Summary

memory corruption when user-interrupts and tcp

Metadata

Description

It seems user interrupts from csi, when reading from tcp ports causes memory corruption. As shown in this paste: http://paste.call-cc.org/paste?id=ae7e3a52e4a71cf9c1c6697234d168a343022309

 ~$ csi -n
 
 CHICKEN
 (c) 2008-2014, The CHICKEN Team
 (c) 2000-2007, Felix L. Winkelmann
 Version 4.9.1 (rev 6a860ab)
 linux-unix-gnu-x86-64 [ 64bit manyargs dload ptables ]
 compiled 2014-09-30 on devpool08 (Linux)
 
 #;1> (use tcp)
 ; loading /home/ckellerm/chickens/master/lib/chicken/7/tcp.import.so ...
 ; loading library tcp ...
 #;2> (define-values (i o) (tcp-connect "127.0.0.1" 1234))
 #;3> (read-char i)
 ^C
 *** user interrupt ***
 #;3> ^[[A
 
 Error: unbound variable: 
 3>  (read-char i)                                        
 ^C
 *** user interrupt ***
 #;3>  (read-char i)
 ^C
 *** user interrupt ***
 #;3> ^C
 *** user interrupt ***
 #;3>  (read-char i)
 ^C
 *** user interrupt ***
 #;3>  (read-char i)
 ^C
 *** user interrupt ***
 #;3>  (read-char i)
 *** Error in `csi': realloc(): invalid next size: 0x0000000001f5beb0 ***
 Aborted

I am reproducing this on Linux 3.18.6.

Changes and comments

[2015-08-27 09:09:20 UTC] sjamaan changed milestone from someday to 4.11.0

[2015-08-27 09:09:20 UTC] sjamaan wrote:

There's a patch for fixing the scheduler, which needs some more attention

[2015-11-19 14:45:11 UTC] sjamaan changed status from new to closed

[2015-11-19 14:45:11 UTC] sjamaan set resolution to fixed

[2015-11-19 14:45:11 UTC] sjamaan wrote:

I can no longer reproduce this bug. It looks like this was fixed at the ICC with 4e32262036806b2fc51964e518135fcd3de3a79b