[Toybox] [PATCH] fix "vmstat 2"
enh at google.com
Wed Dec 3 17:00:08 PST 2014
On Tue, Dec 2, 2014 at 2:09 PM, Rob Landley <rob at landley.net> wrote:
> On 11/25/14 21:14, enh wrote:
>> actually, loop_max =1 gives the right behavior by accident (because of
>> the break in the loop), so setting loop_max = 0 is a better idea.
> Sorry for the delay in replying. I'm hip-deep in deflate
> compression-side logic (implementing the rest of gzip, somebody poked me
> and said they need it now), and the vmstat command is a can of worms for me.
> My most recent cleanup pass on vmstat was:
> And that description ends with the fact that I have no CLUE how several
> of those fields are calculated. What we're doing doesn't match what
> other implementations are doing, and there's no standard or
> documentation on what it _should_ be. (Trying to find that sort of thing
> out is a giant timesink. Gotta do it eventually, but there are higher
> priority todo items at present, at least until somebody pokes me and
> says "I need X now". :)
> Yes, there are some "pending" commands that aren't in the "pending"
> directory, the list of them is in toys/pending/README. Initially I just
> had a list of commands needing cleanup, but it kept getting longer as
> contributions came in faster than I could clean them up, and I added
> toys/pending when the list got too long to manage. I've got it down to
> 11 commands and one lib file, and I keep meaning to just properly review
> and clean up that old list rather than try to explain it to people, but
> the remaining issues in those commands tend to be nontrivial (or I'd
> have fixed them by now). Sometimes, simple is a destination you go
> through a lot of complexity to reach...
> Anyway, your bug is easy to test and thus easy to fix, but I'd actually
> fix it in a different way because:
> 1) I'm uncomfortable assuming INT_MAX will never be reached. (Yeah,
> almost certainly true in this case but INT_MAX != infinity and I don't
> want to get into the habit.)
> 2) Minor bug with "./vmstat 2 3" where it has the 2 second pause right
> before exiting. It'snot going to produce more output, so it should just
> have the pause twice.
> Did the fix I just checked in work for you?
in the sense of making the command-line arguments behave, yes, this works fine.
as for the "what are these fields and what should they mean?", yeah, i
was leaving that for later. Android's vmstat does look a little
different and "toolbox vmstat" and "toybox vmstat" don't produce the
same output, even for the fields they share (like 'in'):
/huge-ssd/aosp-arm64/bionic$ adb shell toolbox vmstat
procs memory system cpu
r b free mapped anon slab in cs flt us ni sy id wa ir
0 0 1172960 12660 13336 36532 113 204 0 0 0 0 99 0 0
/huge-ssd/aosp-arm64/bionic$ adb shell toybox vmstat
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
0 0 0 1173364 4316 345556 0 0 43 28 0 234 1 1 98 0
i don't know what i'm going to do about that, but wanted to at least
get easy stuff sorted out first. i'm likely to have a lot of trouble
with commands where the output is different even if it's clearly
better (like ifconfig) because i don't know what (if anything) is
trying to parse the existing output.
> (Sorry, I ramble a bit when sleep deprived. #include Henry David
> Thoreau's apology for writing a 5 page letter because he didn't have
> time to write a 1 page one...)
More information about the Toybox