Bug #1159

fdisk: If no partition table, partition 4 is a whole-disk DragonFly partition? (plus bonus bug)

Added by floid over 5 years ago. Updated over 5 years ago.

Status:ClosedStart date:
Priority:NormalDue date:
Assignee:-% Done:

0%

Category:-
Target version:-

Description

Hi all. I decided to do a fresh install today, and discovered possible quirks
with fdisk, and also head on ad devices, using the Nov 2 snapshot CD
(2.1.0-DEVELOPMENT, Sun Nov 2 00:57 CET 2008).

As part of this adventure, I installed a new-in-box PATA disk, booted from the
install CD, and for curiosity ran fdisk against it. fdisk properly complains
that the partition table is invalid (the disk should be blank) but to my
surprise then states that partition 4 is a type 165 Free/Net/DragonFly partition
spanning the entire disk.

Just to make sure the disk was blank, I tried a quick "head /dev/ad0 | hd" and
was surprised to find that it hangs. A "cat /dev/ad0" will fill the screen with
nuls until terminated, as expected. A ktrace of head piped through stdout shows
that it's reading data but eventually appears to hang in GIO after the last read
attempt. (I have nowhere to throw the dump output right now, but it should be
easily reproducible later if it's of interest.)

Notes:
* I'm seeing no problem with fdisk against disks that actually have tables, this
is just potentially-confusing output when confronted with one that doesn't.

* If `head adN` is not expected to behave that way and this can't be reproduced
elsewhere, I'm still using the same board with an ATI SB600 for the ATA controllers.

* The Maxtor disk is a "DiamondMax 10."

* head works against the (formatted, partitioned) Western Digital drive I'd been
using with 2.0 (tested under 2.0 and this 2.1 CD) with no problems. Strange!
Perhaps the Maxtor has a firmware bug? Or perhaps all will be well after I
actually write something to it?

I originally had the Maxtor hooked up with a generic PATA-to-SATA adapter, but
moved it to a plain PATA connection to confirm before running the ktrace. The
adapter obviously was not the culprit, but did inspire me to take a look,
because it has a bus activity LED and I noticed it was thrumming softly
when head otherwise appeared hung.

Of course, I removed it and went to a normal PATA connection before trying the
ktrace, but the experience with the adapter's LED would seem to indicate there's
constant bus activity even *after* head reaches the read attempt that never
seems to complete.

-Keeping it confusing,
-Floid

History

#1 Updated by swildner over 5 years ago

Hmm, yeah, for some reason, fdisk(8) will create partition 4 on a blank
disk. I never really got the reason why it does that.

Sascha

#2 Updated by floid over 5 years ago

Update: Yay, I now have a shiny new install with a HAMMER partition to explore. No disk errors or I/O errors, and after partitioning and formatting it's quite possible to peek at the start of the disk with head. So it only occurs with blank disks -- still strange and possibly of interest to someone.

-Cheers,
-Floid

#3 Updated by corecode over 5 years ago

old bsd behavior, not a bug

Also available in: Atom PDF