Re: _Very_ strange problem with compiling
Peter Kovacs [peter@kovax.org] wrote:
> The problem is: you're simply asking for too much stack space. On my
> machine, ulimit -s reports 8129 (that's in Kilobytes). When I write a
> seperate program such as:
>
> #include <stdio.h>
>
> void main(int argc, char **argv)
> {
> double a[209677];
> printf("%d\n", sizeof(a)*5/1024);
> }
>
> it reports a total usage of 8190K in stack space. This is obviously
> over the limit. So just increase the stack space and the program runs
> fine: (ulimit -s <size>)
Thanks for this! I increased the stack size via "ulimit -s <size>"
and the thick4 executable no longer crashes.
However the curious thing is that with test.c, I thought "surely by
increasing N I could make it seg fault too" but I can't, even
#define N 200000000
runs fine. (c.f. #define N 209677 for thick4.c) (Remember that test.c
and thick4.c are identical apart from name --- and now size of N.)
Perhaps my understanding of what the stack is and how it works is
wrong. (Basically it is where local variables are stored --- or is
there more to it than that?)
Perhaps the compiler does things differently depending on how big
the name space is?
Anyway, thanks again for your help.
Mark.
>
> As for the slight difference between the two executables, that's because
> you were storing symbols in the executable. If you run them through
> strip there is no difference between the two files.
>
> Peter
>
>
> On Wed, Mar 07, 2001 at 03:18:56AM +1030, Mark Phillips wrote:
> > Hi,
> >
> > I thought I'd found a gcc compiler error (and still may have). The
> > following program crashes with a segmentation fault.
> >
> > #define N 209677
> >
> > int main(int argv, char **argc) {
> > int
> > i;
> > double
> > a[N],
> > b[N],
> > c[N],
> > d[N],
> > e[N];
> >
> > c[0]=0.0;
> > i=1;
> > c[i]=c[i-1];
> > }
> >
> > If I make N slightly less, or reduce the number of arrays defined, it
> > runs fine. If I write instead "c[1]=c[0]" it runs fine. Basically, I
> > simplified the original program as much as I could in ways which still
> > caused the Seg Fault to happen.
> >
> > So I thought this must be a compiler error. The program was called
> > "thick4.c". I used emacs to paste the file into a new buffer called
> > "test.c". But this one worked fine when I compiled!!!! The two files
> > "thick4.c" and "test.c" are identical, as both diff and cmp prove.
> >
> > The executables "thick4" and "test" differ very slightly however.
> >
> > -rwxrwxr-x 1 mark mark 4731 Mar 7 02:53 test*
> > -rw-rw-r-- 1 mark mark 162 Mar 7 02:53 test.c
> > -rwxrwxr-x 1 mark mark 4733 Mar 7 02:53 thick4*
> > -rw-r--r-- 1 mark mark 162 Mar 7 02:51 thick4.c
> >
> > and
> >
> > $ cmp thick4 test
> > thick4 test differ: char 3061, line 2
> >
> > Why would identical files compile to slightly different executables?
> > I have included gzipped versions of these executables with this email.
> >
> > Has anyone got any idea what might be going on? Or suggestions about
> > what I could investigate next!
> >
> > Also, can anyone else replicate the error I got (using thick4.c as
> > your file name).
> >
> > Thanks,
> >
> > Mark.
> >
> > --
> > _/~~~~~~~~\___/~~~~~~\____________________________________________________
> > ____/~~\_____/~~\__/~~\__________________________Mark_Phillips____________
> > ____/~~\_____/~~\________________________________mark@ist.flinders.edu.au_
> > ____/~~\HE___/~~\__/~~\APTAIN_____________________________________________
> > ____/~~\______/~~~~~~\____________________________________________________
> > __________________________________________________________________________
> > "They told me I was gullible ... and I believed them!"
> >
>
>
> > ???:>
> > ????F%?Dm??Z?^???\?+?@<lwO?kv?lv???%?H??C?SL?h?l??>TJD?%?h0* &????D?!???3g???$???4?3????????K??i????c~F????
?[?E{ ?y???e?PF yu?5?:? ?:e?(/C}??i2??P&}V/??m2? e????b
> > ?)?(?>)????????^K9> ?o?
> > ?????1?~lOl????}???????-?8:????1??R??????)??/??EQ?? _A?QV<?[
> > ?Y???????????Z?za?\
)?#?????YQ~pJ(4????Ob?\??r???w?y???;:??=>?sG????M?{4x???1?[j?$?????ck??
c?pQ_tqm????? ?9J_|??9l???S7?/?/??,<T?]???U???6???s??U,????DD?????o(??[N)??????#?M?k??>k?q????X?FIqB?%??ERo????2 ;4???y\1v;???z%? ^&K??nC;?j?X~C?t?/???z0???b}??x?~?5?????? h????T> ???a? ???V?\?"?~?????>Wt(o???V?a"?????W?? m??+??????s???uZ??&??????????*???h?`\_?h?XXn*aa!T??;?%,????~q???*??L???J?_? 8vJx!?S%???K%,V0????9?[?K8>0??? 'JXD[??8??p#??%,v??^.?O?xE????~~Q??
> > ?????k?^Wp?&u|.?*?w+?:?V???"?7Ht???-e??k?7??"?????A???Z8H?R?#? +??6??c????S
> > ~?????=?}
???|?G??|?`??JG?'{m#??w?y?X???????D?W?2?i???4o?4`?,???????+?x????F??-Z??????z???y????????????V?????x`?'5o>???7+?{W????,pH????a/(??????'@?C?H??;O:????? ????,~??d??? 4",m??gkwh?n?????w0i'??m?????=N???e?sy? 5???? 4;?=?;?Dw53?K&2{?)+?--L??E+??du??D,??M~?3?.??j0?????nN?V??@??u0'??? ?{?????n? ??`??=?L~??????/?h??[??|?W?+@U?i???NkZb?=};;;??????Cf???/?Y???????Q?o???>^?{
>
> --
> Peter D. Kovacs <peter@kovax.org>
> Software Developer
> Webmachines, Inc. http://webmachines.com
--
_/~~~~~~~~\___/~~~~~~\____________________________________________________
____/~~\_____/~~\__/~~\__________________________Mark_Phillips____________
____/~~\_____/~~\________________________________mark@ist.flinders.edu.au_
____/~~\HE___/~~\__/~~\APTAIN_____________________________________________
____/~~\______/~~~~~~\____________________________________________________
__________________________________________________________________________
"They told me I was gullible ... and I believed them!"
Reply to: