Commits
- Commit:
6d3224a4064b7c0db0800d004d6e15f50bd8fb8b- From:
- lex0de <lex0de@gmail.com>
- Date:
Improve Lenix stability and performance
- Commit:
e7410e29eb092d28a616a7e1bf805d39c154d987- From:
- lex0de <lex0de@gmail.com>
- Date:
disk fixes
- Commit:
881c6377d4b550f742935f30340d6a2fc92426f0- From:
- lex0de <lex0de@gmail.com>
- Date:
Fix duplicate shell input delivery
- Commit:
11244300a5210268239c9b0962a5886da0a15d0a- From:
- lex0de <lex0de@gmail.com>
- Date:
Fix x86 boot and toolchain setup
- Commit:
bb549fa703c9691eb1437b83266b9853a7c4c1f2- From:
- int16h <int16h@gmail.com>
- Date:
Documenting yield / perf counters; fixing line.c unused var
- Commit:
a91825ce61ca2e31c3990fcc0084552ebd8f23f4- From:
- int16h <int16h@gmail.com>
- Date:
Shitstorm ending: System now boots to shell with console/tty/pty working (intermittent ramdiskd or related disk I/O issues remain); yield storm tamed. TODO: Work more on ramdiskd/IPC
- Commit:
59b10ca5186d691ee2071af0a3aef145ee195dc2- From:
- int16h <int16h@gmail.com>
- Date:
Tame IPC wait spins - stable
- Commit:
1419dd05375a71fc5791606d438ee109402a909a- From:
- int16h <int16h@gmail.com>
- Date:
Diag: added `yields` to console menu to display process yields/blocks; instrumented further and made service-to-service IPC waits block instead of busy yield
- Commit:
79729c5d4ed16722fe7477cfc4b40df21923ddfa- From:
- int16h <int16h@gmail.com>
- Date:
IPC: Implemented mailbox blocking (instead of polling) to eliminate idle polling and try to reduce context switches; made service-to-service request tokens globally unique so responses can't be delivered to the wrong client.
- Commit:
b5a77f73fc3eda6264c0524368eec0c870e25e86- From:
- int16h <int16h@gmail.com>
- Date:
Sched: reverted pause loop that was blocking input. WIP: polling-based IPC issues with sched_context_switches
- Commit:
836647ae0821cc54f898ee48617f06de87d6a654- From:
- int16h <int16h@gmail.com>
- Date:
Sched: Update PERF_COUNTER_SCHED_TICKS with sched_tick()
- Commit:
68d2febf4860768d02aaa134cfcac0fa3ee35508- From:
- int16h <int16h@gmail.com>
- Date:
-Last commit before process accounting-
- Commit:
b8d611242cec2588168e3fc461c196f1a722bbfb- From:
- int16h <int16h@gmail.com>
- Date:
Fix: tmpfs now skips processing when a short/empty message arrives instead of treating it as a full req; task exit now treats portal revoke as best-effort... if transfer to a child succeeds we drop the handle; if transfer fails we attempt revoke but ignore owner mismatches to avoid noisy warnings when ownership changed
- Commit:
1e18d00d2fa0e623ef0d612616fcb6d01f2a2c10- From:
- int16h <int16h@gmail.com>
- Date:
Fix: stack underflow indicated the 16KB kernel stacks were overflowing (blockd’s saved SP was ~45KB below the top). I’ve increased the per-task kernel stack size to 64KB by bumping SCHED_STACK_PAGES from 4 to 16
- Commit:
9f442e154ffb1bf9c7bf317327fea00ee2d385e8- From:
- int16h <int16h@gmail.com>
- Date:
Debug: instrumentation for an unknown opcode error in ls. Fix: increase ext2 file handles to 256
- Commit:
e8059c1867bc82f1f97014db914b69187ce6c47b- From:
- int16h <int16h@gmail.com>
- Date:
Fix: build errors from last commit relating to debugging options (LENIX_DEBUG_EXIT)
- Commit:
56248200cb9b36d2cb2c595655895b7e03abd637- From:
- int16h <int16h@gmail.com>
- Date:
Disabled automatic mailbox quotas so newly created task mailboxes no longer enable depth limits by default (prevents ENOBUFS-induced EBADF while we debug) (kernel/sched/task.c). Quota support is still available if explicitly turned on.
- Commit:
ace583b69a3b05ca805707befbc795a4db603773- From:
- int16h <int16h@gmail.com>
- Date:
Fix: added per-task stdin spill buffer so partial reads pull from saved input instead of requeueing leftover bytes back into the mailbox - prevents dropped chars when mailbox is under pressure; introduced scheduler helpers to manage spill buffer and clear it on task takedown, keeping stdin attached to owning task
- Commit:
0b2bb3ee51023cbb01c53c7358550324c16c5d21- From:
- int16h <int16h@gmail.com>
- Date:
FIX: Kernel: add exponential backoff to portal_recv busy-wait loop; New IPC_BLOCK_BACKEND_STATUS_IDLE code; blockd: return IDLE instead of BUSY when no work pending; ramdiskd: handle IDLE status with aggressive backoff; virtio-blk: add IDLE status handling; reduced IDLE backoff from 100ms to 10ms.
- Commit:
2b4233ff2aff2093784c67e84ca4def9b306af90- From:
- int16h <int16h@gmail.com>
- Date:
Unblocking portal issue with filetest exit
- Commit:
0a0ef0d75760df941fb4e94320a4f99a1bffe1da- From:
- int16h <int16h@gmail.com>
- Date:
Block service hardening: timeouts, validation, and adaptive back-off
- Commit:
1a155fb7a2b4ca3780be755801dcde3ce94a8034- From:
- int16h <int16h@gmail.com>
- Date:
Added defensive code to write_full() to detect and break out of the infinite loop if write() returns 0 (stdio); other mods while debugging file/mem I/O
- Commit:
6b515a6f644d1aebf0f99f296c2a4e6187d8233a- From:
- int16h <int16h@gmail.com>
- Date:
Debug: Guard VFS logging behind LENIX_VFS_DEBUG
- Commit:
7c5aef6eb2cc265bf634fbc9bcfb007874634449- From:
- int16h <int16h@gmail.com>
- Date:
Fixed filetest failures (though it mostly hangs at end); minor fixes for initrd->rootfs debug logging
- Commit:
2257f59c3fcad7f544c256ad2c936495981a99cd- From:
- int16h <int16h@gmail.com>
- Date:
Fix: Added a spinlock to serialize all console writes
