Linux-2.6.12-rc2v2.6.12-rc2

Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.

Let it rip!

1 files changed, 513 insertions, 0 deletions

diff --git a/drivers/net/wireless/prism54/islpci_mgt.c b/drivers/net/wireless/prism54/islpci_mgt.c
new file mode 100644
index 000000000000..b6f2e5a223be
--- /dev/null
+++ b/drivers/net/wireless/prism54/islpci_mgt.c
@@ -0,0 +1,513 @@
+/*
+ *  
+ *  Copyright (C) 2002 Intersil Americas Inc.
+ *  Copyright 2004 Jens Maurer <Jens.Maurer@gmx.net>
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License
+ *
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program; if not, write to the Free Software
+ *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ *
+ */
+
+#include <linux/config.h>
+#include <linux/netdevice.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+
+#include <asm/io.h>
+#include <asm/system.h>
+#include <linux/if_arp.h>
+
+#include "prismcompat.h"
+#include "isl_38xx.h"
+#include "islpci_mgt.h"
+#include "isl_oid.h"		/* additional types and defs for isl38xx fw */
+#include "isl_ioctl.h"
+
+#include <net/iw_handler.h>
+
+/******************************************************************************
+        Global variable definition section
+******************************************************************************/
+int pc_debug = VERBOSE;
+module_param(pc_debug, int, 0);
+
+/******************************************************************************
+    Driver general functions
+******************************************************************************/
+#if VERBOSE > SHOW_ERROR_MESSAGES
+void
+display_buffer(char *buffer, int length)
+{
+	if ((pc_debug & SHOW_BUFFER_CONTENTS) == 0)
+		return;
+
+	while (length > 0) {
+		printk("[%02x]", *buffer & 255);
+		length--;
+		buffer++;
+	}
+
+	printk("\n");
+}
+#endif
+
+/*****************************************************************************
+    Queue handling for management frames
+******************************************************************************/
+
+/*
+ * Helper function to create a PIMFOR management frame header.
+ */
+static void
+pimfor_encode_header(int operation, u32 oid, u32 length, pimfor_header_t *h)
+{
+	h->version = PIMFOR_VERSION;
+	h->operation = operation;
+	h->device_id = PIMFOR_DEV_ID_MHLI_MIB;
+	h->flags = 0;
+	h->oid = cpu_to_be32(oid);
+	h->length = cpu_to_be32(length);
+}
+
+/*
+ * Helper function to analyze a PIMFOR management frame header.
+ */
+static pimfor_header_t *
+pimfor_decode_header(void *data, int len)
+{
+	pimfor_header_t *h = data;
+
+	while ((void *) h < data + len) {
+		if (h->flags & PIMFOR_FLAG_LITTLE_ENDIAN) {
+			le32_to_cpus(&h->oid);
+			le32_to_cpus(&h->length);
+		} else {
+			be32_to_cpus(&h->oid);
+			be32_to_cpus(&h->length);
+		}
+		if (h->oid != OID_INL_TUNNEL)
+			return h;
+		h++;
+	}
+	return NULL;
+}
+
+/*
+ * Fill the receive queue for management frames with fresh buffers.
