ipath_intr.c

来自「linux 内核源代码」· C语言 代码 · 共 1,209 行 · 第 1/3 页

/*
 * Copyright (c) 2006, 2007 QLogic Corporation. All rights reserved.
 * Copyright (c) 2003, 2004, 2005, 2006 PathScale, Inc. All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include <linux/pci.h>

#include "ipath_kernel.h"
#include "ipath_verbs.h"
#include "ipath_common.h"

/*
 * clear (write) a pio buffer, to clear a parity error.   This routine
 * should only be called when in freeze mode, and the buffer should be
 * canceled afterwards.
 */
static void ipath_clrpiobuf(struct ipath_devdata *dd, u32 pnum)
{
	u32 __iomem *pbuf;
	u32 dwcnt; /* dword count to write */
	if (pnum < dd->ipath_piobcnt2k) {
		pbuf = (u32 __iomem *) (dd->ipath_pio2kbase + pnum *
			dd->ipath_palign);
		dwcnt = dd->ipath_piosize2k >> 2;
	}
	else {
		pbuf = (u32 __iomem *) (dd->ipath_pio4kbase +
			(pnum - dd->ipath_piobcnt2k) * dd->ipath_4kalign);
		dwcnt = dd->ipath_piosize4k >> 2;
	}
	dev_info(&dd->pcidev->dev,
		"Rewrite PIO buffer %u, to recover from parity error\n",
		pnum);
	*pbuf = dwcnt+1; /* no flush required, since already in freeze */
	while(--dwcnt)
		*pbuf++ = 0;
}

/*
 * Called when we might have an error that is specific to a particular
 * PIO buffer, and may need to cancel that buffer, so it can be re-used.
 * If rewrite is true, and bits are set in the sendbufferror registers,
 * we'll write to the buffer, for error recovery on parity errors.
 */
static void ipath_disarm_senderrbufs(struct ipath_devdata *dd, int rewrite)
{
	u32 piobcnt;
	unsigned long sbuf[4];
	/*
	 * it's possible that sendbuffererror could have bits set; might
	 * have already done this as a result of hardware error handling
	 */
	piobcnt = dd->ipath_piobcnt2k + dd->ipath_piobcnt4k;
	/* read these before writing errorclear */
	sbuf[0] = ipath_read_kreg64(
		dd, dd->ipath_kregs->kr_sendbuffererror);
	sbuf[1] = ipath_read_kreg64(
		dd, dd->ipath_kregs->kr_sendbuffererror + 1);
	if (piobcnt > 128) {
		sbuf[2] = ipath_read_kreg64(
			dd, dd->ipath_kregs->kr_sendbuffererror + 2);
		sbuf[3] = ipath_read_kreg64(
			dd, dd->ipath_kregs->kr_sendbuffererror + 3);
	}

	if (sbuf[0] || sbuf[1] || (piobcnt > 128 && (sbuf[2] || sbuf[3]))) {
		int i;
		if (ipath_debug & (__IPATH_PKTDBG|__IPATH_DBG) &&
			dd->ipath_lastcancel > jiffies) {
			__IPATH_DBG_WHICH(__IPATH_PKTDBG|__IPATH_DBG,
					  "SendbufErrs %lx %lx", sbuf[0],
					  sbuf[1]);
			if (ipath_debug & __IPATH_PKTDBG && piobcnt > 128)
				printk(" %lx %lx ", sbuf[2], sbuf[3]);
			printk("\n");
		}

		for (i = 0; i < piobcnt; i++)
			if (test_bit(i, sbuf)) {
				if (rewrite)
					ipath_clrpiobuf(dd, i);
				ipath_disarm_piobufs(dd, i, 1);
			}
		/* ignore armlaunch errs for a bit */
		dd->ipath_lastcancel = jiffies+3;
	}
}


/* These are all rcv-related errors which we want to count for stats */
#define E_SUM_PKTERRS \
	(INFINIPATH_E_RHDRLEN | INFINIPATH_E_RBADTID | \
	 INFINIPATH_E_RBADVERSION | INFINIPATH_E_RHDR | \
	 INFINIPATH_E_RLONGPKTLEN | INFINIPATH_E_RSHORTPKTLEN | \
	 INFINIPATH_E_RMAXPKTLEN | INFINIPATH_E_RMINPKTLEN | \
	 INFINIPATH_E_RFORMATERR | INFINIPATH_E_RUNSUPVL | \
	 INFINIPATH_E_RUNEXPCHAR | INFINIPATH_E_REBP)

