Re: Fw: [Patch] Some updates to serial-5.05

[Stuart MacDonald]

From: "Russell King" <[e-mail]>
> I've merged the simple bits of this by hand with my serial CVS.  As I
> said in a previous mail here, I'm not taking on the maintainence of the
> existing serial.c driver.  Therefore, these comments apply to the
> new serial driver, not the existing drivers.

Copied to you because I thought you might be interested in adding
some of them to the new driver.

> These two I'd rather waited until we've got the driver merged into 2.5,
> at which point I'd rather have a patch against the new driver.

Fair enough.

> I don't actually printk() the serial ports that have been discovered at
> boot time in the new serial CVS.  If people scream enough, I could be
> persuaded.  I'm currently of the opinion that they're noise, and if
> we're really interested in them, we've got a userspace tool to do it
> for us: setserial -bg /dev/ttyS*

From: "Jeff Garzik" <[e-mail]>
> I'll complain ;-)   It seems pretty standard for a driver to print out
> at least one single line for each "interface" it registers; interface in

I agree with Jeff.

> Only the MULTISERIAL support applied - 2.4 has the PCI class definitions,
> so when the new driver is merged, we already have the definitions.

serial_compat.h has more than just missing pci #defs. Although
I suppose you wouldn't need it in the new driver if backwards
compatability isn't being preserved. Compatability of the new driver
with old kernels that is.

..Stu