+ */
+int
+islpci_mgmt_rx_fill(struct net_device *ndev)
+{
+	islpci_private *priv = netdev_priv(ndev);
+	isl38xx_control_block *cb =	/* volatile not needed */
+	    (isl38xx_control_block *) priv->control_block;
+	u32 curr = le32_to_cpu(cb->driver_curr_frag[ISL38XX_CB_RX_MGMTQ]);
+
+#if VERBOSE > SHOW_ERROR_MESSAGES
+	DEBUG(SHOW_FUNCTION_CALLS, "islpci_mgmt_rx_fill \n");
+#endif
+
+	while (curr - priv->index_mgmt_rx < ISL38XX_CB_MGMT_QSIZE) {
+		u32 index = curr % ISL38XX_CB_MGMT_QSIZE;
+		struct islpci_membuf *buf = &priv->mgmt_rx[index];
+		isl38xx_fragment *frag = &cb->rx_data_mgmt[index];
+
+		if (buf->mem == NULL) {
+			buf->mem = kmalloc(MGMT_FRAME_SIZE, GFP_ATOMIC);
+			if (!buf->mem) {
+				printk(KERN_WARNING
+				       "Error allocating management frame.\n");
+				return -ENOMEM;
+			}
+			buf->size = MGMT_FRAME_SIZE;
+		}
+		if (buf->pci_addr == 0) {
+			buf->pci_addr = pci_map_single(priv->pdev, buf->mem,
+						       MGMT_FRAME_SIZE,
+						       PCI_DMA_FROMDEVICE);
+			if (!buf->pci_addr) {
+				printk(KERN_WARNING
+				       "Failed to make memory DMA'able\n.");
+				return -ENOMEM;
+			}
+		}
+
+		/* be safe: always reset control block information */
+		frag->size = cpu_to_le16(MGMT_FRAME_SIZE);
+		frag->flags = 0;
+		frag->address = cpu_to_le32(buf->pci_addr);
+		curr++;
+
+		/* The fragment address in the control block must have
+		 * been written before announcing the frame buffer to
+		 * device */
+		wmb();
+		cb->driver_curr_frag[ISL38XX_CB_RX_MGMTQ] = cpu_to_le32(curr);
+	}
+	return 0;
+}
+
+/*
+ * Create and transmit a management frame using "operation" and "oid",
+ * with arguments data/length.
+ * We either return an error and free the frame, or we return 0 and
+ * islpci_mgt_cleanup_transmit() frees the frame in the tx-done
+ * interrupt.
+ */
+static int
+islpci_mgt_transmit(struct net_device *ndev, int operation, unsigned long oid,
+		    void *data, int length)
+{
+	islpci_private *priv = netdev_priv(ndev);
+	isl38xx_control_block *cb =
+	    (isl38xx_control_block *) priv->control_block;
+	void *p;
+	int err = -EINVAL;
+	unsigned long flags;
+	isl38xx_fragment *frag;
+	struct islpci_membuf buf;
+	u32 curr_frag;
+	int index;
+	int frag_len = length + PIMFOR_HEADER_SIZE;
+
+#if VERBOSE > SHOW_ERROR_MESSAGES
+	DEBUG(SHOW_FUNCTION_CALLS, "islpci_mgt_transmit\n");
+#endif
+
+	if (frag_len > MGMT_FRAME_SIZE) {
+		printk(KERN_DEBUG "%s: mgmt frame too large %d\n",
+		       ndev->name, frag_len);
+		goto error;
+	}
+
+	err = -ENOMEM;
+	p = buf.mem = kmalloc(frag_len, GFP_KERNEL);
+	if (!buf.mem) {
+		printk(KERN_DEBUG "%s: cannot allocate mgmt frame\n",
+		       ndev->name);
+		goto error;
+	}
+	buf.size = frag_len;
+
+	/* create the header directly in the fragment data area */
+	pimfor_encode_header(operation, oid, length, (pimfor_header_t *) p);
+	p += PIMFOR_HEADER_SIZE;
+
+	if (data)
+		memcpy(p, data, length);
+	else
+		memset(p, 0, length);
+
+#if VERBOSE > SHOW_ERROR_MESSAGES
+	{
+		pimfor_header_t *h = buf.mem;
+		DEBUG(SHOW_PIMFOR_FRAMES,
+		      "PIMFOR: op %i, oid 0x%08lx, device %i, flags 0x%x length 0x%x \n",
+		      h->operation, oid, h->device_id, h->flags, length);
+
+		/* display the buffer contents for debugging */
+		display_buffer((char *) h, sizeof (pimfor_header_t));
+		display_buffer(p, length);
+	}
+#endif
+
+	err = -ENOMEM;
+	buf.pci_addr = pci_map_single(priv->pdev, buf.mem, frag_len,
+				      PCI_DMA_TODEVICE);
+	if (!buf.pci_addr) {
+		printk(KERN_WARNING "%s: cannot map PCI memory for mgmt\n",
