Thread: Re: [RTnet-developers] arcnet driver

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Jan,

Thanks for such a quick response! We've made good progress with 
the RTnet Arcnet driver but have a few questions about code which 
doesn't quite match up to the existing examples.

And I have little experience with designing network drivers (or 
any driver!) .. therefore the questions might seem basic.

1.) The original arcnet driver does not call alloc_etherdev but 
rather alloc_netdev (shown below). Can I still change this 
function call like suggested in your porting documenation? What 
effect will this have on the program?

BEFORE

struct net_device *alloc_arcdev(char *name)
{
	struct net_device *dev;

	dev = alloc_netdev(sizeof(struct arcnet_local),
			   name && *name ? name : "arc%d", arcdev_setup);
	if(dev) {
		struct arcnet_local *lp = (struct arcnet_local *) dev->priv;
		spin_lock_init(&lp->lock);
	}

	return dev;
}

AFTER

struct rtnet_device *alloc_arcdev(char *name)
{
	struct rtnet_device *dev;

	/*** RTnet ***/
	dev = rt_alloc_etherdev(sizeof(struct arcnet_local)); // 
etherdev?

	rtdev_alloc_name(dev, "rtarc%d");
	rt_rtdev_connect(dev, &RTDEV_manager);
	RTNET_SET_MODULE_OWNER(dev);
	dev->vers = RTDEV_VERS_2_0;
	/*** RTnet ***/

	if(dev) {
		struct arcnet_local *lp = (struct arcnet_local *) dev->priv;
		rtdm_lock_init(&lp->lock); /*** RTnet ***/
	}

	return dev;
}

2.) In your example, you initialize the skb_pool in the probe 
function and if the return value is less than twice the ring size 
then you release the skb_pool and call a few 'cleanup' functions.

e.g.     pci_free_consistent(lp->pci_dev, sizeof(*lp), lp, 
lp->dma_addr);

    I'm not sure how it works, but I cannot this function because 
the struct arcnet_local does not have a field called 'dma_addr'. 
So my question is .. is dma_addr disguised under another name in 
the arcnet_local struct? and if not do I even need to call this 
function?

static int __devinit com20020pci_probe(struct pci_dev *pdev, const 
struct pci_device_id *id)
{
	struct rtnet_device *dev;
	struct arcnet_local *lp;
	int ioaddr, err;

	/*** RTnet ***/

	cards_found++;
    	if (cards[cards_found] == 0)
     	return -ENODEV;

	/*** RTnet ***/

	if (pci_enable_device(pdev))
		return -EIO;
	dev = alloc_arcdev(device);
	if (!dev)
		return -ENOMEM;
	lp = dev->priv;

	pci_set_drvdata(pdev, dev);

	// SOHARD needs PCI base addr 4
	if (pdev->vendor==0x10B5) {
		BUGMSG(D_NORMAL, "SOHARD\n");
		ioaddr = pci_resource_start(pdev, 4);
	}
	else {
		BUGMSG(D_NORMAL, "Contemporary Controls\n");
		ioaddr = pci_resource_start(pdev, 2);
	}

	if (!request_region(ioaddr, ARCNET_TOTAL_SIZE, "com20020-pci")) 
{
		BUGMSG(D_INIT, "IO region %xh-%xh already allocated.\n",
		       ioaddr, ioaddr + ARCNET_TOTAL_SIZE - 1);
		err = -EBUSY;
		goto out_dev;
	}

	/*** RTnet ***/
	if (rtskb_pool_init(&lp->skb_pool, RX_RING_SIZE*2) < 
RX_RING_SIZE*2) {
		rtskb_pool_release(&lp->skb_pool);
--------->   // pci_free_consistent(lp->pci_dev, sizeof(*lp), lp, 
lp->dma_addr);
		release_region(ioaddr, ARCNET_TOTAL_SIZE);
		rtdev_free(dev);
     	return -ENOMEM;
        }
	/*** RTnet ***/

	// Dummy access after Reset
	// ARCNET controller needs this access to detect bustype
	outb(0x00,ioaddr+1);
	inb(ioaddr+1);

	dev->base_addr = ioaddr;
	dev->irq = pdev->irq;
	dev->dev_addr[0] = node;
	lp->card_name = "PCI COM20020";
	lp->card_flags = id->driver_data;
	lp->backplane = backplane;
	lp->clockp = clockp & 7;
	lp->clockm = clockm & 3;
	lp->timeout = timeout;
	lp->hw.owner = THIS_MODULE;

	if (ASTATUS() == 0xFF) {
		BUGMSG(D_NORMAL, "IO address %Xh was reported by PCI BIOS, "
		       "but seems empty!\n", ioaddr);
		err = -EIO;
		goto out_port;
	}

	if (com20020_check(dev)) {
		err = -EIO;
		goto out_port;
	}

	if ((err = com20020_found(dev, SA_SHIRQ)) != 0)
	        goto out_port;

	return 0;

out_port:
	release_region(ioaddr, ARCNET_TOTAL_SIZE);
out_dev:
	rtdev_free(dev);
	return err;
}

     here's a snapshot of the arcnet_local struct ...

