Re: [PATCH v2] serial: exar: Fix GPIO configuration for Sealevel cards based on XR17V35X

[Matthew Howell <mrhowel@g.clemson.edu>]

On 7/11/20 2:40 AM, Greg KH wrote:
> On Fri, Jul 10, 2020 at 04:33:00PM -0400, Matthew Howell wrote:
>>
>> From: Matthew Howell <mrhowel@g.clemson.edu>
>>
>> Sealevel XR17V35X based devices are inoperable on kernel versions
>> 4.11 and above due to a change in the GPIO preconfiguration introduced in commit
>> 7dea8165f1d. This patch fixes this by preconfiguring the GPIO on Sealevel
>> cards to the value (0x00) used prior to commit 7dea8165f1d
>>
>> Fixes: 7dea8165f1d ("serial: exar: Preconfigure xr17v35x MPIOs as output")
>> Signed-off-by: Matthew Howell <mrhowel@g.clemson.edu>
>> ---
>>
>> This is a revised patch submission based on comments received on
>> the previous submission.
>> See https://www.spinics.net/lists/linux-serial/msg39348.html
>>
>> I am using a different email address to address the email footer issue,
>> and I have attempted to fix the formatting issues.
>
> The footer issues are fixed, but you should probably change the from:
> and signed-off-by to your company address, right?
>

That would be optimal, yes. However, I don't have direct control over
the footer as it is enforced by our email server / group policy. Let
me know if the company email is *required* to be in the from: field
for this patch to be accepted though I will see if there is any way I
can get an exemption in this case.

>>
>> Summary/justification of the patch is below.
>>
>> With GPIOs preconfigured as per commit 7dea8165f1d all ports on Sealevel
>> XR17V35X based devices become stuck in high impedance mode, regardless of
>> dip-switch or software configuration. This causes the device to become
>> effectively unusable. This patch (in various forms) has been distributed
>> to our customers and no issues related to it have been reported.
>
> Why not put that paragraph in the changelog as well?

It is my understanding that the message above signed-off-by is
included as the commit message and should be as short as possible,
while additional information and justification is provided below the
sign-off-by line. Is that not the case? If it is preferable to be
above signed-off-by line I can move it to there.

>
>>
>> Let me know if any changes need to be made.
>>
>> --- linux/drivers/tty/serial/8250/8250_exar.c.orig    2020-07-09 11:05:03.920060577 -0400
>> +++ linux/drivers/tty/serial/8250/8250_exar.c    2020-07-09 11:05:25.275891627 -0400
>> @@ -326,7 +326,7 @@ static void setup_gpio(struct pci_dev *p
>>       * devices will export them as GPIOs, so we pre-configure them safely
>>       * as inputs.
>>       */
>> -    u8 dir = pcidev->vendor == PCI_VENDOR_ID_EXAR ? 0xff : 0x00;
>> +    u8 dir = (pcidev->vendor == PCI_VENDOR_ID_EXAR && pcidev->subsystem_vendor != PCI_VENDOR_ID_SEALEVEL) ? 0xff : 0x00;
>
> That's a horrible line to try to read now, right?
>
> Why not turn it into a real if statement so we can make more sense of it
> over time:
>
>     u8 dir = 0x00;
>
>     if ((pcidev->vendor == PCI_VENDOR_ID_EXAR) &&
>         (pcidev->subsystem_vendor != PCI_VENDOR_ID_SEALEVEL))
>         dir = 0xff;
>
> Looks better, right?
>
> thanks,
>
> greg k-h

Thanks for that feedback. It must have been unclear since the value of
dir in your if statement has the wrong value. Revised patch diff with
added comments is below.

--- linux/drivers/tty/serial/8250/8250_exar.c.orig    2020-07-09 11:05:03.920060577 -0400
+++ linux/drivers/tty/serial/8250/8250_exar.c    2020-07-13 11:54:44.386718167 -0400
@@ -326,7 +326,20 @@ static void setup_gpio(struct pci_dev *p
      * devices will export them as GPIOs, so we pre-configure them safely
      * as inputs.
      */
-    u8 dir = pcidev->vendor == PCI_VENDOR_ID_EXAR ? 0xff : 0x00;
+
+    u8 dir = 0x00;
+
+    if  ((pcidev->vendor == PCI_VENDOR_ID_EXAR) &&
+        (pcidev->subsystem_vendor != PCI_VENDOR_ID_SEALEVEL))
+    {
+       // Configure GPIO as inputs for Commtech adapters
+       dir = 0xff;
+    }
+    else
+    {
+       // Configure GPIO as outputs for SeaLevel adapters
+       dir = 0x00;
+    }
 
     writeb(0x00, p + UART_EXAR_MPIOINT_7_0);
     writeb(0x00, p + UART_EXAR_MPIOLVL_7_0);

--

Thanks,
Matthew Howell