On Sat, Jun 10, 2023 at 02:07:58AM +0800, Sui Jingfeng wrote: > On 2023/6/10 01:52, Bjorn Helgaas wrote: > > On Fri, Jun 09, 2023 at 09:37:02AM +0800, Sui Jingfeng wrote: > > > On 2023/6/9 01:32, Bjorn Helgaas wrote: > > > > On Wed, Jun 07, 2023 at 06:55:49PM +0800, Sui Jingfeng wrote: > > > > > From: Sui Jingfeng <suijingfeng@xxxxxxxxxxx> > > > > > > > > > > This patch adds PCI driver support on top of what we already have. Take > > > > > the GC1000 in LS7A1000/LS2K1000 as the first instance of the PCI device > > > > > driver. There is only one GPU core for the GC1000 in the LS7A1000 and > > > > > LS2K1000. Therefore, component frameworks can be avoided. > > > > > + {PCI_VENDOR_ID_LOONGSON, 0x7a15, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, > > > > > + {PCI_VENDOR_ID_LOONGSON, 0x7a05, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0}, > > > > PCI_VDEVICE() > > > This make it impossible to hook device-specific data in the future. > > > > > > But currently there no device specific data associated with the > > > 0x7a05 and 0x7a15, > > > > > > so it's acceptable for now. Thanks. > > Haha, ISTR having this conversation before, sorry for repeating it. > > > > Indeed, it's fine as-is. But PCI_VDEVICE() actually *does* allow for > > vendor-specific data because it doesn't include the data element, > > which defaults to zero if you don't specify it. > > > > So for example, drivers/net/ethernet/realtek/r8169_main.c has this: > > > > { PCI_VDEVICE(REALTEK, 0x8129) }, > > { PCI_VDEVICE(REALTEK, 0x8136), RTL_CFG_NO_GBIT }, > > > > where 0x8129 has no driver_data (it defaults to zero), but 0x8136 > > does. > > PCI_VDEVICE macro end with two zero. (I thought it was three) No worries, I thought the same thing the first five times I read it :)