struct arcnet_local {
	struct net_device_stats stats;

	uint8_t config,		/* current value of CONFIG register */
		timeout,	/* Extended timeout for COM20020 */
		backplane,	/* Backplane flag for COM20020 */
		clockp,		/* COM20020 clock divider */
		clockm,		/* COM20020 clock multiplier flag */
		setup,		/* Contents of setup1 register */
		setup2,		/* Contents of setup2 register */
		intmask;	/* current value of INTMASK register */
	uint8_t default_proto[256];	/* default encap to use for each host 
*/
	int	cur_tx,		/* buffer used by current transmit, or -1 */
		next_tx,	/* buffer where a packet is ready to send */
		cur_rx;		/* current receive buffer */
	int	lastload_dest,	/* can last loaded packet be acked? */
		lasttrans_dest;	/* can last TX'd packet be acked? */
	int	timed_out;	/* need to process TX timeout and drop packet */
	unsigned long last_timeout;	/* time of last reported timeout */
	char *card_name;	/* card ident string */
	int card_flags;		/* special card features */

	/*** RTnet ***/

        struct rtskb *tx_skbuff[TX_RING_SIZE];
        struct rtskb *rx_skbuff[RX_RING_SIZE];
        struct rtskb_queue skb_pool;

        rtdm_irq_t irq_handle;

	/*** RTnet ***/

	/* On preemtive and SMB a lock is needed */

	rtdm_lock_t lock;

	/*
	 * Buffer management: an ARCnet card has 4 x 512-byte buffers, 
each of
	 * which can be used for either sending or receiving.  The new 
dynamic
	 * buffer management routines use a simple circular queue of 
available
	 * buffers, and take them as they're needed.  This way, we 
simplify
	 * situations in which we (for example) want to pre-load a 
transmit
	 * buffer, or start receiving while we copy a received packet to
	 * memory.
	 *
	 * The rules: only the interrupt handler is allowed to _add_ 
buffers to
	 * the queue; thus, this doesn't require a lock.  Both the 
interrupt
	 * handler and the transmit function will want to _remove_ 
buffers, so
	 * we need to handle the situation where they try to do it at the 
same
	 * time.
	 *
	 * If next_buf == first_free_buf, the queue is empty.  Since 
there are
	 * only four possible buffers, the queue should never be full.
	 */
	atomic_t buf_lock;
	int buf_queue[5];
	int next_buf, first_free_buf;

	/* network "reconfiguration" handling */
	time_t first_recon,		/* time of "first" RECON message to count 
*/
		last_recon;		/* time of most recent RECON */
	int num_recons;		/* number of RECONs between first and last. */
	bool network_down;		/* do we think the network is down? */

	bool excnak_pending;	/* We just got an excesive nak interrupt */

	struct {
		uint16_t sequence;	/* sequence number (incs with each packet) 
*/
		uint16_t aborted_seq;

		struct Incoming incoming[256];	/* one from each address */
	} rfc1201;

	/* really only used by rfc1201, but we'll pretend it's not */
	struct Outgoing outgoing;	/* packet currently being sent */

	/* hardware-specific functions */
	struct {
		struct module *owner;
		void (*command) (struct rtnet_device * dev, int cmd);
		int (*status) (struct rtnet_device * dev);
		void (*intmask) (struct rtnet_device * dev, int mask);
		bool (*reset) (struct rtnet_device * dev, bool really_reset);
		void (*open) (struct rtnet_device * dev);
		void (*close) (struct rtnet_device * dev);

		void (*copy_to_card) (struct rtnet_device * dev, int bufnum, int 
offset,
				      void *buf, int count);
		void (*copy_from_card) (struct rtnet_device * dev, int bufnum, 
int offset,
					void *buf, int count);
	} hw;

	void __iomem *mem_start;	/* pointer to ioremap'ed MMIO */
};

     and if you're curious com20020_found looks like this ...

/* Set up the struct rtnet_device associated with this card.  
Called after
 * probing succeeds.
 */
int com20020_found(struct rtnet_device *dev, int shared)
{
	struct arcnet_local *lp;
	int ioaddr = dev->base_addr;
	int retval;

	/* Initialize the rest of the device structure. */

	lp = dev->priv;

	lp->hw.owner = THIS_MODULE;
	lp->hw.command = com20020_command;
	lp->hw.status = com20020_status;
	lp->hw.intmask = com20020_setmask;
	lp->hw.reset = com20020_reset;
	lp->hw.copy_to_card = com20020_copy_to_card;
	lp->hw.copy_from_card = com20020_copy_from_card;
	lp->hw.close = com20020_close;

