Commit message (Collapse) | Author | Age | |
---|---|---|---|
* | litmus core: don't leak np_flag upon exec | Bjoern B. Brandenburg | 2008-01-24 |
| | |||
* | litmus core: don't leak real-time state across forks and mode transitions | Bjoern B. Brandenburg | 2008-01-24 |
| | | | | Always cleanup stale real-time flags, job numbers, etc. | ||
* | litmus core: remove prepare_task() system call | Bjoern B. Brandenburg | 2008-01-22 |
| | | | | The old real-time task creation has been replaced by the new transitioning method. | ||
* | [litmus core] task_mode_transition() fixes | Bjoern B. Brandenburg | 2008-01-22 |
| | | | | - restore old best-effort priority - allow non-rt tasks to register a np-flag | ||
* | [litmus core] add sys_task_mode_transition() | Bjoern B. Brandenburg | 2008-01-18 |
| | | | This allows real-time tasks to become best-effort tasks and vice versa. | ||
* | [litmus core] implement rt mode transition mechanics | Bjoern B. Brandenburg | 2008-01-17 |
| | | | This will be the basis for the new BE->RT->BE system call. | ||
* | [cleanup] remove trailing whitespace | Bjoern B. Brandenburg | 2008-01-17 |
| | |||
* | ICS: re-worked for nesting support | Bjoern B. Brandenburg | 2007-12-14 |
| | |||
* | ICS: bug fixes and debugging support | Bjoern B. Brandenburg | 2007-11-20 |
| | | | | | - don't disable preemption - don't spin inside the kernel on behalf of user space - give some debug info (can be turned off) | ||
* | ICS: implement rollback mechanics | Bjoern B. Brandenburg | 2007-11-19 |
| | | | rollback the EIP on return to user space if TIF_ROLLBACK_RCS is set | ||
* | ICS: arch independent local task rollback | Bjoern B. Brandenburg | 2007-11-19 |
| | |||
* | ICS: implement writer protocol kernel support | Bjoern B. Brandenburg | 2007-11-19 |
| | | | | Writer implementation mostly complete, reader rollback and registry stuff still missing. | ||
* | FDSO: allow user config for od_open() | Bjoern B. Brandenburg | 2007-11-19 |
| | | | This is required to support ICSs. | ||
* | ICS: start implementation of ICS support | Bjoern B. Brandenburg | 2007-11-19 |
| | | | Add platform independent files and enable in FDSO layer. | ||
* | FDSO: use convenience wrapper functions for invoking class methods | Bjoern B. Brandenburg | 2007-11-19 |
| | | | Less error prone and easier to type. | ||
* | FDSO: remove unnecessary forward declarations | Bjoern B. Brandenburg | 2007-11-19 |
| | | | This is not necessary and just adds extra dependencies. | ||
* | FDSO: Use mutex for inode object list protection | Bjoern B. Brandenburg | 2007-11-16 |
| | | | A task may sleep while holding that lock (due to kmalloc). | ||
* | FMLP: Rework semaphore allocation to work on top of FDSO | Bjoern B. Brandenburg | 2007-11-15 |
| | | | | | Throw out the terribly insecure, wasteful, and brittle static allocations. Many issues remain with this implementation (robustness, etc.) but it is a first step. | ||
* | fdso: introduce shared object ops | Bjoern B. Brandenburg | 2007-11-15 |
| | | | | Creating and destroying objects with type-specific ops will make the code more modular. Also allow easy lookup of correctly typed objects. | ||
* | fdso: introduce core infrastructure | Bjoern B. Brandenburg | 2007-11-15 |
| | | | | | | | | | This patch introduces "file descriptor attached shared objects" that will be used to manage LITMUS^RT PI semaphores, SRP resources, and ICS. FDSO essentially allows tasks to create and share kernel objects in the context (namespace) of a shared file. This is useful for mmap()-based shared objects that need synchronization primitives. One example would be the FMLP library libso. | ||
* | cleanup: tidy up code before release2007.2 | Bjoern B. Brandenburg | 2007-10-27 |
| | | | This gets rid of quite a few FIXMEs and TODOs, as well as some cruft. | ||
* | seems to be working like this | Bjoern B. Brandenburg | 2007-10-19 |
| | |||
* | litmus: Rework scheduler signals implementation | Bjoern B. Brandenburg | 2007-10-17 |
