Re: [pci:for-linus] BUILD REGRESSION f0bfeb2c51e44bee7876f2a0eda3518bd2c30a01

[Bjorn Helgaas <helgaas@kernel.org>]

On Thu, Oct 16, 2025 at 11:35:07AM -0500, Mario Limonciello wrote:
> On 10/16/25 11:28 AM, Bjorn Helgaas wrote:
> > On Thu, Oct 16, 2025 at 11:18:38AM -0500, Mario Limonciello wrote:
> > > On 10/16/25 11:15 AM, Bjorn Helgaas wrote:
> > > > On Thu, Oct 16, 2025 at 07:26:50AM +0800, kernel test robot wrote:
> > > > > tree/branch: https://git.kernel.org/pub/scm/linux/kernel/git/pci/pci.git for-linus
> > > > > branch HEAD: f0bfeb2c51e44bee7876f2a0eda3518bd2c30a01  PCI/VGA: Select SCREEN_INFO on X86
> > > > 
> > > > Just making sure you've seen this, Mario.
> > > 
> > > I didn't see this, thanks for including me.
> > > 
> > > > I *think* f0bfeb2c51e4 is the most recent version, and it was
> > > > on pci/for-linus, so I'll drop it for now.
> > > 
> > > Are you sure the failure is caused by "PCI/VGA: Select
> > > SCREEN_INFO on X86"?
> > 
> > I'm not sure.  I looked briefly for a more detailed report but
> > didn't find it.  Maybe didn't look hard enough.  This email seems
> > like a summary that could possibly have included a link to
> > details.
> 
> I looked at https://lore.kernel.org/oe-kbuild-all/ and don't see one
> there either.
> 
> I think you should keep the patch in.  As it pertains to arch
> specific stuff it behaves identically to pre-337bf13aa9dda.

Yep, added back as 54a880a5af73 ("PCI/VGA: Select SCREEN_INFO on X86"):

commit 54a880a5af73 ("PCI/VGA: Select SCREEN_INFO on X86")
Author: Mario Limonciello (AMD) <superm1@kernel.org>
Date:   Mon Oct 13 17:08:26 2025 -0500

    PCI/VGA: Select SCREEN_INFO on X86
    
    commit 337bf13aa9dda ("PCI/VGA: Replace vga_is_firmware_default() with a
    screen info check") introduced an implicit dependency upon SCREEN_INFO by
    removing the open coded implementation.
    
    If a user didn't have CONFIG_SCREEN_INFO set, vga_is_firmware_default()
    would now return false.  SCREEN_INFO is only used on X86 so add a
    conditional select for SCREEN_INFO to ensure that the VGA arbiter works as
    intended.
    
    Fixes: 337bf13aa9dda ("PCI/VGA: Replace vga_is_firmware_default() with a screen info check")
    Reported-by: Eric Biggers <ebiggers@kernel.org>
    Closes: https://lore.kernel.org/linux-pci/20251012182302.GA3412@sol/
    Suggested-by: Thomas Zimmermann <tzimmermann@suse.de>
    Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
    Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
    Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
    Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
    Tested-by: Eric Biggers <ebiggers@kernel.org>
    Link: https://patch.msgid.link/20251013220829.1536292-1-superm1@kernel.org


diff --git a/drivers/pci/Kconfig b/drivers/pci/Kconfig
index 7065a8e5f9b1..f94f5d384362 100644
--- a/drivers/pci/Kconfig
+++ b/drivers/pci/Kconfig
@@ -306,6 +306,7 @@ config VGA_ARB
 	bool "VGA Arbitration" if EXPERT
 	default y
 	depends on (PCI && !S390)
+	select SCREEN_INFO if X86
 	help
 	  Some "legacy" VGA devices implemented on PCI typically have the same
 	  hard-decoded addresses as they did on ISA. When multiple PCI devices
diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
index b58f94ee4891..436fa7f4c387 100644
--- a/drivers/pci/vgaarb.c
+++ b/drivers/pci/vgaarb.c
@@ -556,10 +556,8 @@ EXPORT_SYMBOL(vga_put);
 
 static bool vga_is_firmware_default(struct pci_dev *pdev)
 {
-#ifdef CONFIG_SCREEN_INFO
-	struct screen_info *si = &screen_info;
-
-	return pdev == screen_info_pci_dev(si);
+#if defined CONFIG_X86
+	return pdev == screen_info_pci_dev(&screen_info);
 #else
 	return false;
 #endif