Re: NetBSD/sun3 9.0_STABLE GENERIC hang

[Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>]

I wrote:

> I wrote on port-sun3@:
> 
> > - (Not Xserver related) recent netbsd-9 GENERIC kernel doesn't work?
> >   NetBSD-daily/netbsd-9/202007161930Z one hangs just after
> >   "enabling interrupts" message
> 
> By dumb manual bisect, it turns out this problem is caused by
> recent src/sys/dev/scsipi/scsiconf.c pullup:
> 
>  https://mail-index.netbsd.org/source-changes/2020/07/13/msg119325.html
>  > Continue scanning a SCSI bus when a LUN is reported not present
> 
> --- src/sys/dev/scsipi/scsiconf.c 2020/05/04 13:50:07 1.284.4.1
> +++ src/sys/dev/scsipi/scsiconf.c 2020/07/13 14:09:10 1.284.4.2
> @@ -887,6 +887,8 @@
>  		break;
>  
>  	case SID_QUAL_LU_NOTPRESENT:
> +		docontinue = 1;
> +		/* FALLTHROUGH */
>  	case SID_QUAL_reserved:
>  	case SID_QUAL_LU_NOT_SUPP:
>  		goto bad;
> 
> Adding printf before the "/* FALLTHROUGH */" line also solves the issue,
> so it looks caused by gcc bug, and actually depends on COPTS:
> -Os NG   (note sun3 uses -Os by default even for GENERIC)
> -O0 OK
> -O1 OK
> -O2 OK
> -O2 -fno-reorder-blocks OK
> -Os -freorder-blocks NG
> -Os -freorder-blocks -freorder-blocks-algorithm=stc OK
> -Os -freorder-blocks-algorithm=stc OK
> -Os -fno-reorder-blocks OK
> 
> Does anyone see the similar problem on other m68k ports?
> (especially on INSTALL kernels built with -Os)

This problem seems resolved in recent netbsd-9, at least 202009020530Z.
(maybe by gcc 7.5.0 update)

---
Izumi Tsutsui