On Wednesday 09 April 2014 22:11:54 Ketil Froyn wrote: > If you set yaffs_auto_checkpoint to 0 it sounds like it'll write a > checkpoint on unmount, so it wouldn't need to check the whole NAND on boot > (if you do clean unmounts on shutdown). If you're aiming to save space, it > might be worth trying 0 instead of the default 1, sounds like you'll get > fewer checkpoints that way. There is only ever, at maximum, one checkpoint per partition, thus turning auto checkpointing on/off has no impact on the space used by the checkpoint data. The auto checkpointing will write a checkpoint if the system is idle for some time. I don't really recommend this as being very useful. In the days before block summaries, a checkpoint was very valuable because the boot time difference between a full scan and a mount from checkpoint could be large. Block summaries have closed the gab considerably, meaning the penalty for not having a checkpoint is a lot less. > > There's also this, from the link I posted: > > skipCheckpointWrite > Yaffs2 only: Flag to skip writing checkpoint on sync or unmount. > > That should probably get rid of all checkpoints if that's what you want, > but since it's only listed under "Initialisation Parameters", maybe you > have to set this on file system creation. Not even sure how, though. > > Sorry I can't be more specific, but my knowledge of yaffs2 is very > theoretical, I have little practical experience using it. Charles would be > the authority on this stuff. > > Regards, Ketil > > On 9 April 2014 11:47, albhuang wrote: > > Hi Ketil, > > A global variable, yaffs_auto_checkpoint, control the checkpoint action, > > just like: > > If 0 only checkpoint on unmount. > > If 1 checkpoint on sync. > > If 2 checkpoint on write_super and sync. > > Default is 1. > > I'd like to know what would happen if it was set to > > zero(yaffs_auto_checkpoint = 0), only a little slow during booting > > because it would check the whole NAND, was it? You can force the checkpoint writing/reading to to skipped with the no-checkpoint-read and no-checkpoint-write mount flags. If there is no checkpoint, or the checkpoint is skipped, then yaffs2 will do a scan (using block summaries if they are available). > > > > Thanks! > > > > ------------------------------ > > albhuang > > > > *From:* Ketil Froyn > > *Date:* 2014-04-09 13:53 > > *To:* albhuang > > *Subject:* Re: Re: [Yaffs] umount YAFFS2 block: the used NAND space was > > increased > > > > Not sure how it can be disabled, but looking at > > http://www.yaffs.net/yaffs-tuning-and-configuration#Initialisation_Parame > >tersit seems possible. > > > > If you redo the math I think you will find that the difference was 128k > > after all, but perhaps you did that already. :) > > > > On 9 Apr 2014 04:05, "albhuang" wrote: > >> Ketil, > >> It seemed the checkpoint action caused the error,could i abandon it, or > >> anything could i do? > >> > >> Best Regards! > >> > >> ------------------------------ > >> albhuang > >> > >> *From:* Ketil Froyn > >> *Date:* 2014-04-08 17:42 > >> *To:* albhuang > >> *CC:* yaffs > >> *Subject:* Re: [Yaffs] umount YAFFS2 block: the used NAND space was > >> increased > >> Could it perhaps be a yaffs2 checkpoint that was written on unmount so > >> that it'll mount quickly next time? > >> > >> Cheers, Ketil > >> > >> On 8 April 2014 10:07, albhuang wrote: > >>> Hi, > >>> I utilized yaffs2 to K9F2G08(1Page=2K+64Bytes), the kernel was > >>> linux2.6.12. All seemed OK except the command, umount, was executed, > >>> the used space would be increased, each time, when a umount command > >>> followed by the mount command. What's wrong with it. > >>> > >>> The following was the infoes. > >>> > >>> Thanks! > >>> > >>> [root@INTUM /]$mount > >>> /dev/mtdblock/5 on /mnt/database type yaffs2 (rw,noatime) > >>> [root@INTUM /]$df > >>> Filesystem 1k-blocks Used Available Use% Mounted on > >>> /dev/mtdblock/5 208896 1292 207604 1% /mnt/database > >>> //the used space was 1292K > >>> > >>> [root@INTUM /]$umount /mnt/database/ > >>> save exit: isCheckpointed 1 > >>> [root@INTUM /]$mount -t yaffs2 /dev/mtdblock/5 /mnt/database/ > >>> yaffs: dev is 32505861 name is "mtdblock5" > >>> yaffs: passed flags "" > >>> yaffs: Attempting MTD mount on 31.5, "mtdblock5" > >>> [root@INTUM /]$df > >>> Filesystem 1k-blocks Used Available Use% Mounted on > >>> /dev/mtdblock/5 208896 1420 207476 1% /mnt/database > >>> //the used space increased to 1420K > >>> > >>> > >>> > >>> > >>> > >>> > >>> _______________________________________________ > >>> yaffs mailing list > >>> yaffs@lists.aleph1.co.uk > >>> http://lists.aleph1.co.uk/cgi-bin/mailman/listinfo/yaffs > >> > >> -- > >> -Ketil