On Mon, Aug 12, 2019 at 12:45 PM Luis de Oliveira
[off-list ref] wrote:

From: Andy Shevchenko <redacted>
Date: Sat, Aug 10, 2019 at 14:09:21

quoted

On Fri, Aug 9, 2019 at 5:38 PM Sakari Ailus [off-list ref] wrote:

quoted

On Tue, Jun 11, 2019 at 09:20:51PM +0200, Luis Oliveira wrote:

quoted

quoted

quoted

+     of_id = of_match_node(dw_mipi_csi_of_match, dev->of_node);
+     if (!of_id)
+             return -EINVAL;

Is it possible to have this asserted?

I will remove it.

But please double check that is really the case.

quoted

quoted

quoted

+     ret = devm_request_irq(dev, csi->ctrl_irq_number,
+                            dw_mipi_csi_irq1, IRQF_SHARED,
+                            dev_name(dev), csi);
+     if (ret) {
+             dev_err(dev, "irq csi %s failed\n", of_id->name);
+
+             goto end;
+     }

devm_*irq() might be a bad idea. Is it race free in your driver?

I never thought about it like that. Should I use request_irq and
free_irq?

It's you, author of the driver, who knows it better :-)

quoted

I told once, can repeat again. Synopsys perhaps needs better reviews
inside company. Each time I see the code, it repeats same mistakes
over and over. Have you, guys, do something about it?

We are working on it. It will get better, sorry.

Thanks! Hope to see a progress!

-- 
With Best Regards,
Andy Shevchenko