Re: where are fsck return values documented?

On Mar 17,  3:14pm, reedreedmedia.net ("Jeremy
C. Reed") wrote:
-- Subject: Re: where are fsck return values documented?

| > They are documented now in fsck.8
| 
| Thank you for documenting these.
| 
| Now I noticed it reboots every time (per exit code 4 and
rc.d/fsck script) 
| after MARKING FILE SYSTEM CLEAN with "Root filesystem
was modified, 
| rebooting ..."
| 
| I don't recall it ever doing that before. I know it had
option to reboot 
| before but I don't remember ever noticing that happening
to me.
| 
| Something else change to cause this?
| 
| I see the new fsck man page says "and updating the
mount was 
| unsuccessful". I just don't think I saw any
indication of that -- or maybe 
| I don't know what that means.
| 
| Is this necessary? Why?

The code in question is main.c:

	if (hotroot()) {
		struct statvfs stfs_buf;
		/*
		 * We modified the root.  Do a mount update on
		 * it, unless it is read-write, so we can continue.
		 */
		if (statvfs("/", &stfs_buf) == 0) {
			long flags = stfs_buf.f_flag;
			struct ufs_args args;

			if (flags & MNT_RDONLY) {
				args.fspec = 0;
				flags |= MNT_UPDATE | MNT_RELOAD;
				if (mount(MOUNT_FFS, "/", flags,
				    &args, sizeof args) == 0)
					return FSCK_EXIT_OK;
			}
		}
		if (!preen)
			pwarn("n***** REBOOT NOW *****n");
		sync();
		return FSCK_EXIT_ROOT_CHANGED;
	}

I don't see anything wrong with it. Can you add some
debugging to see
what's going on?

christos

On Mon, 17 Mar 2008, Christos Zoulas wrote:

> | Now I noticed it reboots every time (per exit code 4
and rc.d/fsck script) 
> | after MARKING FILE SYSTEM CLEAN with "Root
filesystem was modified, 
> | rebooting ..."
> | 
> | I don't recall it ever doing that before. I know it
had option to reboot 
> | before but I don't remember ever noticing that
happening to me.
> | 
> | Something else change to cause this?
> | 
> | I see the new fsck man page says "and updating
the mount was 
> | unsuccessful". I just don't think I saw any
indication of that -- or maybe 
> | I don't know what that means.
> | 
> | Is this necessary? Why?
> 
> The code in question is main.c:
> 
> 	if (hotroot()) {
> 		struct statvfs stfs_buf;
> 		/*
> 		 * We modified the root.  Do a mount update on
> 		 * it, unless it is read-write, so we can continue.
> 		 */
> 		if (statvfs("/", &stfs_buf) == 0) {
> 			long flags = stfs_buf.f_flag;
> 			struct ufs_args args;
> 
> 			if (flags & MNT_RDONLY) {
> 				args.fspec = 0;
> 				flags |= MNT_UPDATE | MNT_RELOAD;
> 				if (mount(MOUNT_FFS, "/", flags,
> 				    &args, sizeof args) == 0)
> 					return FSCK_EXIT_OK;
> 			}
> 		}
> 		if (!preen)
> 			pwarn("n***** REBOOT NOW *****n");
> 		sync();
> 		return FSCK_EXIT_ROOT_CHANGED;
> 	}
> 
> I don't see anything wrong with it. Can you add some
debugging to see
> what's going on?

I will try to. This is on my main laptop so don't have
chance to reboot 
often but I can try to HEAD in bochs or other emulator.

I noticed ffs_reload() in sys/ufs/ffs/ffs_vfsops.c was
updated in January. 
(I didn't start using HEAD on here until 10 days ago.)

Also MNT_RELOAD is not documented in mount(2).


  Jeremy C. Reed