Restoring the opensource AGESA boards takes just 1% of git reverts since their removal

[Mike Banon]

Dear friends, please take a look at this change:

https://review.coreboot.org/c/coreboot/+/76832

This script reverts the opensource AGESA AMD boards removal that happened
after 5e8e911b7caee021faff96c4e82a77a42544ea62 (0 point of history, or 0 PoH)
- by git-reverting:
   1) the "bad commits" (marked as "CBF" = coreboot build failure)
      - that either remove or break a code needed for our boards
   2) the "unlucky commits" (marked as "GRF" = git revert failure)
      - that are a roadblock for git-reverting the "bad commits"

Right now at 3bd83b27afbc840f184b652a1dd8ed539623a258 (3765 PoH), it takes
38 CBF git reverts - just 1% of 3765 commits since the OSS AGESA removal! -
- making this removal look questionable and the idea of restoration viable.

Btw plenty of these git reverts could be avoided if the changes to the
"dropped code" are allowed: i.e. instead of git reverting the four
patches numbered 497-500 at the script above, one could simply add a
single " #define SPEEDSTEP_APIC_MAGIC 0xACAC " somewhere. Therefore
the real number of "bad commits" (which either remove our boards or
break the things for us) isn't that great and the opensource AGESA
boards really can be brought back to coreboot with a bunch of ifdef's
etc. Please let me know if you're interested in such a quest...

SUCCESSFUL TESTS for the opensource AGESA boards which I own (Lenovo G505S -
   - fam15 laptop, ASUS A88XM-E - fam15 desktop, ASUS AM1I-A - fam16 desktop) :
1) only build:
     69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for AM1I-A,
     most likely boot works too but I didn't have the time to test
2) build & boot:
     69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for G505S and A88XM-E
     11ba8ebbcc662ebd1dc8e14372a020eb32f26561 (3741 PoH) for G505S test only

[Mike Banon]

Now coreboot 4.21 boots flawlessly on AMD Lenovo G505S ! That's how we
get coreboot 4.21 there ;-)
a new "restore_agesa.sh" script compatible with coreboot master -
https://review.coreboot.org/c/coreboot/+/76832

On Sat, Aug 5, 2023 at 12:23 AM Mike Banon  wrote:
> Dear friends, please take a look at this change:
>
> https://review.coreboot.org/c/coreboot/+/76832
>
> This script reverts the opensource AGESA AMD boards removal that happened
> after 5e8e911b7caee021faff96c4e82a77a42544ea62 (0 point of history, or 0 PoH)
> by git-reverting:
> the "bad commits" (marked as "CBF" = coreboot build failure)
> that either remove or break a code needed for our boards
> the "unlucky commits" (marked as "GRF" = git revert failure)
> that are a roadblock for git-reverting the "bad commits"
>
> Right now at 3bd83b27afbc840f184b652a1dd8ed539623a258 (3765 PoH), it takes
> 38 CBF git reverts - just 1% of 3765 commits since the OSS AGESA removal! -
> making this removal look questionable and the idea of restoration viable.
>
> Btw plenty of these git reverts could be avoided if the changes to the
> "dropped code" are allowed: i.e. instead of git reverting the four
> patches numbered 497-500 at the script above, one could simply add a
> single " #define SPEEDSTEP_APIC_MAGIC 0xACAC " somewhere. Therefore
> the real number of "bad commits" (which either remove our boards or
> break the things for us) isn't that great and the opensource AGESA
> boards really can be brought back to coreboot with a bunch of ifdef's
> etc. Please let me know if you're interested in such a quest...
>
> SUCCESSFUL TESTS for the opensource AGESA boards which I own (Lenovo G505S -
> fam15 laptop, ASUS A88XM-E - fam15 desktop, ASUS AM1I-A - fam16 desktop) :
> only build:
>   69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for AM1I-A,
>   most likely boot works too but I didn't have the time to test
> build & boot:
>   69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for G505S and A88XM-E
>   11ba8ebbcc662ebd1dc8e14372a020eb32f26561 (3741 PoH) for G505S test only

Merci beaucoup! 
;-)
 Florentin

----- Mail original -----
De: "Mike Banon" 
À: "coreboot" 
Envoyé: Samedi 2 Septembre 2023 00:03:01
Objet: [coreboot] Re: Restoring the opensource AGESA boards takes just 1% of git reverts since their removal

Now coreboot 4.21 boots flawlessly on AMD Lenovo G505S ! That's how we
get coreboot 4.21 there ;-)
a new "restore_agesa.sh" script compatible with coreboot master -
https://review.coreboot.org/c/coreboot/+/76832