| | | | | This approach should both make the scheduler signals list more flexible and fix the locking dependency detected by lockdep. | ||
* | adaptive: Aaron's fixes, clean out debug, and merge header | Bjoern B. Brandenburg | 2007-10-17 |
| | | | | | | This patch is a collection of several changes. It includes Aaron's fixes to prevent division by zero, makes it possible to disable the optimizer debug output, and merges the header file into the scheduler implementation since it is not being used anywhere else. | ||
* | Fixed Adaptive stuff | Aaron Block | 2007-10-10 |
| | |||
* | compile fixes | Bjoern B. Brandenburg | 2007-10-09 |
| | |||
* | adaptive: only send signal when requested. | Bjoern B. Brandenburg | 2007-10-09 |
| | |||
* | litmus-core: add scheduler_signal() infrastructure | Bjoern B. Brandenburg | 2007-10-09 |
| | | | | use it to send weight change notifications | ||
* | adaptive: increase+decrease goodness | Bjoern B. Brandenburg | 2007-10-09 |
| | | | | Complete reweighting rules. | ||
* | fpmath: add _max() | Bjoern B. Brandenburg | 2007-10-09 |
| | |||
* | sched_trace: add sched_trace_weight_error() | Bjoern B. Brandenburg | 2007-10-09 |
| | | | | Export the accuracy of the error prediction to user space. | ||
* | adaptive: add a ton of debugging output | Bjoern B. Brandenburg | 2007-10-09 |
| | | | And fix some small errors. | ||
* | adaptive: work-around weird output behavior. | Bjoern B. Brandenburg | 2007-10-09 |
| | | | This doesn't do what I want it to do. Display fp_t as fraction instead. | ||
* | adaptive: get rid of weight-estimator slope misconception | Bjoern B. Brandenburg | 2007-10-09 |
| | | | | This concept is redundant with the per-service-level weights. Also fix the get_sl() macro. | ||
* | litmus-core: fix debug output | Bjoern B. Brandenburg | 2007-10-08 |
| | | | The little comma wrecked havoc in the log files... | ||
* | adaptive: debugging | Bjoern B. Brandenburg | 2007-10-08 |
| | |||
* | adaptive: check for pending weight changes on scheduler tick | Bjoern B. Brandenburg | 2007-10-08 |
| | |||
* | adaptive: optimizer compiles | Bjoern B. Brandenburg | 2007-10-08 |
| | | | | enacting weight changes is still missing | ||
* | adpative: massive work on the optimizer | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | Not yet complete. | ||
* | litmus: add qsort list manipulation function | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | The adaptive optimizer needs to sort lists. | ||
* | adaptive: use 32bit fixed point math | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | getting 64bit to work is too much of a pain right now | ||
* | adaptive: add linear weight_transfer metric | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | Adds slope and intercept to adaptive tasks and setup code. | ||
* | sched_trace: Store size of entry in header to ease parsing. | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | | This will make it possible to skip entries that are unknown or not of interest. | ||
* | adaptive: error threshold must test against absolute error | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | Negative errors must trigger reweighting, too. | ||
* | Cleanup: Remove job stats that no scheduler uses anyway. | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | No use in supporting something that isn't used. | ||
* | adaptive: introduce fixed point math | Bjoern B. Brandenburg | 2007-10-07 |
| | | | | | Introduces fixed point math header stuff and start of predictor support in sched_adaptive.c | ||
* | Add some infrastructure for reweighting. | Bjoern B. Brandenburg | 2007-10-05 |
| | | | Track error of allocation, prepare service level changes. | ||
* | Add service level changes to sched_trace. | Bjoern B. Brandenburg | 2007-10-05 |
| | | | Record when tasks get a new service level. | ||
* | adaptive: introduce service level | Bjoern B. Brandenburg | 2007-10-04 |
| | | | | This patch adds fields for service levels to the rt_param struct. | ||
* | Wrap task teardown. | Bjoern B. Brandenburg | 2007-10-04 |
| | | | | Provide a place to do generic tear-down work. |