| Age | Commit message (Collapse) | Author |
|
what?
- The main algorithm got improved for real time tasks, and SMP usage.
- The SMP usage was present before, I just reintroduced it after
realizing that it won't scale well from what I have right now.
- Also removed weird implementations quirks from previous sketch.
- Such as the core 0 being reserved for the boot core.
- Also moved FS init code after IDT initalization.
- To avoid weird FS format behavior.
- Wrap HPET signature in a macro.
next?
- Work on the HAL's userspace transition mechanism.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
why:
- After the image's entry is done, and no exit is called.
also:
- Added sched_idle_task to CPS's API.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
ABRT, KILL, TRAP.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
the segment
loading for the long mode stub works.
TODO:
- Stack pointer shall be set when jumping to AP routine.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
implementation.
why?
- The previous one wasn't entierly correct on some parts.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
making the LAPIC work correctly.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
other:
- Removed DmaPool into its own Kit.
- ApplicationProcessor unit has been cleaned up.
- Rename functions of MemoryMgr.
- Use KIB instead of MIBs of stack.
- Cleanup parts of the scheduler, and hw scheduler.
- Use UD handler for INT 6.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
what?
- Add simple generic RTL8139 NIC driver, to be used within a NK device.
- Update IVT accordingly.
- Comment ARM's AP GIC init function, to tell what it's actually doing.
- Cleanup Kernel Main, removed the useless pre_init_scheduler function.
- Prepare new FileMgr with HeFileSystemMgr.
- Fallback to NeFS when trying to format a fileysstem.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
numbers with 0x100, instead of 100h.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
what?
- UserProcessScheduler and CoreProcessScheduler have been
extended for FILE_TREE and also HEAP_TREE structures.
- DDK device's API will use dk_ calls instead of sk_ calls.
- SIGTRAP and Interrupt handlers have been fixed to handle when no
process is being run, and the kernel is instead raising the
interrupt.
- Add file for HeFS formating in DiskImage.fwrk
- Replace generic handler with breakpoint handler in int 3.
why?
- These changes are bug fixes and improvements.
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
compile with another profile than AHCI)
Signed-off-by: Amlal El Mahrouss <amlal@nekernel.org>
|
|
filesystem implementation, with a bootinfo.cfg being added too.
why:
- The LaTeX specs were outdated, some elements had to be reseen.
also:
- Removed useless part of the logger, for the panic message.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
has been improved with better traversal, and inode/ind allocation.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
checks inside the kernel, and cli.
also:
- make use _NeMain for the mk_fwrk tool.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
|
|
why?
- Some parts were causing issues on the filesystem.
- The slot probe code was naive.
- Made the current MBCI implementation clearer.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
why?
- This will cause math exceptions being raised by the CPU if it happens to be zero.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
implementation.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
iterations.
other:
- In16 -> rt_in16 in BootKit.h
why?
- Silent corruption with other disk backends.
- Better off starting with new work based on this backend but with bug fixes.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
regarding memory leaks.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
why?
To keep track of it, avoid security issues in the future.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
need it.
details:
- Reworked HeFS traversal algorithm, fixing last parts of HeFS regarding IND allocation.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
memory footprint (unstable changes)
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
writing an IND. (HeFS)
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
other/related:
- Add new KPC codes.
- Final refactors for HeFS's Format method.
- Dma pool improvements.
- Better standard disk I/O names.
- Add mm_memory_fence function inside HalPagingMgrAMD64.cc
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
why?
the struct were way too big to fit wihin a sector.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
what?
- AHCI now writes to disk, forgot to do it.
- Codebase's architecutre has been used to reuse the Generic+AHCI driver in SysChk for AHCI. (tradeoff is 256K in size instead of 36K)
- DriveMgr now detects EPM.
- And HeFS is still being worked on.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
allocation tree fix.
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|
|
Signed-off-by: Amlal <amlal@nekernel.org>
|