This adds a stub for fnmatch and the following defined values:
- FNM_PATHNAME
- FNM_NOESCAPE
- FNM_PERIOD
- FNM_FILE_NAME
- FNM_LEADING_DIR
- FNM_CASEFOLD
- FNM_EXTMATCH
`--version` always prints "git" for now.
The motivation is that the neofetch port calls `Shell --version` and
adds the output to its output. And if `Shell --version` prints a long
error message about it not knowing the flag, neofetch's output looks a
bit ugly. Per Discord discussion, just add the flag to ArgsParser
instead of only to Shell.
When overriding visit_edges() in a JS::Object subclass, we must make
sure to call the base class visit_edges(), or the object's Shape (and
any properties) will not get marked.
This patch begins the work of sharing types and macros between Kernel
and LibC instead of duplicating them via the kludge in UnixTypes.h.
The basic idea is that the Kernel vends various POSIX headers via
Kernel/API/POSIX/ and LibC simply #include's them to get the macros.
This is no longer needed as per the previous commit, UserspaceEmulator's
malloc tracer now correctly handles functions called from within
`malloc` and `free`. This might also have a benefit on performance
because forcibly inlining all function calls pessimizes cache locality.
It was fragile to use the address of the body of the memory management
functions to disable memory auditing within them. Functions called from
these did not get exempted from the audits, so in some cases
UserspaceEmulator reported bogus heap buffer overflows.
Memory auditing did not work at all on Clang because when querying the
addresses, their offset was taken relative to the base of `.text` which
is not the first segment in the `R/RX/RW(RELRO)/RW(non-RELRO)` layout
produced by LLD.
Similarly to when setting metadata about the allocations, we now use the
`emuctl` system call to selectively suppress auditing when we reach
these functions. This ensures that functions called from `malloc` are
affected too, and no issues occur because of the inconsistency between
Clang and GCC memory layouts.
Previously, we pushed the old `eip` on the stack before reading the new
address, which made us jump to the wrong place if the destination was
relative to the `esp`.
When printing a backtrace, each library's base address is found by
walking through all memory regions in the coredump, and selecting the
address of the first region whose name begins with the library's soname.
This is done to support the Clang toolchain, where .text is not at
offset 0.
However, because the libraries loaded by the emulated process used the
same names, we could not distinguish those with the ones used by
UserspaceEmulator, so the backtrace ended up being garbage.
Using the libraries mapped by UE would not be a sufficient, as the
running application could ask for other libraries too, and doing away
with setting names would make debugging issues within UE code more
difficult.
Some properties, such as `margin`, take multiple values but are not
complicated enough to require special-case handling. These no longer
parsed after my previous StyleValue changes, so this fixes that.
In the future we may want to configure whether to allow this for each
property.
This adds on_tool_activation() to Tool which GuideTool can use
to show guides, if they're hidden, when it's activated. Also
show guides on mousedown since there's no use in drawing invisible
guides.
Tightening of the update rect when firing the marching ants timer
unfortunately meant that only finalized selections would be redrawn,
and any new selection drawn outside of the update rect would not
update.
Apart from now gathering comma-separated font-family names into a
StyleValueList, this also corrects the logic for parsing a single
font-family. Previously, we did not handle unquoted font names at all,
and now they are handled including when they are several words separated
by whitespace.
Modified the logic for `font` to use `parse_font_family_value()`.
`FontStyleValue.font_families()` is now a StyleValueList instead of a
vector, so that it can be better handled in StyleResolver.
We also finally remove the CSS::ParsingContext in
set_property_expanding_shorthands(). :^)
This wraps an x and y background-repeat value. Potentially, we could use
this in place of the background-repeat-x and background-repeat-y
pseudo-properties, but for now StyleResolver splits it into those
properties, like it did before.
This parses the elliptical border-radius values, though currently we
only support circular ones. So, to_length() is overloaded to return the
horizontal-radius. This means the code in Layout/Node.cpp does not need
to change for now.
