This is a new promise that guards access to mmap() with MAP_FIXED.
Fixed-address mappings are rarely used, but can be useful if you are
trying to groom the process address space for malicious purposes.
None of our programs need this at the moment, as the only user of
MAP_FIXED is DynamicLoader, but the fixed mappings are constructed
before the process has had a chance to pledge anything.
We want to make sure these functions actually do get unmapped. If they
were inlined somewhere, the inlined version(s) would remain mapped.
Thanks to "thislooksfun" for the suggestion! :^)
There's no real system here, I just added it to various functions
that I don't believe we ever want to call after initialization
has finished.
With these changes, we're able to unmap 60 KiB of kernel text
after init. :^)
You can now declare functions with UNMAP_AFTER_INIT and they'll get
segregated into a separate kernel section that gets completely
unmapped at the end of initialization.
This can be used for anything we don't need to call once we've booted
into userspace.
There are two nice things about this mechanism:
- It allows us to free up entire pages of memory for other use.
(Note that this patch does not actually make use of the freed
pages yet, but in the future we totally could!)
- It allows us to get rid of obviously dangerous gadgets like
write-to-CR0 and write-to-CR4 which are very useful for an attacker
trying to disable SMAP/SMEP/etc.
I've also made sure to include a helpful panic message in case you
hit a kernel crash because of this protection. :^)
If we have a tracer process waiting for us to exec, we need to release
the ptrace lock before stopping ourselves, since otherwise the tracer
will block forever on the lock.
Fixes#5409.
This enabled trivial ASLR bypass for non-dumpable programs by simply
opening /proc/PID/vm before exec'ing.
We now hold the target process's ptrace lock across the refresh/write
operations, and deny access if the process is non-dumpable. The lock
is necessary to prevent a TOCTOU race on Process::is_dumpable() while
the target is exec'ing.
Fixes#5270.
sys$mmap() and related syscalls must pad to the nearest page boundary
below the base address *and* above the end address of the specified
range. Since we have to do this in many places, let's make a helper.
This bug is a good example why copy-paste code should eventually be eliminated
from the code base: Apparently the code was copied from read.cpp before
c6027ed7cc, so the same bug got introduced here.
To recap: A malicious program can ask the Kernel to prepare sys-ing to
a huge amount of iovecs. The Kernel must first copy all the vector locations
into 'vecs', and before that allocates an arbitrary amount of memory:
vecs.resize(iov_count);
This can cause Kernel memory exhaustion, triggered by any malicious userland
program.
This removes some hard references to the toolchain, some unnecessary
uses of an external install command, and disables a -Werror flag (for
the time being) - only if run inside serenity.
With this, we can build and link the kernel :^)
KASAN is a dynamic analysis tool that finds memory errors. It focuses
mostly on finding use-after-free and out-of-bound read/writes bugs.
KASAN works by allocating a "shadow memory" region which is used to store
whether each byte of memory is safe to access. The compiler then instruments
the kernel code and a check is inserted which validates the state of the
shadow memory region on every memory access (load or store).
To fully integrate KASAN into the SerenityOS kernel we need to:
a) Implement the KASAN interface to intercept the injected loads/stores.
void __asan_load*(address);
void __asan_store(address);
b) Setup KASAN region and determine the shadow memory offset + translation.
This might be challenging since Serenity is only 32bit at this time.
Ex: Linux implements kernel address -> shadow address translation like:
static inline void *kasan_mem_to_shadow(const void *addr)
{
return ((unsigned long)addr >> KASAN_SHADOW_SCALE_SHIFT)
+ KASAN_SHADOW_OFFSET;
}
c) Integrating KASAN with Kernel allocators.
The kernel allocators need to be taught how to record allocation state
in the shadow memory region.
This commit only implements the initial steps of this long process:
- A new (default OFF) CMake build flag `ENABLE_KERNEL_ADDRESS_SANITIZER`
- Stubs out enough of the KASAN interface to allow the Kernel to link clean.
Currently the KASAN kernel crashes on boot (triple fault because of the crash
in strlen other sanitizer are seeing) but the goal here is to just get started,
and this should help others jump in and continue making progress on KASAN.
