Crash on any malloc call in instrumented application

[derchr]

Bug description
Running QBDIPreload on applications that eventually call malloc results in a crash. As virtually any application calls malloc once on execution time, I cannot instrument any application that is more complex than hello world. After adding a single malloc call in a hello world example, this crash also occurs.

Bug log
Crash message:

malloc(): unaligned tcache chunk detected
[1]    2888922 IOT instruction (core dumped)  LD_BIND_NOW=1 LD_PRELOAD=./libtracer_preload_c.so grep

Minimal code to reproduce the bug
Run the official QBDIPreload c example with for example grep or something similar:
LD_BIND_NOW=1 LD_PRELOAD=./libtracer_preload_c.so grep

Specific command to reproduce the bug
LD_BIND_NOW=1 LD_PRELOAD=./libtracer_preload_c.so grep

Version (please complete the following information):

• OS: Manjaro Linux (derivative of Arch Linux)
• Architecture: x64
• QBDI version: current dev-next (48bae1d) and also the official binary at version 0.8.0

Additional context
I have a very similar problem when trying to instrument an application with DynamoRIO. There almost all applications crash with a SEGV even in the most trivial examples.
After some testing I found out that this could have been a result of the recent update of the Arch Linux glibc package from version glibc-2.33-5 to glibc-2.35-2.
So it could be that this issue is reproducible on any recent Arch Linux installs.

[nsurbay]

Hello,

I think it may be the name of the libc. Before, the realname of the libc is libc-2.xx.so but it's now libc.so.6.
QBDI should not instrument the libc, as malloc isn't reentrant. We will update QBDIPreload to blacklist the new name.

Thank for the issue.

[derchr]

Just tested the fix and now everything works.

Thank you so much for the quick fix!