This is not just moving the code from StyleResolver to Parser. The logic
has changed to allow for the `flex-basis` to come before or after the
`flex-grow/shrink` values, as well as handle the special one-value
cases.
Also added test cases to flex.html to check the parsing. It does parse
correctly, but elements with `flex-basis: auto` do not calculate their
width correctly.
Modified text-decoration.html to better test that the values can be in
any order, and that it adopts the color from the `color` property if no
decoration color is specified. Right now, it always does because we do
not support a different decoration color. Later, we need to support the
`currentcolor` special CSS value for this purpose.
Yes, the name is silly, but it's a StyleValue for list-style, so...
yeah. :^)
Since `list-style-type` and `list-style-image` can both have `none` as a
value, and can appear in any order, we have to handle it separately, and
then assign either or both of those to `none` depending on how many
`none`s there are, and whether those sub-properties already have values.
Added some extra test cases to lists.html to cover list-style-image and
list-style-position parts of the list-style shorthand, and the `none`
values.
This one represents one secton of a `background` property, since it can
have multiple background values separated by commas. Eventually, we will
represent that as a List of BackgroundStyleValues.
Also modified some background-foo properties in StyleResolver so that
the is_background_x() functions could be removed.
I realized that our handling of var() in shorthand properties is wrong,
so have been removing the is_builtin_or_dynamic() calls from the parsing
code for shorthands. This broke our var() test page, so I have replaced
the use of 'background' with 'background-color' there.
After working with the code for a while, it makes more sense to put all
the parsing in Parser, instead of some of it living in StyleResolver.
That means our current ValueListStyleValue needs to be replaced with
specific StyleValue types for the properties that are shorthands or
otherwise combine several values together.
Here we implement FontStyleProperty, which represents a `font` CSS
property.
Also adjusted the fonts.html test page so that font-weights are featured
in test cases without things we do not yet support.
StyleValueList is a list of StyleValues of the same type, for use in
properties like `margin` which accept a variable number of arguments.
I had originally hoped to simply swap the old ValueListStyleValue from
being a list of ComponentValues to one of StyleValues, but I can see now
that I will need to have both for a little while, so renamed the old
is_value_list() to is_component_value_list() temporarily.
I had accidentally parsed it in `parse_builtin_or_dynamic_value()`
instead of `parse_length()` before, which was confusing, so now it's
parsed along with other Lengths.
Whether it should be a Length is up for debate, and has been tripping me
up a few times, but a lot of code expects it to be one. For now, an
'auto' Length value (or any other value which overloads `is_auto()`)
also claims to be a `ValueID::Auto` identifier.
A '0' token can be interpreted both as a Number, and as a Length. This
is problematic as in our CSS parser, we often call parse_css_value()
first, to figure out what something is, and then assign it. So we do not
know in advance whether we want a Length or not. Previously, it always
got parsed as a Length, and then every place that expected a
NumericStyleValue had to also check for a Length(0), which is easy to
forget to do.
In particular, this was causing issues with the `flex` property parsing.
To solve this, we now always parse 0 as a NumericStyleValue, and NSVs of
0 pretend to be a Length(0px) when asked. In two places, we were casting
to a LengthStyleValue* based on is_length(), which no longer works, so
those have been adjusted to use `StyleValue::to_length()` instead. They
also now check for `is_numeric()` first, to avoid the extra conversion
to a Length and back.
Possibly this opens up new issues elsewhere. In my testing it seems
fine, but until we can get CSS test suites running, it's hard to know
for certain.
Similarly to the LibCpp parser regression tests, these tests run the
preprocessor on the .cpp test files under
Userland/LibCpp/Tests/preprocessor, and compare the output with existing
.txt ground truth files.
The preprocessor now understands when a function-like macro is defined,
and can also parse calls to such macros.
The actual evaluation of function-like macros will be done in a
separate commit.
