Bug check

We have enough information at this point to see if this might be a known bug, perhaps even with a known patch for it. We gave the SunSolve bug database (available on CD-ROM) two key words to search for: panic and tcp_respond . The search turned up three bug reports to check out.

• Bug Id: 1122457

  Category: kernel

  Subcategory: network

  Release summary: 4.1.3

  Synopsis: trash on net panic box w/"Data Fault" or "Mem addr align" null ptr tcp_respond

• Bug Id: 1151988

  Category: kernel

  Subcategory: tcp-ip

  Release summary: no-v4, 4.1.3

  Synopsis: panic in tcp_respond with data fault

• Bug Id: 1108657

  Category: network

  Subcategory: internet

  Release summary: 4.1.2

  Synopsis: in_cksum() caused system panic

All these look somewhat promising , especially the first two. The last was filed against 4.1.2, however, and from the description of the problem, it appears that it only shows up on FDDI links. We'll rule that one out, since the symptoms don't match.

The other two were filed against 4.1.3, not 4.1.3_U1. However, since the U1 release was primarily an upgrade with just bug fixes in it, they may still be applicable if those particular fixes did not make it into the new release. Let's check the patch database just to see if they were taken care of with an official patch.

Searching the patch description files for occurrences of either of these two bug numbers (1122457 and 1151988) gives us two matches.

• Patch-ID# 100584-06

  Keywords: loopback keepalive tcp_respond panic sigio exponential backoff

  Synopsis: SunOS 4.1.1, 4.1.2, 4.1.3: system freezes using loopback interface.

• Patch-ID# 102010-01

  Keywords: loopback keepalive tcp_respond panic sigio exponential backoff

  Synopsis: SunOS 4.1.3_U1: system freezes using loopback interface.

It looks like that problem was fixed in three releases by patch 100584, and it was ported to SunOS 4.1.3_U1 and given a new number, 102010. Let's try this patch and see if it solves the problem.