On Sat, Aug 5, 2023 at 12:23 AM Mike Banon  wrote:
> Dear friends, please take a look at this change:
>
> https://review.coreboot.org/c/coreboot/+/76832
>
> This script reverts the opensource AGESA AMD boards removal that happened
> after 5e8e911b7caee021faff96c4e82a77a42544ea62 (0 point of history, or 0 PoH)
> by git-reverting:
> the "bad commits" (marked as "CBF" = coreboot build failure)
> that either remove or break a code needed for our boards
> the "unlucky commits" (marked as "GRF" = git revert failure)
> that are a roadblock for git-reverting the "bad commits"
>
> Right now at 3bd83b27afbc840f184b652a1dd8ed539623a258 (3765 PoH), it takes
> 38 CBF git reverts - just 1% of 3765 commits since the OSS AGESA removal! -
> making this removal look questionable and the idea of restoration viable.
>
> Btw plenty of these git reverts could be avoided if the changes to the
> "dropped code" are allowed: i.e. instead of git reverting the four
> patches numbered 497-500 at the script above, one could simply add a
> single " #define SPEEDSTEP_APIC_MAGIC 0xACAC " somewhere. Therefore
> the real number of "bad commits" (which either remove our boards or
> break the things for us) isn't that great and the opensource AGESA
> boards really can be brought back to coreboot with a bunch of ifdef's
> etc. Please let me know if you're interested in such a quest...
>
> SUCCESSFUL TESTS for the opensource AGESA boards which I own (Lenovo G505S -
> fam15 laptop, ASUS A88XM-E - fam15 desktop, ASUS AM1I-A - fam16 desktop) :
> only build:
>   69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for AM1I-A,
>   most likely boot works too but I didn't have the time to test
> build & boot:
>   69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for G505S and A88XM-E
>   11ba8ebbcc662ebd1dc8e14372a020eb32f26561 (3741 PoH) for G505S test only

[Mike Banon]

Behold: new restore_agesa.sh delivers coreboot 4.22.01 to Lenovo G505S
- boots fine! - and to other AMD AGESA boards which I didn't have the
opportunity to test yet ;-)
https://review.coreboot.org/c/coreboot/+/76832

On Sat, Sep 2, 2023 at 3:22 PM  wrote:
> Merci beaucoup!
> ;-)
>  Florentin
>
> ----- Mail original -----
> De: "Mike Banon" 
> À: "coreboot" 
> Envoyé: Samedi 2 Septembre 2023 00:03:01
> Objet: [coreboot] Re: Restoring the opensource AGESA boards takes just 1% of git reverts since their removal
>
> Now coreboot 4.21 boots flawlessly on AMD Lenovo G505S ! That's how we
> get coreboot 4.21 there ;-)
> a new "restore_agesa.sh" script compatible with coreboot master -
> https://review.coreboot.org/c/coreboot/+/76832
>
> On Sat, Aug 5, 2023 at 12:23 AM Mike Banon  wrote:
> > Dear friends, please take a look at this change:
> >
> > https://review.coreboot.org/c/coreboot/+/76832
> >
> > This script reverts the opensource AGESA AMD boards removal that happened
> > after 5e8e911b7caee021faff96c4e82a77a42544ea62 (0 point of history, or 0 PoH)
> > by git-reverting:
> > the "bad commits" (marked as "CBF" = coreboot build failure)
> > that either remove or break a code needed for our boards
> > the "unlucky commits" (marked as "GRF" = git revert failure)
> > that are a roadblock for git-reverting the "bad commits"
> >
> > Right now at 3bd83b27afbc840f184b652a1dd8ed539623a258 (3765 PoH), it takes
> > 38 CBF git reverts - just 1% of 3765 commits since the OSS AGESA removal! -
> > making this removal look questionable and the idea of restoration viable.
> >
> > Btw plenty of these git reverts could be avoided if the changes to the
> > "dropped code" are allowed: i.e. instead of git reverting the four
> > patches numbered 497-500 at the script above, one could simply add a
> > single " #define SPEEDSTEP_APIC_MAGIC 0xACAC " somewhere. Therefore
> > the real number of "bad commits" (which either remove our boards or
> > break the things for us) isn't that great and the opensource AGESA
> > boards really can be brought back to coreboot with a bunch of ifdef's
> > etc. Please let me know if you're interested in such a quest...
> >
> > SUCCESSFUL TESTS for the opensource AGESA boards which I own (Lenovo G505S -
> > fam15 laptop, ASUS A88XM-E - fam15 desktop, ASUS AM1I-A - fam16 desktop) :
> > only build:
> >   69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for AM1I-A,
> >   most likely boot works too but I didn't have the time to test
> > build & boot:
> >   69ffebf5ccf123bc0b3fb28b485985af0597761d (3698 PoH) for G505S and A88XM-E
> >   11ba8ebbcc662ebd1dc8e14372a020eb32f26561 (3741 PoH) for G505S test only