Incorrect UART in /tty/serial/8250

Incorrect UART in /tty/serial/8250

[Joshua Droney]

Hello!

My name is Josh. I'm currently employed at Commtech, Inc.
(https://fastcomproducts.com/). I'm following the instructions at
https://kernelnewbies.org/FoundBug to report this issue. Please let me
know if this is incorrect.

We (Commtech) have several serial boards, but the board in question is
the '422/2-PCIe'. This board does not exist, but someone added it to
the Linux kernel at some point in the past. This wouldn't be a
problem, except that the board ID that was used for this non-existent
board (0x22) was ultimately used for a different REAL board, with a
different UART. Instead of a xr172358, we use a 16C950 on the real
board. But the Linux kernel finds our other board ID, and tries to
initialize it as an exar chip and fails. I don't know much about the
kernel, but I believe the lines in question are:

https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L47
https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L925

You can see that we have a different board with device ID 0x22 here:
https://admin.pci-ids.ucw.cz/read/PC/18f7

The 422/4-PCIe exists (0x20), the 422/8-PCIe exists (0x21), but the
422/2-PCIe was never created and its device id was used for a
different board (0x22, the SuperFSCC/4-LVDS-PCIe).

Josh

Re: Incorrect UART in /tty/serial/8250

[Greg KH]

On Thu, Mar 28, 2024 at 02:34:10PM -0500, Joshua Droney wrote:
> Hello!
> 
> My name is Josh. I'm currently employed at Commtech, Inc.
> (https://fastcomproducts.com/). I'm following the instructions at
> https://kernelnewbies.org/FoundBug to report this issue. Please let me
> know if this is incorrect.
> 
> We (Commtech) have several serial boards, but the board in question is
> the '422/2-PCIe'. This board does not exist, but someone added it to
> the Linux kernel at some point in the past. This wouldn't be a
> problem, except that the board ID that was used for this non-existent
> board (0x22) was ultimately used for a different REAL board, with a
> different UART. Instead of a xr172358, we use a 16C950 on the real
> board. But the Linux kernel finds our other board ID, and tries to
> initialize it as an exar chip and fails. I don't know much about the
> kernel, but I believe the lines in question are:
> 
> https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L47
> https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L925
> 
> You can see that we have a different board with device ID 0x22 here:
> https://admin.pci-ids.ucw.cz/read/PC/18f7
> 
> The 422/4-PCIe exists (0x20), the 422/8-PCIe exists (0x21), but the
> 422/2-PCIe was never created and its device id was used for a
> different board (0x22, the SuperFSCC/4-LVDS-PCIe).

Can you please submit a patch that fixes this?  We'll be glad to take
it.  If you need help making it, please let us know.

thanks,

greg k-h

Re: Incorrect UART in /tty/serial/8250

[Joshua Droney]

I would appreciate some guidance on this. I don't know enough about
the kernel's serial driver to be confident about what changes need to
be made, nor anything about creating/submitting a patch.

Thanks!
Josh

On Fri, Mar 29, 2024 at 1:01 AM Greg KH <gregkh@linuxfoundation.org> wrote:
>
> On Thu, Mar 28, 2024 at 02:34:10PM -0500, Joshua Droney wrote:
> > Hello!
> >
> > My name is Josh. I'm currently employed at Commtech, Inc.
> > (https://fastcomproducts.com/). I'm following the instructions at
> > https://kernelnewbies.org/FoundBug to report this issue. Please let me
> > know if this is incorrect.
> >
> > We (Commtech) have several serial boards, but the board in question is
> > the '422/2-PCIe'. This board does not exist, but someone added it to
> > the Linux kernel at some point in the past. This wouldn't be a
> > problem, except that the board ID that was used for this non-existent
> > board (0x22) was ultimately used for a different REAL board, with a
> > different UART. Instead of a xr172358, we use a 16C950 on the real
> > board. But the Linux kernel finds our other board ID, and tries to
> > initialize it as an exar chip and fails. I don't know much about the
> > kernel, but I believe the lines in question are:
> >
> > https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L47
> > https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L925
> >
> > You can see that we have a different board with device ID 0x22 here:
> > https://admin.pci-ids.ucw.cz/read/PC/18f7
> >
> > The 422/4-PCIe exists (0x20), the 422/8-PCIe exists (0x21), but the
> > 422/2-PCIe was never created and its device id was used for a
> > different board (0x22, the SuperFSCC/4-LVDS-PCIe).
>
> Can you please submit a patch that fixes this?  We'll be glad to take
> it.  If you need help making it, please let us know.
>
> thanks,
>
> greg k-h

Re: Incorrect UART in /tty/serial/8250

[Andy Shevchenko]

Fri, Mar 29, 2024 at 09:23:49AM -0500, Joshua Droney kirjoitti:

First of all, do not top-post in the Linux kernel related mailing lists.

> I would appreciate some guidance on this. I don't know enough about
> the kernel's serial driver to be confident about what changes need to
> be made, nor anything about creating/submitting a patch.

You may start from here:
https://www.kernel.org/doc/html/latest/process/submitting-patches.html
https://www.kernel.org/doc/html/latest/process/submit-checklist.html

> On Fri, Mar 29, 2024 at 1:01 AM Greg KH <gregkh@linuxfoundation.org> wrote:
> >
> > On Thu, Mar 28, 2024 at 02:34:10PM -0500, Joshua Droney wrote:
> > > Hello!
> > >
> > > My name is Josh. I'm currently employed at Commtech, Inc.
> > > (https://fastcomproducts.com/). I'm following the instructions at
> > > https://kernelnewbies.org/FoundBug to report this issue. Please let me
> > > know if this is incorrect.
> > >
> > > We (Commtech) have several serial boards, but the board in question is
> > > the '422/2-PCIe'. This board does not exist, but someone added it to
> > > the Linux kernel at some point in the past. This wouldn't be a
> > > problem, except that the board ID that was used for this non-existent
> > > board (0x22) was ultimately used for a different REAL board, with a
> > > different UART. Instead of a xr172358, we use a 16C950 on the real
> > > board. But the Linux kernel finds our other board ID, and tries to
> > > initialize it as an exar chip and fails. I don't know much about the
> > > kernel, but I believe the lines in question are:
> > >
> > > https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L47
> > > https://github.com/torvalds/linux/blob/master/drivers/tty/serial/8250/8250_exar.c#L925
> > >
> > > You can see that we have a different board with device ID 0x22 here:
> > > https://admin.pci-ids.ucw.cz/read/PC/18f7
> > >
> > > The 422/4-PCIe exists (0x20), the 422/8-PCIe exists (0x21), but the
> > > 422/2-PCIe was never created and its device id was used for a
> > > different board (0x22, the SuperFSCC/4-LVDS-PCIe).
> >
> > Can you please submit a patch that fixes this?  We'll be glad to take
> > it.  If you need help making it, please let us know.

-- 
With Best Regards,
Andy Shevchenko