nmh-workers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Nmh-workers] Updates to nmh


From: Peter Maydell
Subject: Re: [Nmh-workers] Updates to nmh
Date: Fri, 16 Jan 2009 20:56:27 +0000

Ken Hornstein wrote:
>So ... what MTS are you using?  SMTP?  Or sendmail?  Does "-snoop" tell you
>anything useful?  I thought what I did only affected the SMTP MTS, but
>hey, I've been wrong before.

I use mts:sendmail (this is the Debian package default and generally what
I recommend people use because (a) nmh's SMTP talking code doesn't get
the corner cases of the spec right and (b) Unix systems should have a
sendmail binary that can do this kind of thing anyhow).

It turns out that send is calling post, which is segfaulting:

Program received signal SIGSEGV, Segmentation fault.
0xb7e58b35 in memcpy () from /lib/tls/i686/cmov/libc.so.6
(gdb) bt
#0  0xb7e58b35 in memcpy () from /lib/tls/i686/cmov/libc.so.6
#1  0x0804e908 in sm_fgetc (f=0x8093278) at smtp.c:1412
#2  0x0804eaa7 in smhear () at smtp.c:1332
#3  0x0804ff1f in sm_init (client=0x805fa20 "mnementh.archaic.org.uk", 
server=0x0, port=0x8058fe2 "smtp", watch=0,
    verbose=0, debug=0, onex=1, queued=0, sasl=0, saslmech=0x0, user=0x0) at 
smtp.c:394
#4  0x0804ba12 in post (file=0x8061c80 "/home/pm215/Mail/postLEAgOY", bccque=0, 
talk=1) at post.c:1414
#5  0x0804cd7a in main (argc=Cannot access memory at address 0x26
) at post.c:641

This is because you're trying to use sasl_inbuffer, but this is only
allocated in smtp_init() and not sendmail_init(), so if you're using
sendmail then it just crashes...

>>I also noticed a compile warning:
>>./client.c: In function ‘client’:
>>./client.c:88: warning: passing argument 6 of ‘getnameinfo’ makes integer 
>>from pointer without a cast
>
>I guess technically that should be "0".  I forget on some el-lamo systems
>NULL is not defined as 0, but includes a cast to (void *).  Should be
>harmless, though.

I'm vaguely trying to eliminate compiler warnings.

>  Guess I missed it in the bazillion warnings about
>mismatches between unsigned char * and signed char *.

I never notice these because I always compile with CFLAGS including
-Wno-pointer-sign. I suppose I should push that into CVS somehow.

-- PMM




reply via email to

[Prev in Thread] Current Thread [Next in Thread]