Commit Graph
113 Commits
Author SHA1 Message Date
Bananymous cf2e8ffaff Kernel: Remove unnecessary custom RefPtr hashes
RefPtr now exposes its own default hash
2026-04-25 22:10:01 +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 812ae77cd7 Kernel: Make TCP sockets use the new ring buffer
Also fix race condition that sometimes prevented window updates not
being sent after zero window effectively hanging the whole socket
2026-02-28 14:22:08 +02:00
Bananymous 9ddf19f605 Kernel: Optimize networking code
Remove buffering from network layer and rework loopback interface.
loopback now has a separate recieve thread to allow concurrent sends and
prevent deadlocks
2026-02-27 19:08:08 +02:00
Bananymous ff378e4538 Kernel: Cleanup and optimize TCP
We now only send enough data to fill other ends window, not past that.
Previous logic had a but that allowed sending too much data leading to
retransmissions.

When the target sends zero window and later updates window size,
immediately retransmit non-acknowledged bytes.

Don't validate packets through listeing socket twice. The actual socket
will already verify the checksum so the listening socket does not have
to.
2026-02-24 16:20:23 +02:00
Bananymous 2ea0a24795 Kernel: Fix TCP SYN option propagation
Listening socket now forwards TCP options to the newly created socket
2026-02-23 23:00:47 +02:00
Bananymous acf28d8170 Kernel: Use ring buffers for TCP windows
This speeds up TCP networkign a ton as it doesnt have to do unnecessary
memmoves for each send/receive
2026-02-23 21:10:13 +02:00
Bananymous 666a7bb826 Kernel: Rework TCP window size reporting
We now report actually available window size when sending packets. If
the available window size grows significantly we send an ACK to reflect
this to the remote.
2026-02-23 21:10:13 +02:00
Bananymous 1ac20251cf Kernel: Fix TCP stack crash on retransmission 2026-02-23 17:48:16 +02:00
Bananymous 9445332499 Kernel: Remove unnecessary interface lookup
This prevented connecting to local sockets listening on INADDR_ANY
2026-02-23 16:06:48 +02:00
Bananymous 8edd63d115 Kernel: Cleanup {set,get}sockopt debug prints 2026-02-23 16:06:48 +02:00
Bananymous a0211d88e7 Kernel: Don't include TCP header in MSS 2026-02-08 19:44:30 +02:00
Bananymous e216fc7798 Kernel: Fix port allocation endianness 2026-02-08 19:43:08 +02:00
Bananymous c648ea12f2 Kernel: Cleanup and fix UNIX sockets
EPOLLOUT is now sent to the correct socket and buffer is now a ring
buffer to avoid unnecessary memmove on every packet
2026-02-08 19:38:28 +02:00
Bananymous 2e59373a1e Kernel: Fix non blocking sockets blocking :D 2026-02-08 19:33:28 +02:00
Bananymous a51a81b6cd Kernel: Move {set,get}sockopt to sockets
Sockets can now actually implement socket options :D
2026-02-08 19:27:16 +02:00
Bananymous ba6229b92d Kernel: Fix TCP accept bind address
I was accidentally binding the new socket to the target address instead
of the listening socket's address
2026-01-24 00:33:05 +02:00
Bananymous f06e5d33e7 Kernel: Rework socket binding to an address
Sockets are no longer bound to an interface, but an ipv4 address. This
allows servers at 0.0.0.0 talk to multiple different interfaces
2025-12-31 19:28:55 +02:00
Bananymous d60f12d3b8 Kernel: Add support for SCM_CREDENTIALS and fix recvmsg
recvmsg was broken when receiving into more than a single iovec
2025-11-18 05:40:36 +02:00
Bananymous dd636ffcb2 Kernel: Add support for SA_SIGINFO 2025-11-17 05:26:07 +02:00
Bananymous c700d9f714 Kernel: Implement connect for UDP socket 2025-11-17 05:26:07 +02:00
Bananymous 59cfc339b0 Kernel: Ignore MSG_NOSIGNAL and invalid flags 2025-11-17 05:26:07 +02:00
Bananymous bb86520094 Kernel: Set message flags in UDP and TCP recvmsg 2025-11-13 04:20:53 +02:00
Bananymous 7b580b8f56 Kernel: Implement fd passing with SCM_RIGTHS 2025-11-12 00:06:36 +02:00
Bananymous 04d24bce70 Kernel/LibC: Implement {recv,send}msg as syscalls
This also removes the now old recvfrom and sendto syscalls. These are
now implemented as wrappers around recvmsg and sendmsg.

