Commit graph

638 commits

Author SHA1 Message Date
Andreas Kling
a2ec09bc20 Allow the scheduler to unblock the current process.
It's a bit confusing that the "current" process is not actually running
while we're inside the scheduler. Perhaps the scheduler should redirect
"current" to its own dummy Process. I'm not sure.

Regardless, this patch improves responsiveness by allowing the scheduler
to unblock a process right after it calls select() in case it already has
a pending wakeup request.
2019-01-16 17:47:18 +01:00
Andreas Kling
4fef895eda Rework WindowServer to use select() in its main event loop.
The system can finally idle without burning CPU. :^)

There are some issues with scheduling making the mouse cursor sloppy
and unresponsive that need to be dealt with.
2019-01-16 17:20:58 +01:00
Andreas Kling
f7ca6d254d Tear out or duplicate what's unique for WindowServer from Widgets.
This turned into a huge refactoring that somehow also includes
making locks recursive/reentrant.
2019-01-16 16:03:50 +01:00
Andreas Kling
e655aebd70 Window: Update coding style. 2019-01-16 13:50:24 +01:00
Andreas Kling
9dd29f9aa9 Add a PTY multiplexer (/dev/ptmx) device.
When you open /dev/ptmx, you get a file descriptor pointing to one of the
available MasterPTY's. If none are available, you get an EBUSY.

This makes it possible to open multiple (up to 4) Terminals. :^)

To support this, I also added a CharacterDevice::open() that gets control
when VFS is opening a CharacterDevice. This is useful when we want to return
a custom FileDescriptor like we do here.
2019-01-16 13:39:32 +01:00
Andreas Kling
b46ae2bf09 Get rid of Vnode concept.
We already have an abstraction between Process and Inode/CharacterDevice/FIFO
and it's called FileDescriptor. :^)
2019-01-16 12:57:07 +01:00
Andreas Kling
310a5f4199 Let each MasterPTY create its slave. 2019-01-16 02:11:50 +01:00
Andreas Kling
9c51d9dfcd Catch anyone trying to use a DoubleBuffer with interrupts disabled. 2019-01-16 01:58:11 +01:00
Andreas Kling
09ba129bcf Use a CircularQueue for the PS/2 mouse driver. 2019-01-16 01:52:39 +01:00
Andreas Kling
8ad2dfb6e1 Rename FileDescriptor::has_data_available_for_reading() -> can_read(). 2019-01-16 00:47:00 +01:00
Andreas Kling
a8baee4dcd Tidy up memory map a bit and write out the general map in MemoryManager.
There was a bug that given enough supervisor page allocation, we would
eventually start dipping into the kmalloc range.
2019-01-16 00:44:09 +01:00
Andreas Kling
bd3e77cc16 Pass the process to CharacterDevice::read/write.
This is much nicer than grabbing directly at 'current' inside a read().
2019-01-16 00:20:38 +01:00
Andreas Kling
08bfe518f9 Rename CharacterDevice::has_data_available_for_reading() -> can_read(). 2019-01-16 00:10:13 +01:00
Andreas Kling
10387beda7 Implement basic support for POSIX-style select().
Now we can block on both the PTY *and* the GUI event stream in Terminal.
2019-01-16 00:09:58 +01:00
Andreas Kling
46181cf023 Oops, finish WindowServer rename. 2019-01-15 21:46:15 +01:00
Andreas Kling
52c004eb53 Add internal locking to DoubleBuffer. 2019-01-15 21:43:38 +01:00
Andreas Kling
e42f090ed3 Rename WindowComposer -> WindowServer.
I keep referring to it as the windowing server anyway.
2019-01-15 21:16:04 +01:00
Andreas Kling
d0137f0e96 Terminal: optimize repaints a bunch.
We track dirty character cells + pending whole-terminal scrolls.
This drastically reduces the number of pixels pushed.
2019-01-15 10:23:15 +01:00
Andreas Kling
e452303c66 Allow character devices to block write attempts until there is more space. 2019-01-15 09:17:22 +01:00
Andreas Kling
49b63281a0 Make it possible for a process to switch controlling terminals.
Via the TIOCSCTTY and TIOCNOTTY ioctls.
2019-01-15 08:49:24 +01:00
Andreas Kling
10308c6434 Let's not require TERM=ansi for tgetent() to work. 2019-01-15 08:32:53 +01:00
Andreas Kling
6f61eb4a87 Slap an InterruptDisabler on gui$invalidate_window().
This is obviously not a permanent solution but it works now to allow
the windowing system to withstand invalidation spam.
2019-01-15 08:23:13 +01:00
Andreas Kling
14712ad9c5 Let's do dword-at-a-time memcpy() and memset() in userspace as well.
Also fix a dumb bug that showed up when I was memsetting something other
than zeroes.
2019-01-15 08:14:44 +01:00
Andreas Kling
a8e42bacf4 Minor Terminal tweaks. 2019-01-15 07:57:30 +01:00
Andreas Kling
d14ec951ea Terminal: Add some inset and line spacing.
This is starting to feel vaguely usable! :^)
2019-01-15 07:39:51 +01:00
Andreas Kling
c24f543a57 Terminal: basic ANSI color support. 2019-01-15 07:30:24 +01:00
Andreas Kling
78696236d3 Add very basic KeyDown events to the GUI event stream.
The Terminal program now hosts an interactive shell. :^)
2019-01-15 06:51:00 +01:00
Andreas Kling
c0ef060a7c Let's just assume we have 32MB of physical memory to work with.
I should eventually figure out the exact amount of memory but not now.
2019-01-15 06:48:36 +01:00
Andreas Kling
2f74c2f430 Add basic PTY support.
For now, there are four hard-coded PTYs: /dev/pt{m,s}[0123]
Use this in the Terminal to open a pty pair and spawn a shell.
2019-01-15 06:30:19 +01:00
Andreas Kling
ecb4ab0943 Add Terminal/.gitignore 2019-01-15 04:49:01 +01:00
Andreas Kling
dfff2996d2 Factor out individual glyph drawing into Painter::draw_glyph(). 2019-01-15 04:44:47 +01:00
Andreas Kling
6d8043767e Start working on a graphical Terminal program. 2019-01-15 04:30:55 +01:00
Andreas Kling
b673c1a77d Build Painter & friends into LibC. Use it in the GUI test app. 2019-01-14 20:02:51 +01:00
Andreas Kling
bfef4afa6a So long SDL stuff. You were a nice bootstrapping environment. 2019-01-14 15:30:37 +01:00
Andreas Kling
0c5ecd303c Share GraphicsBitmaps between the windowing server and the client process.
This is pretty cool. :^)

GraphicsBitmaps are now mapped into both the server and the client address
space (usually at different addresses but that doesn't matter.)

Added a GUI syscall for getting a window's backing store, and another one
for invalidating a window so that the server redraws it.
2019-01-14 15:25:34 +01:00
Andreas Kling
b0e3f73375 Start refactoring the windowing system to use an event loop.
Userspace programs can now open /dev/gui_events and read a stream of GUI_Event
structs one at a time.

I was stuck on a stupid problem where we'd reenter Scheduler::yield() due to
having one of the has_data_available_for_reading() implementations using locks.
2019-01-14 14:42:49 +01:00
Andreas Kling
b4da4e8fbd Always inline the locks. 2019-01-14 14:14:30 +01:00
Andreas Kling
11331e9639 Add Vector::take_first(). 2019-01-14 02:49:30 +01:00
Andreas Kling
973ff14180 Make QtCreator stop complaining about the ASSERT macro. 2019-01-14 00:35:07 +01:00
Andreas Kling
e7b1101f62 TTY::write() should return the number of bytes written. 2019-01-14 00:19:03 +01:00
Andreas Kling
cc0be2e78a Keep back and front painters around in WindowManager.
The internal state never changes in some meaningful way for these (at this time.)
2019-01-13 07:41:31 +01:00
Andreas Kling
0ead888821 gui$create_widget() shouldn't try to make button corners opaque. 2019-01-13 07:33:10 +01:00
Andreas Kling
9053027dc0 Flush the old and new cursor rects in a single rect. 2019-01-13 07:23:07 +01:00
Andreas Kling
8e96cba7ef Make a nice bitmap cursor.
It's rendered as two CharacterBitmaps right now because I don't have a good
primitive for this and I wanted it right away. I should add 2-color bitmaps..

Also make a neat little effect where the cursor becomes inverted on press.
2019-01-13 07:17:12 +01:00
Andreas Kling
1d914cbd84 Minor GUI API things + make Button corners properly transparent. 2019-01-13 06:26:20 +01:00
Andreas Kling
f7261d7b26 Let's use the existing Rect and Color types in the GUI API.
Any type that doesn't depend on indirect data can probably be used here.
2019-01-13 05:31:07 +01:00
Andreas Kling
b2d86b7597 Add basic GUI API for creating labels and buttons. 2019-01-13 05:03:17 +01:00
Andreas Kling
17c7bf01a5 Fix Userland build. 2019-01-13 04:31:16 +01:00
Andreas Kling
8f8c8d1ca3 Start working on a GUI kernel API. 2019-01-13 02:02:34 +01:00
Andreas Kling
becc2c7fa5 Make GraphicsBitmaps be Region-backed when running in the kernel.
This is a lot better than having them in kmalloc memory. I'm gonna need
a way to keep track of which process owns which bitmap eventually,
maybe through some sort of resource keying system. We'll see.
2019-01-13 00:29:32 +01:00