cvs commit: src/sys/cam/scsi scsi_cd.c scsi_da.c src/sys/geom geom_disk.c geom_disk.h geom_subr.c

[Nate Lawson]

John Polstra wrote:
> jdp         2005-11-18 02:43:49 UTC
> 
>   FreeBSD src repository
> 
>   Modified files:
>     sys/cam/scsi         scsi_cd.c scsi_da.c 
>     sys/geom             geom_disk.c geom_disk.h geom_subr.c 
>   Log:
>   Fix a bug that caused some /dev entries to continue to exist after
>   the underlying drive had been hot-unplugged from the system.  Here
>   is a specific example.  Filesystem code had opened /dev/da1s1e.
>   Subsequently, the drive was hot-unplugged.  This (correctly) caused
>   all of the associated /dev/da1* entries to be deleted.  When the
>   filesystem later realized that the drive was gone it closed the
>   device, reducing the write-access counts to 0 on the geom providers
>   for da1s1e, da1s1, and da1.  This caused geom to re-taste the
>   providers, resulting in the devices being created again.  When the
>   drive was hot-plugged back in, it resulted in duplicate /dev entries
>   for da1s1e, da1s1, and da1.
>   
>   This fix adds a new disk_gone() function which is called by CAM when a
>   drive goes away.  It orphans all of the providers associated with the
>   drive, setting an error condition of ENXIO in each one.  In addition,
>   we prevent a re-taste on last close for writing if an error condition
>   has been set in the provider.
>   
>   Sponsored by:   Isilon Systems
>   Reviewed by:    phk
>   MFC after:      1 week
>   
>   Revision  Changes    Path
>   1.94      +1 -0      src/sys/cam/scsi/scsi_cd.c
>   1.181     +1 -0      src/sys/cam/scsi/scsi_da.c
>   1.98      +12 -0     src/sys/geom/geom_disk.c
>   1.5       +1 -0      src/sys/geom/geom_disk.h
>   1.88      +3 -2      src/sys/geom/geom_subr.c
> 
> 
> Index: src/sys/geom/geom_disk.c
> diff -u src/sys/geom/geom_disk.c:1.97 src/sys/geom/geom_disk.c:1.98
> --- src/sys/geom/geom_disk.c:1.97	Fri Sep 30 17:32:08 2005
> +++ src/sys/geom/geom_disk.c	Fri Nov 18 02:43:49 2005
> @@ -419,6 +419,18 @@
>  	g_post_event(g_disk_destroy, dp, M_WAITOK, NULL);
>  }
>  
> +void
> +disk_gone(struct disk *dp)
> +{
> +	struct g_geom *gp;
> +	struct g_provider *pp;
> +
> +	gp = dp->d_geom;
> +	if (gp != NULL)
> +		LIST_FOREACH(pp, &gp->provider, provider)
> +			g_orphan_provider(pp, ENXIO);
> +}
> +

Does there need to be locking for this list traversal?  Couldn't 
disk_gone() race in parallel with a taste event if someone plugs/unplugs 
quickly, especially for a slow device (i.e. floppy)?

-- 
Nate