Date: Mon, 28 Oct 2002 16:24:22 +0000 From: Ian Dowse <iedowse@maths.tcd.ie> To: pawmal@unia.3lo.lublin.pl Cc: freebsd-bugs@FreeBSD.org Subject: Re: kern/42260: 4.6-STABLE kernel panic, fxp related Message-ID: <200210281624.aa28226@salmon.maths.tcd.ie> In-Reply-To: Your message of "Wed, 04 Sep 2002 00:17:23 %2B0200." <20020903221718Z2129491-22726%2B13@mx1.lublin.pl>
next in thread | previous in thread | raw e-mail | index | archive | help
In message <20020903221718Z2129491-22726+13@mx1.lublin.pl>, =?ISO-8859-2?Q?Pawe
=B3_Ma=B3achowski?= writes:
>On 3 Sep 02, at 16:22, Ian Dowse wrote:
>> Could you try using the latest 4.6-STABLE source, but with the old
>> version of if_fxp.c? Grab it from cvsweb at
>
>But let's go back to this PR. Todays RELENG_4, 4.7-PRERELASE
>with older version of if_fxp.c (1.110.2.23) looks stable for
>me. I'm happy.
>
>What now?
>Problem is specific only to one combination of my machines.
>I have no idea, how this combinaction could affect the newer
>fxp driver and cause a kernel panic...
Hi,
If possible, could you try the following patch against the latest
revision of if_fxp.c. This should limit the abuse of the reserved
status bit to the DEVICE_POLLING case. I haven't tested this patch
beyond compiling it, so be prepared for problems. I'll push to get
this committed if it solves the problem for you, as I hate to see
this kind of hack remaining in the driver for one of FreeBSD's most
recommended NICs.
Ian
Index: if_fxp.c
===================================================================
RCS file: /home/iedowse/CVS/src/sys/dev/fxp/if_fxp.c,v
retrieving revision 1.110.2.25
diff -u -r1.110.2.25 if_fxp.c
--- if_fxp.c 24 Sep 2002 23:45:16 -0000 1.110.2.25
+++ if_fxp.c 28 Oct 2002 15:57:22 -0000
@@ -1288,15 +1288,22 @@
* Process receiver interrupts. If a no-resource (RNR)
* condition exists, get whatever packets we can and
* re-start the receiver.
+ */
+
+#ifdef DEVICE_POLLING
+ /*
* When using polling, we do not process the list to completion,
* so when we get an RNR interrupt we must defer the restart
* until we hit the last buffer with the C bit set.
* If we run out of cycles and rfa_headm has the C bit set,
* record the pending RNR in an unused status bit, so that the
* info will be used in the subsequent polling cycle.
+ *
+ * XXX there is absolutely no guarantee that this reserved bit
+ * will be ignored by the hardware!
*/
-
#define FXP_RFA_RNRMARK 0x4000 /* used to mark a pending RNR intr */
+#endif
for (;;) {
m = sc->rfa_headm;
@@ -1311,8 +1318,10 @@
if ( (rfa->rfa_status & FXP_RFA_STATUS_C) == 0)
break;
+#ifdef DEVICE_POLLING
if (rfa->rfa_status & FXP_RFA_RNRMARK)
rnr = 1;
+#endif
/*
* Remove first packet from the chain.
*/
@@ -1346,15 +1355,19 @@
}
}
if (rnr) {
+#ifdef DEVICE_POLLING
if (rfa->rfa_status & FXP_RFA_STATUS_C)
rfa->rfa_status |= FXP_RFA_RNRMARK;
else {
+#endif
fxp_scb_wait(sc);
CSR_WRITE_4(sc, FXP_CSR_SCB_GENERAL,
vtophys(sc->rfa_headm->m_ext.ext_buf) +
RFA_ALIGNMENT_FUDGE);
fxp_scb_cmd(sc, FXP_SCB_COMMAND_RU_START);
+#ifdef DEVICE_POLLING
}
+#endif
}
}
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200210281624.aa28226>