Re: octave memory

[Thomas Weber]

Am Mittwoch, den 11.02.2009, 15:37 -0500 schrieb John W. Eaton:
> On 11-Feb-2009, Miroslaw Kwasniak wrote:
> 
> | Current kernels have an entry /proc/*/status:
> | 
> | $ grep Vm /proc/`pidof octave-3.0.1`/status
> | VmPeak:   419416 kB <- max allocated
> | VmSize:   321288 kB
> | VmLck:         0 kB
> | VmHWM:    106716 kB <- max used (HWM = High Water Mark)
> | VmRSS:    105828 kB 
> | VmData:   253276 kB
> | VmStk:       312 kB
> | VmExe:         4 kB
> | VmLib:     47508 kB
> | VmPTE:       288 kB
> | 
> | I assume VmHWM is a true max usage when not swapped.
> | 
> | Below the same process listed by top:
> | 
> |   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> | 18177 mirek     39  19  399m 103m  24m S  1.6  6.8  32:49.40 octave-3.0.1
> 
> So why isn't this information returned by the getrusage library
> function?  

Missing functionality in the Linux kernel:
http://zwillow.blogspot.com/2006/05/broken-getrusage-in-linux.html

>From getrusage(2):
The structure definition shown at the start of this page was taken from
4.3BSD Reno.  Not all fields are meaningful under Linux.  In Linux  2.4
only the fields ru_utime, ru_stime, ru_minflt, and ru_majflt are
maintained.  Since Linux 2.6, ru_nvcsw and ru_nivcsw are also
maintained.

There's hope, however:
http://marc.info/?t=123203053700006&r=1&w=2

Jeff, it might be helpful if you weighted in on the kernel discussion
and give your specific use case. It seems kernel developers don't see a
use case for the getrusage information.

        Thomas