Date:	Fri, 20 May 2011 11:07:14 +0200 (CEST)
From:	Lukas Czerner <lczerner@...hat.com>
To:	Eric Sandeen <sandeen@...hat.com>
cc:	Lukas Czerner <lczerner@...hat.com>, linux-ext4@...r.kernel.org,
	tytso@....EDU, adilger@...ger.ca
Subject: Re: [PATCH v2 1/2] e2fsprogs: Fix how we treat user-spcified filesystem
 size

On Thu, 19 May 2011, Eric Sandeen wrote:

> On 2/9/11 7:03 AM, Lukas Czerner wrote:
> > (v2: do not spell units - mebibytes etc... - but rather use just abbrev.)
> > 
> > mke2fs and resize2fs allows specifying filesystem size as a parameter,
> > former called as blocks-count. However it has been a little bit messy so
> > this commit fixes it mainly by updating man pages.
> > 
> > We can not specify filesystem size in blocks count without specifying
> > blocksize as well. It is because we need blocks count to determine
> > filesystem type, and we need filesystem type to determine blocksize. So
> 
> how does blocks count determine filesystem type?  (because we need room
> for the journal?)  This isn't clear from the commit message...

It was not my intention to explain specifically why we need block count
to determine fs type etc. But the "type" is probably not the best name.
The thing is that, we need it to determine filesystem "type" like
"floppy,small,default,big,huge" and then use appropriate settings from
mke2fs.conf. DO you think it is worth it to mention that in the
descritpion ?

> 
> > it should not be allowed, however due to compatibility reason it should
> > be still possible, so at least print warning message for now, so we can
> > easily restrict that later.
> > 
> > Filesystem size can be specified using units as suffixes. This was not
> > documented for mke2fs, so this commits adds proper documentation into
> > mke2fs man page.
> > 
> > For the sake of completeness add 'b' (blocks count) unit.
> 
> Some notes below
> 
> > Signed-off-by: Lukas Czerner <lczerner@...hat.com>
> > ---
> >  lib/e2p/parse_num.c   |    9 +++++++++
> >  misc/mke2fs.8.in      |   32 ++++++++++++++++++++++++++------
> >  resize/resize2fs.8.in |   19 ++++++-------------
> >  3 files changed, 41 insertions(+), 19 deletions(-)
> > 
> > diff --git a/lib/e2p/parse_num.c b/lib/e2p/parse_num.c
> > index 83a329a..5e7924b 100644
> > --- a/lib/e2p/parse_num.c
> > +++ b/lib/e2p/parse_num.c
> > @@ -10,6 +10,7 @@
> >   */
> >  
> >  #include "e2p.h"
> > +#include "../misc/nls-enable.h"
> >  
> >  #include <stdlib.h>
> >  
> > @@ -37,6 +38,14 @@ unsigned long long parse_num_blocks2(const char *arg, int log_block_size)
> >  		num >>= (1+log_block_size);
> >  		break;
> >  	case '\0':
> > +	case 'b':
> > +		if (!log_block_size) {
> > +			fprintf(stderr,
> > +				_("Warning: You can not specify blocks count "
> > +				"without specifying block size '-b'. Will "
> > +				"assume kilobytes instead of blocks count!.\n"
> > +				));
> > +		}
> >  		break;
> >  	default:
> >  		return 0;
> > diff --git a/misc/mke2fs.8.in b/misc/mke2fs.8.in
> > index 2eead17..e093be0 100644
> > --- a/misc/mke2fs.8.in
> > +++ b/misc/mke2fs.8.in
> > @@ -108,7 +108,7 @@ mke2fs \- create an ext2/ext3/ext4 filesystem
> >  ]
> >  .I device
> >  [
> > -.I blocks-count
> > +.I filesystem-size
> >  ]
> >  @JDEV@.sp
> >  @JDEV@.B "mke2fs \-O journal_dev"
> > @@ -136,7 +136,7 @@ mke2fs \- create an ext2/ext3/ext4 filesystem
> >  @JDEV@]
> >  @JDEV@.I external-journal
> >  @JDEV@[
> > -@...V@.I blocks-count
> > +@...V@.I filesystem-size
> >  @JDEV@]
> >  .SH DESCRIPTION
> >  .B mke2fs
> > @@ -145,10 +145,30 @@ partition.
> >  .I device
> >  is the special file corresponding to the device (e.g
> >  .IR /dev/hdXX ).
> > -.I blocks-count
> > -is the number of blocks on the device.  If omitted,
> > -.B mke2fs
> > -automagically figures the file system size.  If called as
> > +.I filesystem-size
> > +is the size of the filesystem you want to create.
> > +If no units are specified, the units of the
> > +.I filesystem-size
> > +parameter shall be the blocks count of the filesystem. Keep in mind that
> 
> should be "shall be equal to the block size of the filesystem" right?
> 
> (blocks count implies the number of blocks in the fs, not the size of each block)
> 
> Oh, I see.
> 
> The number does represent the count of blocks.
> Therefore the -unit- of that number is the blocksize.
> 
> So saying "the units ... shall be the blocks count" isn't correct.

