[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: What on Earth is JFFS2 GC doing?



David Woodhouse wrote:

> Hmmm. That's a little bit more than the amount of hysteresis I'd expect
> from jffs2_statfs(). We make stuff up, sure - but 600KiB difference is more
> than I'd expect.
>
> What did the debug log say all the size variables were during both of those?

Blimey! I didn't have debug turned on at that time :(

I'm redoing the test- filling the fs- but this time I'll provide you with your
debug log
with the "df" and the "JitterTest" going on.


But I have some interesting results already. To "hurry" things along I
"prefilled" the flash
with two kernel images, each about 600KB.

Check this out:
<~34% full, original results with just JitterTEst and binary file being copied
and deleted>

<SNIP>
  100.002000 ms   0.002000 ms
CHIP_ERASE:Erasing_at_address: 0x300000
 2959.247000 ms   2859.247000 ms
 610.814000 ms   510.814000 ms
 107.808000 ms   7.808000 ms
<SNIP>


<Now with those two "prefilled" files on the fs, again at ~34% full>

<SNIP>
 100.004000 ms   0.004000 ms
 99.998000 ms   -0.002000 ms
 100.002000 ms   0.002000 ms
CHIP_ERASE:Erasing_at_address: 0xc0000
 570.311000 ms   470.311000 ms
 609.694000 ms   509.694000 ms
 99.996000 ms   -0.004000 ms
<SNIP>


See, how much *less* time it blocks during a sector erase with the flash
prefilled with some other (static) files!
What's going on???

Vipin



To unsubscribe from this list: send the line "unsubscribe jffs-dev" in
the body of a message to majordomo@xxxxxxx.com