Rev 40 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 40 | pj | 1 | - The vm86 reflection code contains a dirty hack: the vm86_exc1 function |
| 2 | in vm86-exc.c hardcodes the int to be reflected... |
||
| 3 | This should be fixed in some way!!! |
||
| 4 | FIXED (and seems to be working very well...) |
||
| 5 | |||
| 6 | - Now, the registers structure should be fixed: move the flags field to the |
||
| 7 | other side of the structure (adding the CS and EIP fields), so that |
||
| 8 | we will not have to push flags and explicitly move it anymore... |
||
| 9 | DONE |
||
| 10 | |||
| 2 | pj | 11 | - Fix problems with ASs and INTs (see kl/cxsw-1.s and xlib/exc.s) |
| 12 | Done! At least it doesn't crash :) |
||
| 13 | Must check in some way if the stack is correct or not... |
||
| 40 | pj | 14 | I checked it, and it seems to be ok... |
| 2 | pj | 15 | |
| 16 | - verify that only the needed .o files are linked (see -DPROFILE); provide |
||
| 17 | a lightweight implementation of ``message'' (currently remapped on |
||
| 18 | cprintf) that does not require linking libm |
||
| 19 | |||
| 20 | - profile and optimize the event management code |
||
| 21 | |||
| 22 | - rewrite cons.c isolating the parts making assumptions on memory spaces and |
||
| 23 | improving efficency (window-based functions can be implemented |
||
| 24 | separately) |
||
| 25 | |||
| 26 | - pentium timestamp conter code... Integrate it in timer code and event |
||
| 27 | management |
||
| 28 | Assigned to Gerardo |
||
| 29 | |||
| 30 | - provide some macro for int-based and call gate-based system calls |
||
| 40 | pj | 31 | Done for INT: see examples/syscalls.c |
| 2 | pj | 32 | |
| 33 | - begin to implement some serious code using OSLib: |
||
| 34 | 1) some simple multithreading executive based on cooperative and |
||
| 35 | preemptive scheduling, message-based or based on shared |
||
| 36 | memory and semaphores... |
||
| 37 | 2) some real small kernel, separating it from application code |
||
| 38 | 3) as 2), but with different address spaces for kernel and |
||
| 39 | application |
||
| 40 | ... |
||
| 41 | |||
| 42 | - fix the eXtender in order to be fully multiboot compliant (no extension to |
||
| 43 | the standard multiboot structure: we can use memory maps). Upgrade it |
||
| 44 | to the latest version of the standard. Implement modules loading |
||
| 45 | == Some work has been done... Now the eXtender can load MB compiant images |
||
| 46 | and manages modules. |
||
| 40 | pj | 47 | == Now everything is compatible with the latest MB standard (the loader |
| 48 | name is used) |
||
| 2 | pj | 49 | o Still to do: the memory map stuff & module parameters |
| 40 | pj | 50 | |
| 51 | - what about fixing that _NOH4_ story? |