Hi Michael, On 8/1/25 2:29 PM, Michael S. Tsirkin wrote: > On Mon, Jul 28, 2025 at 04:56:37PM +0200, Eric Auger wrote: >> Hi Michael, >> >> On 7/28/25 11:07 AM, Michael Tokarev wrote: >>> On 27.07.2025 14:48, Michael Tokarev wrote: >>>> On 14.07.2025 11:04, Eric Auger wrote: >>>>> gpex build_host_bridge_osc() and x86 originated >>>>> build_pci_host_bridge_osc_method() are mostly identical. >>>>> >>>>> In GPEX, SUPP is set to CDW2 but is not further used. CTRL is >>>>> same as Local0. >>>>> >>>>> So let gpex code reuse build_pci_host_bridge_osc_method() and remove >>>>> build_host_bridge_osc(). >>>>> >>>>> Also add an imply ACPI_PCI clause along with >>>>> PCI_EXPRESS_GENERIC_BRIDGE to compile hw/acpi/pci.c when its >>>>> dependency is resolved (ie. CONFIG_ACPI_PCI). This is requested >>>>> to link qemu-system-mips64el. >>>> This patch somehow broke microvm-only build which is used by debian. >>>> >>>> configure options used (trimmed a bit): >>>> >>>> ../configure \ --without-default-features \ --target-list=x86_64- >>>> softmmu \ --enable-kvm --disable-tcg \ --enable-pixman \ --enable- >>>> vnc \ --audio-drv-list="" \ --without-default-devices \ --with- >>>> devices-x86_64=microvm \ --enable-vhost-user >>>> >>>> configs/devices/x86_64-softmmu/microvm.mak: ---- cut ---- # see >>>> configs/devices/i386-softmmu/default.mak # for additional devices >>>> which can be disabled # CONFIG_PCI_DEVICES=n >>>> >>>> CONFIG_MICROVM=y >>>> >>>> CONFIG_VIRTIO_BLK=y CONFIG_VIRTIO_SERIAL=y CONFIG_VIRTIO_INPUT=y >>>> CONFIG_VIRTIO_INPUT_HOST=y CONFIG_VHOST_USER_INPUT=y >>>> CONFIG_VIRTIO_NET=y CONFIG_VIRTIO_SCSI=y CONFIG_VIRTIO_RNG=y >>>> CONFIG_VIRTIO_CRYPTO=y CONFIG_VIRTIO_BALLOON=y CONFIG_VIRTIO_GPU=y >>>> CONFIG_VHOST_USER_GPU=y ---- cut ---- >>>> >>>> The result: FAILED: qemu-system-x86_64 cc -m64 @qemu-system- >>>> x86_64.rsp /usr/bin/ld: libsystem.a.p/hw_pci-host_gpex-acpi.c.o: in >>>> function `acpi_dsdt_add_host_bridge_methods': hw/pci-host/gpex- >>>> acpi.c:83:(.text+0x274): undefined reference to >>>> `build_pci_host_bridge_osc_method' collect2: error: ld returned 1 >>>> exit status >>> In hw/i386/Kconfig, MICROVM selects ACPI_HW_REDUCED (not ACPI or >>> ACPI_PCI), *and* PCI_EXPRESS_GENERIC_BRIDGE. But in hw/pci/meson.build, >>> pci.c (where the function in question is defined) is built when >>> ACPI_PCI is enabled. Something's not right here. >> This commit added an imply ACPI_PCI to the PCI_EXPRESS_GENERIC_BRIDGE >> config in hw/pci-host/Kconfig: >> >> config PCI_EXPRESS_GENERIC_BRIDGE >> bool >> select PCI_EXPRESS -> selects PCI >> imply ACPI_PCI >> >> In hw/acpi/Kconfig we have: >> >> config ACPI_PCI >> bool >> depends on ACPI && PCI >> >> ACPI is selected by ACPI_HW_REDUCED >> >> config ACPI_HW_REDUCED >> bool >> select ACPI >> select ACPI_MEMORY_HOTPLUG >> select ACPI_NVDIMM >> >> so logically the ACPI && PCI dependencies for ACPI_PCI should be >> resolved. What do I miss? >> regards >> >> Eric >> > > Not sure I get what are you proposing? I have not proposed anything ;-) - sorry I am on vacation - I was trying to enumerate the various config settings and to me dependencies should be resolved so I fail to understand. Unfortunately I cannot test because the problem happens with a custom microvm.mak. I would need this latter to reproduce and try to fix.
if the microvm.mak selects PCI_EXPRESS_GENERIC_BRIDGE (which selects PCI_EXPRESS which itself selects PCI) and ACPI_HW_REDUCED (which selects ACPI), the imply ACPI_PCI should end up as a selection because PCI and ACPI are resolved. At least that's my understanding. Reading it again, please can you check whether ACPI_HW_REDUCED is selected *before* PCI_EXPRESS_GENERIC_BRIDGE in the custom microvm.mak. Depending on the order of dependency resolution it might make a difference. Otherwise please can you send the full microvm.mak and configure command so that I can try reproducing. Thanks Eric > >> >> >> >> >> >> >>> Addig kraxel for 24db877ab6e6 "microvm: add pcie support". >>> >>> Thanks >>> >>> /mjt >>>
