RE: [PATCH 0/2] Fix boot hang issue on Ampere Emag server

[Justin He <Justin.He@arm.com>]

Hi Ard

> -----Original Message-----
> From: Ard Biesheuvel <ardb@kernel.org>
> Sent: Tuesday, February 7, 2023 4:54 PM
> To: Justin He <Justin.He@arm.com>
> Cc: Huacai Chen <chenhuacai@kernel.org>; linux-efi@vger.kernel.org;
> linux-kernel@vger.kernel.org; Alexandru Elisei <Alexandru.Elisei@arm.com>;
> Jason A. Donenfeld <Jason@zx2c4.com>; nd <nd@arm.com>
> Subject: Re: [PATCH 0/2] Fix boot hang issue on Ampere Emag server
> 
> On Tue, 7 Feb 2023 at 09:49, Justin He <Justin.He@arm.com> wrote:
> >
> >
> >
> > > -----Original Message-----
> > [..]
> > > > The root cause of the hung IMO might be similar to commit
> > > > 550b33cfd445296868a478e8413ffb2e963eed32
> > > > Author: Ard Biesheuvel <ardb@kernel.org>
> > > > Date:   Thu Nov 10 10:36:20 2022 +0100
> > > >
> > > >     arm64: efi: Force the use of SetVirtualAddressMap() on Altra
> > > > machines
> > > >
> > > > Do you agree with the idea if I add Ampere ”eMAG” machine into the
> > > > list of Using SetVirtualAddressMap() forcibly?
> > > >
> > > > Please note that even in previous kernel patch, the efibootmgr -t
> > > > 10 will make kernel hung if I passed "efi=novamap" to the boot
> parameter.
> > > >
> > >
> > > Interesting. What does dmidecode return for the family in the type 1
> record?
> >
> > # dmidecode |grep -i family
> >         Family: eMAG
> >         Family: ARMv8
> >
> > The full dmidecode log is at https://pastebin.com/M3MAJtUG
> >
> 
> OK please try this:
> 
> diff --git a/drivers/firmware/efi/libstub/arm64.c
> b/drivers/firmware/efi/libstub/arm64.c
> index ff2d18c42ee74979..fae930dec82be7c6 100644
> --- a/drivers/firmware/efi/libstub/arm64.c
> +++ b/drivers/firmware/efi/libstub/arm64.c
> @@ -22,7 +22,8 @@ static bool system_needs_vamap(void)
>          * Ampere Altra machines crash in SetTime() if
> SetVirtualAddressMap()
>          * has not been called prior.
>          */
> -       if (!type1_family || strcmp(type1_family, "Altra"))
> +       if (!type1_family ||
> +           (strcmp(type1_family, "Altra") && strcmp(type1_family,
> + "eMAG")))
>                 return false;
> 
>         efi_warn("Working around broken SetVirtualAddressMap()\n");

Yes, it works on my eMAG server: the kernel boots.
Other than efibootmgr failure. But I noticed this efibootmgr failure even before
Commit d3549a938b7 ("avoid SetVirtualAddressMap() when possible ")

root@:~/linux# efibootmgr -t 9; efibootmgr -t 5;
Could not set Timeout: Input/output error
Could not set Timeout: Input/output error


--
Cheers,
Justin (Jia He)