Re: [PATCH 1/5] usb: xhci: add firmware loader for uPD720201 and uPD720202 w/o ROM

[Christian Lamparter <chunkeey@gmail.com>]

On Thursday, June 20, 2019 7:03:58 PM CEST Vinod Koul wrote:
> On 20-06-19, 14:19, Greg Kroah-Hartman wrote:
> > On Thu, Jun 20, 2019 at 03:51:50PM +0530, Vinod Koul wrote:
> > > From: Christian Lamparter <chunkeey@googlemail.com>
> > > 
> > > This patch adds a firmware loader for the uPD720201K8-711-BAC-A
> > > and uPD720202K8-711-BAA-A variant. Both of these chips are listed
> > > in Renesas' R19UH0078EJ0500 Rev.5.00 "User's Manual: Hardware" as
> > > devices which need the firmware loader on page 2 in order to
> > > work as they "do not support the External ROM".
> > > 
> > > The "Firmware Download Sequence" is describe in chapter
> > > "7.1 FW Download Interface" R19UH0078EJ0500 Rev.5.00 page 131.
> > > 
> > > The firmware "K2013080.mem" is available from a USB3.0 Host to
> > > PCIe Adapter (PP2U-E card) "Firmware download" archive. An
> > > alternative version can be sourced from Netgear's WNDR4700 GPL
> > > archives.
> > > 
> > > The release notes of the PP2U-E's "Firmware Download" ver 2.0.1.3
> > > (2012-06-15) state that the firmware is for the following devices:
> > >  - uPD720201 ES 2.0 sample whose revision ID is 2.
> > >  - uPD720201 ES 2.1 sample & CS sample & Mass product, ID is 3.
> > >  - uPD720202 ES 2.0 sample & CS sample & Mass product, ID is 2.
> > > 
> > > If someone from Renesas is listening: It would be great, if these
> > > firmwares could be added to linux-firmware.git.
> > 
> > That paragraph does not need to be in the changelog :)
> 
> Sure will drop :)

... those this mean that there is a firmware now? Do you have a link to it?

> 
> > >  #include <linux/slab.h>
> > >  #include <linux/module.h>
> > >  #include <linux/acpi.h>
> > > +#include <linux/firmware.h>
> > > +#include <asm/unaligned.h>
> > 
> > asm/ in a driver?  Are you sure???
> 
> Not sure :D, will check and remove

I think, as long as there is a "get_unaligned_le16" defined somewhere
it should be fine.

This was a loong ago, the loader was developped on a PowerPC 464, but
from what I remember it was checkpatch that didn't like the "unaligned"
poking around in the firmware below.

> > > +static int renesas_fw_download_image(struct pci_dev *dev,
> > > +				     const u32 *fw,
> > > +				     size_t step)
> > > +{
> > > +	size_t i;
> > > +	int err;
> > > +	u8 fw_status;
> > > +	bool data0_or_data1;
> > > +
> > > +	/*
> > > +	 * The hardware does alternate between two 32-bit pages.
> > > +	 * (This is because each row of the firmware is 8 bytes).
> > > +	 *
> > > +	 * for even steps we use DATA0, for odd steps DATA1.
> > > +	 */
> > > +	data0_or_data1 = (step & 1) == 1;
> > > +
> > > +	/* step+1. Read "Set DATAX" and confirm it is cleared. */
> > > +	for (i = 0; i < 10000; i++) {
> > > +		err = pci_read_config_byte(dev, 0xF5, &fw_status);
> > > +		if (err)
> > > +			return pcibios_err_to_errno(err);
> > > +		if (!(fw_status & BIT(data0_or_data1)))
> > > +			break;
> > > +
> > > +		udelay(1);
> > > +	}
> > > +	if (i == 10000)
> > > +		return -ETIMEDOUT;
> > > +
> > > +	/*
> > > +	 * step+2. Write FW data to "DATAX".
> > > +	 * "LSB is left" => force little endian
> > > +	 */
> > > +	err = pci_write_config_dword(dev, data0_or_data1 ? 0xFC : 0xF8,
> > > +				     (__force u32) cpu_to_le32(fw[step]));
> > > +	if (err)
> > > +		return pcibios_err_to_errno(err);
> > > +
> > > +	udelay(100);
> > > +
> > > +	/* step+3. Set "Set DATAX". */
> > > +	err = pci_write_config_byte(dev, 0xF5, BIT(data0_or_data1));
> > > +	if (err)
> > > +		return pcibios_err_to_errno(err);
> > > +
> > 
> > Shouldn't you just do a read after the write to be sure the write
> > actually went out on the wire?  Then you shouldn't have to do the
> > udelay, right?
> 
> Well I am not sure that is how it works. The register is a DATA register
> on the controller. We are writing to the memory of the controller here
> and after writing DATA0 and DATA1 we check the Set DATA0 & Set DATA1
> bits and write subsequenly only when controller is ready to accept more
> data.
> 
> I do recall at least for ROM load (writing to NOR flash attached to
> controller), we need to wait considerably more before the SetData0/1 was
> set and ready for subsequent write

