<Not exactly a problem> Running binaries + debugger inside docker

[sivaplaysmC]

NOTE: Since discussions were disabled, raising an issue to discuss about this.

Instead of the current setup which uses glibc_run.sh (which compiles glibc from scratch and then compiles the binary and then links to it), what if we just use a dockerfile that provides the needed glibc version?

Like so:

# for glibc 2.39
FROM docker.io/library/ubuntu:24.04

RUN apt-mark hold libc6 && apt update && apt install -y gdb wget curl xz-utils binutils gcc
RUN curl -qsL 'https://install.pwndbg.re' | sh -s -- -t pwndbg-gdb

RUN mkdir -p /how2heap
WORKDIR /how2heap

COPY compile-all.sh .

COPY *.c .

RUN sh compile-all.sh

# disables ASLR
ENTRYPOINT ["setarch", "-R", "/usr/bin/bash"]

compile-all.sh

#!/usr/bin/env sh

set -eu

CFILES=$(ls *.c)

for cfile in $CFILES
do
    # set -x
    # put each C file into it's own dir.
    dirname=$(echo $cfile | sed s/.c\$//)
    mkdir "$dirname"
    mv $cfile "$dirname/"

    cmd="gcc -no-pie -g $dirname/$cfile -o $dirname/$dirname"
    echo $cmd
    $cmd
done

This scripts creates a neat setup like so:

This allows for a pre-made environment similar to ret2system's web ui, but in the terminal with pwndbg.

[Kyle-Kyle]

So, I agree that this is the best approach (I personally do it as well when debugging techniques for different glibc versions).
However, there is no best way of implementing this because this setup is very personal. For example, my favorite setup is gef+angelheap, which is quite different from the pwndbg setup.
Considering that different Ubuntu docker images have different glibc versions is kinda widely known and it is just a matter of from ubuntu:xxxx and then install wanted tools, I'll leave it out from this repo.

[Kyle-Kyle]

btw, angelheap is this: https://github.com/scwuaptx/Pwngdb