Forum

Hi Hugo,

It's been a long time


Hiawatha crashes with a segfault after pointing to a directory with a .hiawatha file containing:

"ShowIndex=yes"


Whether this command is incorrect or not; Is this normal behavior?


Hiawatha version: Latest
Operating System: Fortress Linux

ShowIndex seems to work fine here. Please follow these instructions.

Ok, you could not reproduce it. In that case:

- HW configure options used: --enable chroot --prefix=/usr

- I think you could better adjust that valgrind instruction to:
valgrind --tool=memcheck --leak-check=full ./(location-to)/hiawatha -c /(location-to-config-file) -d 2> /var/log/valgrind.txt

The logfile can be found here:
http://www.fortresslinux.org/download/hiawatha-valgrind-log.txt

Thanks for your bug report. I'll take a look at it.

Can you please download and try this 7.5 beta [www.leisink.org] tarball? Maybe it's something I already fixed.

The provided file is NOT an official release. So please, do not distribute that file.

Sure, you can remove the tarball by now (if you want).

First of all; Before this (In version 7.3) I had a system website-wide "500 internal server error" after after pointing to a directory with a "invalid" .hiawatha file. The only way to solve this, was restarting Hiawatha. I don't know exactly what I put into that file. I think it had something to do with IP blocking rules in Apache style.

Anyway, thank you for the file. It's now running on fortresslinux.org (yes, it a beta, but it is also secured by an LXC container.)

First time I received an 500 error and the second time it crashed again. Do you need an other report?

More information is always welcome.

I'll will create another report next monday.

Same problem and a new error log for you:

http://www.fortresslinux.org/download/hiawatha-valgrind-log1.txt

Weird... Does this also happen if you place the "ShowIndex = yes" in the main configuration file hiawatha.conf?

and... the same problem. It may have something to do with xml or libc if it works fine with you. I'll upgrade the server as soon as the first Alpha of Fortress Linux is finished this month. I'm now too busy writing an automatic compile/kernel configurator.

Could it be some sort of kernel security patch that is causing the problem?

No, had it tested on a clean Xen kernel w/o grsecurity, SELinux, Pax and not even with with stack_smashing protection at home. (stage 1 test).

Do you have a VMware image available of your Fortress Linux? I can't reproduce the crash on my system. Maybe it's something in Fortress Linux that triggers it.