Skip to contentSkip to conversation

Hello list!

DISCUSSION gianluca 2 participants 3 messages Feb 12, 2014 2 messages not in the archive NILFS2 filesystem

lore
Share & tools
lore

gianluca

author
#0 lore raw

Hello list!

I am working on a selfmade board with ARM9 and eMMC as storage.
The final product will be a logger (from solar-inverter, to bus-logger, etc.,...) so I choose a rootfilesystem with NILFS2. The distro is a deboostrap based for ARMEL Debian Wheezy 7.2.

I changed the critical filesystem folders (/var, /var/log, /tmp...) to be tempfs in ram for obviously reasons and the rootfilesystem is nilfs2 based with a noatime option activated.

The protection period is set at 60 seconds because the application can write data even in shorter periods, but 60 seconds it is a average value to have a disk-full problem away from normal-user.

What about the wear-leveling of this configuration of the eMMC device?

Can you point me on some tools, utilities to check out this issue?

Best regards,

Signature · 26 lines
           ,,,
          (o o)
======oOO==(_)==OOo======

Gianluca Renzi
R&D
phone: +39.0542.609120
fax:   +39.0542.609212

      .oooO  Oooo.
======(   )==(   )=======
       \ (    ) /
        \_)  (_/

===================
| I would like to |
| fix the world,  |
| but they're not |
| giving me the   |
 \ source code!  /
  ---------------
    ¯\_(".)_/¯
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

A message that is not in the archive (sent off-list or not yet received): 52FB49EA.30004-RMR/+jklkADqD7ZJT3/[email protected]

#2 lore raw

Hi Gianluca,

Quoted 1 line

What about the wear-leveling of this configuration of the eMMC device?

Actually nilfs2 is by design as friendly to flash devices with a weak controller as possible.
It only performs sequential writes, yielding a perfectly equal distribution of writes across the device.
Nilfs2 even overwrites static data which won't be written to when using other filesystems - which is great for devices supporting only dynamic wear leveling (usually SD, mmc, emmc, ...).

However, there is one issue left - the (two) superblocks are written to periodically in short intervals.
There is currently an ongoing discussion about this and Andreas has posted an experimental patch which works well on my raspberry pi.

Quoted 1 line

Can you point me on some tools, utilities to check out this issue?

What I did to observe the write-patterns of various file-systems was to create a ram-disk using BUSE: https://github.com/acozzette/BUSE
It delivers an example-program which can be modified to log each write-access, which I imported as CSV into LibreOffice for visualization.

Regards, Clemens
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html

A message that is not in the archive (sent off-list or not yet received): CAFvQSYTuM7Zjfz47UGU3KnTgH1iSRWiiZ3iSqXUiu7MKFnOmFA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org

author
#4 lore raw
Quoted 4 lines · Clemens Eisserer

Hi Gianluca,

Quoted 1 line

What about the wear-leveling of this configuration of the eMMC device?

Quoted 8 lines

Actually nilfs2 is by design as friendly to flash devices with a weak controller as possible.
It only performs sequential writes, yielding a perfectly equal distribution of writes across the device.
Nilfs2 even overwrites static data which won't be written to when using other filesystems - which is great for devices supporting only dynamic wear leveling (usually SD, mmc, emmc, ...).

Ok. good news to me.

Quoted 5 lines

However, there is one issue left - the (two) superblocks are written to periodically in short intervals.
There is currently an ongoing discussion about this and Andreas has posted an experimental patch which works well on my raspberry pi.

What patch? Can you give some additional info? I am running kernel 3.12.2 vanilla.

Quoted 7 lines
Quoted 1 line

Can you point me on some tools, utilities to check out this issue?

What I did to observe the write-patterns of various file-systems was to create a ram-disk using BUSE: https://github.com/acozzette/BUSE
It delivers an example-program which can be modified to log each write-access, which I imported as CSV into LibreOffice for visualization.

can you explain better?

Regards,

Signature · 26 lines
           ,,,
          (o o)
======oOO==(_)==OOo======

Gianluca Renzi
R&D
phone: +39.0542.609120
fax:   +39.0542.609212

      .oooO  Oooo.
======(   )==(   )=======
       \ (    ) /
        \_)  (_/

===================
| I would like to |
| fix the world,  |
| but they're not |
| giving me the   |
 \ source code!  /
  ---------------
    ¯\_(".)_/¯
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html