> there is no way for me to compare AMD patches directly since it appears AMD 
> doesn't publish them

There is a way: a bit later I will privately share a small list of AMD
boards that are still getting the UEFI updates (to help you to obtain
these microcodes by yourself), and will also share a small C program
that converts a hexadecimal arrays provided by my patches (could be
copy-pasted) back to a binary. After you'll successfully get both of
these, you could SHA256 compare them between each other by yourself -
to see that they are indeed 1:1 matching.

> I'm willing to accept the patches if they are harvested, reviewed and then 
> signed by the coreboot project

Both the new AMD microcodes and AtomBIOS binaries haven't been
released officially yet and waiting for the official release by AMD to
get them merged to coreboot master. They can't be merged until the
official release. Currently we are in the talks with AMD, but these
matters are advancing slowly - so the people who don't want to wait
and need them now, could be using them locally and unofficially. Some
of these patches are almost 1 year old already, I guess this is enough
time for the concerned people from a coreboot community to quickly
look through theses patches to at least see that there is nothing
harmful. Also you could see that I'm a coreboot community member for >
3 years and of course not going to ruin my hard earned reputation by
intentionally submitting something harmful :P So, if you trust - you
can use these patches, but if don't trust - can wait, perhaps a lot...
And I don't think any extra signature is necessary, also because these
tiny scripts which are downloading/extracting the patches - also check
their SHA256.

> I don't know when or if they will be merged

We don't know too, Chris, it all depends on AMD...

> I don't know which patches are considered necessary and which are listed 
> because they are nice to have.

Perhaps all of these patches could be considered as optional, since
the people somehow built and used coreboot on their G505S before these
patches even existed. However, you told that you are going to use a
QubesOS which relies on good function of low level virtualization,
that means a new AMD microcode is required for you - otherwise you'll
run into the freezing problems.

Looking through a list of patches at our DangerousPrototypes "Lenovo
G505S hacking" page:

1) AMD microcode updates - required for you, could get by yourself to check

2) Discrete GPU support - optional, and you can verify these 10+67+20
= 97 lines of source code by yourself

3) AMD GPU AtomBIOS blobs - perhaps the AtomBIOS blob for integrated
GPU is required for you - because it seems you don't want to run G505S
in a headless mode - but you could easily get it by yourself ; also
could for a discrete GPU, however it is significantly more difficult
and time consuming

4) tint build system - optional, however it adds the important
checksum verification for a tint archive that is downloaded from FSF
server. Sorry that I forgot to write a readme at DP wiki for this one,
still it is available at my tint patch commit message. And "tint" is a
small opensource tetris game that will be available at your SeaBIOS
boot menu, to have a lot of fun and maybe to show off to your friends
what your new awesome BIOS can do ;-)

5) Unofficial SeaBIOS patches - optional because it seems you are not
going to have more than 10 menu entries, however your mind could
change if you'd also become interested at these floppy-based operating
systems. Mostly for fun (e.g. MichalOS has a cool built-in piano), but
some of their features could be useful to your for the real purposes:
e.g. as soon as the KolibriOS networking driver will be completed for
the network controller of our G505S, it will be possible to access the
Internet and chat with your friends using IRCC. And it seems that all
these listed floppy-based OS, with the exception of a plop bootloader
floppy, are 100% open source which already gives some trust to them

6) Sample G505S .config - optional, since you could configure by
yourself, but of course this config is 100% open source and you could
look through it to verify that there are no harmful options enabled,
and I'm using such a config by myself without any problems.

Best regards,
Mike Banon


On Mon, May 20, 2019 at 4:32 AM Chris Laprise <[email protected]> wrote:
>
> On 5/16/19 2:35 PM, Mike Banon wrote:
> > Hi Chris, if you'd like to verify the microcodes inside my AMD ucode
> > patch: convert the hexadecimal arrays at their .c files back to
> > binary, extract the microcodes from proprietary UEFI updates for those
> > few AMD boards that are still getting them ( or get them already
> > extracted by platomav from platomav's CPUMicrocodes repository -
> > https://github.com/platomav/CPUMicrocodes ), and compare. They will
> > match 1:1. And if you have any questions about any other parts of my
> > patches, I'll try my best to address them.
>
> Thanks. I'm a neophyte when it comes to firmware, and I'm just now
> inferring that there is no way for me to compare AMD patches directly
> since it appears AMD doesn't publish them.
>
> I'm willing to accept the patches if they are harvested, reviewed and
> then signed by the coreboot project. But I don't know when or if they
> will be merged and available this way in the upcoming 4.10 release.
>
> I also don't know which patches are considered necessary and which are
> listed because they are nice to have.
>
> For reference, I intend to run Qubes OS, so I don't need discrete
> graphics, but it appears I'll need AtomBIOS. Will AtomBIOS be merged
> with the upcoming 4.10 release? I can't tell. Going down the list,
> "tint" is indicated but there is no What or Why or a link, and I can't
> turn up any background info by searching. OTOH, it looks like I can skip
> the SeaBIOS patch.
>
> --
>
> Chris Laprise, [email protected]
> https://github.com/tasket
> https://twitter.com/ttaskett
> PGP: BEE2 20C5 356E 764A 73EB  4AB3 1DC4 D106 F07F 1886
_______________________________________________
coreboot mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to