Also replace unnecessary spinlocks from unix socket with mutexes
2025-11-10 01:40:33 +02:00
Bananymous 0066b20413 Kernel: Fix spinlock leaks with unix sockets 2025-08-19 16:23:30 +03:00
Bananymous ef6ee78fd1 Kernel/LibC: Implement chroot 2025-08-11 14:07:37 +03:00
Bananymous 5b587d199e Kernel/LibC: Implement FIONREAD for tcp and udp sockets 2025-08-10 19:57:31 +03:00
Bananymous 95cfac471a Kernel: Rename loopback adapter lo0 -> lo 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 8a0269d29e Kernel: Remove kernel processes
Kernel can just use raw threads, pretty muchs the only thing that
process provides is syscalls which kernel threads of course don't
need.

Also this makes init process have pid 1 :D
2025-07-02 01:54:03 +03:00
Bananymous 4695fa061d Kernel: Don't panic if UNIX socket is already bound but deleted 2025-06-28 16:55:13 +03:00
Bananymous 125f8b591d Kernel: Don't crash if socket tries to reconnect
:D
2025-06-11 01:54:41 +03:00
Bananymous e92f039a17 Kernel: Fix tcp sending with already sent unacknowledged bytes 2025-06-06 11:55:44 +03:00
Bananymous 66726090ec Kenrel: Fix TCP connection closing
If TCP socket was connected with connect() instead of accept() it would
never send FIN to other end when it was closed.
2025-06-06 11:10:29 +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 96d5ed9cc7 Kernel: Fix E1000 driver interrupts
I have no idea why, but E1000 seems to be missing interrupts frequently
when clearing ICR only at the end of the interrupt handler
2025-06-06 03:45:35 +03:00
Bananymous 8812704601 Kernel: Make sure IPv4 and ARP packets are not missed
This is a really hacky solution but will do until i get atomic
unblocking with respect to locks
2025-06-05 22:03:15 +03:00
Bananymous 0f189d410e Kernel: Fix unix socket recv from
If connection on unix socket was closed and other end tries to recvfrom,
the thread would enter a fucked up state where it held the socket's
spinlock when returning to userspace.
2025-06-05 22:02:53 +03:00
Bananymous bf41b448d6 Kernel: Fix TCP sending
- Fix race condition when adding packet to send buffer before other end
  has acknowledged it
- Allow sending multiple packets before receiving ACK for previous ones
2025-06-04 22:21:35 +03:00
Bananymous 943069b2a3 Kernel: Socket EPIPE and send SIGPIPE on write after hang up 2025-06-03 10:24:44 +03:00
Bananymous 692ba43182 Kernel: Fix spinlock bugs found by the new spinlock security 2025-06-01 13:48:03 +03:00
Bananymous 89c9bfd052 Kernel/LibC: Implement socketpair for UNIX sockets 2025-05-28 03:10:01 +03:00
Bananymous 12b93567f7 Kernel/LibC: Implement getpeername 2025-05-28 03:10:01 +03:00
Bananymous 2f37776614 Kernel: Notify epoll on new TCP connections 2025-05-28 03:10:01 +03:00
Bananymous 7f04b2c96c Kernel: Fix E1000 interrupt handling
E1000 does not support MSI-X and thus does not generate RxQ0 interrupts.
2025-05-28 03:10:01 +03:00
Bananymous 0e0d7016b3 Kernel: Rename has_hangup -> has_hungup 2025-05-17 12:39:23 +03:00
Bananymous 1bcd1edbf5 Kernel/LibC: Implement basic epoll
This implementation is on top of inodes instead of fds as linux does it.
If I start finding ports/software that relies on epoll allowing
duplicate inodes, I will do what linux does.

I'm probably missing multiple epoll_notify's which may cause hangs but
the system seems to work fine :dd:
2025-05-13 10:18:05 +03:00
Bananymous 8e0a56b49a Kernel: Implement loopback network interface 2025-05-13 10:16:21 +03:00
Bananymous b853d29992 Kernel: Fix unix domain socket close detection 2025-04-22 08:36:44 +03:00