boot0cfg and kern.geom.debugflags (was: problem with boot0cfg on a twe?)

On 06/09/05 09:02, Randy Bush wrote:
>>I looked in my archives (well, it's actually at gmane):
>>
>>I got this from Doug White:
>>
>>>This is a erroneous message. The actual problem is:
>>>
>>>>   484 boot0cfg NAMI  "/dev/twed0"
>>>>   484 boot0cfg RET   open -1 errno 1 Operation not permitted
>>>>
>>>>This is a known problem with certain MBR layouts. To work around this
>>>>problem, set:
>>>>
>>>>sysctl kern.geom.debugflags=16
>>>>then try your boot0cfg. There's a protection mechanism that sometimes gets
>>>>confused by certain partition table layouts. Flag 16 disables that
>>>>protection.  I don't recommend running this unless you are explicitly
>>>>trying to updating something in a partition table-like area; its very easy
>>>>to destroy your system with the flag set!
>>
>>Can you try this?
> 
> bingo!!!
> 
>     # sysctl kern.geom.debugflags=16
>     kern.geom.debugflags: 0 -> 16
>     # boot0cfg -B -d 1 -s 1 -v twed0
>     #   flag     start chs   type       end chs       offset         size
>     1   0x80      0:  1: 1   0xa5   1023:254:63           63     72292437
> 
>     version=1.0  drive=0x1  mask=0xf  ticks=182
>     options=packet,update,nosetdrv
>     default_selection=F1 (Slice 1)

 From what I gather from Poul-Henning Kamp's posts on the matter, this 
is a design feature and not a bug.  If a disk is mounted in any way 
(including read-only), you may not update the MBR to prevent foot 
shooting.  The real problem is that the error that is returned gives 
little information.  There has not been a consensus on how to make 
things easier for the user.  Various ways to print friendly error 
messages have been proposed and shot down.

This issue is documented in boot0cfg(8) as the first entry in the BUGS 
section:
"Protection mechanisms in the geom(4) subsystem might prevent boot0cfg 
from being able to update the MBR on a mounted disk.  Instructions for 
temporarily disabling these protection mechanisms can be found in the 
geom(4) manpage."

Under the DIAGNOSTICS section of geom(4) describing the use of the 
kern.geom.debugflags sysctl:
"0x10 (allow foot shooting)
Allow writing to Rank 1 providers.  This would, for example, allow the
super-user to overwrite the MBR on the root disk or write random sectors 
elsewhere to a mounted disk.  The implications are obvious."

I'm not sure what "tracing" is so I don't understand why 0x02 and 0x04 
are necessary (to give us 0x16).

-- 
Jonathan Noack | noackjr_at_alumni.rice.edu | OpenPGP: 0x991D8195