+		       ndev->name);
+		goto error_free;
+	}
+
+	/* Protect the control block modifications against interrupts. */
+	spin_lock_irqsave(&priv->slock, flags);
+	curr_frag = le32_to_cpu(cb->driver_curr_frag[ISL38XX_CB_TX_MGMTQ]);
+	if (curr_frag - priv->index_mgmt_tx >= ISL38XX_CB_MGMT_QSIZE) {
+		printk(KERN_WARNING "%s: mgmt tx queue is still full\n",
+		       ndev->name);
+		goto error_unlock;
+	}
+
+	/* commit the frame to the tx device queue */
+	index = curr_frag % ISL38XX_CB_MGMT_QSIZE;
+	priv->mgmt_tx[index] = buf;
+	frag = &cb->tx_data_mgmt[index];
+	frag->size = cpu_to_le16(frag_len);
+	frag->flags = 0;	/* for any other than the last fragment, set to 1 */
+	frag->address = cpu_to_le32(buf.pci_addr);
+
+	/* The fragment address in the control block must have
+	 * been written before announcing the frame buffer to
+	 * device */
+	wmb();
+	cb->driver_curr_frag[ISL38XX_CB_TX_MGMTQ] = cpu_to_le32(curr_frag + 1);
+	spin_unlock_irqrestore(&priv->slock, flags);
+
+	/* trigger the device */
+	islpci_trigger(priv);
+	return 0;
+
+      error_unlock:
+	spin_unlock_irqrestore(&priv->slock, flags);
+      error_free:
+	kfree(buf.mem);
+      error:
+	return err;
+}
+
+/*
+ * Receive a management frame from the device.
+ * This can be an arbitrary number of traps, and at most one response
+ * frame for a previous request sent via islpci_mgt_transmit().
+ */
+int
+islpci_mgt_receive(struct net_device *ndev)
+{
+	islpci_private *priv = netdev_priv(ndev);
+	isl38xx_control_block *cb =
+	    (isl38xx_control_block *) priv->control_block;
+	u32 curr_frag;
+
+#if VERBOSE > SHOW_ERROR_MESSAGES
+	DEBUG(SHOW_FUNCTION_CALLS, "islpci_mgt_receive \n");
+#endif
+
+	/* Only once per interrupt, determine fragment range to
+	 * process.  This avoids an endless loop (i.e. lockup) if
+	 * frames come in faster than we can process them. */
+	curr_frag = le32_to_cpu(cb->device_curr_frag[ISL38XX_CB_RX_MGMTQ]);
+	barrier();
+
+	for (; priv->index_mgmt_rx < curr_frag; priv->index_mgmt_rx++) {
+		pimfor_header_t *header;
+		u32 index = priv->index_mgmt_rx % ISL38XX_CB_MGMT_QSIZE;
+		struct islpci_membuf *buf = &priv->mgmt_rx[index];
+		u16 frag_len;
+		int size;
+		struct islpci_mgmtframe *frame;
+
+		/* I have no idea (and no documentation) if flags != 0
+		 * is possible.  Drop the frame, reuse the buffer. */
+		if (le16_to_cpu(cb->rx_data_mgmt[index].flags) != 0) {
+			printk(KERN_WARNING "%s: unknown flags 0x%04x\n",
+			       ndev->name,
+			       le16_to_cpu(cb->rx_data_mgmt[index].flags));
+			continue;
+		}
+
+		/* The device only returns the size of the header(s) here. */
+		frag_len = le16_to_cpu(cb->rx_data_mgmt[index].size);
+
+		/*
+		 * We appear to have no way to tell the device the
+		 * size of a receive buffer.  Thus, if this check
+		 * triggers, we likely have kernel heap corruption. */
+		if (frag_len > MGMT_FRAME_SIZE) {
+			printk(KERN_WARNING
+				"%s: Bogus packet size of %d (%#x).\n",
+				ndev->name, frag_len, frag_len);
+			frag_len = MGMT_FRAME_SIZE;
+		}
+
+		/* Ensure the results of device DMA are visible to the CPU. */
+		pci_dma_sync_single_for_cpu(priv->pdev, buf->pci_addr,
+					    buf->size, PCI_DMA_FROMDEVICE);
+
+		/* Perform endianess conversion for PIMFOR header in-place. */
+		header = pimfor_decode_header(buf->mem, frag_len);
+		if (!header) {
+			printk(KERN_WARNING "%s: no PIMFOR header found\n",
+			       ndev->name);
+			continue;
+		}
+
+		/* The device ID from the PIMFOR packet received from
+		 * the MVC is always 0.  We forward a sensible device_id.
+		 * Not that anyone upstream would care... */
+		header->device_id = priv->ndev->ifindex;
+
+#if VERBOSE > SHOW_ERROR_MESSAGES
+		DEBUG(SHOW_PIMFOR_FRAMES,
+		      "PIMFOR: op %i, oid 0x%08x, device %i, flags 0x%x length 0x%x \n",
+		      header->operation, header->oid, header->device_id,
+		      header->flags, header->length);
+
+		/* display the buffer contents for debugging */
+		display_buffer((char *) header, PIMFOR_HEADER_SIZE);
+		display_buffer((char *) header + PIMFOR_HEADER_SIZE,
+			       header->length);
+#endif
+
+		/* nobody sends these */
+		if (header->flags & PIMFOR_FLAG_APPLIC_ORIGIN) {
+			printk(KERN_DEBUG
+			       "%s: errant PIMFOR application frame\n",
+			       ndev->name);
+			continue;
+		}
+
+		/* Determine frame size, skipping OID_INL_TUNNEL headers. */
+		size = PIMFOR_HEADER_SIZE + header->length;
+		frame = kmalloc(sizeof (struct islpci_mgmtframe) + size,
+				GFP_ATOMIC);
+		if (!frame) {
+			printk(KERN_WARNING
+			       "%s: Out of memory, cannot handle oid 0x%08x\n",
+			       ndev->name, header->oid);
+			continue;
+		}
+		frame->ndev = ndev;
+		memcpy(&frame->buf, header, size);
+		frame->header = (pimfor_header_t *) frame->buf;
+		frame->data = frame->buf + PIMFOR_HEADER_SIZE;
+
+#if VERBOSE > SHOW_ERROR_MESSAGES
+		DEBUG(SHOW_PIMFOR_FRAMES,
+		      "frame: header: %p, data: %p, size: %d\n",
+		      frame->header, frame->data, size);
+#endif
+
+		if (header->operation == PIMFOR_OP_TRAP) {
+#if VERBOSE > SHOW_ERROR_MESSAGES
+			printk(KERN_DEBUG
+			       "TRAP: oid 0x%x, device %i, flags 0x%x length %i\n",
+			       header->oid, header->device_id, header->flags,
+			       header->length);
+#endif
+
+			/* Create work to handle trap out of interrupt
+			 * context. */
+			INIT_WORK(&frame->ws, prism54_process_trap, frame);
+			schedule_work(&frame->ws);
+
+		} else {
+			/* Signal the one waiting process that a response
+			 * has been received. */
+			if ((frame = xchg(&priv->mgmt_received, frame)) != NULL) {
+				printk(KERN_WARNING
+				       "%s: mgmt response not collected\n",
+				       ndev->name);
+				kfree(frame);
+			}
+#if VERBOSE > SHOW_ERROR_MESSAGES
+			DEBUG(SHOW_TRACING, "Wake up Mgmt Queue\n");
+#endif
+			wake_up(&priv->mgmt_wqueue);
+		}
+
+	}
+
+	return 0;
+}
+
+/*
+ * Cleanup the transmit queue by freeing all frames handled by the device.
+ */
+void
+islpci_mgt_cleanup_transmit(struct net_device *ndev)
+{
+	islpci_private *priv = netdev_priv(ndev);
+	isl38xx_control_block *cb =	/* volatile not needed */
+	    (isl38xx_control_block *) priv->control_block;
+	u32 curr_frag;
+
+#if VERBOSE > SHOW_ERROR_MESSAGES
+	DEBUG(SHOW_FUNCTION_CALLS, "islpci_mgt_cleanup_transmit\n");
+#endif
+
+	/* Only once per cleanup, determine fragment range to
+	 * process.  This avoids an endless loop (i.e. lockup) if
+	 * the device became confused, incrementing device_curr_frag
+	 * rapidly. */
+	curr_frag = le32_to_cpu(cb->device_curr_frag[ISL38XX_CB_TX_MGMTQ]);
+	barrier();
+
+	for (; priv->index_mgmt_tx < curr_frag; priv->index_mgmt_tx++) {
+		int index = priv->index_mgmt_tx % ISL38XX_CB_MGMT_QSIZE;
+		struct islpci_membuf *buf = &priv->mgmt_tx[index];
+		pci_unmap_single(priv->pdev, buf->pci_addr, buf->size,
+				 PCI_DMA_TODEVICE);
+		buf->pci_addr = 0;
+		kfree(buf->mem);
+		buf->mem = NULL;
+		buf->size = 0;
+	}
+}
+
+/*
+ * Perform one request-response transaction to the device.
+ */
+int
+islpci_mgt_transaction(struct net_device *ndev,
+		       int operation, unsigned long oid,
+		       void *senddata, int sendlen,
+		       struct islpci_mgmtframe **recvframe)
+{
+	islpci_private *priv = netdev_priv(ndev);
+	const long wait_cycle_jiffies = (ISL38XX_WAIT_CYCLE * 10 * HZ) / 1000;
+	long timeout_left = ISL38XX_MAX_WAIT_CYCLES * wait_cycle_jiffies;
+	int err;
+	DEFINE_WAIT(wait);
+
+	*recvframe = NULL;
+
+	if (down_interruptible(&priv->mgmt_sem))
+		return -ERESTARTSYS;
+
+	prepare_to_wait(&priv->mgmt_wqueue, &wait, TASK_UNINTERRUPTIBLE);
+	err = islpci_mgt_transmit(ndev, operation, oid, senddata, sendlen);
+	if (err)
+		goto out;
+
+	err = -ETIMEDOUT;
+	while (timeout_left > 0) {
+		int timeleft;
+		struct islpci_mgmtframe *frame;
+
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		timeleft = schedule_timeout(wait_cycle_jiffies);
+		frame = xchg(&priv->mgmt_received, NULL);
+		if (frame) {
+			if (frame->header->oid == oid) {
+				*recvframe = frame;
+				err = 0;
+				goto out;
+			} else {
+				printk(KERN_DEBUG
+				       "%s: expecting oid 0x%x, received 0x%x.\n",
+				       ndev->name, (unsigned int) oid,
+				       frame->header->oid);
+				kfree(frame);
+				frame = NULL;
+			}
+		}
+		if (timeleft == 0) {
+			printk(KERN_DEBUG
+				"%s: timeout waiting for mgmt response %lu, "
+				"triggering device\n",
+				ndev->name, timeout_left);
+			islpci_trigger(priv);
+		}
+		timeout_left += timeleft - wait_cycle_jiffies;
+	}
+	printk(KERN_WARNING "%s: timeout waiting for mgmt response\n",
+	       ndev->name);
+
+	/* TODO: we should reset the device here */     
+ out:
+	finish_wait(&priv->mgmt_wqueue, &wait);
+	up(&priv->mgmt_sem);
+	return err;
+}
+