On Sun, Jun 10, 2012 at 7:43 PM, Francois Pussault
<fpussa...@contactoffice.fr> wrote:
>
>
>> ----------------------------------------
>> From: Tomas Bodzar <tomas.bod...@gmail.com>
>> Sent: Sun Jun 10 19:19:57 CEST 2012
>> To: Francois Pussault <fpussa...@contactoffice.fr>
>> Subject: Re: how normal is this ?
>>
>>
>> On Sun, Jun 10, 2012 at 7:05 PM, Francois Pussault
>> <fpussa...@contactoffice.fr> wrote:
>> > hi all,
>> >
>> > here is my default memory setup :
>> >
>> > hw.machine=i386
>> > hw.model=Intel(R) Core(TM)2 Duo CPU T8300 @ 2.40GHz ("GenuineIntel"
686-class)
>> > [..]
>> > hw.physmem=2136907776
>> > hw.usermem=2062499840
>> > hw.ncpufound=2
>> > hw.allowpowerdown=1
>> >
>> > top info :
>> > Memory: Real: 437M/956M act/tot Free: 1039M Cache: 376M Swap: 0K/2047M
>> >
>> > for some reasons, I don't know yet, some applications securely answer
"Cannot allocate memory"
>> > until memory in Real decrease to about 300M/956M. (...even if I guess
those applications are written with foot...this is strange).
>> >
>> > but I have 1039M free & some swap free so why that ?
>> >
>> > Is this range of memory protected or reserved to something ?
>> > Any method to force memory "Real" to be higher & have less Free memory in
default should be a solution too..
>>
>> ulimit -a
>> man login.conf
>> on i386 max 1GB/process (~700MB real memory/process) per design
>> check /var/log/messages for some uvm errors
>>
>> >
>> > thanks
>> > Regards
>> >
>
> # ulimit -a
> time(cpu-seconds) unlimited
> file(blocks) unlimited
> coredump(blocks) unlimited
> data(kbytes) 2097152
> stack(kbytes) 8192
> lockedmem(kbytes) 680574
> memory(kbytes) 2035316
> nofiles(descriptors) 128
> processes 1310
> #
>
> I don't see any thing anormal in dmesg & I use the default login.conf
setup.
> So i guess this is due a linux binary "bad coded" so not very well supported
even with
> kern.emul.linux=1 # enable running Linux binaries
> line uncommented in sysctl.conf...
It's not default, because such a datasize is not possible on i386. It
will not work. So something modified that (either you or that app
you're trying).
>
> because this binary runs well in a qemu Virtual Machine running debian...
> so I guess it is application issue then problem is "solved" because is is
isolated.
