[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
(usagi-users 00988) Problem with USAGI on Linux ARM
- To: usagi-users@xxxxxxxxxxxxxx
- Subject: (usagi-users 00988) Problem with USAGI on Linux ARM
- From: Guido Barzini <guido.barzini@xxxxxxxxxx>
- Date: Wed, 07 Nov 2001 18:04:08 -0000 (GMT)
- In-reply-to: <200111080244.FMLAAB17402.usagi-users@linux-ipv6.org>
- Organization: Virata
- Reply-to: usagi-users@xxxxxxxxxxxxxx
- Sender: gnb@xxxxxxxxxx
Hello. I've run into what seems to be a problem with the USAGI stack running
on the ARM CPU. The setup is as follows:
1 PC running 2.4.7 with linux24-2.4.7-usagi-20010816.patch applied
1 ARM system running 2.4.7 with linux24-2.4.7-usagi-20010816.patch and rmk-3
The PC has MAC address 00:60:97:CF:F0:99 and the ARM box is 00:20:2b:00:54:30
I send a few pings from each machine and then do a 'cat /proc/net/ndisc'. On
the PC I get:
fe8000000000000002202bfffe005430 0b 80 01 04 00001c04 00001c04 00000fb7 1770
0001 0002 eth0 00202b005430
which is correct, but on the ARM side I see
fe80000000000000026097fffecff099 10 80 01 04 00010588 00010588 0000117b 1770
0001 0002 eth0 97cff0991819
the MAC address is the last 4 bytes of the PC's MAC address, followed by two
bytes of garbage. Anything I try to send from the ARM system to the PC goes out
to that H/W address.
The looks like a problem with gcc/ARM padding structures differently from
gcc/i386 (the ARM works better with aligned quantities, so the compiler is more
likely to insert padding), but I haven't been able to diagnose it further. I
believe that it isn't a compile bug, since the default IPv6 support in 2.4.7
behaves correctly.
Has anyone seen a problem like this before? Is the fix known? Thank you.
Guido
--
Guido Barzini gnb@xxxxxxxxxx