OffTopic: There's some leeway here. From what I remember you could just push
the data through DATA0 and cut down on the logic. But this was slower than
using both DATA0 and DATA1.

The udelay was placed because I vaguely remember that polling SET DATA0
over and over slowed down the firmware download.
So the intention was to have the 100µs as a baseline and then we don't
slow down and waste more cycles in "step+1".

> 
> > > +static int renesas_hw_check_run_stop_busy(struct pci_dev *pdev)
> > > +{
> > > +#if 0
> > > +	u32 val;
> > > +
> > > +	/*
> > > +	 * 7.1.3 Note 3: "... must not set 'FW Download Enable' when
> > > +	 * 'RUN/STOP' of USBCMD Register is set"
> > > +	 */
> > > +	val = readl(hcd->regs + 0x20);
> > > +	if (val & BIT(0)) {
> > > +		dev_err(&pdev->dev, "hardware is busy and can't receive a FW.");
> > > +		return -EBUSY;
> > > +	}
> > > +#endif
> > > +	return 0;
> > > +}
> > > +
> > 
> > Is this function still really needed anymore?
> 
> Nope I will drop it unless Christian objects

You can drop it. From what I remember it was used for a minimal backup
solution that would simply prevent stuck the xhci-pci modules. (never
heard from Greg or Filipe)

> > > +	/*
> > > +	 * 11. After finishing writing the last data of FW, the
> > > +	 * System Software must clear "FW Download Enable"
> > > +	 */
> > > +	err = pci_write_config_byte(pdev, 0xF4, 0);
> > > +	if (err)
> > > +		return pcibios_err_to_errno(err);
> > > +
> > > +	/* 12. Read "Result Code" and confirm it is good. */
> > > +	for (i = 0; i < 10000; i++) {
> > > +		err = pci_read_config_byte(pdev, 0xF4, &fw_status);
> > > +		if (err)
> > > +			return pcibios_err_to_errno(err);
> > > +		if (fw_status & BIT(4))
> > > +			break;
> > > +
> > > +		udelay(1);
> > > +	}
> > 
> > 1000 reads???  I've heard of having to read a few times to ensure
> > something "latched" in the device, but not 1000.  Why so many?
> 
> For ROM load it did need significant time, I will check if we can go down
> to 100 here
yes, it takes a while! Though you could use a bigger udelay here and do
less retries. 

> > > +	if (i == 10000) {
> > > +		/* Timed out / Error - let's see if we can fix this */
> > > +		err = renesas_fw_check_running(pdev);
> > > +		switch (err) {
> > > +		case 0: /*
> > > +			 * we shouldn't end up here.
> > > +			 * maybe it took a little bit longer.
> > > +			 * But all should be well?
> > > +			 */
> > > +			break;
> > > +
> > > +		case 1: /* (No result yet? - we can try to retry) */
> > > +			if (retry_counter < 10) {
> > > +				retry_counter++;
> > > +				dev_warn(&pdev->dev, "Retry Firmware download: %d try.",
> > > +					  retry_counter);
> > > +				return renesas_fw_download(pdev, fw,
> > > +							   retry_counter);
> > 
> > recursion?
> 
> I didnt encounter the need, we should remove it unless Christian objects

Sure, I think it should be safe to just say that there was a timeout and then abort.

Cheers,
Christian