| Summary: | vsftpd stack overflow on incorrect config file | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 7 | Reporter: | Paulo Andrade <pandrade> |
| Component: | vsftpd | Assignee: | Martin Sehnoutka <msehnout> |
| Status: | CLOSED DUPLICATE | QA Contact: | BaseOS QE Security Team <qe-baseos-security> |
| Severity: | low | Docs Contact: | |
| Priority: | low | ||
| Version: | 7.2 | CC: | sbroz |
| Target Milestone: | rc | ||
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | If docs needed, set a value | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2016-12-15 08:51:39 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
Hi, Thank you for you report, did you really try it with our version vsftpd package? From my point of view our version of vsftpd is slightly modified and this bug does not work for it. I am uncertain how the customer managed to generated the
coredump using vsftpd-3.0.2-11.el7_2.x86_64.
The issue can be reproduced with:
---8<---
# echo "a'/." >> /etc/vsftpd/vsftpd.conf
$ cat fcntl.c
#include <stdio.h>
#include <string.h>
#include <dlfcn.h>
#include <sys/types.h>
/* Function pointers to hold the value of the glibc functions */
static int (*real_fcntl)(int fd, int cmd, ...) = NULL;
int fcntl(int fd, int cmd, ...)
{
if (real_fcntl == NULL)
real_fcntl = dlsym(RTLD_NEXT, "write");
if (fd == 0)
return -1;
return real_fcntl(fd, cmd);
}
$ gcc -D_GNU_SOURCE=1 -o fcntl.so -shared -fPIC fcntl.c -ldl
$ LD_PRELOAD=$PWD/fcntl.so vsftpd
---8<---
*** This bug has been marked as a duplicate of bug 1666380 *** |
Backtrace looks like: #0 0x00007fa18db538f1 in __openat_nocancel (mode=<optimized out>, oflag=-128, file=0x3 <Address 0x3 out of bounds>, fd=0) at ../sysdeps/unix/sysv/linux/openat.c:94 #1 __GI___openat (fd=0, file=0x3 <Address 0x3 out of bounds>, oflag=<optimized out>) at ../sysdeps/unix/sysv/linux/openat.c:170 #2 0x00007fa18f0f9023 in vsf_sysutil_activate_noblock (fd=-902463408, fd@entry=0) at sysutil.c:766 #3 0x00007fa18f0e874b in bug (p_text=p_text@entry=0x7fa18f1013d4 "fcntl") at utility.c:44 #4 0x00007fa18f0e8799 in die (p_text=p_text@entry=0x7fa18f1013d4 "fcntl") at utility.c:19 #5 0x00007fa18f0f9070 in vsf_sysutil_activate_noblock (fd=fd@entry=0) at sysutil.c:769 #6 0x00007fa18f0e874b in bug (p_text=p_text@entry=0x7fa18f1013d4 "fcntl") at utility.c:44 #7 0x00007fa18f0e8799 in die (p_text=p_text@entry=0x7fa18f1013d4 "fcntl") at utility.c:19 #8 0x00007fa18f0f9070 in vsf_sysutil_activate_noblock (fd=fd@entry=0) at sysutil.c:769 ... basically, it enters an infinite recursion where vsf_sysutil_activate_noblock() calls die() that calls bug() that calls vsf_sysutil_activate_noblock(). Reproducer and possible patch listed at: https://lists.freebsd.org/pipermail/freebsd-ports-bugs/2005-October/068040.html