If the stored symbol is custom (non-global), it is allocated on the
heap, and may not be visited by anyone else, so we must register it in
order to keep it alive.
This was previously indirectly forcing Heap/Handle.h to include it
instead. This will let us include Handle.h from PropertyKey, which will
allow us to solve a different issue.
The BootFramebufferConsole class maps the framebuffer using the
MemoryManager, so to be able to draw the logo, we need to get this
mapped framebuffer. This commit adds a unsafe API for that.
The MemoryManager now works, so we can use the same code as on x86 to
map the framebuffer. Since it uses the MemoryManager, the initialization
of the BootFramebufferConsole has to happen after the MemoryManager is
working.
For the initial page tables we only need to identity map the kernel
image, the rest of the memory will be managed by the MemoryManager. The
linker script is updated to get the kernel image start and end
addresses.
The page table and page directory formats are architecture specific, so
move the headers into the Arch directory. Also move the aarch64 page
table constants from aarch64/MMU.cpp to aarch64/PageDirectory.h.
When an exception happens it is sometimes hard to figure out where
exactly the exception happened, so use the frame pointer of the trap
frame to print a backtrace.
Previously Welcome relied on a bogus executable key value to disable
startup. This always printed an error on login and littered the config
file with a useless entry. Adding/removing the group as needed seems
a bit nicer.
Plumbs synchronous calls for adding and removing group entries to
config files. This is useful for services like SystemServer which
default to group names for executable paths, and for removing all
keys at once.
While this loses quite a bit of accuracy (although to no apparent
decrease in emulation quality) , it helps avoiding the additional
overhead of the `clock_gettime` syscall (as `CLOCK_MONOTONIC_COARSE`
is forwarded using the mapped time page) and we don't have to do a
HPET timer read for each tick.
This results in a decrease of Serenity boot time from 1h16m down to
42m when running on Serenity.
I thought the spec listing out the elements again was an oversight, but
it isn't, as simply inverting "is_actually_disabled" makes :enabled
apply to every element.
Previously we only considered an element disabled if it was an <input>
element with the disabled attribute, but there's way more elements that
apply with more nuanced disabled/enabled rules.
This check happens very often in LibGUI code. 25% of time spent
layouting the emoji input dialog was wasted on RTTI. Adding a simple
fast_is<Widget>() melts almost all of that away.
