#9455 NORM 1.5-F11: maybe use data=ordered for root mount
Zarro Boogs per Child
bugtracker at laptop.org
Mon Nov 16 18:53:52 EST 2009
#9455: maybe use data=ordered for root mount
---------------------------------+------------------------------------------
Reporter: cjb | Owner: cjb
Type: defect | Status: new
Priority: normal | Milestone: 1.5-F11
Component: distro | Version: Development build as of this date
Resolution: | Keywords:
Next_action: never set | Verified: 0
Deployment_affected: | Blockedby:
Blocking: |
---------------------------------+------------------------------------------
Changes (by Quozl):
* version: not specified => Development build as of this date
* component: not assigned => distro
Comment:
Related to #9612 which describes Sugar's response to zero-filled or
uninitialised files after a crash.
''data=writeback'' is what we are using on os40. mount(8) says this is
''rumoured to be the highest-throughput option''. This rumour may be true
only for storage with a rotational latency.
''data=ordered'' is the default for mount(8), and other systems use this
as the default, and it would have hidden the symptom in #9612.
Applications seem to be written to match this common behaviour. It is
effectively how jffs2 on 802 worked as well.
I'm ambivalent. It's a huge debate, plenty of community discussion,
especially around ext4 and solid state disks.
We could also consider fiddling with ''commit=nrsec'' ... or add a
''sync'' to the system's response to a power button down event, so that a
user forced power down would be preceeded with an attempt to write data to
the SD.
--
Ticket URL: <http://dev.laptop.org/ticket/9455#comment:2>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system
More information about the Bugs
mailing list