On Fri, Apr 17, 2015 at 12:06 AM, <
jean-christophe.gillet@schneider-electric.com> wrote:

> Hello Charles,
>
> Thank you for your answer.
> Just to be sure. Are you talking about flags skip_checkpt_rd and
> skip_checkpt_wr
>
Yes, that is correct.
You need to either use skip_checkpt_wr when unmounting before the change or
skip_checkpt_rd when mounting after the change.

>
> Do you think this sequence is correct for resizing up ?
> for the new partition size
>
1/  create yaffs device structure with  skip_checkpt_rd=1 and
> skip_checkpt_wr=1 and mount partition
>
Just skip_checkptr_wr is enough here.

At this point you will need to make some change to invalidate any existing
checkpoint. For example, create and delete a 0 byte file.
yaffs_open("/mntpoint/junk_file", O_RDWR | O_CREAT, 0666);
yaffs_unlink("/mntpoint/junk_file");


> 2/ unmount partition and delete yaffs device structure
> 3/ same as 1 but without skipping checkpoint
>
You could alternatively do skip_checkpt_rd here


> 4/ yaffs_sync on partition to have checkpoint saved
>
> I tested with verbose messages (YAFF_TRACE_VERIFY, YAFFS_TRACE_CHECKPOINT and
> YAFFS_TRACE_ALWAYS) and I didn't find strange messages
>
> Best regards
> Jean-Christophe
>
> _____________________________________________________________________________________
>
>
> * Jean-Christophe GILLET*  |  * Schneider Electric **  |  Industry
> Business - Process Automation R&D*  |
> *Firmware Engineer Phone:* +33 (0)4 92 13 3780 or 543780
> * Email:* *jean-christophe.gillet@schneider-electric.com*
> <%20jean-christophe.gillet@schneider-electric.com>  |   *Site:*
> * www.schneider-electric.com* <http://www.schneider-electric.com/>  |
> *Address:* 8ème Rue, ZI de Carros BP 147, 06516 Carros Cedex - France
> *** Please consider the environment before printing this e-mail
>
>
> [image: Inactive hide details for Charles Manning ---15/04/2015
> 09:11:47---Hello Jean-Christophe On Tue, Apr 14, 2015 at 8:36 PM, <]Charles
> Manning ---15/04/2015 09:11:47---Hello Jean-Christophe On Tue, Apr 14, 2015
> at 8:36 PM, <
>
> De : Charles Manning <cdhmanning@gmail.com>
> A : Jean-Christophe Gillet/Aut/Schneider@Europe,
> Cc : YAFFS ML <yaffs@lists.aleph1.co.uk>
> Date : 15/04/2015 09:11
> Objet : Re: [Yaffs] yaffs2 - resize up
> ------------------------------
>
>
>
> Hello Jean-Christophe
>
>
> On Tue, Apr 14, 2015 at 8:36 PM, <
> *jean-christophe.gillet@schneider-electric.com*
> <jean-christophe.gillet@schneider-electric.com>> wrote:
>
>
>    Hi,
>
>    I would like to resize up a partition in a special use case.
>
>    I have a 256MB nand chip with 2 yaffs partitions
>    /boot of 32MB - nand blocks from 1 to 261
>    /usr of 128MB - nand blocks from 262 to 1290
>    nand blocks from 1291 to 2047 was never used and are clean
>    The resize will be done only once and at boot time. /usr will be
>    increased to 224MB (blocks from 262 to 2047)
>
>    Because of yaffs2 architecture, do you think it is possible to resize
>    up "/usr" partition without any issue on filesytem ?
>
>    Jean-Christophe
>    PS: my operating system is vxWorks
>
>
>    Yes it is very possible to do this, but there are some important
>    issues to doing this properly.
>
>    Let us assume you're doing this while the Yaffs partition in NOT
>    mounted, just to make things simpler...
>
>    1) Make sure the blocks you add are erased (or bad blocks). If you add
>    blocks with data in them, then that will cause confusion.
>    2) Make sure the first time you mount after the change that the
>    checkpoint is not used. This can be achieved by one of:
>    a) Set flags so that the checkpoint is NOT saved.
>    b) Set flags so that the checkpoint is ingroed when booting.
>
>    If your system never writes a checkpoint (eg you always kill power and
>    never do unmounts or other checkpoint writes) then you do not need to worry
>    about (2). If that is the case, and the blocks were never written, then
>    just expand the partition size and your job is done.
>
>    You can even do a resize down of the file system, but that is a little
>    more complicated.
>
>    -- Charles
>
>
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud service.
> ______________________________________________________________________
>
>