/* These are all send-related errors which we want to count for stats */
#define E_SUM_ERRS \
	(INFINIPATH_E_SPIOARMLAUNCH | INFINIPATH_E_SUNEXPERRPKTNUM | \
	 INFINIPATH_E_SDROPPEDDATAPKT | INFINIPATH_E_SDROPPEDSMPPKT | \
	 INFINIPATH_E_SMAXPKTLEN | INFINIPATH_E_SUNSUPVL | \
	 INFINIPATH_E_SMINPKTLEN | INFINIPATH_E_SPKTLEN | \
	 INFINIPATH_E_INVALIDADDR)

/*
 * this is similar to E_SUM_ERRS, but can't ignore armlaunch, don't ignore
 * errors not related to freeze and cancelling buffers.  Can't ignore
 * armlaunch because could get more while still cleaning up, and need
 * to cancel those as they happen.
 */
#define E_SPKT_ERRS_IGNORE \
	 (INFINIPATH_E_SDROPPEDDATAPKT | INFINIPATH_E_SDROPPEDSMPPKT | \
	 INFINIPATH_E_SMAXPKTLEN | INFINIPATH_E_SMINPKTLEN | \
	 INFINIPATH_E_SPKTLEN)

/*
 * these are errors that can occur when the link changes state while
 * a packet is being sent or received.  This doesn't cover things
 * like EBP or VCRC that can be the result of a sending having the
 * link change state, so we receive a "known bad" packet.
 */
#define E_SUM_LINK_PKTERRS \
	(INFINIPATH_E_SDROPPEDDATAPKT | INFINIPATH_E_SDROPPEDSMPPKT | \
	 INFINIPATH_E_SMINPKTLEN | INFINIPATH_E_SPKTLEN | \
	 INFINIPATH_E_RSHORTPKTLEN | INFINIPATH_E_RMINPKTLEN | \
	 INFINIPATH_E_RUNEXPCHAR)

static u64 handle_e_sum_errs(struct ipath_devdata *dd, ipath_err_t errs)
{
	u64 ignore_this_time = 0;

	ipath_disarm_senderrbufs(dd, 0);
	if ((errs & E_SUM_LINK_PKTERRS) &&
	    !(dd->ipath_flags & IPATH_LINKACTIVE)) {
		/*
		 * This can happen when SMA is trying to bring the link
		 * up, but the IB link changes state at the "wrong" time.
		 * The IB logic then complains that the packet isn't
		 * valid.  We don't want to confuse people, so we just
		 * don't print them, except at debug
		 */
		ipath_dbg("Ignoring packet errors %llx, because link not "
			  "ACTIVE\n", (unsigned long long) errs);
		ignore_this_time = errs & E_SUM_LINK_PKTERRS;
	}

	return ignore_this_time;
}

/* generic hw error messages... */
#define INFINIPATH_HWE_TXEMEMPARITYERR_MSG(a) \
	{ \
		.mask = ( INFINIPATH_HWE_TXEMEMPARITYERR_##a <<    \
			  INFINIPATH_HWE_TXEMEMPARITYERR_SHIFT ),   \
		.msg = "TXE " #a " Memory Parity"	     \
	}
#define INFINIPATH_HWE_RXEMEMPARITYERR_MSG(a) \
	{ \
		.mask = ( INFINIPATH_HWE_RXEMEMPARITYERR_##a <<    \
			  INFINIPATH_HWE_RXEMEMPARITYERR_SHIFT ),   \
		.msg = "RXE " #a " Memory Parity"	     \
	}

static const struct ipath_hwerror_msgs ipath_generic_hwerror_msgs[] = {
	INFINIPATH_HWE_MSG(IBCBUSFRSPCPARITYERR, "IPATH2IB Parity"),
	INFINIPATH_HWE_MSG(IBCBUSTOSPCPARITYERR, "IB2IPATH Parity"),

	INFINIPATH_HWE_TXEMEMPARITYERR_MSG(PIOBUF),
	INFINIPATH_HWE_TXEMEMPARITYERR_MSG(PIOPBC),
	INFINIPATH_HWE_TXEMEMPARITYERR_MSG(PIOLAUNCHFIFO),

