Re: top io patch

[Antonio Huete Jimenez <ahuete.devel@xxxxxxxxx>]

Hi Pratyush,

It didn't apply cleanly in master:

Patching file usr.bin/top/m_dragonfly.c using Plan A...

Hunk #1 succeeded at 63.

Hunk #2 succeeded at 85.

Hunk #3 failed at 98.

Hunk #4 succeeded at 178 (offset -1 lines).

Hunk #5 failed at 198.

Hunk #6 failed at 210.

Hunk #7 failed at 222.

Hunk #8 succeeded at 348 (offset 2 lines).

Hunk #9 failed at 390.

Hunk #10 succeeded at 473 (offset -2 lines).

Hunk #11 succeeded at 554 (offset 2 lines).

Hunk #12 succeeded at 560 (offset -2 lines).

Hunk #13 succeeded at 606 (offset 2 lines).

Hunk #14 succeeded at 617 (offset -2 lines).

Hunk #15 succeeded at 670 (offset 2 lines).

Hunk #16 failed at 747.

Hunk #17 succeeded at 797 (offset -4 lines).

Hunk #18 succeeded at 1077 (offset 31 lines).

Hunk #19 succeeded at 1076 (offset -4 lines).

6 out of 19 hunks failed--saving rejects to usr.bin/top/m_dragonfly.c.rej

done

Can you please fix it?

Cheers,

Antonio Huete

2011/10/14 Pratyush Kshirsagar <pratyush.kshirsagar@gmail.com>

Hello,
Here is the top io patch attached.
This patch is just for testing purpose.
I have made some changes to PRES print and sorting stuff.
As, I was testing this on my old system, cause its graphical and gives me some good read and write results while fire-fox is running.
I talked on IRC and Alex told me that the IVCSW  (In voluntary context switches) values are not stored properly but on my system I always see the IVCSW values zero both with the values from the struct kp_ru and kp_cru i.e. I tried to check the values of process resource usage and childs resource usage and the IVCSW is allways zero. May be their are no involuntary context switches on my system, may be the processes are kind enough to utilise cpu in bounds.
Also, I added a function to sort the reads which is not sorting the processes properly, although the logic is same the way with other sort ORDERKEY's.

Please test this patch with the top -m io command and give me some suggestions on it.

Note: This patch is just to get some feedback. The only things missing in this patch is sorting the output depending on write, total io, vcsw, ivcsw and faults.

- Pratyush Kshirsagar.