Commit Graph
137 Commits
Author SHA1 Message Date
Bananymous 89d3ee7b92 Kernel: Use per cpu fast page when clearing SMO pages 2026-07-04 19:18:31 +03:00
Bananymous 954898b14d Kernel: Fix file backed mmap syncing, my if condition was inverted 2026-05-25 02:14:04 +03:00
Bananymous 32206069bc Kernel: Use per cpu fast page for {Memory,File}BackedRegion 2026-05-20 02:51:48 +03:00
Bananymous 46a1903f8d Kernel: Use per cpu fast page for PMM 2026-05-20 02:46:50 +03:00
Bananymous d7865b2929 Kernel: Don't msync file backed pages that were never mapped writable 2026-05-17 03:21:08 +03:00
Bananymous dd8a9b1793 Kernel: clamp msync address range instead of calling contains 2026-05-13 20:03:37 +03:00
Bananymous 9f4271f6d8 Kernel: Remove the big inode lock
This moves locking to the inodes themselves which allows reducing lock
times significantly. Main inodes (ext2 and tmpfs) still do contain a
single big mutex that gets locked during operations but now we have the
architecture to optimize these.
2026-05-09 23:28:00 +03:00
Bananymous a7356716ff Kernel: Reduce locking FileBackedRegions 2026-05-09 15:08:26 +03:00
Bananymous 3ac955714b Kernel: Optimize FileBackedRegion CoW
Just copy data from the existing page instead of locking inode's shared
data, copying to temporary buffer and finally copying to destination
2026-05-06 17:30:43 +03:00
Bananymous 62f5292f38 Kernel: Send only one TLB invalidation when creating a ByteRingBuffer 2026-05-06 17:30:43 +03:00
Bananymous 7553ede3b4 Kernel: Send only one TLB invalidation when creating a VirtualRange 2026-05-06 00:27:26 +03:00
Bananymous 93e1091252 Kernel: Rewrite kmalloc
Kmalloc is now a bitmap allocator with dynamic resizing and we dont need
to allocate 64 MiB static block of memory reserved for kmalloc :^)
2026-05-04 20:26:02 +03:00
Bananymous f4e2e62d04 Kernel: Remove kmalloc API for identity mapped results
This is no longer used. This finally allows me to rewrite kmalloc :^)
2026-05-04 20:26:02 +03:00
Bananymous f37d9dbdb1 Kernel: Remove kmalloc_vaddr_of
This is no longer needed. It was only used for x86_64 paging and AP
stack initialization
2026-05-04 20:26:01 +03:00
Bananymous ab8bcbec3e Kernel: Allow mapping dma regions as not uncached 2026-04-27 19:36:32 +03:00
Bananymous b74812d669 Kernel: Remove unnused features from VirtualRange
On-demand paging has not been used ever since I made userspace stack be
a normal MemoryRegion.
2026-04-21 19:58:09 +03:00
Bananymous 5b7b2d7ac3 Kernel: Fix memory leak when cleaning up inodes shared page cache 2026-04-19 17:52:21 +03:00
Bananymous 7e907b70f6 Kernel: Store memory region size as uint64_t
On 32 bit target, we were storing 32 bit physical region sizes which
would truncate regions > 4 GiB
2026-04-07 03:41:25 +03:00
Bananymous e4c6539964 Kernel: Be more clever with physical memory
Initially allocate all physical memory except kernel memory and boot
modules. Before we just skipped all memory before kernel boot modules.
Also release memory used by boot modules after the kernel is up and
running. Once the boot modules are loaded, there is no need to keep them
in memory.
2026-04-06 19:29:34 +03:00
Bananymous 9fce114e8e Kernel: Don't clone entire kernel stack on fork
We only need to copy area between [ret_sp, stack_end]. This range is
always very small compared to the whole stack (64 KiB).
2026-04-04 23:48:43 +03:00
Bananymous 47d85eb281 Kernel: Pass the actual vaddr range to reserve pages 2026-04-04 23:48:43 +03:00
Bananymous 675c215e6a Kernel: Add CoW support to MemoryBackedRegion
This speeds up fork by A LOT. Forking WindowServer took ~90 ms before
this and now its ~5 ms.
2026-04-03 01:54:59 +03:00
Bananymous 493b5cb9b1 Kernel: Implement byte ring buffer
This maps the ring twice right next to each other so we don't have to
care about wrapping around when doing memcpy or accessing the data
2026-02-28 14:18:23 +02:00
Bananymous aef536fff3 Kernel: Fix SharedMemoryObject cloning on deleted keys 2026-01-25 01:42:17 +02:00
Bananymous 1143dc3cae Kernel: Rework syscall memory validation and locking
Process's memory regions are now behind an rwlock instead of using the
full process lock. This allows most pointer validations to not block as
write operations to memory regions are rare.

Thread's userspace stack is now part of process's memory regions. This
simplifies code that explicitly looped over threads to see if the
accessed address was inside a thread's stack.

Only drawback of this is that MemoryRegions don't support guard pages,
so userspace stackoverflow will be handeled as cleanly as it was prior
to this.