Instead of keeping a separate Vector<Event> for signposts, let them live
in the main event stream. For fast iteration, we instead keep a cache of
the signpost event indices.
In #9373, /usr/local/bin was added to the unveiled directories to make
symbolization work on ports. This directory only exists if at least one
port is installed, so unveil would fail with ENOENT if we had none.
Previously the system would assume the socket was connected after the
file descriptor became writeable. Just because the fd is signaled as
ready for output does not necessarily indicate the socket is connected.
Instead, we should check the status of the socket with SO_ERROR and
handle successes/errors accordingly.
Adds netstat command line argument to display slash-separated pair of
the id and name of the process that owns the socket. User must have
superuser privileges to see information on non-owned sockets.
Also check for the most common event type (sample) first instead of
leaving it as the fallback. This avoids a lot of string comparisons
while parsing profiles.
Heredocs have a different parse end condition than double-quoted
strings. parse_doublequoted_string_inner would assume that a string
would always end in a double quote, so let's generalize it to
parse_string_inner and have it take a StringEndCondition enum which
specifies how the string terminates.
Instead of neatly searching for all framebuffer device nodes and
changing ownership of them, let's generalize this function so we can
apply the same pattern on tty nodes.
GCC implements `fputc`, `fputs` and `fwrite` as builtin functions, whose
`FILE*` argument is implicitly marked `__attribute__((nonnull))`. This
causes our `VERIFY(stream)` statements to be removed. This does not
happen with Clang, as they do not use the `nonnull` attribute in this
way.
The `nonnull` attribute may delete null checks in the generated code, as
per the [GCC documentation]:
> The compiler may also perform optimizations based on the knowledge
> that nonnul parameters cannot be null. This can currently not be
> disabled other than by removing the nonnull attribute.
Disassembling the function as compiled by GCC, we can see that there is
no branch based on if `tv` is null. This means that `gettimeofday`
would produce UB if passed a null parameter, even if we wanted to
predictably return an error. Clang refuses to compile this due to a
`pointer-bool-conversion` warning.
In this commit, `settimeofday` is changed as well to match
`gettimeofday`'s null argument handling.
[GCC documentation]:
https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html#index-nonnull-function-attribute
We only need to know the initial bounds, which we calculate by default
when the interpreter is constructed.
This cuts down on syscalls and makes wasm calls a lot cheaper.
The variant member already contains enough information to give us the
type when needed, so remove the type member and synthesize it when
needed, this allows lots of optimisation opportunaties when copying and
moving Values around.
This should make debugging and profiling much better, at little to no
runtime cost.
Also moves off the operator definitions to a separate header, so it
should also improve the editing experience quite a bit.
Since the InodeIndex encapsulates a 64 bit value, it is correct to
ensure that the Kernel is exposing the entire value and the LibC is
aware of it.
This commit requires an entire re-compile because it's essentially a
change in the Kernel ABI, together with a corresponding change in LibC.
This extracts the sampler functionality into its own class.
Beginning with OpenGL 3 samplers are actual objects, separate
from textures. It makes sense to do this already as it also
cleans up code organization quite a bit.
The POSIX C regex functions are expected to live in the C standard
library, but Serenity split off its regex library into LibRegex. Make a
compromise by implementing stub forwarders for the C regex library that
load libregex.so and call the real implementation.
This is needed for ports that expect these C functions to be available
inside the standard C library without introducing a strong coupling
between LibC and LibDl or LibRegex. The non-standard Serenity C++ regex
API still lives inside LibRegex as before.
28b1e66b51 made that
the m_all_editor_wrappers vector is cleared everytime a project path
is changed (the m_project if check is just for the app launch --
the vector is empty there anyway), making the code never execute.
Previously when user wanted to save an uncreated file, the program
would just quietly ignore the save request, without giving any message.
This can be seen when creating a new editor in split view mode.
Making userspace provide a global string ID was silly, and made the API
extremely difficult to use correctly in a global profiling context.