	INFINIPATH_HWE_RXEMEMPARITYERR_MSG(RCVBUF),
	INFINIPATH_HWE_RXEMEMPARITYERR_MSG(LOOKUPQ),
	INFINIPATH_HWE_RXEMEMPARITYERR_MSG(EAGERTID),
	INFINIPATH_HWE_RXEMEMPARITYERR_MSG(EXPTID),
	INFINIPATH_HWE_RXEMEMPARITYERR_MSG(FLAGBUF),
	INFINIPATH_HWE_RXEMEMPARITYERR_MSG(DATAINFO),
	INFINIPATH_HWE_RXEMEMPARITYERR_MSG(HDRINFO),
};

/**
 * ipath_format_hwmsg - format a single hwerror message
 * @msg message buffer
 * @msgl length of message buffer
 * @hwmsg message to add to message buffer
 */
static void ipath_format_hwmsg(char *msg, size_t msgl, const char *hwmsg)
{
	strlcat(msg, "[", msgl);
	strlcat(msg, hwmsg, msgl);
	strlcat(msg, "]", msgl);
}

/**
 * ipath_format_hwerrors - format hardware error messages for display
 * @hwerrs hardware errors bit vector
 * @hwerrmsgs hardware error descriptions
 * @nhwerrmsgs number of hwerrmsgs
 * @msg message buffer
 * @msgl message buffer length
 */
void ipath_format_hwerrors(u64 hwerrs,
			   const struct ipath_hwerror_msgs *hwerrmsgs,
			   size_t nhwerrmsgs,
			   char *msg, size_t msgl)
{
	int i;
	const int glen =
	    sizeof(ipath_generic_hwerror_msgs) /
	    sizeof(ipath_generic_hwerror_msgs[0]);

	for (i=0; i<glen; i++) {
		if (hwerrs & ipath_generic_hwerror_msgs[i].mask) {
			ipath_format_hwmsg(msg, msgl,
					   ipath_generic_hwerror_msgs[i].msg);
		}
	}

	for (i=0; i<nhwerrmsgs; i++) {
		if (hwerrs & hwerrmsgs[i].mask) {
			ipath_format_hwmsg(msg, msgl, hwerrmsgs[i].msg);
		}
	}
}

/* return the strings for the most common link states */
static char *ib_linkstate(u32 linkstate)
{
	char *ret;

	switch (linkstate) {
	case IPATH_IBSTATE_INIT:
		ret = "Init";
		break;
	case IPATH_IBSTATE_ARM:
		ret = "Arm";
		break;
	case IPATH_IBSTATE_ACTIVE:
		ret = "Active";
		break;
	default:
		ret = "Down";
	}

	return ret;
}

void signal_ib_event(struct ipath_devdata *dd, enum ib_event_type ev)
{
	struct ib_event event;

	event.device = &dd->verbs_dev->ibdev;
	event.element.port_num = 1;
	event.event = ev;
	ib_dispatch_event(&event);
}

static void handle_e_ibstatuschanged(struct ipath_devdata *dd,
				     ipath_err_t errs, int noprint)
{
	u64 val;
	u32 ltstate, lstate;

	/*
	 * even if diags are enabled, we want to notice LINKINIT, etc.
	 * We just don't want to change the LED state, or
	 * dd->ipath_kregs->kr_ibcctrl
	 */
	val = ipath_read_kreg64(dd, dd->ipath_kregs->kr_ibcstatus);
	lstate = val & IPATH_IBSTATE_MASK;

