Just became aware of this patch because Wolfram listed open i801 patches. http://patchwork.ozlabs.org/project/linux-i2c/patch/20230514103634.235917-2-mail@xxxxxxxxxxxxx/ I understand that BIOS is broken and assigns an interrupt to the SMBUS PCI host controller that is actually used by another device. Dealing with internal low-level flag irq_managed in a device driver seems to be fundamentally wrong. Also we may be in trouble if pirq_disable_irq() should ever be called for the SMBUS host controller driver with irq_managed being set and irq set to the one of the other device. Not sure whether setting pci_dev->irq to 255 before calling pci_enable_device() would help. As this isn't an i2c/smbus issue but a PCI/BIOS issue, maybe the PCI folks have an idea what the least ugly quirk could be. And to ask the obvious question: How about a BIOS update from DELL fixing the broken interrupt assignment?
