calloc() was internally calling malloc_impl() which would scrub out
all the allocated memory with the scrub byte (0xdc). We would then
immediately zero-fill the memory.
This was obviously a waste of time, and our hash tables were doing
it all the time. :^)
In line wrapping mode, the same column refers both to the end of a
visual line and the beginning of the subsequent visual line. This
meant that the cursor couldn't move to the beginning of a visual
line. Now, we only let the cursor go to the end of a physical line,
not the visual line.
Closes#4786
This will make constructing (and destructing) Positions a lot cheaper
(as it no longer needs to ref() and unref() a String).
Resulted from #5483, but doesn't fix it.
Let's help our future selves find this problem sooner next time it
happens. Hopefully we'll come up with a nicer loader before then,
but who knows. :^)
Instead of having a special case in the dynamic loader where we ignore
TM-related GCC symbols, just stub them out in LibC like we already do
for various other things we don't support.
A new operator, operator""sv was added as of C++17 to support
string_view literals. This allows string_views to be constructed
from string literals and with no runtime cost to find the string
length.
See: https://en.cppreference.com/w/cpp/string/basic_string_view/operator%22%22sv
This change implements that functionality in AK::StringView.
We do have to suppress some warnings about implementing reserved
operators as we are essentially implementing STL functions in AK
as we have no STL :).
Instead of writing to the userspace utsname struct one field at a time,
build up a utsname on the kernel stack and copy it out to userspace
once it's finished. This is both simpler and gets validity checking
built-in for free.
Found by KUBSAN! :^)
Fixes#5499.
If you drag-select a slice of the profile off of the side of the
Profiler window, the profiler will try to render a negative start time,
which will overflow. This commit fixes that bug by clamping timestamps
to the start/end of the profile before rendering.
For some reason I don't yet understand, building the kernel with -O2
produces a way-too-large kernel on some people's systems.
Since there are some really nice performance benefits from -O2 in
userspace, let's do a compromise and build Userland with -O2 but
put Kernel back into the -Os box for now.
Due to the non-standard way the boot assembler code is linked into
the kernel (not and actual dependency, but linked via linker.ld script)
both make and ninja weren't re-linking the kernel when boot.S was
changed. This should theoretically work since we use the cmake
`add_dependencies(..)` directive to express a manual dependency
on boot from Kernel, but something is obviously broken in cmake.
We can work around that with a hack, which forces a dependency on
a file we know will always exist in the kernel (init.cpp). So if
boot.S is rebuilt, then init.cpp is forced to be rebuilt, and then
we re-link the kernel. init.cpp is also relatively small, so it
compiles fast.
https://tc39.es/ecma262/#sec-properties-of-the-regexp-prototype-object
The RegExp prototype object:
- is an ordinary object.
- is not a RegExp instance and does not have a [[RegExpMatcher]]
internal slot or any of the other internal slots of RegExp instance
objects.
In other words: no need to have RegExpPrototype inherit from
RegExpObject (we weren't even calling its initialize()).
https://tc39.es/ecma262/#sec-properties-of-the-array-prototype-object
The Array prototype object: [...] is an Array exotic object and has the
internal methods specified for such objects.
NOTE: The Array prototype object is specified to be an Array exotic
object to ensure compatibility with ECMAScript code that was created
prior to the ECMAScript 2015 specification.
We were only 448 KiB away from filling up the old slot size we reserve
for the kernel above the 3 GiB mark. This expands the slot to 16 MiB,
which allows us to continue booting the kernel until somebody takes
the time to improve our loader.
(...and ASSERT_NOT_REACHED => VERIFY_NOT_REACHED)
Since all of these checks are done in release builds as well,
let's rename them to VERIFY to prevent confusion, as everyone is
used to assertions being compiled out in release.
We can introduce a new ASSERT macro that is specifically for debug
checks, but I'm doing this wholesale conversion first since we've
accumulated thousands of these already, and it's not immediately
obvious which ones are suitable for ASSERT.
The name-to-section lookup table was only used in a handful of places,
and none of them were calling it nearly enough to justify building
a cache for it in the first place. So let's get rid of it and reduce
startup time by a little bit. :^)
With the kernel command line issue fixed, we can now enable these
KUBSAN options without getting triple faults on startup:
* alignment
* null
* pointer-overflow
When building the kernel with -O2, we somehow ended up with the kernel
command line outside of the lower 8MB of physical memory. Since we don't
map that area in our initial page table setup, we would triple fault
when trying to parse the command line.
This patch sidesteps the issue by copying the (first 4KB of) the kernel
command line to a buffer in a known safe location at boot.
It's a lot faster to iterate the GNU hash tables if we don't have to
compute the length of every symbol name before rejecting it anyway while
comparing the first character. :^)
When performing a global symbol lookup, we were recomputing the symbol
hashes once for every dynamic object searched. The hash function was
at the very top of a profile (15%) of program startup.
With this change, the hash function is no longer visible among the top
stacks in the profile. :^)
This logging mode was unusable anyway since it spams way too much.
The dynamic loader is in a pretty good place now anyway, so I think
it's okay for us to drop some of the bring-up debug logging. :^)
Also, we have to be careful with dbgln_if(FOO_DEBUG, "{}", foo())
where foo() is something expensive, since it might get evaluated
even if !FOO_DEBUG.
