Re: Problems creating and writing to disk slices
In message <[EMAIL PROTECTED]>, Darren Pilgrim writes: >I can understand the usefulness of preventing modifications applying to >mounted partitions, and I can see the logic in unliaterally preventing >them, but preventing modifying slice table and disk label entries for >unmounted portions of the disk gets more in the way than helps. You should be able to modify any non-open partition/slice provided you don't try to make them overlap an open one. You have to use the right options to the right programs on the right devices though. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Problems creating and writing to disk slices
[EMAIL PROTECTED] wrote: In message <[EMAIL PROTECTED]>, Joerg Wunsch writes: Darren Pilgrim <[EMAIL PROTECTED]> wrote: The above practices have worked fine for a long time in 4.x and still do even in 4.7p4, which is on this same machine. Get Matthew N. Dodd's patch at: ftp://ftp.jurai.net/users/winter/patches/geom-foot.patch (Hint: sysctl kern.geom.allow_foot_shooting=1) Then help me convincing phk that there might be legitimate needs for such an option. :-/ I can understand the usefulness of preventing modifications applying to mounted partitions, and I can see the logic in unliaterally preventing them, but preventing modifying slice table and disk label entries for unmounted portions of the disk gets more in the way than helps. Does this safety measure prevent slicing and labeling a completely empty disk or just those with the boot partition or mounted filesystems on them? You're out of date, one such has been committed to current. Is the patch appropriate for RELENG_5_0, though? To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Problems creating and writing to disk slices
In message <[EMAIL PROTECTED]>, Joerg Wunsch writes: >Darren Pilgrim <[EMAIL PROTECTED]> wrote: > >> The above practices have worked fine for a long time in 4.x and still do >> even in 4.7p4, which is on this same machine. > >Get Matthew N. Dodd's patch at: > >ftp://ftp.jurai.net/users/winter/patches/geom-foot.patch > >(Hint: sysctl kern.geom.allow_foot_shooting=1) > >Then help me convincing phk that there might be legitimate needs >for such an option. :-/ You're out of date, one such has been committed to current. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Problems creating and writing to disk slices
Darren Pilgrim <[EMAIL PROTECTED]> wrote: > The above practices have worked fine for a long time in 4.x and still do > even in 4.7p4, which is on this same machine. Get Matthew N. Dodd's patch at: ftp://ftp.jurai.net/users/winter/patches/geom-foot.patch (Hint: sysctl kern.geom.allow_foot_shooting=1) Then help me convincing phk that there might be legitimate needs for such an option. :-/ -- cheers, J"org .-.-. --... ...-- -.. . DL8DTL http://www.sax.de/~joerg/NIC: JW11-RIPE Never trust an operating system you don't have sources for. ;-) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Problems creating and writing to disk slices
I've been trying to move the installed OSes around on my hard disk, but am having a huge amount of trouble doing so. The task involves dd'ing the slices off the disk for safe-keeping, modifying the on-disk slice table, then dd'ing the slices back onto the disk in their new locations. However, sysinstall doesn't work for manipulating the slice table and labels. It gives me errors saying it couldn't write to the device. Manually doing the work using fdisk and disklabel yield multiple problems: - Fdisk returns device busy errors when trying to write to the slice table to make a modification to a slice that has no mounted filesystems. I can only modify the slice table if I boot from either DOS or an off-disk OS (like the Fix-It CD). - After the slice is created and I've rebooted back into 5.0p1, the lack of a /dev/ad4s1c device node means I have to use `disklabel -w -r ad4s1 auto` to create the c partition, then `disklabel -e -r ad4s1` to modify the label. The editor comes up with a label with the c partition offset of 0. I add the other partitions making calculations to ensure the partitions start/end on boundaries. Exiting the editor, disklabel complains that the 'c' partition doesn't begin with sector 0 and doesn't cover the whole disk, and also that my last partition goes past the end of the unit. Reeditting the label, the offsets are now shifted by 63. Re-adjusting the partition sizes and offsets doesn't fix the problem even though the numbers indicate that everything is aligned properly and within the slice boundaries. After growing quite frustrated with this, I gave up and booted into DOS and was able to create the slice. Booting back into 5.0p1, fdisk tells me the slice is fine, and the size and geometry are an exact match to what I was trying to make fdisk use. All that's left is to dd the original slice back into place, but `dd if=win98.ad4s1 of=/dev/ad4s1 bs=1024k` returns an error saying that writing to ad4s1 failed because the device is busy. How can the device be busy? The above practices have worked fine for a long time in 4.x and still do even in 4.7p4, which is on this same machine. The disklabel in 4.7p4 doesn't complain about the state of the labels, and sysinstall can manipulate the slice table and labels just fine on the live disk. What's changed in 5.x to make this not work, and what do I need to do to accomplish my task? To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message