References:
* ASAN Paper: https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/37752.pdf
* KASAN Docs: https://github.com/google/kasan
* NetBSD KASAN Blog: https://blog.netbsd.org/tnf/entry/kernel_address_sanitizer_part_3
* LWN KASAN Article: https://lwn.net/Articles/612153/
* Tracking Issue #5351
There is no reason to call a getter without observing the result, doing
so indicates an error in the code. Mark these methods as [[nodiscard]]
to find these cases.
There is no reason to call a getter without observing the result, doing
so indicates an error in the code. Mark these methods as [[nodiscard]]
to find these cases.
`UserOrKernelBuffer` objects should always be observed when created, in
turn there is no reason to call a getter without observing the result.
Doing either of these indicates an error in the code. Mark these methods
as [[nodiscard]] to find these cases.
There is no reason to call a getter without observing the result, doing
so indicates an error in the code. Mark these methods as [[nodiscard]]
to find these cases.
There is no reason to call a getter without observing the result, doing
so indicates an error in the code. Mark these methods as [[nodiscard]]
to find these cases.
In the majority of cases we want to force callers to observe the
result of a blocking operation as it's not grantee to succeed as
they expect. Mark BlockResult as [[nodiscard]] to force any callers
to observe the result of the blocking operation.
In preparation for marking BlockingResult [[nodiscard]], there are a few
places that perform infinite waits, which we never observe the result of
the wait. Instead of suppressing them, add an alternate function which
returns void when performing and infinite wait.
You can now use the READONLY_AFTER_INIT macro when declaring a variable
and we will put it in a special ".ro_after_init" section in the kernel.
Data in that section remains writable during the boot and init process,
and is then marked read-only just before launching the SystemServer.
This is based on an idea from the Linux kernel. :^)
Since kernel stacks are much smaller (64 KiB) than userspace stacks,
we only add a small bit of randomness here (0-256 bytes, 16b aligned.)
This makes the location of the task context switch buffer not be
100% predictable. Note that we still also add extra randomness upon
syscall entry, so this patch primarily affects context switching.
This patch adds a random offset between 0 and 4096 to the initial
stack pointer in new processes. Since the stack has to be 16-byte
aligned, the bottom bits can't be randomized.
Yet another thing to make things less predictable. :^)
We were doing stack and syscall-origin region validations before
taking the big process lock. There was a window of time where those
regions could then be unmapped/remapped by another thread before we
proceed with our syscall.
This patch closes that window, and makes sys$get_stack_bounds() rely
on the fact that we now know the userspace stack pointer to be valid.
Thanks to @BenWiederhake for spotting this! :^)
If we try to align a number above 0xfffff000 to the next multiple of
the page size (4 KiB), it would wrap around to 0. This is most likely
never what we want, so let's assert if that happens.
Let's be a little more expressive when inducing a kernel panic. :^)
PANIC(...) passes any arguments you give it to dmesgln(), then prints
a backtrace and hangs the machine.
Now that we no longer need to support the signal trampolines being
user-accessible inside the kernel memory range, we can get rid of the
"kernel" and "user-accessible" flags on Region and simply use the
address of the region to determine whether it's kernel or user.
This also tightens the page table mapping code, since it can now set
user-accessibility based solely on the virtual address of a page.
The signal trampoline was previously in kernelspace memory, but with
a special exception to make it user-accessible.
This patch moves it into each process's regular address space so we
can stop supporting user-allowed memory above 0xc0000000.
If we're flushing user space pointers and the process only has one
thread, we do not need to broadcast this to other processors as
they will all discard that request anyway.
We were failing to round down the base of partial VM ranges. This led
to split regions being constructed that could have a non-page-aligned
base address. This would then trip assertions in the VM code.
Found by fuzz-syscalls. :^)
If a program attempts to write from more than a million different locations,
there is likely shenaniganery afoot! Refuse to write to prevent kmem exhaustion.
Found by fuzz-syscalls. Can be reproduced by running this in the Shell:
$ syscall writev 1 [ 0 ] 0x08000000
Found by fuzz-syscalls. Can be reproduced by running this in the Shell:
$ syscall exit_thread
This leaves the process in the 'Dying' state but never actually removes it.
Therefore, avoid this scenario by pretending to exit the entire process.
Since the payload size is user-controlled, this could be used to
overflow the kernel stack.
We should probably also be breaking things into smaller packets at a
higher level, e.g TCPSocket::protocol_send(), but let's do that as
a separate exercise.
Fixes#5310.