	/*
	 * this is confusing enough when it happens that I want to always put it
	 * on the console and in the logs.  If it was a requested state change,
	 * we'll have already cleared the flags, so we won't print this warning
	 */
	if ((lstate != IPATH_IBSTATE_ARM && lstate != IPATH_IBSTATE_ACTIVE)
		&& (dd->ipath_flags & (IPATH_LINKARMED | IPATH_LINKACTIVE))) {
		dev_info(&dd->pcidev->dev, "Link state changed from %s to %s\n",
				 (dd->ipath_flags & IPATH_LINKARMED) ? "ARM" : "ACTIVE",
				 ib_linkstate(lstate));
		/*
		 * Flush all queued sends when link went to DOWN or INIT,
		 * to be sure that they don't block SMA and other MAD packets
		 */
		ipath_cancel_sends(dd, 1);
	}
	else if (lstate == IPATH_IBSTATE_INIT || lstate == IPATH_IBSTATE_ARM ||
	    lstate == IPATH_IBSTATE_ACTIVE) {
		/*
		 * only print at SMA if there is a change, debug if not
		 * (sometimes we want to know that, usually not).
		 */
		if (lstate == ((unsigned) dd->ipath_lastibcstat
			       & IPATH_IBSTATE_MASK)) {
			ipath_dbg("Status change intr but no change (%s)\n",
				  ib_linkstate(lstate));
		}
		else
			ipath_cdbg(VERBOSE, "Unit %u link state %s, last "
				   "was %s\n", dd->ipath_unit,
				   ib_linkstate(lstate),
				   ib_linkstate((unsigned)
						dd->ipath_lastibcstat
						& IPATH_IBSTATE_MASK));
	}
	else {
		lstate = dd->ipath_lastibcstat & IPATH_IBSTATE_MASK;
		if (lstate == IPATH_IBSTATE_INIT ||
		    lstate == IPATH_IBSTATE_ARM ||
		    lstate == IPATH_IBSTATE_ACTIVE)
			ipath_cdbg(VERBOSE, "Unit %u link state down"
				   " (state 0x%x), from %s\n",
				   dd->ipath_unit,
				   (u32)val & IPATH_IBSTATE_MASK,
				   ib_linkstate(lstate));
		else
			ipath_cdbg(VERBOSE, "Unit %u link state changed "
				   "to 0x%x from down (%x)\n",
				   dd->ipath_unit, (u32) val, lstate);
	}
	ltstate = (val >> INFINIPATH_IBCS_LINKTRAININGSTATE_SHIFT) &
		INFINIPATH_IBCS_LINKTRAININGSTATE_MASK;
	lstate = (val >> INFINIPATH_IBCS_LINKSTATE_SHIFT) &
		INFINIPATH_IBCS_LINKSTATE_MASK;

	if (ltstate == INFINIPATH_IBCS_LT_STATE_POLLACTIVE ||
	    ltstate == INFINIPATH_IBCS_LT_STATE_POLLQUIET) {
		u32 last_ltstate;

		/*
		 * Ignore cycling back and forth from Polling.Active
		 * to Polling.Quiet while waiting for the other end of
		 * the link to come up. We will cycle back and forth
		 * between them if no cable is plugged in,
		 * the other device is powered off or disabled, etc.
		 */
		last_ltstate = (dd->ipath_lastibcstat >>
				INFINIPATH_IBCS_LINKTRAININGSTATE_SHIFT)
			& INFINIPATH_IBCS_LINKTRAININGSTATE_MASK;
		if (last_ltstate == INFINIPATH_IBCS_LT_STATE_POLLACTIVE
		    || last_ltstate ==
		    INFINIPATH_IBCS_LT_STATE_POLLQUIET) {
			if (dd->ipath_ibpollcnt > 40) {
				dd->ipath_flags |= IPATH_NOCABLE;
				*dd->ipath_statusp |=
					IPATH_STATUS_IB_NOCABLE;
			} else
				dd->ipath_ibpollcnt++;
			goto skip_ibchange;
		}
	}
	dd->ipath_ibpollcnt = 0;	/* some state other than 2 or 3 */
	ipath_stats.sps_iblink++;
	if (ltstate != INFINIPATH_IBCS_LT_STATE_LINKUP) {
		if (dd->ipath_flags & IPATH_LINKACTIVE)
			signal_ib_event(dd, IB_EVENT_PORT_ERR);
		dd->ipath_flags |= IPATH_LINKDOWN;
		dd->ipath_flags &= ~(IPATH_LINKUNK | IPATH_LINKINIT
				     | IPATH_LINKACTIVE |
				     IPATH_LINKARMED);
		*dd->ipath_statusp &= ~IPATH_STATUS_IB_READY;
		dd->ipath_lli_counter = 0;
		if (!noprint) {
			if (((dd->ipath_lastibcstat >>
			      INFINIPATH_IBCS_LINKSTATE_SHIFT) &
			     INFINIPATH_IBCS_LINKSTATE_MASK)
			    == INFINIPATH_IBCS_L_STATE_ACTIVE)
				/* if from up to down be more vocal */
				ipath_cdbg(VERBOSE,
					   "Unit %u link now down (%s)\n",
					   dd->ipath_unit,
					   ipath_ibcstatus_str[ltstate]);