lvresize.c: use kilobytes when passing sizes to fsadm

lvresize.c: use kilobytes when passing sizes to fsadm

[Bryn M. Reeves]

Hi all,

The code in lvresize.c computes a new file system size to pass to 
fsadm when resizing as (extents * extent_size), giving a result in 
kilobytes.

This is then passed on to fsadm without a suffix, meaning that fsadm 
interprets it as a count of the number of file system blocks instead.

This breaks fsadm for all file system block sizes other than 1K.

The attached patch adds a "K" suffix and fixes the problem for me - see:

https://bugzilla.redhat.com/show_bug.cgi?id=480022

For testing details.

Regards,
Bryn.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: lvm2-lvresize-units.patch
URL: <http://listman.redhat.com/archives/lvm-devel/attachments/20090114/26d3d877/attachment.ksh>

lvresize.c: use kilobytes when passing sizes to fsadm

[Milan Broz]

Bryn M. Reeves wrote:
> The code in lvresize.c computes a new file system size to pass to 
> fsadm when resizing as (extents * extent_size), giving a result in 
> kilobytes.
> 
> This is then passed on to fsadm without a suffix, meaning that fsadm 
> interprets it as a count of the number of file system blocks instead.
> 
> This breaks fsadm for all file system block sizes other than 1K.
> 
> The attached patch adds a "K" suffix and fixes the problem for me - see:
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=480022

Reviewed-by: Milan Broz <mbroz@redhat.com>
Tested-by: Milan Broz <mbroz@redhat.com>

(btw lvresize -r still works but is is not mentioned in man page, it is
intentional to prefer using fsadm --lvresize now?)

Milan

p.s.
Jaroslav, please can you add some tests internally for fsadm/lvresize?
(test in bugzilla, just extend/reduce for 1024,2048,4096 fs blocksize,
mke2fs -b <bs> is enough (or maybe xfs, reiserfs tests too))

lvresize.c: use kilobytes when passing sizes to fsadm

[Bryn M. Reeves]

Milan Broz wrote:
> Bryn M. Reeves wrote:
>> The code in lvresize.c computes a new file system size to pass to 
>> fsadm when resizing as (extents * extent_size), giving a result in 
>> kilobytes.
>>
>> This is then passed on to fsadm without a suffix, meaning that fsadm 
>> interprets it as a count of the number of file system blocks instead.
>>
>> This breaks fsadm for all file system block sizes other than 1K.
>>
>> The attached patch adds a "K" suffix and fixes the problem for me - see:
>>
>> https://bugzilla.redhat.com/show_bug.cgi?id=480022
> 
> Reviewed-by: Milan Broz <mbroz@redhat.com>
> Tested-by: Milan Broz <mbroz@redhat.com>
> 

Thanks Milan!

> 
> p.s.
> Jaroslav, please can you add some tests internally for fsadm/lvresize?
> (test in bugzilla, just extend/reduce for 1024,2048,4096 fs blocksize,
> mke2fs -b <bs> is enough (or maybe xfs, reiserfs tests too))

Sorry, I meant to take a look at the test suite yesterday to see if I 
could figure out adding tests for this but ran out of time.

Cheers,
Bryn.