Hmm, I think the -unit- is really "block". I would say that filesystem
is, for example, 1024 blocks big (or long?) - hence the unit is blocks.

> 
> > +you can not specify
> 
> how about:
> 
> you should not specify ...

Good point, thanks:). We are trying to be polite and user-friendly:)

> 
> > +.I filesystem-size
> > +in blocks count without specifying the blocksize as well. This will be
> > +allowed for now (1024 byte blocks will be assumed)
> > +due to compatibility reasons, however it may be restricted
> > +in the future.
> 
> and then something like: "If blocksize is not specified on the
> commandline and filesystem-size is specified in block units,
> 1024-byte blocks will be assumed for this purpose, regardless
> of the filesystem blocksize ultimately chosen by mke2fs."
> 
> (which is what it does today, right?)

yes it is, I'll add the sentence thanks.

> 
> > +Optionally, the
> > +.I filesystem-size
> > +parameter may be suffixed by one of the following units
> > +designators: 'b', 's', 'K', 'M', 'G', or 'T',
> > +for blocks count, 512 byte sectors, KiB (2^10 Bytes), MiB (2^20 Bytes),
> 
> for blocks, 512 byte sectors ....
> 
> > +GiB (2^30 Bytes), or TiB (2^40 Bytes) respectively, which are binary (power-of-2),
> > +not decimal, units. The
> > +.I filesystem-size
> > +may never be larger than the size of the partition.
> > +If
> 
> If the 
> 
> > +.I filesystem-size
> > +parameter is not specified, it will default to the size of the partition.
> > +If called as
> >  .B mkfs.ext3
> >  a journal is created as if the
> >  .B \-j
> > diff --git a/resize/resize2fs.8.in b/resize/resize2fs.8.in
> > index e02345d..769fd14 100644
> > --- a/resize/resize2fs.8.in
> > +++ b/resize/resize2fs.8.in
> > @@ -38,27 +38,20 @@ The
> >  parameter specifies the requested new size of the filesystem.
> >  If no units are specified, the units of the
> >  .I size
> > -parameter shall be the filesystem blocksize of the filesystem.
> > +parameter shall be the blocks count of the filesystem.
> 
> shall be equal to the filesystem blocksize ...
> 
> >  Optionally, the 
> >  .I size
> > -parameter may be suffixed by one of the following the units 
> > -designators: 's', 'K', 'M', or 'G',
> > -for 512 byte sectors, kilobytes, megabytes, or gigabytes, respectively.
> > -The 
> > +parameter may be suffixed by one of the following the units
> > +designators: 'b', 's', 'K', 'M', 'G', or 'T',
> > +for blocks count, 512 byte sectors, KiB (2^10 Bytes), MiB (2^20 Bytes),
> 
> for blocks, 512 byte sectors ....
> 
> > +GiB (2^30 Bytes), or TiB (2^40 Bytes) respectively, which are binary (power-of-2),
> > +not decimal, units. The
> >  .I size
> >  of the filesystem may never be larger than the size of the partition.
> >  If 
> >  .I size
> >  parameter is not specified, it will default to the size of the partition.
> >  .PP
> > -Note: when kilobytes is used above, I mean
> > -.IR real ,
> > -power-of-2 kilobytes, (i.e., 1024 bytes), which some politically correct
> > -folks insist should be the stupid-sounding ``kibibytes''.  The same
> > -holds true for megabytes, also sometimes known as ``mebibytes'', or
> > -gigabytes, as the amazingly silly ``gibibytes''.  Makes you want to
> > -gibber, doesn't it?
> > -.PP
> >  The
> >  .B resize2fs
> >  program does not manipulate the size of partitions.  If you wish to enlarge
> 
> 

Thanks a lot for the review Eric! I'll repost ASAP.

-Lukas
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists