[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: [dm-devel] dmsetup on uClibc based system
- From: Kevin Corry <kevcorry us ibm com>
- To: DevMapper <dm-devel redhat com>
- Cc:
- Subject: Re: [dm-devel] dmsetup on uClibc based system
- Date: Fri, 19 Mar 2004 09:46:15 -0600
On Thursday 18 March 2004 6:45 pm, Pete Flugstad wrote:
> Greetings
>
> (resend now that I'm subscribed and sending from the right
> account...)
>
> I'm trying to get dmsetup running on an embedded uClibc based
> system - linux kernel 2.6.4 - X86 architecture, GCC 3.3.1. The first
> problem seems to be that when reading the table file, dmsetup uses the
> 'a' conversion modifier to have sscanf malloc space for the resulting
> string. uClibc doesn't appear to support this. No biggy, fairly easy
> fix (I believe - I assume ttype won't be larger than the line read in).
> Patch attached.
>
> Next problem is that I'm getting an error when I try and setup a
> "chunk" of a partition:
>
> [root denis root]# dmsetup create hdc5.000
> 0 204800 linear /dev/hdc5 0<cr>
> <ctrl-D>
> device-mapper ioctl cmd 9 failed: Invalid argument
> Command failed
> [root denis root]# dmsetup ls
> hdc5.000 (254, 0)
> [root denis root]# ls /dev/mapper/
> control
>
> The command appears to have worked, but the entry in /dev/mapper is
> not there.
>
> Does dmsetup not work on partitions? Or is this probably something
> else going on? Does it matter if the partition in question is mounted?
Actually, it does matter. You won't be able to activate a DM device on a
partition that contains a mounted filesystem (2.6 kernels don't allow that).
Try "dmsetup info hdc5.000" and/or "dmsetup table hdc5.000" to see if the
table was actually loaded into the device. Remember that a DM device can
exist without any mapping (all I/O to that device will obviously fail).
--
Kevin Corry
kevcorry us ibm com
http://evms.sourceforge.net/
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]