I got a report that the pcnet32 module is mangling things upon rmmod
followed by insmod/rmmod of any other PCI driver (ksymoops mis-implicated
rtc.o hence it was sent to me). I think I found the problem regardless,
but don't have the hardware to test it on.
IIRC Jeff had a patch brewing for pcnet32 to clean things up a bit (dump
MOD_USE_COUNT and ifdef USE_PCI_REGISTER_DRIVER, fix VLbus probing which
is currently implicitly disabled - see "return 0;" etc.) so I did not
get into that - this is only a minimal patch to (hopefully) fix the oops.
Paul.
--- drivers/net/pcnet32.c~ Sat Dec 16 04:01:43 2000
+++ drivers/net/pcnet32.c Mon Jan 29 16:15:13 2001
@@ -161,6 +161,7 @@
* v1.25kf Added No Interrupt on successful Tx for some Tx's <kaf@fc.hp.com>
* v1.26 Converted to pci_alloc_consistent, Jamey Hicks / George France
* <jamey@crl.dec.com>
+ * v1.26p Fix oops: was missing pci_unregister_driver - Paul Gortmaker.
*/
@@ -1494,6 +1495,8 @@
static int debug = -1;
static int tx_start_pt = -1;
+static int pcnet32_pci_present;
+
static int __init pcnet32_init_module(void)
{
int cards_found = 0;
@@ -1510,6 +1513,8 @@
#ifdef USE_PCI_REGISTER_DRIVER
if ((err = pci_module_init(&pcnet32_driver)) < 0 )
return err;
+ else
+ pcnet32_pci_present = 1;
#else
{
struct pci_device_id *devid = pcnet32_pci_tbl;
@@ -1542,6 +1547,8 @@
kfree(pcnet32_dev);
pcnet32_dev = next_dev;
}
+ if (pcnet32_pci_present)
+ pci_unregister_driver(&pcnet32_driver);
}
module_init(pcnet32_init_module);
_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
