We used to build with -Os in order to fit within a certain size, but
there isn't really a good reason for that kind of restriction.
Switching to -O2 yields a significant improvement in throughput,
for example `test-js` is roughly 20% faster on my machine. :^)
Now that the userland has a compatiblity wrapper for select(), the
kernel doesn't need to implement this syscall natively. The poll()
interface been around since 1987, any code still using select()
should be slapped silly.
Note: the SerenityOS source tree mostly uses select() and not poll()
despite SerenityOS having support for poll() since early 2019...
Like what happened with the PCI and USB code, this feels like the right
thing to do because we can improve on the ATA capabilities and keep it
distinguished from the rest of the subsystem.
We create a base class called GenericFramebufferDevice, which defines
all the virtual functions that must be implemented by a
FramebufferDevice. Then, we make the VirtIO FramebufferDevice and other
FramebufferDevice implementations inherit from it.
The most important consequence of rearranging the classes is that we now
have one IOCTL method, so all drivers should be committed to not
override the IOCTL method or make their own IOCTLs of FramebufferDevice.
All graphical IOCTLs are known to all FramebufferDevices, and it's up to
the specific implementation whether to support them or discard them (so
we require extensive usage of KResult and KResultOr, together with
virtual characteristic functions).
As a result, the interface is much cleaner and understandable to read.
A VirtIO graphics adapter is really the VirtIO GPU, so the virtualized
hardware has no distinction between both components so there's no
need to put such distinction in software.
We might need to split things in the future, but when we do so, we must
take proper care to ensure that the interface between the components
is correct and use the usual codebase patterns.
By enabling LTO for the kernel_heap object too, we open the door for
optimization opportunities that come from (partially) inlining `::new`
or kmalloc. Every software spends a non-trivial amount of its run time
on allocating memory, so hopefully this change will make LTO builds even
faster.
This commit updates the Clang toolchain's version to 13.0.0, which comes
with better C++20 support and improved handling of new features by
clang-format. Due to the newly enabled `-Bsymbolic-functions` flag, our
Clang binaries will only be 2-4% slower than if we dynamically linked
them, but we save hundreds of megabytes of disk space.
The `BuildClang.sh` script has been reworked to build the entire
toolchain in just three steps: one for the compiler, one for GNU
binutils, and one for the runtime libraries. This reduces the complexity
of the build script, and will allow us to modify the CI configuration to
only rebuild the libraries when our libc headers change.
Most of the compile flags have been moved out to a separate CMake cache
file, similarly to how the Android and Fuchsia toolchains are
implemented within the LLVM repo. This provides a nicer interface than
the heaps of command-line arguments.
We no longer build separate toolchains for each architecture, as the
same Clang binary can compile code for multiple targets.
The horrible mess that `SERENITY_CLANG_ARCH` was, has been removed in
this commit. Clang happily accepts an `i686-pc-serenity` target triple,
which matches what our GCC toolchain accepts.
LLD fails to define the _GLOBAL_OFFSET_TABLE_ symbol if all inputs to it
are LLVM bitcode files (i.e. those used for LTO). To allow the kernel to
be built with ThinLTO, the workaround suggested in the original LLVM bug
report (<https://bugs.llvm.org/show_bug.cgi?id=39634>) is added in this
commit.
There's basically no real difference in software between a SATA harddisk
and IDE harddisk. The difference in the implementation is for the host
bus adapter protocol and registers layout.
Therefore, there's no point in putting a distinction in software to
these devices.
This change also greatly simplifies and removes stale APIs and removes
unnecessary parameters in constructor calls, which tighten things
further everywhere.
This singleton simplifies many aspects that we struggled with before:
1. There's no need to make derived classes of Device expose the
constructor as public anymore. The singleton is a friend of them, so he
can call the constructor. This solves the issue with try_create_device
helper neatly, hopefully for good.
2. Getting a reference of the NullDevice is now being done from this
singleton, which means that NullDevice no longer needs to use its own
singleton, and we can apply the try_create_device helper on it too :)
3. We can now defer registration completely after the Device constructor
which means the Device constructor is merely assigning the major and
minor numbers of the Device, and the try_create_device helper ensures it
calls the after_inserting method immediately after construction. This
creates a great opportunity to make registration more OOM-safe.
