| Bug #33854 | client can crash during mysql_real_connect if the connection dies too fast | ||
|---|---|---|---|
| Submitted: | 13 Jan 2008 19:49 | Modified: | 30 Jun 2009 18:14 |
| Reporter: | Shane Bester (Platinum Quality Contributor) | Email Updates: | |
| Status: | Can't repeat | Impact on me: | |
| Category: | MySQL Server: C API (client library) | Severity: | S3 (Non-critical) |
| Version: | 5.1.23 | OS: | Windows |
| Assigned to: | CPU Architecture: | Any | |
| Tags: | mysql_real_connect | ||
[14 Jan 2008 13:48]
MySQL Verification Team
Hmm, my simple test program doesn't crash. I'll have to spend the time debugging the original complex program to find out what's going wrong.
[6 Mar 2009 17:11]
Andrew Hutchings
Could be a duplicate of bug #25532
Description: If the connection is made, but dies quick enough due to whatever reason, a client crash can happen with stack trace: libmysql.dll!my_real_read libmysql.dll!my_net_read libmysql.dll!cli_safe_read libmysql.dll!mysql_real_connect gypsy.exe!db_connect gypsy.exe!db_pthread In net_serv.cc the crash is here: /* First read is done with non blocking mode */ if ((long) (length= vio_read(net->vio, pos, remain)) <= 0L) { my_bool interrupted = vio_should_retry(net->vio); <---------- vio_should_retry is a macro that dereferences vio, which is null in this crash. In addition, the debug client prints out these messages: Error: Freeing pointer out of range at line 231, '..\vio\vio.c' Error: Freeing unallocated data at line 232, '..\vio\vio.c' How to repeat: I will upload a simple program that: 1) spawns some threads 2) each thread has auto-reconnect enabled 3) each thread tries to kill another thread using KILL <id>