1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-10 08:17:35 +00:00
Commit graph

6017 commits

Author SHA1 Message Date
Andreas Kling
a29788e58a Kernel: Don't record sys$perf_event() if profiling is not enabled
If you want to record perf events, just enable profiling. This allows us
to add random perf events to programs without littering the file system
with perfcore files.
2021-08-12 00:03:40 +02:00
Andreas Kling
1e90a3a542 Kernel: Make sys$perf_register_string() generate the string ID's
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.)
2021-08-12 00:03:39 +02:00
Andreas Kling
4657c79143 Kernel+LibC: Add sys$perf_register_string()
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. :^)
2021-08-12 00:03:39 +02:00
Andreas Kling
0d997d48ea Kernel+LibC: Add PERF_EVENT_SIGNPOST
This event will be used by userspace programs wanting to mark
interesting high-level events in the profile. :^)
2021-08-12 00:03:38 +02:00
Andreas Kling
7f50805903 Kernel/SMP: Fix RecursiveSpinLock remembering the wrong CPU when locking
We have to disable interrupts before capturing the current Processor*,
or we risk storing the wrong one if we get preempted and resume on a
different CPU.

Caught by the VERIFY in RecursiveSpinLock::unlock()
2021-08-11 12:34:55 +02:00
Andreas Kling
5e27861c2e Kernel: Don't ask RTC for current time when generating coredumps
Use kgettimeofday() like everybody else. :^)
2021-08-11 12:34:55 +02:00
Gunnar Beutner
9964c106ad Kernel: Disambiguate instruction size for mov in read_gs_ptr
Previously we allowed using immediate values here which is ambiguous
as to which specific mov instruction the compiler should choose.
2021-08-11 12:34:47 +02:00
Timothy Flynn
c16aca7abf AK+Kernel: Add StringBuilder::append overload for UTF-16 views
Currently, to append a UTF-16 view to a StringBuilder, callers must
first convert the view to UTF-8 and then append the copy. Add a UTF-16
overload so callers do not need to hold an entire copy in memory.
2021-08-10 23:07:50 +02:00
Gunnar Beutner
3322efd4cd Kernel: Fix kernel panic when blocking on the process' big lock
Another thread might end up marking the blocking thread as holding
the lock before it gets a chance to finish invoking the scheduler.
2021-08-10 22:33:50 +02:00
LuK1337
e1ed1d9176 VirtualFileSystem: Return early in rename() when old_path==new_path
This change fixes disappearing symlink after trying to do the following
thing:
  cp /res/icons/16x16/add-event.png .
  ln -s add-event.png a
  mv a a
2021-08-10 21:56:10 +02:00
Jean-Baptiste Boric
0286160b62 Kernel: Add syscall performance event type
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.
2021-08-10 21:55:48 +02:00
Jean-Baptiste Boric
4c4b8ea443 Kernel: Properly handle non-blocking I/O on pipes
Previously, non-blocking read operations on pipes returned EOF instead
of EAGAIN and non-blocking write operations blocked. This fixes
pkgsrc's bmake "eof on job pipe!" error message when running in
non-compatibility mode.
2021-08-10 21:55:35 +02:00
Andreas Kling
b7dae4f90e Kernel: Add CLOCK_MONOTONIC_COARSE to the kernel time page
This allows clock_gettime(CLOCK_MONOTONIC_COARSE) without syscalls.
Core::EventLoop takes advantage of this automatically. :^)
2021-08-10 21:51:05 +02:00
Andreas Kling
aaead6f332 Kernel: Only expose CLOCK_REALTIME_COARSE via the kernel time page
Non-COARSE clock sources may probably still require a syscall.
2021-08-10 21:51:05 +02:00
Andreas Kling
11456ebc00 Kernel: Close race window in timestamp update mechanism
As pointed out by 8infy, this mechanism is racy:

    WRITER:

        1. ++update1;
        2. write_data();
        3. ++update2;

    READER:

        1. do { auto saved = update1;
        2. read_data();
        3. } while (saved != update2);

The following sequence can lead to a bogus/partial read:

    R1      R2  R3
        W1  W2      W3

We close this race by incrementing the second update counter first:

    WRITER:

        1. ++update2;
        2. write_data();
        3. ++update1;
2021-08-10 21:51:05 +02:00
brapru
342e1f0a84 Kernel: Properly implement SO_ERROR option
This fixes the placeholder stub for the SO_ERROR via getsockopt. It
leverages the m_so_error value that each socket maintains. The SO_ERROR
option obtains and then clears this field, which is useful when checking
for errors that occur between socket calls. This uses an integer value
to return the SO_ERROR status.

