There were changes after Jul 27, can you catch up the latest
commits? The current segfault should be more deterministic.
Пн, 15 авг 2016, Thomas Costigliola написал(а):
> That's strange. Mine doesn't crash in jtprobe. It crashes in jtsymfree being
> called from jtsymfreeh where u is invalid memory on these lines:
>
> u=*wv+jt->sympv;
> RZ(symfree(u));
>
> I'm not using the latest commit though. I am using this one:
>
> commit 05d24010d132df1927c621b376b60135add93b36
> Author: HenryHRich <henryhr...@gmail.com>
> Date: Wed Jul 27 18:01:36 2016 -0400
>
>
> On 08/13/2016 11:49 PM, bill lam wrote:
> > I don't know what has happened or any idea about symbol table.
> > It crashed in s.c , back trace:
> >
> > #0 0x00007ffff7065ff4 in jtprobe (jt=jt@entry=0x643e70,
> > a=<optimized out>,
> > g=0x7ffff7f315a0) at /home/bill/gitdev/jsource/jsrc/s.c:167
> > #1 0x00007ffff7066467 in jtprobelocal (jt=jt@entry=0x643e70,
> > a=a@entry=0x7ffff7f31620)
> > at /home/bill/gitdev/jsource/jsrc/s.c:202
> > #2 0x00007ffff70667f4 in jtsyrd (jt=jt@entry=0x643e70,
> > a=0x7ffff7f31620,
> > symb=symb@entry=0x0) at
> > /home/bill/gitdev/jsource/jsrc/s.c:314
> > #3 0x00007ffff706b2ec in jtex (jt=jt@entry=0x643e70,
> > w=<optimized out>)
> > at /home/bill/gitdev/jsource/jsrc/sn.c:205
> > #4 0x00007ffff717b09a in jttest (jt=0x643e70, w=0x7ffff7f317a0)
> > at /home/bill/gitdev/jsource/jsrc/x.c:262
> > #5 0x00007ffff701a09b in jtdfs1 (jt=jt@entry=0x643e70,
> > w=<optimized out>,
> > self=<optimized out>) at
> > /home/bill/gitdev/jsource/jsrc/au.c:32
> > #6 0x00007ffff705bdbd in jtparsea (jt=jt@entry=0x643e70,
> > w=w@entry=0x7ffff7f31320)
> > at /home/bill/gitdev/jsource/jsrc/p.c:376
> > #7 0x00007ffff705c890 in jtparse (jt=jt@entry=0x643e70,
> > w=0x7ffff7f31320)
> > at /home/bill/gitdev/jsource/jsrc/p.c:126
> > #8 0x00007ffff705ea5a in jtimmex (jt=jt@entry=0x643e70,
> > w=w@entry=0x7ffff7f312a0)
> > at /home/bill/gitdev/jsource/jsrc/px.c:37
> > #9 0x00007ffff7052f0b in jdo (jt=jt@entry=0x643e70,
> > lp=<optimized out>)
> > at /home/bill/gitdev/jsource/jsrc/io.c:148
> > #10 0x00007ffff7052f4a in JDo (jt=0x643e70, lp=<optimized out>)
> > at /home/bill/gitdev/jsource/jsrc/io.c:203
> > #11 0x0000000000402422 in jedo ()
> > #12 0x000000000040240c in main ()
> >
> > list starting from 161 (crashed at line 167), perhaps you may
> > want to reinstate line 166 for checking.
> >
> > 161 // a is A for name, g is symbol table
> > 162 // result is L* address of the symbol-table entry for
> > the name, or 0 if not found
> > 163 L*jtprobe(J jt,A a,A g){C*s;I*hv,m;L*v;NM*u;
> > 164 RZ(a&&g);u=NAV(a);
> > 165 // If there is bucket information,
> > 166 // obsolete
> > ASSERTSYS(jt->local||(u->bucket==0&&u->bucketx==0),"bucket");
> > 167 m=u->m; s=u->s;
> > hv=AV(g)+SYMHASH(u->hash,AN(g)-SYMLINFOSIZE); // get bucket
> > number among the hash tables
> > 168 if(!*hv)R 0; /* (0) empty
> > slot */
> > 169 v=*hv+jt->sympv;
> > 170 while(1){
> > 171 u=NAV(v->name);
> > 172 if(m==u->m&&!memcmp(s,u->s,m))R v->val?v:0; // (1)
> > exact match - if there is a value, use this slot, else say not
> > found
> > 173 if(!v->next)R 0; /* (2) link
> > list end */
> > 174 v=v->next+jt->sympv;
> > 175 }}
> >
> >
> > Пт, 12 авг 2016, Thomas Costigliola написал(а):
> > > I am having trouble with custom code that used to work but no longer does
> > > in
> > > the latest beta sources. The trouble seems to be related to symbol tables
> > > and/or reference counting etc. I managed to reduce the issues to a small
> > > piece of code which crashes under some circumstances but does not in the
> > > pre-805 J sources.
> > >
> > > // in x.c
> > > // Foreign 104 !: 1
> > > ...
> > > case XC(104,1): R CDERIV(CIBEAM, jttest, 0, RMAX, 0, 0 );
> > > ...
> > >
> > > static F1(jttest){A z, n, loc=jt->local;
> > > RZ(w);
> > > RZ(jt->local=stcreate(2,1L,0L,0L));
> > > n=nfs(1,"A");
> > > IS(n,w);
> > > ex(box(cstr("A")));
> > > symfreeh(jt->local,0L);
> > > jt->local=loc;
> > > R w;
> > > }
> > >
> > > The code does the following:
> > > 1. Create a local symbol table
> > > 2. Assign the argument w to A in the local symbol table
> > > 3. Erase A in the local symbol table
> > > 4. Erase the local symbol table
> > > 5. Return the argument
> > >
> > > Doing the following in the J session causes a crash in symfreeh due to
> > > accessing invalid memory in jt->sympv
> > >
> > > 104!:1 ] 1 2 3
> > > 1 2 3
> > > B=. 1 2 3
> > > 104!:1 B
> > > 1 2 3
> > > 104!:1 B
> > > 1 2 3
> > > 104!:1 B
> > > 1 2 3
> > > 104!:1 B
> > > 1 2 3
> > > 104!:1 B
> > > BOOM!
> > >
> > > Is this a bug or is there a new way to deal with local symbol tables?
> > > ----------------------------------------------------------------------
> > > For information about J forums see http://www.jsoftware.com/forums.htm
> >
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
--
regards,
====================================================
GPG key 1024D/4434BAB3 2008-08-24
gpg --keyserver subkeys.pgp.net --recv-keys 4434BAB3
gpg --keyserver subkeys.pgp.net --armor --export 4434BAB3
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