Instead, simply make the kernel do the string ID allocation for us.
This also allows us to convert the string storage to a Vector in the
kernel (and an array in the JSON profile data.)
The first perf_event argument to a PERF_EVENT_SIGNPOST is now
interpreted as a string ID (in the profile strings set.)
This allows us to generate signposts with custom strings. :^)
This syscall allows userspace to register a keyed string that appears in
a new "strings" JSON object in profile output.
This will be used to add custom strings to profile signposts. :^)
Signposts generated by perf_event(PERF_EVENT_SIGNPOST) now show up in
profile timelines, and if you hover them you get a tooltip with the two
arguments passed with the event.
This file contains the last properties that LibUnicode is not parsing.
Much of the data in this file is not currently used; that is left as a
FIXME for when String.prototype.normalize is implemented. Until then,
only the code point properties are utilized for regular expression
pattern escapes.
These script extensions have some peculiar behavior in the Unicode spec.
The UCD ScriptExtension file does not contain these scripts. Rather, it
is implied the code points which have these scripts as an extension are
the code points that both:
1. Have Common or Inherited as their primary script value
2. Do not have any other script value in their script extension lists
Because these are not explictly listed in the UCD, we must manually form
these script extensions.
Notice that unlike the note in populate_general_category_unions(),
script extension do indeed have code point ranges which overlap. Thus,
this commit adds code to handle that, and hooks it into the GC unions.
Previously, each code point's General Category was part of the generated
UnicodeData structure. This ultimately presented two problems, one
functional and one performance related:
* Some General Categories are applied to unassigned code points, for
example the Unassigned (Cn) category. Unassigned code points are
strictly excluded from UnicodeData.txt, so by relying on that file,
the generator is unable to handle these categories.
* Lookups for General Categories are slower when searching through the
large UnicodeData hash map. Even though lookups are O(1), the hash
function turned out to be slower than binary searching through a
category-specific table.
So, now a table is generated for each General Category. When querying a
code point for a category, a binary search is done on each code point
range in that category's table to check if code point has that category.
Further, General Categories are now parsed from the UCD file
DerivedGeneralCategory.txt. This file is a normal "prop list" file and
contains the categories for unassigned code points.
This was originally used for the "is_final_code_point" algorithm in
LibUnicode/CharacterTypes.cpp. However, it has since been superseded by
DerivedCoreProperties and is now unused. Remove it as it is currently a
waste of time to process the data, and is trivial to add back if we need
it again.
Prior this change, the action opened a File Picker
in user home directory.
Changing the startup path to a project path might make correcting
the path or switching between different projects a bit faster,
as you don't have to go through the subdirectories all over again.
It's also the path that's showed in the project tree view.
When appending two strings together to form a new string, if both of the
strings are already UTF-16, create the new string as UTF-16 as well.
This shaves about 0.5 seconds off the following test262 test:
RegExp/property-escapes/generated/General_Category_-_Decimal_Number.js
The test262 tests under RegExp/property-escapes/generated will invoke
Reflect.apply with up to 10,000 arguments at a time. In LibJS, when the
call stack reached VM::call_internal, we transfer those arguments from
a MarkedValueList to the execution context's arguments Vector.
Because these types differ (MarkedValueList is a Vector<Value, 32>), the
arguments are copied rather than moved. By changing the arguments vector
to a MarkedValueList, we can properly move the passed arguments over.
This shaves about 2 seconds off the following test262 test (from 15sec):
RegExp/property-escapes/generated/General_Category_-_Decimal_Number.js
In addition to invoking js_string() with existing UTF-16 strings when
possible, RegExpExec now takes a Utf16String instead of a Utf16View. The
view was previously fully copied into the returned result object, so
this prevents potentially large copies of string data.
The primary themes here are invoking js_string() with existing instances
of Utf16String when possible, and not creating entire UTF-8 copies when
not needed.
This commit does not go out of its way to reduce copying of the string
data yet, but is a minimum set of changes to compile LibJS after making
PrimitiveString hold a Utf16String.
To help alleviate memory usage when creating and copying large strings,
create a simple wrapper around a Vector<u16> to reference count UTF-16
strings.
TreeView now prints columns mostly like it used to. The paddings are now
properly applied, though. focus_rect drawing has been gated behind a
selection_behavior() check to make sure we don't draw a focus rect
around the column text when we're supposed to draw it over the entire
row.
AbstractTableView (which TreeView inherits from) sets the selection
behavior of the view to SelectRows. This is not how TreeViews are used
in most of the system, and TreeView::paint_event actually always draws
with the assumption of selecting individual items. This commit defines
the expected selection behavior for TreeViews. Users of TreeView can
still override this via TreeView::set_selection_behavior.
This allows tracing the syscalls made by a thread through the kernel's
performance event framework, which is similar in principle to strace.
Currently, this merely logs a stack backtrace to the current thread's
performance event buffer whenever a syscall is made, if profiling is
enabled. Future improvements could include tracing the arguments and
the return value, for example.
WindowServer was not able to utilize any other framebuffer device in the
/dev directory due to wrong group ownership of other framebuffer
devices. Therefore we need to ensure that when SystemServer starts,
it checks all directory entries in /dev, searching for framebuffer
devices, and then apply the correct ownership for them.
This patch adds a vDSO-like mechanism for exposing the current time as
an array of per-clock-source timestamps.
LibC's clock_gettime() calls sys$map_time_page() to map the kernel's
"time page" into the process address space (at a random address, ofc.)
This is only done on first call, and from then on the timestamps are
fetched from the time page.
This first patch only adds support for CLOCK_REALTIME, but eventually
we should be able to support all clock sources this way and get rid of
sys$clock_gettime() in the kernel entirely. :^)
Accesses are synchronized using two atomic integers that are incremented
at the start and finish of the kernel's time page update cycle.
This can happen if the symbol is part of a switch-case, and not
a function, which would previously have made the disassembly view
appear empty.
Now we disassemble the containing function, starting at the given label
and continuing up until the last captured instruction.
These integer => pointer => integer conversions were technically prone
to UB, since they were used as offsets (which are perfectly fine to be
zero), but we calculated them with pointer arithmetic. This made Clang
insert pointer overflow UBSAN checks, which trigger in case of a zero
result.
With this change, LaunchServer will always return an empty list of file
handlers for special files e.g. sockets and devices. Before this change,
TextEditor was always returned as a default handler for these files.
This is a tiny difference and only changes anything for primitives in
strict mode. However this is tested in test262 and can be noticed by
overriding toString of primitive values.
This does now require one to wrap an object in a Value to call invoke
but all code using invoke has been migrated.
Earlier, we were using 0 value for characters not found in "map".
We should return failure for invalid inputs.
So, I have changed the return type of function to Optional<size_t>.
Also changed caller to handle Optional return.
Fixed convert_from_string() function to return correct output.
The value of a number is equal to sum of each digit multiplied by it's
positional weight.
This adds support for the Tools in PixelPaint to use different cursors
within ImageEditor. For now most of them get the crosshair cursor since
it's the most fitting, but in the future we will want to add custom
cursors.
Add a JS_ENUMERATE_INTL_OBJECTS macro and use it to generate:
- Forward declarations
- CommonPropertyNames class name members
- Constructor and prototype GlobalObject members, getters, visitors,
and initialize_constructor() calls
This is the start of implementing ECMA-402 in LibJS, better known as the
ECMAScript Internationalization API.
Much like Temporal this gets its own subdirectory (Runtime/Intl/) as
well as a new C++ namespace (JS::Intl) so we don't have to prefix all
the files and classes with "Intl".
https://tc39.es/ecma402/
This commit adds the PlainYearMonth object itself, its constructor and
prototype (currently empty), and the CreateTemporalYearMonth and
ISOYearMonthWithinLimits abstract operations.
