Datasets:
Devilishcode
/
hacking

Dataset card Data Studio Files
xet
 Community
text
stringlengths
0
1.99k
sane and allowed. If so it then performs an action on behalf of that
ring 3 request.
---[ 2.1.2 - Page protections
A page is a physical region of memory that can be mapped to one or more
virtual addresses [3]. These mappings have several flags that determine
how the mapped page can be accessed. For this article we only care
about the following 3 flags:
P - Present
Is this page mapped at all?
R/W - Read/Write
Is this page read-only or writable?
U/S - User/Supervisor
Is this page accessible from ring 3 or only ring 0?
I want to explicitly point out that these flags exist for each separate
mapping of a page. The same physical page can be mapped at multiple
virtual addresses with different permissions.
---[ 2.1.3 - KASLR - Kernel Address Space Layout Randomization
The kernel version of user space ASLR you may already be familiar with.
What this effectively means is that you don't know ahead of time where
in memory the kernel will be mapped.
---[ 2.1.4 - SMEP/SMAP - Supervisor Mode Execution/Access Prevention
These two mitigations prevent the kernel from accessing userspace
memory directly through a pointer. Any data access has to instead go
through special functions that will temporarily disable the mitigation.
Any execution access of userspace memory in kernel mode is completely
disallowed. When the kernel returns to userspace it has to also switch
to user mode at the same time.
---[ 2.1.5 - ToaruOS mitigations overview
+ CPU rings
+ Page Protections
- KASLR
- SMEP/SMAP
On ToaruOS the first two mitigations exist and the latter two don't.
This is more or less expected since the first two are mainly enforced
by the hardware architecture rather than the OS.
The first three of those are the ones you should keep in mind for the
rest of this article.
---[ 3 - Kernel bugs
We are all very used to the security guarantees that our OS provides
and most of us probably take them for granted.
Of course, you can't open /etc/shadow as a normal user.
Of course, you can't just attach a debugger to a root process and alter
what it does.
Of course, you can't change the owner of an suid executable and keep
the suid flag.
But all of those things are enforced by the operating system.
It is common to become root or SYSTEM to demonstrate a kernel exploit,
but the truth is that you effectively have even higher privileges.
If the OS, specifically the kernel, isn't stopping you, you can do
anything.
(Yes, I am ignoring hypervisor based security for dramatic reasons)
All this to say: Kernel bugs may have the same root causes as many user
space bugs, but there are also entirely different bug classes that can
only really exist in a kernel.
So, I encourage you to challenge your preconceptions and question even
those "obvious" security concepts. ToaruOS has quite a few
similarities to Linux, so it is tempting to assume it provides all of
the same guarantees.
---[ 4 - Searching for a bug
Since the kernels' job is to enforce security guarantees it makes sense
to start by looking at how exactly it does that. Our goal is simply to
read a file, so let's look at how we may be able to open it.
---[ 4.1 - How to open a file
If you want to open a file in C you call the libc open() function.
This function internally then issues the corresponding system call.
The kernel side code of ToaruOS that handles the syscall is sys_open()