[prev in list] [next in list] [prev in thread] [next in thread] 

List:       olpc-linux-mm-cc
Subject:    Re: [linux-mm-cc] /proc/swaps reports more ramzswap used than
From:       Nitin Gupta <ngupta () vflare ! org>
Date:       2009-11-10 4:10:25
Message-ID: d760cf2d0911091958p378cc1adpa1cbf2acdb88a501 () mail ! gmail ! com
[Download RAW message or body]

Hi,

On Sat, Nov 7, 2009 at 12:19 PM, John McCabe-Dansted <gmatht@gmail.com> wrote:
> Hi, I have read
>  http://code.google.com/p/compcache/wiki/StatsExplained
> and I can see why OrigDataSize could be larger than the figure
> reported in /proc/swaps. However I am getting the reverse and I don't
> know why I would have a smaller OrigDataSize than the figure reported
> in /proc/swaps.
>
> From gpl2mem_150 (Attached):
>
> --
>
> Filename                                Type            Size    Used    Priority
> /dev/ramzswap0                          partition       376116  376084  100
> /dev/sdb1                               partition       2096440 54344   -1
>
> DiskSize:         376120 kB
> NumReads:           1296
> NumWrites:         94343
> FailedReads:           0
> FailedWrites:          0
> InvalidIO:             0
> NotifyFree:            0
> PagesDiscard:          0
> ZeroPages:         69734
> GoodCompress:         76 %
> NoCompress:            1 %
> PagesStored:       24295
> PagesUsed:          8658
> OrigDataSize:      97180 kB
> ComprDataSize:     33636 kB
> MemUsedTotal:      34632 kB
>

OrigDataSize should probably be renamed to "NonZeroDataSize".
Data stored in ramzswap is actually: OrigDataSize + ZeroPages*4.

So, in your case this is: 97180 + 69734*4 = 376116KB
which is less than 376084KB as reported by /proc/swaps

The reason for not adding ZeroPages when reporting OrigDataSize
is that it can give false impression of very high compressibility of memory
data.

Thanks,
Nitin
_______________________________________________
linux-mm-cc mailing list
linux-mm-cc@lists.laptop.org
http://lists.laptop.org/listinfo/linux-mm-cc

[prev in list] [next in list] [prev in thread] [next in thread]