On 18:36 Mon 09 Oct , David Miller wrote: > From: Will Simoneau <[email protected]> > Date: Thu, 5 Oct 2006 17:15:44 -0400 > > > The first one I seem to be able to fix by adding a get_unaligned() at > > lines 1679-1680 of eth1394.c around eth->h_dest; the second one seems to > > be triggered by this code: > > Right. > > > (gdb) list *ether1394_data_handler+0x914 > > 0xc94 is in ether1394_data_handler (drivers/ieee1394/eth1394.c:1264). > > 1259 priv->stats.rx_dropped++; > > 1260 dev_kfree_skb_any(skb); > > 1261 goto bad_proto; > > 1262 } > > 1263 > > 1264 if (netif_rx(skb) == NET_RX_DROP) { > > 1265 priv->stats.rx_errors++; > > 1266 priv->stats.rx_dropped++; > > 1267 goto bad_proto; > > 1268 } > > Actually, I think this one is in eth1394_parse_encap(). > > Can you test this patch and tell me if it makes both unaligned > access problems go away? Thanks. > > diff --git a/drivers/ieee1394/eth1394.c b/drivers/ieee1394/eth1394.c > index 8a7b8fa..78abb9b 100644 > --- a/drivers/ieee1394/eth1394.c > +++ b/drivers/ieee1394/eth1394.c > @@ -64,6 +64,7 @@ #include <linux/bitops.h> > #include <linux/ethtool.h> > #include <asm/uaccess.h> > #include <asm/delay.h> > +#include <asm/unaligned.h> > #include <net/arp.h> > > #include "config_roms.h" > @@ -894,6 +895,7 @@ static inline u16 ether1394_parse_encap( > u16 maxpayload; > struct eth1394_node_ref *node; > struct eth1394_node_info *node_info; > + __be64 guid; > > /* Sanity check. MacOSX seems to be sending us 131 in this > * field (atleast on my Panther G5). Not sure why. */ > @@ -902,8 +904,9 @@ static inline u16 ether1394_parse_encap( > > maxpayload = min(eth1394_speedto_maxpayload[sspd], (u16)(1 << (max_rec + 1))); > > + guid = get_unaligned(&arp1394->s_uniq_id); > node = eth1394_find_node_guid(&priv->ip_node_list, > - be64_to_cpu(arp1394->s_uniq_id)); > + be64_to_cpu(guid)); > if (!node) { > return 0; > } > @@ -1675,8 +1678,9 @@ #endif > if (max_payload < dg_size + hdr_type_len[ETH1394_HDR_LF_UF]) > priv->bc_dgl++; > } else { > + __be64 guid = get_unaligned((u64 *)eth->h_dest); > node = eth1394_find_node_guid(&priv->ip_node_list, > - be64_to_cpu(*(u64*)eth->h_dest)); > + be64_to_cpu(guid)); > if (!node) { > ret = -EAGAIN; > goto fail; I still get: Kernel unaligned access at TPC[10162164] ether1394_reset_priv+0x2c/0xe0 [eth1394] Kernel unaligned access at TPC[10163148] ether1394_data_handler+0xdd0/0x1060 [eth1394] The second one triggers on every packet received, the first only triggers once in a while. If you want more gdb info or a disassembly just ask.
Attachment:
pgpQFsUCzJVqh.pgp
Description: PGP signature
- Follow-Ups:
- Re: [sparc64] 2.6.18 unaligned accesses in eth1394
- From: David Miller <[email protected]>
- Re: [sparc64] 2.6.18 unaligned accesses in eth1394
- References:
- [sparc64] 2.6.18 unaligned accesses in eth1394
- From: Will Simoneau <[email protected]>
- Re: [sparc64] 2.6.18 unaligned accesses in eth1394
- From: David Miller <[email protected]>
- [sparc64] 2.6.18 unaligned accesses in eth1394
- Prev by Date: Re: [PATCH] firmware/efivars: handle error
- Next by Date: Re: [PATCH] Introduce vfs_listxattr
- Previous by thread: Re: [sparc64] 2.6.18 unaligned accesses in eth1394
- Next by thread: Re: [sparc64] 2.6.18 unaligned accesses in eth1394
- Index(es):
 |