Resolves #146
2021-08-10 20:59:53 +02:00
brapru
0095c7cb7d Kernel: Add so_error to keep track of the socket's error state
This sets the m_so_error variable every time the socket returns an
error.
2021-08-10 20:59:53 +02:00
Andreas Kling
fdfc66db61 Kernel+LibC: Allow clock_gettime() to run without syscalls
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.
2021-08-10 19:21:16 +02:00
Andreas Kling
f02d73db4d LibC+Kernel: Use an enum for clockid_t values 2021-08-10 13:01:39 +02:00
Andreas Kling
595ed59eb7 Kernel: Alphabetize the syscall list 2021-08-10 13:01:39 +02:00
Andreas Kling
fa64ab26a4 Kernel+UserspaceEmulator: Remove unused sys$gettimeofday()
Now that LibC uses clock_gettime() to implement gettimeofday(), we can
get rid of this entire syscall. :^)
2021-08-10 13:01:39 +02:00
Andreas Kling
0a02496f04 Kernel/SMP: Change critical sections to not disable interrupts
Leave interrupts enabled so that we can still process IRQs. Critical
sections should only prevent preemption by another thread.

Co-authored-by: Tom <tomut@yahoo.com>
2021-08-10 02:49:37 +02:00
Andreas Kling
364134ad4b Kernel/SMP: Skip thread registers in core dump if there is no trap frame
We can only get thread registers if there's a trap frame.
2021-08-10 02:49:37 +02:00
Andreas Kling
9babb92a4b Kernel/SMP: Make entering/leaving critical sections multi-processor safe
By making these functions static we close a window where we could get
preempted after calling Processor::current() and move to another
processor.

Co-authored-by: Tom <tomut@yahoo.com>
2021-08-10 02:49:37 +02:00
Liav A
d16d116b65 Kernel/USB: Tighten up USBManagement enumeration code a bit 2021-08-09 22:52:09 +02:00
Luke
7a86a8df90 Kernel/USB: Create controller base class and introduce USBManagement
This removes Pipes dependency on the UHCIController by introducing a
controller base class. This will be used to implement other controllers
such as OHCI.

Additionally, there can be multiple instances of a UHCI controller.
For example, multiple UHCI instances can be required for systems with
EHCI controllers. EHCI relies on using multiple of either UHCI or OHCI
controllers to drive USB 1.x devices.

This means UHCIController can no longer be a singleton. Multiple
instances of it can now be created and passed to the device and then to
the pipe.

To handle finding and creating these instances, USBManagement has been
introduced. It has the same pattern as the other management classes
such as NetworkManagement.
2021-08-09 21:05:25 +02:00
Jesse Buhagiar
c5974d3518 USB: Fix wrong port read/write in portscan daemon
Port2 logic was errantly using `portsc1` registers, meaning that
the port wouldn't be reset properly. In effect, this puts devices
connected to Port2 in an undefined state.
2021-08-09 21:04:55 +02:00
Andreas Kling
74e6a70958 Kernel/SMP: Don't process SMP messages in non-SMP mode
Processing SMP messages outside of non-SMP mode is a waste of time,
and now that we don't rely on the side effects of calling the message
processing function, let's stop calling it entirely. :^)
2021-08-09 13:39:08 +02:00
Andreas Kling
a971de89d3 Kernel/SMP: Process the deferred call queue in exit_trap()
We were previously relying on a side effect of the critical section in
smp_process_pending_messages(): when exiting that section, it would
process any pending deferred calls.

Instead of relying on that, make the deferred invocations explicit by
calling deferred_call_execute_pending() in exit_trap().

This ensures that deferred calls get processed before entering the
scheduler at the end of exit_trap(). Since thread unblocking happens
via deferred calls, the threads don't have to wait until the next
scheduling opportunity when they could be ready *now*. :^)

This was the main reason Tom's SMP branch ran slowly in non-SMP mode.
2021-08-09 13:35:52 +02:00
Andreas Kling
57a7dfbd28 Kernel/SMP: Don't process SMP messages in exit_trap() in non-SMP mode 2021-08-09 13:23:42 +02:00
Andreas Kling
f27e7bbbf4 Kernel/SMP: Don't enable interrupts in Processor::exit_trap
Enter a critical section in Processor::exit_trap so that processing
SMP messages doesn't enable interrupts upon leaving. We need to delay
this until the end where we call into the Scheduler if exiting the
trap results in being outside of a critical section and irq handler.

Co-authored-by: Tom <tomut@yahoo.com>
2021-08-09 13:22:22 +02:00
Andreas Kling
cd0fc7f52c Kernel/SMP: Mark s_smp_enabled READONLY_AFTER_INIT
We can't enter/leave SMP mode once the kernel is up and running.
2021-08-09 13:19:26 +02:00
Andreas Kling
f357fd91da Kernel: Don't hog MM lock in Memory::Region::clone()
We don't need to hold the MM lock across all of Region::clone().
2021-08-09 11:46:31 +02:00
Andreas Kling
08b4d8f0de Kernel: Fix deadlock in ~Memory::Region()
First off: unregister the region from MemoryManager before unmapping it.
The order of operations here was a bit strange, presumably to avoid a
situation where a fault would happen while unmapping, and the fault
handler would find the MemoryManager region list in an invalid state.

