Re: [PATCH RESEND 1/1] serial: 8250_pci: fix support for MosChip 98xx boards

[Greg KH <gregkh@linuxfoundation.org>]

On Fri, Feb 28, 2014 at 03:57:19PM -0800, Ira W. Snyder wrote:
> Commit 7808edcd306f22aeb23775d34e70b7fa2f58b852 "Basic support for
> Moschip 9900 family I/O chips" broke support for the 98xx boards. This
> is due to a missing check for the 99xx family inside the newly added
> pci_netmos_9900_setup() function, which is now used for all boards in
> the Moschip family.
> 
> The code for skipping BARs is incorrect for the 98xx boards. Using it
> causes two serial ports to be left undetected on my 9865 board.
> 
> By checking for the 99xx boards and using the new code exclusively for
> them, all of my serial ports are now detected.
> 
> Signed-off-by: Ira W. Snyder <iws@ovro.caltech.edu>
> ---
> 
> Here is the lspci output for my 9865 board. On the 05:00.2 device, the
> middle two serial ports (I/O ports 0x1010 and 0x1008) are left unused
> and undetected without this patch.
> 
> After the patch, they work perfectly.
> 
> 05:00.0 Serial controller: MosChip Semiconductor Technology Ltd. PCI 9865 Multi-I/O Controller (prog-if 02 [16550])
> 	Subsystem: Device a000:1000
> 	Flags: bus master, fast Back2Back, medium devsel, latency 32, IRQ 21
> 	I/O ports at 1028 [size=8]
> 	Memory at e0104000 (32-bit, non-prefetchable) [size=4K]
> 	Memory at e0103000 (32-bit, non-prefetchable) [size=4K]
> 	Capabilities: <access denied>
> 	Kernel driver in use: serial
> 	Kernel modules: parport_pc
> 
> 05:00.1 Serial controller: MosChip Semiconductor Technology Ltd. PCI 9865 Multi-I/O Controller (prog-if 02 [16550])
> 	Subsystem: Device a000:1000
> 	Flags: bus master, fast Back2Back, medium devsel, latency 32, IRQ 22
> 	I/O ports at 1020 [size=8]
> 	Memory at e0102000 (32-bit, non-prefetchable) [size=4K]
> 	Memory at e0101000 (32-bit, non-prefetchable) [size=4K]
> 	Capabilities: <access denied>
> 	Kernel driver in use: serial
> 	Kernel modules: parport_pc
> 
> 05:00.2 Communication controller: MosChip Semiconductor Technology Ltd. PCI 9865 Multi-I/O Controller
> 	Subsystem: Device a000:3004
> 	Flags: bus master, fast Back2Back, medium devsel, latency 32, IRQ 23
> 	I/O ports at 1018 [size=8]
> 	I/O ports at 1010 [size=8]
> 	I/O ports at 1008 [size=8]
> 	I/O ports at 1000 [size=8]
> 	Memory at e0100000 (32-bit, non-prefetchable) [size=4K]
> 	Capabilities: <access denied>
> 	Kernel driver in use: serial
> 
>  drivers/tty/serial/8250/8250_pci.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/tty/serial/8250/8250_pci.c b/drivers/tty/serial/8250/8250_pci.c
> index 50228eed3b6f..374551f196c5 100644
> --- a/drivers/tty/serial/8250/8250_pci.c
> +++ b/drivers/tty/serial/8250/8250_pci.c
> @@ -783,7 +783,8 @@ static int pci_netmos_9900_setup(struct serial_private *priv,
>  {
>  	unsigned int bar;
>  
> -	if ((priv->dev->subsystem_device & 0xff00) == 0x3000) {
> +	if ((priv->dev->subsystem_device & 0xff00) == 0x3000 &&
> +		(priv->dev->device & 0xff00) == 0x9900) {
>  		/* netmos apparently orders BARs by datasheet layout, so serial
>  		 * ports get BARs 0 and 3 (or 1 and 4 for memmapped)
>  		 */

This patch doesn't apply at all due to a change to this statement in
3.14-rc3.  Actually, is the patch even needed after commit
333c085e4b0cd9df04b087d29f87969a4e26dfa9 went into Linus's tree?  If so,
please refresh it against the latest tree so I can apply it.

thanks,

greg k-h