This patch also fixes some unnecessary locking of the process lock and
moves locking to the internal helper functions instead of asserting that
the lock is held. Also we now make sure loaded ELF regions are in sorted
order as we previously expected.
2026-01-16 16:09:38 +02:00
Bananymous 4b5a8196c3 Kernel: Fix bitmap byte index calculation in PMM
No idea how I had not crashed here earlier, but running on real hw with
a bit initrd ended up crashing :D
2026-01-04 01:16:43 +02:00
Bananymous 05d59a05df Kernel: Remove obsolete kprint
This hasn't been in use in 3 years :)
2025-11-22 06:21:50 +02:00
Bananymous 9537922acc Kernel: Implement proper memory region splitting
Memory regions are now splitted when they get munmapped, mprotected, or
mmapped with MAP_FIXED. This is used by couple of ports, and without
this we were just leaking up memory or straight up crashing programs.
2025-11-13 04:20:53 +02:00
Bananymous f1d12c330e Kernel/LibC: Implement MMAP_FIXED_NOREPLACE
This is a handy thing from linux

Also fix MMAP_FIXED validation and error reporting
2025-11-12 00:06:36 +02:00
Bananymous 706cfeb443 Kernel: Allow file backed mapping be larger than inode size
This is only allowed if the mapping does **not** exceed a page boundary.
Some port was doing an exactly two-page-mapping on a file that was one
and a half page long
2025-08-28 15:55:40 +03:00
Bananymous 804cbeb1a7 Kernel: Increment kmalloc storage size to 64 MiB
I really don't want to do this, but rewriting kmalloc to be dynamic
would require me to rewrite 32 bit paging and I really don't want to.
2025-08-15 17:02:15 +03:00
Bananymous 72f85dce2b Kernel: Make userspace stack on-demand allocated
Also bump the hardlimit of stack size from 512 KiB->32 MiB. This still
feels quite low but is much better than before :D
2025-08-07 16:43:05 +03:00
Bananymous f1369c8fd6 Kernel/LibC: Implement mprotect
There may be some race conditions with this but i think this is good
enough to start with
2025-08-05 03:09:24 +03:00
Bananymous 6084aae603 Kernel: Add guard pages to kernel and userspace stacks 2025-07-02 23:12:36 +03:00
Bananymous 17f1ac10e3 Kernel: Don't ignore modules passed with multiboot2 2025-07-02 00:17:42 +03:00
Bananymous eecdad50a6 Kernel: Fix most of mutex + block race conditions
All block functions now take an optional mutex parameter that is
atomically unlocked instead of having the user unlock it before hand.
This prevents a ton of race conditions everywhere in the code!
2025-06-06 03:59:22 +03:00
Bananymous 60d1c26114 Kernel: Fix race condition in FileBackedRegion creation 2025-06-01 13:48:03 +03:00
Bananymous d54489bbcb Kernel: Implement MemoryRegion pinning
This allows process to pin a MemoryRegion into users memory space for
syscall duration without allowing user to munmap the region.
2025-04-22 04:30:07 +03:00
Bananymous 76bad31dd5 Kernel: Reorder locking with VirtualRange 2025-04-01 23:09:30 +03:00
Bananymous e6a2f55a59 Kernel: Closing file backed mmap wont expand the file size 2024-12-03 16:12:26 +02:00
Bananymous 64c52012df Kernel: Fix MemoryRegion overlap comparison
This made some calls to `munmap` unmap extra regions.
2024-11-13 17:26:42 +02:00
Bananymous f0b18da881 Kernel: Add kmalloc helper APIs 2024-10-14 11:39:04 +03:00
Bananymous 2b43569927 Kernel: Use enums in boot info instead of magic values 2024-10-14 11:36:51 +03:00
Bananymous 55fbd09e45 Kernel: Rewrite physical memory allocation with PageTable::fast_pages 2024-10-14 11:32:54 +03:00
Bananymous 54732edff4 Kernel: Fix bugs in memory regions
MemoryRegion::initialize() required size to be page aligned, this is not
necessary.

MemoryBackedRegion::copy_data_to_region() required user-write
permissions for underlying data. This did not matter as memory backed
regions dont support COW memory, but it could lead to bugs later on.
2024-09-15 23:18:05 +03:00
Bananymous c77ad5fb34 Kernel: Implement copy-on-write memory for file backed mmaps 2024-09-11 19:33:50 +03:00
Bananymous de35cec2e1 Kernel: Allow private file mappings past file end
This is just to make memory mapping ELF files easier :D
2024-09-05 14:48:42 +03:00
Bananymous 2f241e1b61 Kernel: Implement FileBackedRegion::clone 2024-09-05 13:59:09 +03:00
Bananymous 1138ec77ca Kernel: Fix MemoryBackedRegion cloning for non page aligned size 2024-09-04 21:14:26 +03:00
Bananymous a134d16070 Kernel: Fix MemoryRegion test for contains_fully 2024-08-30 15:33:58 +03:00