[U-Boot] [PATCH] net: asix: add support for AX88772B
Marek Vasut
marek.vasut at gmail.com
Mon Aug 20 17:15:16 CEST 2012
Dear Lucas Stach,
> Hello Marek,
>
> Am Montag, den 20.08.2012, 16:13 +0200 schrieb Marek Vasut:
> > Dear Lucas Stach,
> >
> > > There are multiple changes needed to make AX88772B work:
> > >
> > > 1. add vendor and product ID (trivial)
> > >
> > > 2. We need to read out the MAC address from the EEPROM and write
> > > it into the NodeID register manually.
> > >
> > > 3. The packet length check has to be adjusted, as all ASIX chips
> > > only use 11 bits to indicate the length. AX88772B uses the other
> > > bits to indicate unrelated things, which cause the check to fail.
> > > This fix is based on a fix for the Linux kernel by Marek Vasut.
> > > Linux upstream commit: bca0beb9363f8487ac902931a50eb00180a2d14a
> >
> > Yea, I enjoyed my share with this one. Though the linux patch still
> > doesn't work on my hardware completely ... I get some random trouble. Do
> > you get any mess in dmesg?
>
> Currently I'm mostly doing U-Boot work, so I didn't test Linux more than
> some simple boot tests.
Ok, it might have been just my hardware though.
> > > 4. AX88772B provides several bulk endpoints. Only the first
> > > IN/OUT endpoints work in the default configuration. So stop
> > > enumeration after we found them to avoid overwriting the
> > > endpoint config with a non-working one.
> >
> > ;-)
> >
> > > This was tested to work on a Colibri T20 board.
> > > Patch is based on u-boot-net/next.
> > >
> > > Signed-off-by: Lucas Stach <dev at lynxeye.de>
> > > ---
> > >
> > > drivers/usb/eth/asix.c | 62
> > >
> > > +++++++++++++++++++++++++++++++++++++++++--------- 1 Datei geändert, 51
> > > Zeilen hinzugefügt(+), 11 Zeilen entfernt(-)
> > >
> > > diff --git a/drivers/usb/eth/asix.c b/drivers/usb/eth/asix.c
> > > index 8fb7fc8..86e0f0f 100644
> > > --- a/drivers/usb/eth/asix.c
> > > +++ b/drivers/usb/eth/asix.c
> > > @@ -31,10 +31,12 @@
> > >
> > > #define AX_CMD_READ_MII_REG 0x07
> > > #define AX_CMD_WRITE_MII_REG 0x08
> > > #define AX_CMD_SET_HW_MII 0x0a
> > >
> > > +#define AX_CMD_READ_EEPROM 0x0b
> > >
> > > #define AX_CMD_READ_RX_CTL 0x0f
> > > #define AX_CMD_WRITE_RX_CTL 0x10
> > > #define AX_CMD_WRITE_IPG0 0x12
> > > #define AX_CMD_READ_NODE_ID 0x13
> > >
> > > +#define AX_CMD_WRITE_NODE_ID 0x14
> > >
> > > #define AX_CMD_READ_PHY_ID 0x19
> > > #define AX_CMD_WRITE_MEDIUM_MODE 0x1b
> > > #define AX_CMD_WRITE_GPIOS 0x1f
> > >
> > > @@ -322,6 +324,7 @@ static int asix_init(struct eth_device *eth, bd_t
> > > *bd)
> > >
> > > int timeout = 0;
> > >
> > > #define TIMEOUT_RESOLUTION 50 /* ms */
> > >
> > > int link_detected;
> > >
> > > + int i;
> > >
> > > debug("** %s()\n", __func__);
> > >
> > > @@ -359,6 +362,35 @@ static int asix_init(struct eth_device *eth, bd_t
> > > *bd)
> > >
> > > rx_ctl = asix_read_rx_ctl(dev);
> > > debug("RX_CTL is 0x%04x setting to 0x0000\n", rx_ctl);
> > >
> > > + /*
> > > + * For AX88772B to work we have to read the MAC address from the
> > > EEPROM + * and set the node ID manually.
> > > + */
> > > + if ((dev->pusb_dev->descriptor.idVendor == 0x0b95) &&
> > > + (dev->pusb_dev->descriptor.idProduct == 0x772b)) {
> > > +
> > > + ALLOC_CACHE_ALIGN_BUFFER(unsigned char, read_buf, 2);
> > > + memset(buf, 0, ETH_ALEN);
> > > +
> > > + for (i = 0; i < (ETH_ALEN >> 1); i++) {
> > > + memset(read_buf, 0, 2);
> > > + if (asix_read_cmd(dev, AX_CMD_READ_EEPROM,
> > > + 0x04 + i, 0, 2, read_buf)) {
> > > + debug("Failed to read SROM address 04h.\n");
> > > + goto out_err;
> > > + }
> > > + memcpy((buf + i*2), read_buf, 2);
> > > + }
> > > +
> > > + debug("MAC read from EEPROM: %02x:%02x:%02x:%02x:%02x:%02x\n",
> > > + buf[0], buf[1], buf[2], buf[3], buf[4], buf[5]);
> > > +
> > > + if (asix_write_cmd(dev, AX_CMD_WRITE_NODE_ID, 0, 0, ETH_ALEN,
> >
> > buf)) {
> >
> > > + printf("Failed to set MAC address.\n");
> > > + goto out_err;
> > > + }
> > > + }
> >
> > Can this be split into separate function?
