On Tue, 29 Jan 2013, Andrey Zonov wrote:

On 1/28/13 6:51 PM, Bruce Evans wrote:
On Mon, 28 Jan 2013, Andrey Zonov wrote:

Log:
 - Show page faults requiring I/O on vmstat display.

No space is available there for showing it.

Yep, you're right.

+    mvprintw(VMSTATROW, VMSTATCOL + 9, "ioflt");

Putting it first unsorts the fields a bit and makes the diff large.

It is not documented in the man page.

Fixed in attached systat1.patch.txt.

OK.

...
"buf" is even more useless with zfs.  So are some of the other fields
...

I totally agree with you, 'buf' should go away from systat and top. I
removed 'buf' from systat.  Please review systat2.patch.txt.

I'd just like it to be replaced by a useful buf field someday.

Since the field in row 23 (starting at row 0) is now useful, omitting it
is not so good so I I agree with your patch removing the special code
to avoid printing it on 24-row terminals.

To count 'disk cache' we have to add new counter which should track all
pages with OBJT_VNODE type.  It doesn't look hard to implement this.  I
wrote utility [1] which allows me to inspect memory and find what is in
disk cache.

I think I would like at least 2 fields:
- total disk space mapped in VMIO buffers
- total disk space mapped in the buffer cache.

I also found that %ozfod is not useful for me and I removed it to not
mangle 'free' on 24-line terminals (systat3.patch.txt).

Certainly no space is available for the luxury of both ozfod and %ozfod.

That would also allow leaving buf and its 24-column support alone.  Only
3 fields would move relative to the old version (not just 1 field changing
for swapping %ozfod with the new field, since you want to put the new field
first).  Are the other fields in the best order?  I think they are.  If not,
it would be good to move some when adjusting all the row numbers.

Thanks for comments!

[1] https://github.com/z0nt/meminfo/

Bruce
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "[email protected]"

Reply via email to