Change Semaphore -> ThreadBlocker
This was not a semaphore, I just named it one because I didn't know
what semaphore was. I have meant to change this sooner, but it was in
no way urgent :D
Implement SMP events. Processors can now be sent SMP events through
IPIs. SMP events can be sent either to a single processor or broadcasted
to every processor.
PageTable::{map_page,map_range,unmap_page,unmap_range}() now send SMP
event to invalidate TLB caches for the changed pages.
Scheduler no longer uses a global run queue. Each processor has its own
scheduler that keeps track of the load on the processor. Once every
second schedulers do load balancing. Schedulers have no access to other
processors' schedulers, they just see approximate loads. If scheduler
decides that it has too much load, it will send a thread to another
processor through a SMP event.
Schedulers are currently run using the timer interrupt on BSB. This
should be not the case, and each processor should use its LAPIC timer
for interrupts. There is no reason to broadcast SMP event to all
processors when BSB gets timer interrupt.
Old scheduler only achieved 20% idle load on qemu. That was probably a
very inefficient implementation. This new scheduler seems to average
around 1% idle load. This is much closer to what I would expect. On my
own laptop idle load seems to be only around 0.5% on each processor.
realpath is implemented as a syscall. This is not really required but it
was the easiest way to get it working as there is already path
canonicalization at kernel level.
These are pretty much dummy functions in the kernel side. Only case that
is handled is SOL_SOCKET with SO_ERROR. This is hard coded to return no
error. Network stack is currently synchronous, so all errors are already
reported through synchronous network functions.
These can allocate memory that can be shared between processes using
a global key. There is currenly no safety checks meaning anyone can
map any shared memory object just by trying to map every possible key.
This fixes a bug where userspace provided address is not fully mapped
and the kernel tries to read/write it while using PageTable fast page.
In the future userspace input should be copied on syscall entry, so
userspace could not modify the input during syscall. Currently there
is change that userspace input passes kernel syscall validation and
after that userspace could modify the input before the value is
actually used.
This patch implements basic support for power button using ACPI
fixed events. I still need to implement general purpose events
and embedded controller for full power button support.
Current context saving was very hacky and dependant on compiler
behaviour that was not consistent. Now we always use iret for
context saving. This makes everything more clean.
Scheduler now has its own data SchedulerQueue which holds active nad
blocking thread lists. This removes need for BAN/Errors.h and making
current thread separate element instead of iterator into linked list.
This makes it possible to have current_thread on each processor
instead of a global one in Scheduler.
This allows us to allocate processor stacks, and other per processor
structures dynamically in runtime. Giving processor stack to
ap_trampoline feels super hacky, but it works for now.
This code has very ugly file parsing code. I have to create API
for reading files line by line in kernel space...
This allows users to open framebuffer/input files without root.
Mounting has to be moved to userspace soon. It makes no sense to
hard code permissions for every (device) file.
Signal handling code was way too complex. Now everything is
simplified and there is no need for ThreadBlockers.
Only complication that this patch includes is that blocking syscalls
have to manually be made interruptable by signal. There might be some
clever solution to combat this is make this happen automatically.
Before reserving address space in SYS_EXEC verify that ELF address
space is actually loadable. For example when trying to execute the
kernel binary in userspace, binarys address space would overlap with
current kernel address space. Now kernel won't crash anymore and
will just send SIGKILL to the process calling exec*().
Every inode holds a weak pointer to shared file data. This contains
physical addresses of pages for inode file data. Physical addresses
are allocated and read on demand.
When last shared mapping is unmapped. The inodes shared data is freed
and written to the inode.
MemoryBackedRegion now inherits from this and is used for private
anonymous mappigs. This will make shared mappings and file backed
mappings much easier to implement.
All executable files are now read from disk and paged on demand.
This was a big rewrite of the old ELF library but in the end
everything seems much cleaner, since all the old functionality was
not actually needed for execution.
I have to do some measurements, but I feel like memory usage dropped
quite a bit after this change.
Userspace programs can call tty_ctrl() to disable/enable tty from
handling input and displaying output.
This API is probably going to change in the future to ioctl calls
but I'm not sure how ioctl is used and what functionality should it
have. I decided to create whole new function and syscall for now.
Next I will expose framebuffer in /dev/fb0 and then I can start work
on graphical environment! :D
We now validate pointers passed by the user, to forbid arbitary
memory read/write. Now the user is only allowed to pass in pointers
in their own mapped memory space (or null).
I could delete the whole FixedWidthAllocator as it was now obsolete.
GeneralAllocator is still used by kmalloc. Kmalloc cannot actually
use it since, GeneralAllocator depends on SpinLock and kmalloc runs
without interrupts.
I have no idea why I had made PATH environment variable parsing
to be part of the kernel. Now the shell does the parsing and
environment syscall is no longer needed.
Bananymous
Bananymous