Re: [GIT PULL 0/2] EFI fixes for v4.11

From: Ingo Molnar <mingo@kernel.org>
To: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>,
	"linux-efi@vger.kernel.org" <linux-efi@vger.kernel.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	"H . Peter Anvin" <hpa@zytor.com>,
	Leif Lindholm <leif.lindholm@linaro.org>,
	Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Eugene Cohen <eugene@hp.com>,
	Matt Fleming <matt@codeblueprint.co.uk>,
	Peter Jones <pjones@redhat.com>
Subject: Re: [GIT PULL 0/2] EFI fixes for v4.11
Date: Wed, 5 Apr 2017 12:26:42 +0200	[thread overview]
Message-ID: <20170405102641.GA13342@gmail.com> (raw)
In-Reply-To: <CAKv+Gu_VaKNDmaWc2Gqpg1JB6pBthCVUOGEsONb8sj2Uos7qcg@mail.gmail.com>

* Ard Biesheuvel <ard.biesheuvel@linaro.org> wrote:

> On 5 April 2017 at 11:08, Bartlomiej Zolnierkiewicz
> <b.zolnierkie@samsung.com> wrote:
> >
> > Hi,
> >
> > On Tuesday, April 04, 2017 04:27:42 PM Ard Biesheuvel wrote:
> >> Hello all,
> >>
> >> Please pull these fixes for EFI framebuffer support on ARM/arm64 systems.
> >>
> >> The following changes since commit 822f5845f710e57d7e2df1fd1ee00d6e19d334fe:
> >>
> >>   efi/esrt: Cleanup bad memory map log messages (2017-03-17 18:53:12 +0000)
> >>
> >> are available in the git repository at:
> >>
> >>   git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi.git tags/efi-urgent
> >>
> >> for you to fetch changes up to e73c2811538bd36ec1340d01bafdc080af31914e:
> >>
> >>   efifb: Avoid reconfiguration of BAR that covers the framebuffer (2017-04-04 15:56:43 +0100)
> >>
> >> ----------------------------------------------------------------
> >> Two fixes related to the EFI framebuffer driver:
> >> - Ignore Graphics Output Protocol (GOP) implementations that are marked as
> >>   BLT-only -- the framebuffer base address is invalid in this case, and the
> >>   Blt() method is not accessible to the kernel.
> >> - If the GOP framebuffer base address coincides with a memory BAR of a PCI
> >>   device that has memory decoding enabled, claim the memory resource so that
> >>   the PCI core will not attempt to move it later on.
> >>
> >> ----------------------------------------------------------------
> >> Ard Biesheuvel (1):
> >>       efifb: Avoid reconfiguration of BAR that covers the framebuffer
> >
> > This patch breaks build if PCI support is not enabled:
> >
> > drivers/video/fbdev/efifb.c: In function ‘claim_efifb_bar’:
> > drivers/video/fbdev/efifb.c:386:2: error: implicit declaration of function ‘pci_claim_resource’ [-Werror=implicit-function-declaration]
> >
> > (x86 ifdefs are not enough, the patch should also check for PCI support)
> >
> > Also please cc: linux-fbdev mailing list & me on fbdev related patches.
> >
> 
> Thanks for the report. The patch was tested successfully on an
> impressive list of configurations by kbuild test robot, but
> apparently, none of those has PCI disabled.
> 
> Ingo, since you have queued this already, how would you like to
> proceed? I don't think we need anything beyond
> 
> diff --git a/drivers/video/fbdev/efifb.c b/drivers/video/fbdev/efifb.c
> index 758960b6aec9..b827a8113e26 100644
> --- a/drivers/video/fbdev/efifb.c
> +++ b/drivers/video/fbdev/efifb.c
> @@ -364,7 +364,7 @@ static struct platform_driver efifb_driver = {
> 
>  builtin_platform_driver(efifb_driver);
> 
> -#ifndef CONFIG_X86
> +#if defined(CONFIG_PCI) && !defined(CONFIG_X86)

I've rebased it all, it's still pretty fresh (two hours old) and the delta fix 
would look pretty ugly in the urgent branch.

Thanks,

	Ingo