Re: [PATCH 4/5] PCI: tegra: Add Tegra264 support

[Bjorn Helgaas <helgaas@kernel.org>]

On Fri, Mar 20, 2026 at 11:34:53AM +0100, Thierry Reding wrote:
> On Thu, Mar 19, 2026 at 12:46:39PM -0500, Bjorn Helgaas wrote:
> > On Thu, Mar 19, 2026 at 05:01:08PM +0100, Thierry Reding wrote:
> > > From: Thierry Reding <treding@nvidia.com>
> > > 
> > > Add a driver for the PCIe controller found on NVIDIA Tegra264 SoCs. The
> > > driver is very small, with its main purpose being to set up the address
> > > translation registers and then creating a standard PCI host using ECAM.

> > > +#define PCIE_LINK_UP_DELAY	10000	/* 10 msec */
> > > +#define PCIE_LINK_UP_TIMEOUT	1000000	/* 1 s */
> > 
> > Use something from drivers/pci/pci.h if possible.  If not, please add
> > units suffixes to the name, e.g., it looks like these are in "_US".
> > 
> > PCIE_LINK_WAIT_MAX_RETRIES and PCIE_LINK_WAIT_SLEEP_MS look like
> > they're used in similar ways in other drivers.
> 
> I see a bunch of implementations that use custom defines, if you don't
> mind I'll have a go and unifying this a little. Most implementations
> seem to use usleep_range(90000, 100000) with 10 retries.

That would be awesome!

> > > +		err = gpiod_to_irq(pcie->wake_gpio);
> > > +		if (err < 0) {
> > > +			dev_err(pcie->dev, "failed to get wake IRQ: %d\n", err);
> > 
> > Does %pe work here (and below)?
> > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/core-api/printk-formats.rst?id=v6.19#n96
> 
> It won't work in this particular case because %pe needs an ERR_PTR-
> encoded error code. For the other messages it's probably better to move
> to dev_err_probe() as Krzysztof suggested.

We could use dev_err(dev, "..., %pe\n", ERR_PTR(err)).  It's kind of ugly,
but seems like a fairly common style.  I wish we had a simpler way to do
this; it seems like a shame to print a negative number when we have the
symbols available.

Bjorn