Re: Same core dump writen to /var/crash at every boot

On Wed, 1 Oct 2025, pin wrote:

> I don't remember the system crashing but, the problem is that every time I start-up my system, I get a core dump written to /var/crash
>
> ~> ls /var/crash
> [...]
> .rw------- root wheel 97 MB Wed Oct  1 09:33:34 2025  netbsd.6.core.gz
> .rw------- root wheel  0 B  Wed Oct  1 09:33:34 2025  netbsd.6.gz
> .rw------- root wheel 97 MB Wed Oct  1 10:10:20 2025  netbsd.7.core.gz
> .rw------- root wheel  0 B  Wed Oct  1 10:10:21 2025  netbsd.7.gz
>
> Anyone who knows how to clean this up?
>

I know you've already fixed this, but, I think I can offer an explanation,
given those file sizes. Usually, when savecore(8) finishes successfully, it
"clears" the coredump and you get file sizes like:

```
$ ls -l
total 797284
-rw------- 1 root wheel         2 Jul 26 04:39 bounds
-rw------- 1 root wheel         5 Jun  3 20:25 minfree
-rw------- 1 root wheel  47920856 May 28 03:41 netbsd.0.core.gz
-rw------- 1 root wheel    505090 May 28 03:41 netbsd.0.gz
-rw------- 1 root wheel  48055906 May 30 04:47 netbsd.1.core.gz
-rw------- 1 root wheel    505090 May 30 04:47 netbsd.1.gz
-rw------- 1 root wheel  51140776 Jun  4 07:55 netbsd.2.core.gz
-rw------- 1 root wheel    505352 Jun  4 07:55 netbsd.2.gz
-rw------- 1 root wheel 297434961 Jul 18 04:21 netbsd.3.core.gz
-rw------- 1 root wheel    505431 Jul 18 04:21 netbsd.3.gz
-rw------- 1 root wheel 368701658 Jul 26 04:40 netbsd.4.core.gz
-rw------- 1 root wheel    505361 Jul 26 04:40 netbsd.4.gz
$
```

I think, in your case, savecore(8) is encountering some error (reading the
coredump from swap?); trying to write a diagnostic; but, as all this takes time,
and because of the way the rc scripts are run, gets killed with a SIGPIPE as
soon as it writes a single char. See here for an example:

https://mail-index.netbsd.org/netbsd-users/2025/04/04/msg032519.html

Should be easy to fix this part, at least (just ignore SIGPIPE).

-RVP