Segmentation fault for concurrent > 1

[joshuamorris3]

This is related in part to my PR #42. Once I did get siege built in the docker-alpine base, when I ran siege with more than 1 concurrent users it would seg fault. I tried using gdb to get the backtrace (bt or where), but no trace was available. When I tried using the debian:jessie docker image as the base, I could use apt-get install siege (no build needed) and that worked just fine.

To recreate the segmentation fault:

1. docker run --rm -it alpine sh
2. Add the build dependencies: apk add --update alpine-sdk
3. wget http://download.joedog.org/siege/siege-latest.tar.gz
4. tar -zxvf siege-latest.tar.gz
5. cd siege-3.1.3
6. ./configure
7. Add #include <sys/types.h> below line 24 of src/hash.c
8. make
9. Run the following: siege -r 1 -c 2 www.google.com

[JoeDog]

Would you mind testing this:

http://download.joedog.org/siege/beta/siege-3.1.4-beta2.tar.gz

(That version also includes an overhaul of cookie handling, namely stored
persistence)

Jeff

On Sat, Dec 12, 2015 at 12:29 AM, Josh [email protected] wrote:

This is related in part to my PR #42
#42. Once I did get siege built in
the docker-alpine base, when I ran siege with more than 1 concurrent users
it would seg fault. I tried using gdb to get the backtrace (bt or where),
but no trace was available. When I tried using the debian:jessie docker
image as the base, I could use apt-get install siege (no build needed) and
that worked just fine.

To recreate the segmentation fault:

1. docker run --rm -it alpine sh
2. Add the build dependencies: apk add --update alpine-sdk
3. wget http://download.joedog.org/siege/siege-latest.tar.gz
4. tar -zxvf siege-latest.tar.gz
5. cd siege-3.1.3
6. ./configure
7. Add #include below line 24 of src/hash.c
8. make
9. Run the following: siege -r 1 -c 2 www.google.com

—
Reply to this email directly or view it on GitHub
#43.

[joshuamorris3]

Actually worse if I try it with the version you requested. Now it seg faults with concurrent = 1. I went back and tried it again with 3.1.3 and it behaved as I specified, with concurrent = 1 it works, > 1 and it seg faults.

[JoeDog]

I'm not sure what do. I don't have docker-alpine (honestly, I'm not even
sure what that is.)

J.

On Sun, Dec 13, 2015 at 1:57 AM, Josh [email protected] wrote:

Actually worse if I try it with the version you requested. Now it seg
faults with concurrent = 1. I went back and tried it again with 3.1.3 and
it behaved as I specified, with concurrent = 1 it works, > 1 and it seg
faults.

—
Reply to this email directly or view it on GitHub
#43 (comment).

[joshuamorris3]

No problem. Thanks for trying, for now I can use a debian docker image. It's much larger (170MB vs. 20MB), but it works. Docker alpine (https://github.com/gliderlabs/docker-alpine / http://alpinelinux.org/about/) is a minimal linux distro to make small Docker images. I might ask the Alpine guys if they've come across something similar. It's probably just a missing package.

[JoeDog]

Let me know what you find out.

J.

On Tue, Dec 15, 2015 at 2:38 PM, Josh [email protected] wrote:

No problem. Thanks for trying, for now I can use a debian docker image.
It's much larger (170MB vs. 20MB), but it works. Docker alpine (
https://github.com/gliderlabs/docker-alpine /
http://alpinelinux.org/about/) is a minimal linux distro to make small
Docker images. I might ask the Alpine guys if they've come across something
similar. It's probably just a missing package.

—
Reply to this email directly or view it on GitHub
#43 (comment).

[jstarcher]

Here is the trace:

Program terminated with signal SIGSEGV, Segmentation fault.
Core was generated by `siege -d1 -r10 -c25 example.com'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x00007f3aa8a7539e in new_socket (C=C@entry=0x7f3aaa39d060, hostparam=<optimized out>, portparam=portparam@entry=80) at sock.c:193
193	  memcpy(&cli.sin_addr, hp->h_addr, hp->h_length);
[Current thread is 1 (LWP 118)]

Valgrind:

/ # valgrind siege -r1 -c1 example.com
==14== Memcheck, a memory error detector
==14== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==14== Using Valgrind-3.12.0 and LibVEX; rerun with -h for copyright info
==14== Command: siege -r1 -c1 example.com
==14==
==14== Invalid free() / delete / delete[] / realloc()
==14==    at 0x4C939EA: free (vg_replace_malloc.c:530)
==14==    by 0x405792D: ??? (in /lib/ld-musl-x86_64.so.1)
==14==  Address 0x4e9b180 is in a rw- mapped file /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so segment
==14==
==15== Warning: invalid file descriptor 1009024139 in syscall close()
==15== Invalid read of size 1
==15==    at 0x403E271: ??? (in /lib/ld-musl-x86_64.so.1)
==15==    by 0xFFEFFED17: ???
==15==    by 0xC88949C1894D5148: ???
==15==  Address 0xc88949c1894d5141 is not stack'd, malloc'd or (recently) free'd
==15==
==15==
==15== Process terminating with default action of signal 11 (SIGSEGV)
==15==  General Protection Fault
==15==    at 0x403E271: ??? (in /lib/ld-musl-x86_64.so.1)
==15==    by 0xFFEFFED17: ???
==15==    by 0xC88949C1894D5148: ???
==15==
==15== HEAP SUMMARY:
==15==     in use at exit: 771 bytes in 4 blocks
==15==   total heap usage: 4 allocs, 2 frees, 771 bytes allocated
==15==
==15== LEAK SUMMARY:
==15==    definitely lost: 0 bytes in 0 blocks
==15==    indirectly lost: 0 bytes in 0 blocks
==15==      possibly lost: 0 bytes in 0 blocks
==15==    still reachable: 771 bytes in 4 blocks
==15==         suppressed: 0 bytes in 0 blocks
==15== Rerun with --leak-check=full to see details of leaked memory
==15==
==15== For counts of detected and suppressed errors, rerun with: -v
==15== ERROR SUMMARY: 3 errors from 2 contexts (suppressed: 0 from 0)
==14== Warning: invalid file descriptor -1 in syscall close()
** SIEGE 4.0.3rc4
** Preparing 1 concurrent users for battle.
The server is now under siege...
==14==
==14== Process terminating with default action of signal 11 (SIGSEGV)
==14==  Bad permissions for mapped region at address 0x4085EC8
==14==    at 0x1134BF: cookies_header (cookies.c:172)
==14==
==14== Process terminating with default action of signal 11 (SIGSEGV)
==14==  Bad permissions for mapped region at address 0x4085E70
==14==    at 0x4A8D609: _vgnU_freeres (vg_preloaded.c:83)
==14==
==14== HEAP SUMMARY:
==14==     in use at exit: 114,804 bytes in 46 blocks
==14==   total heap usage: 1,300 allocs, 1,256 frees, 859,639 bytes allocated
==14==
==14== LEAK SUMMARY:
==14==    definitely lost: 424 bytes in 2 blocks
==14==    indirectly lost: 0 bytes in 0 blocks
==14==      possibly lost: 0 bytes in 0 blocks
==14==    still reachable: 114,380 bytes in 44 blocks
==14==         suppressed: 0 bytes in 0 blocks
==14== Rerun with --leak-check=full to see details of leaked memory
==14==
==14== For counts of detected and suppressed errors, rerun with: -v
==14== ERROR SUMMARY: 2 errors from 1 contexts (suppressed: 0 from 0)
Segmentation fault