Replace the old logic where we would start with a host build, and swap
all the CMake compiler and target variables underneath it to trick
CMake into building for Serenity after we configured and built the Lagom
code generators.
The SuperBuild creates two ExternalProjects, one for Lagom and one for
Serenity. The Serenity project depends on the install stage for the
Lagom build. The SuperBuild also generates a CMakeToolchain file for the
Serenity build to use that replaces the old toolchain file that was only
used for Ports.
To ensure that code generators are rebuilt when core libraries such as
AK and LibCore are modified, developers will need to direct their manual
`ninja` invocations to the SuperBuild's binary directory instead of the
Serenity binary directory.
This commit includes warning coalescing and option style cleanup for the
affected CMakeLists in the Kernel, top level, and runtime support
libraries. A large part of the cleanup is replacing USE_CLANG_TOOLCHAIN
with the proper CMAKE_CXX_COMPILER_ID variable, which will no longer be
confused by a host clang compiler.
This change removes the halt and reboot syscalls, and create a new
mechanism to change the power state of the machine.
Instead of how power state was changed until now, put a SysFS node as
writable only for the superuser, that with a defined value, can result
in either reboot or poweroff.
In the future, a power group can be assigned to this node (which will be
the GroupID responsible for power management).
This opens an opportunity to permit to shutdown/reboot without superuser
permissions, so in the future, a userspace daemon can take control of
this node to perform power management operations without superuser
permissions, if we enforce different UserID/GroupID on that node.
Both should reside in the SysFS firmware directory which is normally
located in /sys/firmware.
Also, apply some OOM-safety patterns when creating the BIOS and ACPI
directories.
This will somwhat help unify them also under the same SysFS directory in
the commit.
Also, it feels much more like this change reflects the reality that both
ACPI and the BIOS are part of the firmware on x86 computers.
These interfaces are broken for about 9 months, maybe longer than that.
At this point, this is just a dead code nobody tests or tries to use, so
let's remove it instead of keeping a stale code just for the sake of
keeping it and hoping someone will fix it.
To better justify this, I read that OpenBSD removed loadable kernel
modules in 5.7 release (2014), mainly for the same reason we do -
nobody used it so they had no good reason to maintain it.
Still, OpenBSD had LKMs being effectively working, which is not the
current state in our project for a long time.
An arguably better approach to minimize the Kernel image size is to
allow dropping drivers and features while compiling a new image.
Let's remove the DynamicParser class, as it really did nothing yet in
the Kernel. Instead, when we add support for AML parsing, we can figure
out how to do it properly without the need of a derived class that just
complicates everything for no good reason.
The current implementation of DevFS resembles the linux devtmpfs, and
not the traditional DevFS, so let's rename it to better represent the
direction of the development in regard to this filesystem.
The abbreviation for DevTmpFS is still "dev", because it doesn't add
value as a commandline option to make it longer.
In quick summary - DevFS in unix OSes is simply a static filesystem, so
device nodes are generated and removed by the kernel code. DevTmpFS
is a "modern reinvention" of the DevFS, so it is much more like a TmpFS
in the sense that not only it's stored entirely in RAM, but the userland
is responsible to add and remove devices nodes as it sees fit, and no
kernel code is directly being involved to keep the filesystem in sync.
A couple of things were changed:
1. Semantic changes - PCI segments are now called PCI domains, to better
match what they are really. It's also the name that Linux gave, and it
seems that Wikipedia also uses this name.
We also remove PCI::ChangeableAddress, because it was used in the past
but now it's no longer being used.
2. There are no WindowedMMIOAccess or MMIOAccess classes anymore, as
they made a bunch of unnecessary complexity. Instead, Windowed access is
removed entirely (this was tested, but never was benchmarked), so we are
left with IO access and memory access options. The memory access option
is essentially mapping the PCI bus (from the chosen PCI domain), to
virtual memory as-is. This means that unless needed, at any time, there
is only one PCI bus being mapped, and this is changed if access to
another PCI bus in the same PCI domain is needed. For now, we don't
support mapping of different PCI buses from different PCI domains at the
same time, because basically it's still a non-issue for most machines
out there.
2. OOM-safety is increased, especially when constructing the Access
object. It means that we pre-allocating any needed resources, and we try
to find PCI domains (if requested to initialize memory access) after we
attempt to construct the Access object, so it's possible to fail at this
point "gracefully".
