Re: [bug-mailutils] Bug in sieve engine

[Kostas Zorbadelos]

On Tue, Nov 01, 2005 at 10:18:03AM +0200, Kostas Zorbadelos wrote:

Sorry I forgot to send the gdb backtrace

#0  push_source (name=0xbffffd66 "filters.sieve") at sieve.l:297
#1  0xb7f7f311 in sieve_lex_begin (name=0xbffffd66 "filters.sieve") at 
sieve.l:534
#2  0xb7f7d5cd in mu_sieve_compile (mach=0x804dea0, name=0xbffffd66 
"filters.sieve") at sieve.y:541
#3  0x08048fdc in main (argc=3, argv=0xbfffde94) at sieve_eng_test.c:96


> Hello to everyone.
> I believe I found a bug in the implementation of the sieve engine. I
> include a program that demonstrates the faulty behaviour. My program
> opens a mailbox and for each message it creates a new sieve machine,
> compiles the same sieve file, calls sieve_message() and destroys the
> machine. 
> The program is called as 
> 
> ./sieve_eng_test mbox:test.mbox filters.sieve
> 
> From: Smara Frementiti <address@hidden>
>  Subject: SE THELO TORA
>  Sieve flag: 0
> filters.sieve:21: recursive inclusion
> Error compile sieve script: Operation not permitted
> 
> The sieve engine produces the error "recursive inclusion". I believe I
> have traced the error and has to do with the variables
> 
> char *sieve_filename;  
> int sieve_line_num;  
> ino_t sieve_source_inode;
> 
> in sieve.l that are not initialized. However, I hate to admit, my
> lex/yacc skills are quite rusty (I definetely should go through the
> dragon book again).
> 
> I include my program, the makefile, the sieve file and the test
> mailbox. All tests are against snapshot 20050929 (with a patch of mine
> that has to do with the reject action, already sent to the list).
> Let me know if you need anything else to trace the problem.
> 
> Kostas 
> 


-- 
  Kostas Zorbadelos
  Systems Designer/Developer, Otenet SA 
  address@hidden contact: kzorba (at) otenet.gr
  
  Out there in the darkness, out there in the night
  out there in the starlight, one soul burns brighter
  than a thousand suns.