[Trac #617] JFFS2 garbage collection thread starts at boot time.

[Zarro Boogs per Child]

#617: JFFS2 garbage collection thread starts at boot time.
--------------------------+-------------------------------------------------
 Reporter:  jg            |        Owner:  dwmw2  
     Type:  defect        |       Status:  new    
 Priority:  normal        |    Milestone:  BTest-3
Component:  display       |   Resolution:         
 Keywords:  performance.  |  
--------------------------+-------------------------------------------------
Comment (by marcelo):

 dwmw2 says:
 >The same can be achieved with 'killall -STOP jffs2_gcd_mtd0' immediately
 >after remounting the root file system read/write. When you're done,
 >'killall -CONT jffs2_gcd_mtd0'.

 blizzard says:

 > I'm loathe to add another shell command to startup since we're on a huge
 > binge to try and remove as much of that as possible.  This also feels
 > really fragile to me.  i.e. it seems very likely that we might never
 > -CONT the thread and then something strange would happen over time.
 >
 > Can't we add a parameter to tune this or something else?  Startup time
 > is really important to us, but just works is important too.  Feels
 hacky.

 marcelo says:

 > Well, the gc thread already supports being stopped/start via signals, so
 > the hack I proposed is (other than being a hack) redundant.
 > I don't see any reason why we can't do the STOP/CONT reliably from the
 > init scripts... (and invoking "kill" is pretty cheap).
 > Other than that, I think that gc thread should be limited to a certain
 > percentage of CPU, otherwise the system will be sluggish for the initial
 > minutes of execution, which can result in bad impressions from users
 > evaluating the system.

-- 
Ticket URL: <http://dev.laptop.org/ticket/617#comment:2>
One Laptop Per Child <http://laptop.org/>