>
> I thought about this, but decided not to do so as I felt it doesn't do
> much harm in there. But I can certainly change this if you think it
> really should be split out.
It should, since there'll eventually land other such cards with this issue or
other issues.
> > > /* Get the MAC address */
> > > if (asix_read_cmd(dev, AX_CMD_READ_NODE_ID,
> > >
> > > 0, 0, ETH_ALEN, buf) < 0) {
> > >
> > > @@ -493,13 +525,13 @@ static int asix_recv(struct eth_device *eth)
> > >
> > > }
> > > memcpy(&packet_len, buf_ptr, sizeof(packet_len));
> > > le32_to_cpus(&packet_len);
> > >
> > > - if (((packet_len >> 16) ^ 0xffff) != (packet_len & 0xffff)) {
> > > + if (((~packet_len >> 16) & 0x7ff) != (packet_len & 0x7ff)) {
> > >
> > > debug("Rx: malformed packet length: %#x (%#x:%#x)\n",
> > >
> > > - packet_len, (packet_len >> 16) ^ 0xffff,
> > > - packet_len & 0xffff);
> > > + packet_len, (~packet_len >> 16) & 0x7ff,
> > > + packet_len & 0x7ff);
> > >
> > > return -1;
> > >
> > > }
> > >
> > > - packet_len = packet_len & 0xffff;
> > > + packet_len = packet_len & 0x7ff;
> > >
> > > if (packet_len > actual_len - sizeof(packet_len)) {
> > >
> > > debug("Rx: too large packet: %d\n", packet_len);
> > > return -1;
> > >
> > > @@ -546,6 +578,7 @@ static struct asix_dongle asix_dongles[] = {
> > >
> > > { 0x13b1, 0x0018 }, /* Linksys 200M v2.1 */
> > > { 0x1557, 0x7720 }, /* 0Q0 cable ethernet */
> > > { 0x2001, 0x3c05 }, /* DLink DUB-E100 H/W Ver B1 Alternate */
> > >
> > > + { 0x0b95, 0x772b }, /* ASIX 88772B */
> >
> > Maybe this list should be extended by caps as it is in linux?
>
> Hm, I don't know what that would buy us. It's not like we are trying to
> make use of the maximum feature set of each chip to get the last bit of
> performance. IMHO it's better to just use a safe subset of the features
> to keep this driver small and understandable.
That's right, but this chip is really weird, right? It's not like the others, so
we should flag it behaves slightly different. What do you think?
> > > { 0x0000, 0x0000 } /* END - Do not remove */
> > >
> > > };
> > >
> > > @@ -555,6 +588,7 @@ int asix_eth_probe(struct usb_device *dev, unsigned
> > > int ifnum, {
> > >
> > > struct usb_interface *iface;
> > > struct usb_interface_descriptor *iface_desc;
> > >
> > > + int ep_in_found = 0, ep_out_found = 0;
> > >
> > > int i;
> > >
> > > /* let's examine the device now */
> > >
> > > @@ -591,13 +625,19 @@ int asix_eth_probe(struct usb_device *dev,
> > > unsigned int ifnum, /* is it an BULK endpoint? */
> > >
> > > if ((iface->ep_desc[i].bmAttributes &
> > >
> > > USB_ENDPOINT_XFERTYPE_MASK) == USB_ENDPOINT_XFER_BULK) {
> > >
> > > - if (iface->ep_desc[i].bEndpointAddress & USB_DIR_IN)
> > > - ss->ep_in = iface->ep_desc[i].bEndpointAddress &
> > > - USB_ENDPOINT_NUMBER_MASK;
> > > - else
> > > - ss->ep_out =
> > > - iface->ep_desc[i].bEndpointAddress &
> > > - USB_ENDPOINT_NUMBER_MASK;
> > > + if (iface->ep_desc[i].bEndpointAddress & USB_DIR_IN) {
> > > + if (!ep_in_found) {
> > > + ss->ep_in = iface-
> > >
> > >ep_desc[i].bEndpointAddress &
> > >
> > > + USB_ENDPOINT_NUMBER_MASK;
> > > + ep_in_found = 1;
> > > + }
> > > + } else {
> > > + if (!ep_out_found) {
> > > + ss->ep_out = iface-
> > >
> > >ep_desc[i].bEndpointAddress &
> > >
> > > + USB_ENDPOINT_NUMBER_MASK;
> > > + ep_out_found = 1;
> > > + }
> > > + }
> > >
> > > }
> > >
> > > /* is it an interrupt endpoint? */
> >
> > Best regards,
> > Marek Vasut
More information about the U-Boot
mailing list