Not sure why this was 4 MiB in the first place, but that's a lot of
memory to reserve for each thread when we're running with 512 MiB
total in the default testing setup. :^)
* We don't have to lock the "all IPv4 sockets" in exclusive mode, shared mode is
enough for just reading the list (as opposed to modifying it).
* We don't have to lock socket's own lock at all, the IPv4Socket::did_receive()
implementation takes care of this.
* Most importantly, we don't have to hold the "all IPv4 sockets" across the
IPv4Socket::did_receive() call(s). We can copy the current ICMP socket list
while holding the lock, then release the lock, and then call
IPv4Socket::did_receive() on all the ICMP sockets in our list.
These changes fix a deadlock triggered by receiving ICMP messages when using tap
networking setup (as opposed to QEMU's default user/SLIRP networking) on the host.
The way we read/write directories is very inefficient, and this doesn't
solve any of that. It does however reduce memory usage of directory
entry vectors by 25% which has nice immediate benefits.
CLion doesn't understand that we switch compilers mid-build (which I
can understand since it's a bit unusual.) Defining __serenity__ makes
the majority of IDE features work correctly in the kernel context.
sys$fork() already takes care of children inheriting the parent's root
directory, so there was no need to do the same thing when creating a
new user process.
Add a per-process ptrace lock and use it to prevent ptrace access to a
process after it decides to commit to a new executable in sys$execve().
Fixes#5230.
This patch adds Space, a class representing a process's address space.
- Each Process has a Space.
- The Space owns the PageDirectory and all Regions in the Process.
This allows us to reorganize sys$execve() so that it constructs and
populates a new Space fully before committing to it.
Previously, we would construct the new address space while still
running in the old one, and encountering an error meant we had to do
tedious and error-prone rollback.
Those problems are now gone, replaced by what's hopefully a set of much
smaller problems and missing cleanups. :^)
Wrap thread creation in a Thread::try_create() helper that first
allocates a kernel stack region. If that allocation fails, we propagate
an ENOMEM error to the caller.
This avoids the situation where a thread is half-constructed, without a
valid kernel stack, and avoids having to do messy cleanup in that case.
We now build the kernel with partial UBSAN support.
The following -fsanitize sub-options are enabled:
* nonnull-attribute
* bool
If the kernel detects UB at runtime, it will now print a debug message
with a stack trace. This is very cool! I'm leaving it on by default for
now, but we'll probably have to re-evaluate this as more options are
enabled and slowdown increases.
This achieves two things:
- Programs can now intentionally perform arbitrary syscalls by calling
syscall(). This allows us to work on things like syscall fuzzing.
- It restricts the ability of userspace to make syscalls to a single
4KB page of code. In order to call the kernel directly, an attacker
must now locate this page and call through it.
Leaking macros across headers is a terrible thing, but I can't think of
a better way of achieving this.
- We need some way of modifying debug macros from CMake to implement
ENABLE_ALL_THE_DEBUG_MACROS.
- We need some way of modifying debug macros in specific source files
because otherwise we need to rebuild too many files.
This was done using the following script:
sed -i -E 's/#cmakedefine01 ([A-Z0-9_]+)/#ifndef \1\n\0\n#endif\n/' AK/Debug.h.in
sed -i -E 's/#cmakedefine01 ([A-Z0-9_]+)/#ifndef \1\n\0\n#endif\n/' Kernel/Debug.h.in
There's no need for this to be generic and support running from an
arbitrary thread context. Perf events are always generated from within
the thread being profiled, so take advantage of that to simplify the
code. Also use Vector capacity to avoid heap allocations.
We were checking for size_t (unsigned) overflow but the current offset
is actually stored as off_t (signed). Fix this, and also fail with
EOVERFLOW correctly.
This change can be actually seen as two logical changes, the first
change is about to ensure we only read the ATA Status register only
once, because if we read it multiple times, we acknowledge interrupts
unintentionally. To solve this issue, we always use the alternate Status
register and only read the original status register in the IRQ handler.
The second change is how we handle interrupts - if we use DMA, we can
just complete the request and return from the IRQ handler. For PIO mode,
it's more complicated. For PIO write operation, after setting the ATA
registers, we send out the data to IO port, and wait for an interrupt.
For PIO read operation, we set the ATA registers, and wait for an
interrupt to fire, then we just read from the data IO port.
