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 :^)
This required a bit of rearchitecture, as pthread_atfork() required a
mutex, and duplicating a mutex impl for it was silly.
As such, this patch moves some standalone bits of pthread into LibC and
uses those to implement atfork().
It should be noted that for programs that don't use atfork(), this
mechanism only costs two atomic loads (as opposed to the normal mutex
lock+unlock) :^)
This implementation is pretty damn dumb, and probably has more bugs than
features.
But for the time being, it seems to work. however, we should definitely
replace it with a good implementation sometime very soon :^)
Now that we no longer depend on the textual IPC format, we can pass IP addresses
in the format most code actually has and needs it: in binary. The only places we
actually have to deal with textual address representation is:
* When reading /etc/hosts, we have to parse textual addresses & convert them to
binary;
* When doing reverse lookups, we have to form a pseudo-hostname of the form
x.x.x.x.in-addr.arpa.
So we do the conversion in those two cases.
This also increases uniformity between how we handle A (IPv4 address) and other
resource record types. Namely, we now store the raw binary data as received from
a DNS server.
The ad-hoc IPC we were doing with LookupServer was kinda gross. With this,
LookupServer is a regular IPC server. In the future, we want to add more APIs
for LookupServer to talk to its clients (such as DHCPClient telling LookupServer
about the DNS server discovered via DHCP, and DNS-SD client browsing for
services), which calls for a more expressive IPC format; this is what LibIPC is
perfect for.
While the LookupServer side is using the regular LibIPC mechanics and patterns,
the LibC side has to hand-roll LibIPC format serialization without actually
using LibIPC. We might be able to get rid of this in the future, but for now it
has to be like that. The good news is the format is not that bad at all.
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.
Just ignore all these environment flags if the AT_SECURE flag is set in
the program's auxiliary vector.
This prevents a user from tricking set-uid programs into dumping debug
information via environment flags.
load_from_image() becomes map() and link(). This allows us to map
an object before mapping its dependencies.
This solves an issue where fixed-position executables (like GCC)
would clash with the ASLR placement of their own shared libraries.
Refactor DynamicLoader construction with a try_create() helper so that
we can call mmap() before making a loader. This way the loader doesn't
need to have an "mmap failed" state.
This patch also takes care of determining the ELF file size in
try_create() instead of expecting callers to provide it.
Remap the list of atexit handlers as read-only while we're not actively
writing to it. This prevents an attacker from using a memory write
primitive to gain code execution via the atexit list.
This is based on a technique used in OpenBSD. :^)
This can be used to request random VM placement instead of the highly
predictable regular mmap(nullptr, ...) VM allocation strategy.
It will soon be used to implement ASLR in the dynamic loader. :^)
A C++ source file containing just
#include <LibFoo/Bar.h>
should always compile cleanly.
This patch adds missing header inclusions that could have caused weird error
messages if they were used in a different context. Also, this confused QtCreator.
This allows us to implement mkstemp() with open() directly, instead of
first lstat()'ing, and then open()'ing the filename.
Also implement tmpfile() in terms of mkstemp() instead of mktemp().
..and allow implicit creation of KResult and KResultOr from ErrnoCode.
This means that kernel functions that return those types can finally
do "return EINVAL;" and it will just work.
There's a handful of functions that still deal with signed integers
that should be converted to return KResults.
Quoting POSIX:
https://pubs.opengroup.org/onlinepubs/009695399/functions/tzset.html
The tzset() function also shall set the external variable daylight
to 0 if Daylight Savings Time conversions should never be applied
for the timezone in use; otherwise, non-zero.
We're already pretending to be in UTC+0 and setting timezone to 0
accordingly, we can also fake the absence of Daylight Savings Time.
Since tzset() itself pretends to succeed (it just sets timezone = 0 for
now), it seems unwise to leave tzname uninitialized. Since Serenity
already assumes UTC pretty much everywhere time is used, let's continue
that trend here. Quoting POSIX:
https://pubs.opengroup.org/onlinepubs/009695399/functions/tzset.html
The tzset() function shall use the value of the environment variable
TZ to set time conversion information used by ctime(), localtime(),
mktime(), and strftime(). If TZ is absent from the environment,
implementation-defined default timezone information shall be used.
So we still don't care about TZ at all, but the program doesn't need to
know! :^)
This matches what musl libc ("UTC") and glibc ("GMT") do, see:
- https://sourceware.org/git/?p=glibc.git;a=blob;f=time/tzset.c
- https://git.musl-libc.org/cgit/musl/tree/src/time/__tz.c
This adds support for FUTEX_WAKE_OP, FUTEX_WAIT_BITSET, FUTEX_WAKE_BITSET,
FUTEX_REQUEUE, and FUTEX_CMP_REQUEUE, as well well as global and private
futex and absolute/relative timeouts against the appropriate clock. This
also changes the implementation so that kernel resources are only used when
a thread is blocked on a futex.
Global futexes are implemented as offsets in VMObjects, so that different
processes can share a futex against the same VMObject despite potentially
being mapped at different virtual addresses.
All users of this mechanism have been switched to anonymous files and
passing file descriptors with sendfd()/recvfd().
Shbufs got us where we are today, but it's time we say good-bye to them
and welcome a much more idiomatic replacement. :^)