	// dev->set_multicast_list = com20020_set_mc_list; /*** RTNET 
***/

	if (!dev->dev_addr[0])
		dev->dev_addr[0] = inb(ioaddr + BUS_ALIGN*8);	/* FIXME: do this 
some other way! (not written by me - Chris) */

	SET_SUBADR(SUB_SETUP1);
	outb(lp->setup, _XREG);

	if (lp->card_flags & ARC_CAN_10MBIT)
	{
		SET_SUBADR(SUB_SETUP2);
		outb(lp->setup2, _XREG);

		/* must now write the magic "restart operation" command */
		mdelay(1);
		outb(0x18, _COMMAND);
	}

	lp->config = 0x20 | (lp->timeout << 3) | (lp->backplane << 2) | 
1;
	/* Default 0x38 + register: Node ID */
	SETCONF;
	outb(dev->dev_addr[0], _XREG);

	/*** RTnet ***/

	/* reserve the irq */
	/*if (request_irq(dev->irq, &arcnet_interrupt, shared,
			"arcnet (COM20020)", dev)) {
		BUGMSG(D_NORMAL, "Can't get IRQ %d!\n", dev->irq);
		return -ENODEV;
	}*/

	if (dev->irq == 0)
     	return -EAGAIN;

	rt_stack_connect(dev, &STACK_manager);
	retval = rtdm_irq_request(&lp->irq_handle, dev->irq, 
arcnet_interrupt, 0,
                         "rt_arcnet", dev);
	if (retval)
     	return retval;

	rtdm_irq_enable(&lp->irq_handle);

	/*** RTnet ***/

	dev->base_addr = ioaddr;

	BUGMSG(D_NORMAL, "%s: station %02Xh found at %03lXh, IRQ %d.\n",
	       lp->card_name, dev->dev_addr[0], dev->base_addr, 
dev->irq);

	if (lp->backplane)
		BUGMSG(D_NORMAL, "Using backplane mode.\n");

	if (lp->timeout != 3)
		BUGMSG(D_NORMAL, "Using extended timeout value of %d.\n", 
lp->timeout);

	BUGMSG(D_NORMAL, "Using CKP %d - data rate %s.\n",
	       lp->setup >> 1,
	       clockrates[3 - ((lp->setup2 & 0xF0) >> 4) + ((lp->setup & 
0x0F) >> 1)]);

	/*** RTnet ***/

	/*if (register_netdev(dev)) {
		free_irq(dev->irq, dev);
		return -EIO;
	}*/

	if (rt_register_rtnetdev(dev))
     {
     	rtdm_irq_free(&lp->irq_handle);
     	return -EIO;
	}

	/*** RTnet ***/

	return 0;
}

3.) Ok. last question. It seems like I have a problem with module 
dependencies. The arcnet driver is set up to be loaded as four 
separate modules: arcnet.ko (the foundation), arc-rawmode.ko 
(provides rawmode encapsulation), com20020.ko (support for 
chipset), and com20020-pci.ko (support for pci-board).

I've done my homework and created four rt modules that compile and 
load into the kernel.. but the problem is that I cannot get them 
to load with 'rtnet start'. How do I specify dependencies in the 
'rtnet.conf' file? My dilemna is that if I specify com20020-pci 
without any extra work.. the program will not run because it 
cannot load the module without its dependencies.. BUT I cannot 
load rt_arcnet.ko and rt_com20020.ko before running 'rtnet start' 
because they both depend on the module rtnet!!

So I feel like I'm in a bit of a catch-22. Please tell me how to 
get around this problem.

Thank you so much for your help.

Chris Lightcap
University of Florida

On Tue Jul 11 04:03:33 EDT 2006, Jan Kiszka <jan...@we...> 
wrote:

> Chris Lightcap wrote:
>> Hello all,
>> 
>> I'd like to implement a real-time arcnet driver to communicate 
>> with
>> our mitsubishi PA-10 robot using RTAI and RTnet. I found in the
> 
> You are welcome!
> 
>> mailing list archive that there was some interest in this topic 
>> about
>> a year ago.
>> 
>> .. Has anyone made progress with a real-time arcnet driver? is it
> 
> I'm personally not aware of any effort.
> 
>> possible to simply modify the existing driver as explained in the
>> porting documentation? are there any additional problems I should
>> know of beforehand?
> 
> I think to remember once browsing through the Linux arcnet stack 
> without
> finding anything that's fundamentally different from the "normal"
> networking stack. So all the pattern of RTnet /should/ apply to 
> arcnet
> as well. If you encounter something that doesn't fit on first 
> sight,
> feel free to discuss it here.
> 
> Please build your port on top of latest RTnet and only use RTDM 
> for
> interfacing the real-time Linux services - this will help to 
> integrate
> your work into the official version later.
> 
> Jan
> 
> 

--
LIGHTCAP,CHRISTOPHER A

Thread: Re: [RTnet-developers] arcnet driver

rtnet-developers