Ross, Daniel wrote: > > I have a Solaris 8 with all the latest patches installed and 4.1.23 > ipfilter with the lates pfil 2.1.13 > > > > > > Im getting a kernel panic every day similar to the following: > > Jun 19 11:06:54 server unix: [ID 836849 kern.notice] > > Jun 19 11:06:54 server ^Mpanic[cpu0]/thread=2a1001f7d20: > > Jun 19 11:06:54 server unix: [ID 103648 kern.notice] recursive > mutex_enter, lp=30009525828 owner=2a1001f7d20 > > thread=2a1001f7d20 > > Jun 19 11:06:55 server unix: [ID 100000 kern.notice] > > Jun 19 11:06:55 server genunix: [ID 723222 kern.notice] > 000002a1001f6c10 unix:mutex_panic+5c (10415fe8, 3000 > > 9525828, 20, 2a1001f71c0, 0, 26) > > Jun 19 11:06:55 server genunix: [ID 179002 kern.notice] %l0-3: > 000000000000003c 000000000000003c 000003000 > > 1941c90 00000300040fbe64 > > Jun 19 11:06:55 server %l4-7: 0000030004209180 000002a1001f71c0 > 000000000000003c 000002a100771ba0 > > Jun 19 11:06:55 server genunix: [ID 723222 kern.notice] > 000002a1001f6cc0 unix:mutex_vector_enter+250 (300095 > > 25828, 10415fc0, 10424590, 300040fbe3c, 0, 1) > > ... > > I found that a Solaris 9 thread suggested updating the pfil but my > pfil version seems to have that fix. Also that perhaps adding: > > block return-rst in quick proto tcp from any to any > > would help. > > I am not familiar with kernel dump debugging but I am open to suggestions? >
If you can, send the crash dump files into Sun. I've seen an analysis of a panic like this where the problem turns out to be interrupt handling not restoring things the way they should in some part of the NIC driver (be it the driver or GLD, I'm not sure.) Darren
