Running javac on linux 32 bit (carrs) from specjvm98 with product VM gets an intermittent crash with si_code = 128 which is SI_KERNEL. The code that the crash points to is perfectly valid. I don't know why we get these signals and other products get them also, but we should either warn and/or ignore when we get these so we don't confuse this with a real bug.

From digging in the kernel, I think this is the source of the signal. I don't know why things have gone wrong with the linux signal handler.

http://www.takatan.net/lxr/source/kernel/signal.c#L1250
1262 void
1263 force_sig(int sig, struct task_struct *p)
1264 {
1265 force_sig_info(sig, SEND_SIG_PRIV, p);
1266 }
1267
1268 /*
1269 * When things go south during signal handling, we
1270 * will force a SIGSEGV. And if the signal that caused
1271 * the problem was already a SIGSEGV, we'll want to
1272 * make sure we don't even try to deliver the signal..
1273 */
1274 int
1275 force_sigsegv(int sig, struct task_struct *p)
1276 {
1277 if (sig == SIGSEGV) {
1278 unsigned long flags;
1279 spin_lock_irqsave(&p->sighand->siglock, flags);
1280 p->sighand->action[sig - 1].sa.sa_handler = SIG_DFL;
1281 spin_unlock_irqrestore(&p->sighand->siglock, flags);
1282 }
1283 force_sig(SIGSEGV, p);
1284 return 0;
1285 }