SCTP Discussions

Text archives Help


[SCTP-Discussion] Possible bug in sci_timer_expired


Chronological Thread 
  • From: "Oleg Levy" <buskila AT gmail.com>
  • To: discussion AT sctp.de
  • Subject: [SCTP-Discussion] Possible bug in sci_timer_expired
  • Date: Mon Oct 23 00:29:10 2006
  • Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:to:subject:mime-version:content-type; b=uI7k2S1K8G/2jzeYkrVl4Jdm305eulPctU5B/0SHFG7EPzo6rdxZgCnveSOqtZ6waRunnreciCyg0N7qgNZqxzwHjclZ8uOQyiLUzys0pe6U40kqSCzR6plOceWP3gmRCGgY9uGtatLX4MvX//wboZfuTW8N/f+dIbY4dgXiSIE=
  • List-id: SCTP Discussions <discussion.sctp.de>

Hello
 
I think there is a possible bug in the sci_timer_expired() function in SCTP-control.c (i'm working with version 1.0.5 of sctplib)
 
When the timer expires, the association state is COOKIE_WAIT and the maximum number of retransmissions was reached, the function calls mdi_communicationLostNotif() to notify the upper layer. The upper layer calls sctp_deleteAssociation() which in turn calls mdi_removeAssociation() which deletes the SCTP_control instance.
 
The sci_timer_expired() function proceeds and tries to free the initChunk member of the localData (which was set to the SCTP_control instance previously removed by sctp_deleteAssociation()) and this leads to the crash.
 
My question if it is correct to call sctp_deleteAssociation() from within the notification.
 
regards
oleg
 
 



Archive powered by MHonArc 2.6.19+.

Top of page