Re: Kernel V5.7-rc1 doesn't boot (EFI?)

From: Ingo Molnar
Date: Wed Apr 15 2020 - 03:45:52 EST



* Ard Biesheuvel <ardb@xxxxxxxxxx> wrote:

> On Tue, 14 Apr 2020 at 18:32, Jörg Otte <jrg.otte@xxxxxxxxx> wrote:
> >
> > Am Di., 14. Apr. 2020 um 12:50 Uhr schrieb Ard Biesheuvel <ardb@xxxxxxxxxx>:
> > >
> > > On Tue, 14 Apr 2020 at 12:40, Jörg Otte <jrg.otte@xxxxxxxxx> wrote:
> > > >
> > > > Booting my notebook with kernel V57-rc1 I get following
> > > > display:
> > > >
> > > > exit_boot() failed!
> > > > efi_main() failed!
> > > > StartImage failed: Buffer Too Small
> > > >
> > > > Booting Kernel V5.6 works well.
> > > >
> > > > From dmesg (kernel V5.6):
> > > > efi: EFI v2.31 by Phoenix Technologies Ltd.
> > > > efi: ACPI=0xdcffe000 ACPI 2.0=0xdcffe014 SMBIOS=0xdce80000 RNG=0xdc3cd198
> > > > efi: seeding entropy pool
> > > > efi: [Firmware Bug]: Invalid EFI memory map entries:
> > > > efi: mem47: [Reserved | | | | | | | | | | | |
> > > > | ] range=[0x0000000000000000-0x0000000000000000] (invalid)
> > > > efi: mem48: [Reserved | | | | | | | | | | | |
> > > > | ] range=[0x0000000000000000-0x0000000000000000] (invalid)
> > > > efi: mem49: [Reserved | | | | | | | | | | | |
> > > > | ] range=[0x0000000000000000-0x0000000000000000] (invalid)
> > > > efi: mem50: [Reserved | | | | | | | | | | | |
> > > > | ] range=[0x0000000000000000-0x0000000000000000] (invalid)
> > > > efi: mem51: [Reserved | | | | | | | | | | | |
> > > > | ] range=[0x0000000000000000-0x0000000000000000] (invalid)
> > > > efi: Removing 5 invalid memory map entries.
> > > >
> > >
> > > Thanks for the report.
> > >
> > > Can you try booting with efi=no_disable_early_pci_dma passed via the
> > > kernel command line? [*]
> > >
> > Yes, that works!
> >
> > > If that does not help, can you try to reproduce with this branch?
> > >
> > > https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/log/?h=efi/urgent
> >
> > Should I give that branch a try anyway?
> >
>
> Your test proves that BSS is not being cleared correctly, so I have
> the answer I was looking for. However, I would appreciate it if you
> could test that branch, just to double check.

I'll send efi/urgent to Linus within the next couple of hours, to get the
fix to Linus:

105cb9544b16: ("efi/x86: Move efi stub globals from .bss to .data")

Thanks,

Ingo