Kernel bug

[Christophe Dupre <duprec@bloodwolf.org>]

Running Red Hat 7.2 with Red Hat's kernel 2.4.18-19.7.x, I  get this 
error whenever there's a lot of disk activity (such as a backup):

kernel BUG at inode.c:565!
invalid operand: 0000
esp cyclades ip_conntrack_ftp eepro10 3c59x ipt_REHECT iptable_nat 
ip_conntrack
iptable_filter ip_tables st ext3 jbd aic7xxx sd_mod scsi_mod
CPI: 0
EIP: 0010:[<c014c689>]   Not tainted
EFLAGS: 00010206

EIP is at clear_inode [kernel] 0x19 (2.4.19-19.7.x)
eax: cb841060   ebx: cb841040   ecx: cb841040   edx: 00000800
esi: c170ffa8   edi: c170e000   ebp: 0000d6a9   esp: c170ff70
ds: 0018  es: 0018  ss: 0018
Process kswapd (pid: 5, stackpage=c170f000)
Cal Trace: [<c014c797>] dispose_list [kernel] 0x47 (0xc170ff7c))
[<c014c790>] dispose_list [kernel] 0x50 (0xc170ff84))
[<c014ca53>] prune_icache [kernel] 0x123 (0xc170ff98))
[<c014ca90>] shrink_icache_memory [kernel] 0x20 (0xc170ffc0))
[<c0131c9d>] kswapd [kernel] 0x30d (0xc170ffc8))
[<c0105000>] stext [kernel] 0x0 (0xc170ffe8))
[<c0107146>] kernel_thread [kernel] 0x26 (0xc170fff0))
[<c0131990>] kswapd [kernel] 0x0 (0xc170fff8))

Code: 0f 0b 35 02 3d e0 22 c0 8b 83 0c 01 00 00 a9 10 00 00 00 75

The tape backup is SCSI connected to an Adaptec 29160. The disk is IDE. 
The filesystem is ext2fs.

Help would be appreciated.