This patch adds sys$msyscall() which is loosely based on an OpenBSD
mechanism for preventing syscalls from non-blessed memory regions.
It works similarly to pledge and unveil, you can call it as many
times as you like, and when you're finished, you call it with a null
pointer and it will stop accepting new regions from then on.
If a syscall later happens and doesn't originate from one of the
previously blessed regions, the kernel will simply crash the process.
We had two ways of creating a new Ext2FS inode. Either they were empty,
or they started with some pre-allocated size.
In practice, the pre-sizing code path was only used for new directories
and it didn't actually improve anything as far as I can tell.
This patch simplifies inode creation by simply always allocating empty
inodes. Block allocation and block list generation now always happens
on the same code path.
This prevented from dmidecode to get the right PAGE_SIZE when using the
sysconf syscall.
I found this bug, when I tried to figure why dmidecode fails to mmap
/dev/mem when I passed --no-procfs, and the conclusion is that it tried
to mmap unaligned physical address 0xf5ae0 (SMBIOS data), and that was
caused by a wrong value returned after using the sysconf syscall to get
the plaform page size, therefore, allowing to send an unaligned address
to the mmap syscall.
Because it was 'static const' and also shared with userland programs,
the default keymap was defined in multiple places. This commit should
save several kilobytes! :^)
The enumeration code is already enumerating all buses, recursively
enumerating bridges (which are buses) makes devices on bridges being
enumerated multiple times. Also, the PCI code was incorrectly mixing up
terminology; let's settle down on bus, device and function because ever
since PCIe came along "slots" isn't really a thing anymore.
We had an exception that allowed SOL_SOCKET + SO_PEERCRED on local
socket to support LibIPC's PID exchange mechanism. This is no longer
needed so let's just remove the exception.
It's useful for programs to change their thread names to say something
interesting about what they are working on. Let's not require "thread"
for this since single-threaded programs may want to do it without
pledging "thread".
This used to be in Kernel/, next to the build-root-filesystem.sh script,
which was then moved to Meta/ during the transition to CMake but has the
working directory set to Build/, effectively expecting it there - which
seems silly.
TL;DR: Very confusing. Use an explicit path relative to SERENITY_ROOT
instead and update the .gitignore files.
This replaces the current disk detection and disk access code with
code based on https://wiki.osdev.org/IDE
This allows the system to boot on VirtualBox with serial debugging
enabled and VMWare Player.
I believe there were several issues with the current code:
- It didn't utilise the last 8 bits of the LBA in 24-bit mode.
- {read,write}_sectors_with_dma was not setting the obsolete bits,
which according to OSdev wiki aren't used but should be set.
- The PIO and DMA methods were using slightly different copy
and pasted access code, which is now put into a single
function called "ata_access"
- PIO mode doesn't work. This doesn't fix that and should
be looked into in the future.
- The detection code was not checking for ATA/ATAPI.
- The detection code accidentally had cyls/heads/spt as 8-bit,
when they're 16-bit.
- The capabilities of the device were not considered. This is now
brought in and is currently used to check if the device supports
LBA. If not, use CHS.
This prevents sys$mmap() and sys$mprotect() from creating executable
memory mappings in pledged programs that don't have this promise.
Note that the dynamic loader runs before pledging happens, so it's
unaffected by this.
The random address proposals were not checked with the size so it was
increasely likely to try to allocate outside of available space with
larger and larger sizes.
Now they will be ignored instead of triggering a Kernel assertion
failure.
This is a continuation of: c8e7baf4b8
This adds another layer of defense against introducing new code into a
running process. The only permitted way of doing so is by mmapping an
open file with PROT_READ | PROT_EXEC.
This does make any future JIT implementations slightly more complicated
but I think it's a worthwhile trade-off at this point. :^)
This patch adds enforcement of two new rules:
- Memory that was previously writable cannot become executable
- Memory that was previously executable cannot become writable
Unfortunately we have to make an exception for text relocations in the
dynamic loader. Since those necessitate writing into a private copy
of library code, we allow programs to transition from RW to RX under
very specific conditions. See the implementation of sys$mprotect()'s
should_make_executable_exception_for_dynamic_loader() for details.
When mounting an Ext2FS, a block device source is required. All other
filesystem types are unaffected, as most of them ignore the source file
descriptor anyway.
Fixes#5153.