Deleting a symlink via "FileOperation Delete" should not recursively
delete whatever the symlink is pointing to. Same basic idea applies to
the Copy and Move operations.
FileSystemModel will now react to specific events from Core::FileWatcher
in order to granularly update its data based on addition or removal of
files from the tree. Metadata changes are currently not handled, but in
the future they can be used to re-stat() a file to get its updated
statistics.
SortingProxyModel always expected the source model to have the same
number of rows as long as it has not been invalidated. Now that we have
granular updates, this assumption is no longer true, and we need to
resize the source/proxy_rows vectors to the new size. This is safe since
the values in the vector are overwritten right afterwards.
These can be used by Model subclasses to signal the exact operations
that happened to a model, so that persistent model indices in that
area are not invalidated.
This patch adds persistent indices to models. A PersistentModelIndex is
a ModelIndex that will survive most model updates (provided that the
data the PersistentModelIndex points to has not been removed by the
model's data store). PersistentModelIndex objects can be safely held
by objects outside the model they originated from.
Previously HeaderView would just assume that each column or row could
have a minimum size of 2. This makes it so that AbstractTableView
subclasses can provide a new minimum value for a specific column.
This tightens the update rects for EraseTool, BrushTool And the marching
ants update in Selection. Inflate Selection update rect by 10x10 to
avoid misalignment when zoomed out.
Previously using PenTool while zoomed in could cause rendering glitches
since the update rect was misaligned with the image, most likely because
of rounding errors.
Loosening the rect with 1 pixel on either side takes care of this.
After the update -> invalidate change a couple places broke when
update() was supposed to be manually called. This instance was not
marked virtual or override, which made it hard to detect. This commit
makes sure that update() on the original model is called when the
RunningProcessesModel needs an update.
Although this is not spec-compliant, we don't have a way to represent
objects larger than `NumericLimits<size_t>::max()`. Since this abstract
operation is only used when dealing with object size, we don't lose any
functionality by taking that limit into account too.
This fixes a UBSAN error when compiling with Clang.
When the system calls return `NumericLimits<ptrdiff_t>::min()`, negating
the return code would produce `NumericLimits<ptrdiff_t>::max() + 1`
since we are on a two's complement architecture. Because this value
cannot be stored, signed overflow occurs which is UB. This can be fixed
by applying the negation to `EMAXERRNO` since that's known to contain a
relatively small value.
Found when running tests with Clang.
If we call these two functions on a negative value, undefined behavior
occurs due to casting a negative double to an unsigned integer. These
functions are defined to perform modular arithmetic, so negative values
can be fixed up by adding 2^8/2^16.
The reason why this step is not mentioned in ECMA-262 is that it defines
modular arithmetic so that `x mod m` had the same sign as `m`, while
LibM's `fmod(x, m)` copies `x`'s sign.
This issue was found by UBSAN with the Clang toolchain.
Casting a floating point number to an integer and comparing that against
the original value is not a good way to test if it is a whole number. It
may cause unnecessary narrowing conversion issues and UB. This was the
case, which was caught be Clang's `-fsanitize=float-cast-overflow`.
This commit changes the code to use `is_integral_number`, as suggested
in ECMA-262.
If the argument to this function is greater then or equal to 2^32, the
`double` => `u32` cast produces undefined behavior, which Clang catches.
To fix this, we now use `ToUint32` for getting the integer argument, as
specified by ECMA-262.
... segment
This happens with binaries build with Clang or with a custom linker
script. If this is the case, offsets should be calculated not from the
base address of `.text`, but from the first section loaded for the
library.
This commit moves all UserspaceEmulator symbolication into a common
helper function and fixes a FIXME.
This helps us avoid weird truncation issues and fixes a bug on Clang
builds where truncation while reading caused the DIE offsets following
large LEB128 numbers to be incorrect. This removes the need for the
separate `LongUnsignedNumber` type.
