On Mon, Jul 26, 2021 at 06:00:57PM +0200, Ard Biesheuvel wrote:
> (cc Bjorn)
>
> On Mon, 26 Jul 2021 at 11:08, Philippe Mathieu-Daudé <phi...@redhat.com>
> wrote:
> >
> > On 7/26/21 12:56 AM, Guenter Roeck wrote:
> > > On 7/25/21 3:14 PM, Michael S. Tsirkin wrote:
> > >> On Sat, Jul 24, 2021 at 11:52:34AM -0700, Guenter Roeck wrote:
> > >>> Hi all,
> > >>>
> > >>> starting with qemu v6.0, some of my aarch64 efi boot tests no longer
> > >>> work. Analysis shows that PCI devices with IO ports do not instantiate
> > >>> in qemu v6.0 (or v6.1-rc0) when booting through efi. The problem affects
> > >>> (at least) ne2k_pci, tulip, dc390, and am53c974. The problem only
> > >>> affects
> > >>> aarch64, not x86/x86_64.
> > >>>
> > >>> I bisected the problem to commit 0cf8882fd0 ("acpi/gpex: Inform os to
> > >>> keep firmware resource map"). Since this commit, PCI device BAR
> > >>> allocation has changed. Taking tulip as example, the kernel reports
> > >>> the following PCI bar assignments when running qemu v5.2.
> > >>>
> > >>> [ 3.921801] pci 0000:00:01.0: [1011:0019] type 00 class 0x020000
> > >>> [ 3.922207] pci 0000:00:01.0: reg 0x10: [io 0x0000-0x007f]
> > >>> [ 3.922505] pci 0000:00:01.0: reg 0x14: [mem 0x10000000-0x1000007f]
>
> IIUC, these lines are read back from the BARs
>
> > >>> [ 3.927111] pci 0000:00:01.0: BAR 0: assigned [io 0x1000-0x107f]
> > >>> [ 3.927455] pci 0000:00:01.0: BAR 1: assigned [mem
> > >>> 0x10000000-0x1000007f]
> > >>>
>
> ... and this is the assignment created by the kernel.
>
> > >>> With qemu v6.0, the assignment is reported as follows.
> > >>>
> > >>> [ 3.922887] pci 0000:00:01.0: [1011:0019] type 00 class 0x020000
> > >>> [ 3.923278] pci 0000:00:01.0: reg 0x10: [io 0x0000-0x007f]
> > >>> [ 3.923451] pci 0000:00:01.0: reg 0x14: [mem 0x10000000-0x1000007f]
> > >>>
>
> The problem here is that Linux, for legacy reasons, does not support
> I/O ports <= 0x1000 on PCI, so the I/O assignment created by EFI is
> rejected.
>
> This might make sense on x86, where legacy I/O ports may exist, but on
> other architectures, this makes no sense.
Fixing Linux makes sense but OTOH EFI probably shouldn't create mappings
that trip up existing guests, right?
>
> > >>> and the controller does not instantiate. The problem disapears after
> > >>> reverting commit 0cf8882fd0.
> > >>>
> > >>> Attached is a summary of test runs with various devices and qemu v5.2
> > >>> as well as qemu v6.0, and the command line I use for efi boots.
> > >>>
> > >>> Did commit 0cf8882fd0 introduce a bug, do I now need need some different
> > >>> command line to instantiate PCI devices with io ports, or are such
> > >>> devices
> > >>> simply no longer supported if the system is booted with efi support ?
> > >>>
> > >>> Thanks,
> > >>> Guenter
> > >>
> > >>
> > >> So that commit basically just says don't ignore what efi did.
> > >>
> > >> The issue's thus likely efi.
> > >>
> > >
> > > I don't see the problem with efi boots on x86 and x86_64.
> > > Any idea why that might be the case ?
> > >
> > > Thanks,
> > > Guenter
> > >
> > >> Cc the maintainer. Philippe can you comment pls?
> >
> > I'll have a look. Cc'ing Ard for EDK2/Aarch64.
> >
>
> So a potential workaround would be to use a different I/O resource
> window for ArmVirtPkg, that starts at 0x1000. But I would prefer to
> fix Linux instead.
>
>
> > >>
> > >>> ---
> > >>> Command line (tulip network interface):
> > >>>
> > >>> CMDLINE="root=/dev/vda console=ttyAMA0"
> > >>> ROOTFS="rootfs.ext2"
> > >>>
> > >>> qemu-system-aarch64 -M virt -kernel arch/arm64/boot/Image -no-reboot \
> > >>> -m 512 -cpu cortex-a57 -no-reboot \
> > >>> -device tulip,netdev=net0 -netdev user,id=net0 \
> > >>> -bios QEMU_EFI-aarch64.fd \
> > >>> -snapshot \
> > >>> -device virtio-blk-device,drive=d0 \
> > >>> -drive file=${ROOTFS},if=none,id=d0,format=raw \
> > >>> -nographic -serial stdio -monitor none \
> > >>> --append "${CMDLINE}"
> > >>>
> > >>> ---
> > >>> Boot tests with various devices known to work in qemu v5.2.
> > >>>
> > >>> v5.2 v6.0 v6.0
> > >>> efi non-efi efi
> > >>> e1000 pass pass pass
> > >>> e1000-82544gc pass pass pass
> > >>> e1000-82545em pass pass pass
> > >>> e1000e pass pass pass
> > >>> i82550 pass pass pass
> > >>> i82557a pass pass pass
> > >>> i82557b pass pass pass
> > >>> i82557c pass pass pass
> > >>> i82558a pass pass pass
> > >>> i82559b pass pass pass
> > >>> i82559c pass pass pass
> > >>> i82559er pass pass pass
> > >>> i82562 pass pass pass
> > >>> i82801 pass pass pass
> > >>> ne2k_pci pass pass fail <--
> > >>> pcnet pass pass pass
> > >>> rtl8139 pass pass pass
> > >>> tulip pass pass fail <--
> > >>> usb-net pass pass pass
> > >>> virtio-net-device
> > >>> pass pass pass
> > >>> virtio-net-pci pass pass pass
> > >>> virtio-net-pci-non-transitional
> > >>> pass pass pass
> > >>>
> > >>> usb-xhci pass pass pass
> > >>> usb-ehci pass pass pass
> > >>> usb-ohci pass pass pass
> > >>> usb-uas-xhci pass pass pass
> > >>> virtio pass pass pass
> > >>> virtio-blk-pci pass pass pass
> > >>> virtio-blk-device
> > >>> pass pass pass
> > >>> nvme pass pass pass
> > >>> sdhci pass pass pass
> > >>> dc390 pass pass fail <--
> > >>> am53c974 pass pass fail <--
> > >>> lsi53c895ai pass pass pass
> > >>> mptsas1068 pass pass pass
> > >>> lsi53c810 pass pass pass
> > >>> megasas pass pass pass
> > >>> megasas-gen2 pass pass pass
> > >>> virtio-scsi-device
> > >>> pass pass pass
> > >>> virtio-scsi-pci pass pass pass
> > >>
> > >
> >
