[3147] in SIPB bug reports
rn core dump, programmer stupidity
daemon@ATHENA.MIT.EDU (John Carr)
Sat Sep 5 15:32:35 1992
To: bug-sipb@Athena.MIT.EDU
Date: Sat, 05 Sep 1992 15:32:03 EDT
From: John Carr <jfc@Athena.MIT.EDU>
Machine type: IBM RT.
The bug:
I killed some articles while in rn. This is what happened:
End of article 26205 (of 26418)--what next? [npq] /upcom/=j
Searching...
26219 UPCOMING RELEASES #2 Junked
26220 UPCOMING RELEASES #3 Junked
26221 UPCOMING RELEASES #4 Junked
26222 UPCOMING RELEASES #5 Junked
26266 Re: UPCOMING RELEASES #5 Junked
26275 Re: UPCOMING RELEASES #5 Junked
Returning 1 Marked article...
Caught a SIGSEGV--.newsrc restored
load 2.89 2.31 1.89, pid 2625, %cpu 27.95, 978k of 1672k, running
load 2.89 2.31 1.89, pid 2625, %cpu 30.64, 978k of 1672k, running
It continued to use up lots of CPU time, so I got a core dump with gcore.
The programmer stupidity:
$c
_.abort() from _.sig_catcher+0x56
_.sig_catcher(r2=0x00000410) signal handler
Signal 5 (trace/BPT), with sigframe at 0x1fffa464, which interrupted ...
_.abort() from _.finalize+0x100
_.finalize(r2=0x00000400) from _.sig_catcher+0xec
_.sig_catcher(r2=0x00000400) signal handler
Signal 11 (memory fault), with sigframe at 0x1fffaaf4, which interrupted ...
_.malloc(r2=0x0009883c) from _.realloc+0xc4
_.realloc(r2=0x0009883c, r3=0x1000e354) from _.saferealloc+0x10
_.saferealloc(r2=0x0009883c, r3=0x1000e354) from _.fetchsubj+0x2fc
_.fetchsubj(r2=0x0009883c, r3=0x1000e354, r4=0x0000000a) from _.wanted+0x26
_.wanted(r2=0x0009883c, r3=0x1000e354, r4=0x0000000a) from _.art_search+0x5bc
_.art_search(r2=0x0009883c, 0x10003ce4, 0x00006689) from _.art_switch+0x316
_.art_switch() from _.art_switch+0x316
If you catch a signal, you MUST correct the event that caused it before
returning, or disable the signal handler. The program was in an infinite
loop cycling between the abort function and the signal handler.
The core file is /mit/jfc/core.rn. You will need to log into my workstation
or load my kernel onto one of the SIPB RTs to debug this core file.