Unregistering it before unmapping sidesteps the whole problem, and
allows us to easily fix another problem: a deadlock could occur due
to inconsistent acquisition order (PageDirectory must come before MM.)
2021-08-09 11:46:31 +02:00
Andreas Kling
95c8e421ae Kernel: Don't hog MM lock in find_region_from_vaddr()
We don't want to be holding the MM lock if it's a user region and we
have to consult the page directory, since that can lead to a deadlock
if we don't already have the page directory lock.
2021-08-09 11:46:31 +02:00
Andreas Kling
96c7b70de3 Kernel: Remove unused list of user regions in MemoryManager
We were putting all the user Region objects on an intrusive list
but not using it for anything.
2021-08-09 11:46:31 +02:00
Andreas Kling
ab5c422a29 Kernel/SMP: Make SMP message queueing work correctly
- Use the receiver's per-CPU entry in the message, instead of the
  sender's. (Using the sender's entry wasn't safe for broadcast
  messages since the same entry ended up on multiple message queues.)

- Retry the CAS until it *succeeds* instead of *fails*. This closes a
  race window, and also ensures a correct return value. The return value
  is used by the caller to decide whether to broadcast an IPI.
  This was the main reason smp=on was so slow. We had CPUs busy-waiting
  until someone else triggered an IPI and moved things along.

- Add a CPU pause hint to the spin loop. :^)
2021-08-09 11:46:31 +02:00
Andreas Kling
6283acb6b7 Kernel/SMP: Don't panic in inode fault if other CPU already paged it in
It may happen that CPU A manages to page in from the same inode
while we're just entering the same page fault handler on CPU B.

Handle it gracefully by checking if the data has already been paged in
(instead of VERIFY'ing that it hasn't) and then remap the page if that's
the case.
2021-08-09 11:46:31 +02:00
Andreas Kling
00bbbdeda6 Kernel/SMP: Always take PageDirectory lock before the MemoryManager lock
This prevents deadlocking due to inconsistent acquisition order.
2021-08-09 11:46:31 +02:00
Andreas Kling
d21b8f9013 Kernel/SMP: Fix ProcessorMessage deallocation bug
Due to a boolean mistake in smp_return_to_pool(), we didn't retry
pushing the message onto the freelist after a failed attempt.

This caused the message pool to eventually become completely empty
after enough contentious access attempts.

This patch also adds a pause hint to the CPU in the failed attempt
code path.
2021-08-09 11:46:30 +02:00
Andreas Kling
f3fed411d4 Kernel: Rename Processor::smp_queue_message() => smp_enqueue_message() 2021-08-09 11:46:30 +02:00
Andreas Kling
46215a8183 Kernel: Add Processor::pause() and use it to give the CPU a rest
On x86, the "pause" instruction is a "spin loop hint".
2021-08-09 11:46:30 +02:00
Andreas Kling
b7129c57df Kernel: Only get register dump when we have a trap
Co-authored-by: Tom <tomut@yahoo.com>
2021-08-09 11:46:30 +02:00
Andreas Kling
cebdb0b575 Kernel: Fix logic typo in ConsoleManagement::is_initialized()
Regressed in 2ff3c54153.
2021-08-09 11:46:30 +02:00
Brian Gianforcaro
8e93815846 Kernel: Fix panic loop when encountering an unknown boot_mode
The kernel panic handler now parses the kernels boot_mode to decide how
to handle the panic. So the previous logic could end up in an panic loop
until we blew out the kernel stack.

Instead only validate the kernel's boot mode once per boot, after
initializing the kernel command line.
2021-08-08 23:11:50 +02:00
Gunnar Beutner
aabbfa78e2 DynamicLoader: Make sure we don't link against libgcc_s
This bug was reintroduced by the removal of -fbuilding-gcc.
2021-08-08 16:41:51 +02:00
Andreas Kling
2be368e4c4 Kernel: Rename queue_runnable_thread() => enqueue_runnable_thread() 2021-08-08 14:24:55 +02:00
Andreas Kling
0910979dec Kernel: Remove unused Process::FileDescriptions::fd_flags() 2021-08-08 14:24:55 +02:00
Andreas Kling
9ccc77092d Kernel: Remove unused cruft from FileDescriptionAndFlags 2021-08-08 14:24:55 +02:00
Andreas Kling
d959b5d9dd Kernel: Remove unused template variant of create_kernel_thread() 2021-08-08 14:24:55 +02:00