Hello, Damien Zammit, le ven. 24 oct. 2025 00:51:29 +0000, a ecrit: > Forwarding this to the list so we don't lose your patches.
Yes, please, better discuss things on lists so it's not lost... > From: Michael Kelly <[email protected]> > the change to use pci_enable_device rather than > pci_enable_device_mem is a 'hack'. Hurd implements an architecture > specific version of the Linux function pci_enable_device (in > libdde-linux26/lib/src/arch/l4/pci.c) to interface into libddekit. I'd > guess that one or more additional functions might need to be added there > to make all drivers work without any changes to the drivers themselves. Yes, see that function in libdde_linux26/lib/src/arch/l4/pci.c, it should be very trivial to add a pci_enable_device_mem function that just does the same. Actually, pci_enable_device is supposed to be doing more than pci_enable_device_mem (by enabling IORESOURCE_IO too), so the existing implementation of pci_enable_device should be just fine for pci_enable_device_mem. And please post it as patches with changelog etc. so it can be properly commited/submitted etc. Samuel