3. All PCI API functions are now separated into a different header file,
which means only "clients" of the PCI subsystem API will need to include
that header file.
4. Functional changes - we only allow now to enumerate the bus after
a hardware scan. This means that the old method "enumerate_hardware"
is removed, so, when initializing an Access object, the initializing
function must call rescan on it to force it to find devices. This makes
it possible to fail rescan, and also to defer it after construction from
both OOM-safety terms and hotplug capabilities.
This change adds a static lock hierarchy / ranking to the Kernel with
the goal of reducing / finding deadlocks when running with SMP enabled.
We have seen quite a few lock ordering deadlocks (locks taken in a
different order, on two different code paths). As we properly annotate
locks in the system, then these facilities will find these locking
protocol violations automatically
The `LockRank` enum documents the various locks in the system and their
rank. The implementation guarantees that a thread holding one or more
locks of a lower rank cannot acquire an additional lock with rank that
is greater or equal to any of the currently held locks.
Add a dummy Arch/aarch64/boot.S that for now does nothing but
let all processor cores sleep.
For now, none of the actual Prekernel code is built for aarch64.
The pattern of having Prekernel inherit all of the build flags of the
Kernel, and then disabling some flags by adding `-fno-<flag>` options
to then disable those options doesn't work in all scenarios. For example
the ASAN flag `-fasan-shadow-offset=<offset>` has no option to disable
it once it's been passed, so in a future change where this flag is added
we need to be able to disable it cleanly.
The cleaner way is to just allow the Prekernel CMake logic to filter out
the COMPILE_OPTIONS specified for that specific target. This allows us
to remove individual options without trashing all inherited options.
Now that the old PCI::Device was removed, we can complete the PCI
changes by making the PCI::DeviceController to be named PCI::Device.
Really the entire purpose and the distinction between the two was about
interrupts, but since this is no longer a problem, just rename it to
simplify things further.
I created this class a long time ago just to be able to quickly make a
PCI device to also represent an interrupt handler (because PCI devices
have this capability for most devices).
Then after a while I introduced the PCI::DeviceController, which is
really almost the same thing (a PCI device class that has Address member
in it), but is not tied to interrupts so it can have no interrupts, or
spawn interrupt handlers however it wants to seems fit.
However I decided it's time to say goodbye for this class for
a couple of reasons:
1. It made a whole bunch of weird patterns where you had a PCI::Device
and a PCI::DeviceController being used in the topic of implementation,
where originally, they meant to be used mutually exclusively (you
can't and really don't want to use both).
2. We can really make all the classes that inherit from PCI::Device
to inherit from IRQHandler at this point. Later on, when we have MSI
interrupts support, we can go further and untie things even more.
3. It makes it possible to simplify the VirtIO implementation to a great
extent. While this commit almost doesn't change it, future changes
can untangle some complexity in the VirtIO code.
For UHCIController, E1000NetworkAdapter, NE2000NetworkAdapter,
RTL8139NetworkAdapter, RTL8168NetworkAdapter, E1000ENetworkAdapter we
are simply making them to inherit the IRQHandler. This makes some sense,
because the first 3 devices will never support anything besides IRQs.
For the last 2, they might have MSI support, so when we start to utilize
those, we might need to untie these classes from IRQHandler and spawn
IRQHandler(s) or MSIHandler(s) as needed.
The VirtIODevice class is also a case where we currently need to use
both PCI::DeviceController and IRQHandler classes as parents, but it
could also be untied from the latter.
The number of UHCI related files is starting to expand to the point
where it's best if we move this into their own subdirectory. It'll
also make it easier to manage when we decide to add some more
controller types (whenever that may be)
The `-z,text` linker flag causes the linker to reject shared libraries
and PIE executables that have textrels. Our code mostly did not use
these except in one place in LibC, which is changed in this commit.
This makes GNU ld match LLD's behavior, which has this option enabled by
default.
TEXTRELs pose a security risk, as performing these relocations require
executable pages to be written to by the dynamic linker. This can
significantly weaken W^X hardening mitigations.
Note that after this change, TEXTRELs can still be used in ports, as the
dynamic loader code is not changed. There are also uses of it in the
kernel, removing which are outside the scope of this PR. To allow those,
`-z,notext` is added.
