[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#362056: More test results: different bugs in different Tulip drivers



	This is a preliminary report.  I can do more detailed tests and file
a new bug report, if desired.

	A quick test with debian-installer on floppies, which uses a 2.4.27
kernel, shows that the Tulip board works with two 3C905B boards also
present, if I unload the de4x5 driver and load the tulip driver by hand. 
Then the installer sees the Tulip board at the next step, I can configure
it, and download installer components through that interface.
	If I let the installer attempt to load the de4x5 driver, a box pops
up with an message that there was an error when attempting to load the
driver.

	On 2.6.8, I removed the 2104x driver before attempting to bring up
the interface, and loaded the tulip driver by hand.  This time, the command
		ifconfig eth2 192.168.136.55
popped up a message saying "no such interface or device", but didn't hang
the machine.

	On 2.6.15, the de4x5 and de2104x modules _both_ get loaded
automatically at boot time.  This sounds bogus in itself.
	If I remove both of these, and command:
		modprobe tulip
		ifconfig eth2 192.168.136.55
the error messages read:
SIOCSIFFADDR: No such device
eth2: ERROR while getting interface flags: No such device
	Next, if I command:
		rmmod tulip
		modprobe de4x5
		ifconfig eth2 192.168.136.55
		ifconfig
all 3 Ethernet ports are shown as configured.
		cat /proc/interrupts
shows "DE-450CA (0000:00:12.0)", not "eth2", added to IRQ 10 along with
eth0.
The port is able to ping 192.168.136.43 over the coax interface.

	(In addition, I can now command:
		modprobe smc-ultra
		ifconfig eth3 192.168.254.141
		ping 192.168.254.254
and get successful pings.  The SMC Ultra board is assigned IRQ 5.  So I can
get IRQs assigned to boards in all three PCI slots, but only one IRQ is
available for an ISA-PNP network board even if only one PCI board is
present.)

	In summary, different Tulip drivers show different bugs at different
kernel versions.  It appears that at 2.6.15 two drivers are fighting for the
same board, and one of them has a serious bug.



Reply to: