On 9 Jul 2007 at 19:37, Chris Doré wrote: > > A current patch can be found here: > ftp://ftp.connecttech.com/pub/misc/bhbltnptm-2.6.21.patch > I'm just another by-stander, I don't have any vote, let me just present my opinion... +++ linux-2.6.21-cti/include/linux/serial_core.h2007-07-06 03:44:52.000000000 - 0400 @@ -274,6 +274,9 @@ struct uart_port { unsigned intline;/* port index */ unsigned longmapbase;/* for ioremap */ struct device*dev;/* parent device */ +intlmode; +int(*lmode_fn)(struct uart_port *port, int ioctl, unsigned int *value); +int(*ptm_fn)(struct uart_port *port, int ioctl, unsigned int *value); As far as patching the vanilla kernel is concerned: You seem to suggest an addition of rather proprietary data (members) to the neatly generic struct uart_port. I'd suggest adding something like void* priv; // device-specific driver private data to the generic struct uart_port of the vanilla kernel, which you could then use to hook your own private data struct, regardless of whether the rest of your driver lives in 8250_pci.c or in a stand- alone module. This is the way it's done in a number of other generic device structs (or maybe the more modern way is to use a nested struct with "parent struct decoding"). Frank Rysanek
Attachment:
WPM$0431.PM$
Description: Mail message body