4, ast2700-a1-image.tar.gz
5. ast2700-dcscm-image.tar.gz
6. ast2700-a1-dcscm-image.tar.gz

ASPEED will release v11.01 this week. After the v11.01 release, I will update 
these tarballs and the QEMU functional tests accordingly.
Sorry for the late reply and for any inconvenience caused.

Jamin

> Thanks,
> 
> C.
> 
> 
> 
> 

Reply via email to