Re: [Ngspice-devel] pss crashes after converging

[Paolo N. <pne...@gm...>]

Hi Shouri,

Thanks for the report. PSS is still experimental. 
On Monday I will discuss this bug report with Stefano (pss developer).

Ciao,
Paolo

On Jun 9, 2012, at 8:50 PM, Shouri Chatterjee wrote:

> 
> I forgot to include the circuit that I was trying to simulate:
> _______
> 
> Ordinary oscillator
> .include mosis.mdl
> minv1p oscnode2 oscnode1 vdd_ana vdd_ana CMOSP w=30u l=0.18u
> minv1n oscnode2 oscnode1 gnd_ana gnd_ana CMOSN w=10u l=0.18u
> minv2p oscnode1 oscnode2 vdd_ana vdd_ana CMOSP w=30u l=0.18u
> minv2n oscnode1 oscnode2 gnd_ana gnd_ana CMOSN w=10u l=0.18u
> lres  oscnode1 oscnode2 1n ic=1e-3
> rloss oscnode1 oscnode2 6300
> cres  oscnode1 oscnode2 1.6p
> rdd vdd vdd_ana 70m
> rgg 0   gnd_ana 70m
> vpower vdd 0 dc=1.8 pwl 0 0 10e-12 1.8
> .save v(oscnode1) v(oscnode2)
> .pss 4G 12n oscnode1 1024 11 200 1e-3 uic
> .end
> 
> Where, mosis.mdl is a model file, copied from:
> https://www.mosis.com/cgi-bin/cgiwrap/umosis/swp/params/tsmc-018/t73d_mm_non_epi_thk-params.txt
> 
> Shouri
> 
> 
> Shouri Chatterjee wrote:
>> 
>> Hi,
>> 
>> I am trying to use the pss experimental analysis. I have been trying both,
>> with ngspice-24, as well as the git version.
>> 
>> The test circuit is a quad coupled oscillator (like the one in the pss
>> examples folder, but with different l, c etc values and bsim3 MOS models.)
>> 
>> The pss analysis converges and then, before giving me the prompt, crashes:
>> 
>> Convergence reached. Final circuit time is 1.263414241e-08 s and predicted
>> fundamental frequency is 3.87808e+09 Hz.
>> ----------------
>> 
>> throwing out permanent breakpoint times <= current time (brk pt:
>> 1.26341e-08)
>> ckt_time: 1.26341e-08    ckt_min_break: 1.25e-18
>> throwing out permanent breakpoint times <= current time (brk pt:
>> 1.2892e-08)
>> ckt_time: 1.2892e-08    ckt_min_break: 1.25e-18
>> 
>> 
>> No. of Data Rows : 1024
>> 
>> 
>> No. of Data Rows : 11
>> *** glibc detected *** ngspice: double free or corruption (!prev):
>> 0x000000000408bd20 ***
>> ======= Backtrace: =========
>> /lib/x86_64-linux-gnu/libc.so.6(+0x7e626)[0x7fdabc7a5626]
>> 
>> followed by a lengthy backtrace.
>> 
>> I have used gdb to find out what is going on:
>> The crash occurs in the function plotAddRealValue (frontend/outitf.c).
>> After going through this function several times (roughly 8*1024 times),
>> the function for the first time enters the section with the "real parading
>> as a VF_COMPLEX". As soon as the TREALLOC is attempted, the program
>> crashes.
>> 
>> Let me know what I can do to help debug this further, or to clean this up.
>> 
>> Shouri
>> 
>> 
>> _________
>> 
>> Asst Professor
>> Department of Electrical Engineering
>> IIT Delhi, Hauz Khas
>> New Delhi 110016, India
>> 
> 
> 
> 
> _________
> 
> Asst Professor
> Department of Electrical Engineering
> IIT Delhi, Hauz Khas
> New Delhi 110016, India
> 
> 
> ------------------------------------------------------------------------------
> Live Security Virtual Conference
> Exclusive live event will cover all the ways today's security and 
> threat landscape has changed and how IT managers can respond. Discussions 
> will include endpoint security, mobile security and the latest in malware 
> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
> _______________________________________________
> Ngspice-devel mailing list
> Ngs...@li...
> https://lists.sourceforge.net/lists/listinfo/ngspice-devel