| Rev |
Age |
Author |
Path |
Log message |
Diff |
| 4369 |
5485 d 3 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/ |
Kernel stack and userspace window buffer pointers are stored in scratchpad registers instead of memory. |
|
| 4130 |
5541 d 2 h |
rimsky |
/branches/sparc/ |
Cleanup of the Niagara port. The main purpose of these changes is to make it easy to find out which code is generic, which is sun4u-specific and which is sun4v-specific. |
|
| 4129 |
5541 d 22 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/ |
Implemented missing features in Niagara memory management, minor cleanup. |
|
| 4073 |
5544 d 2 h |
rimsky |
/branches/sparc/ |
Implemented input and output drivers for Niagara. Written in the old-fashion style (do not use the new DDI infrastructure nor the new keyboard driver). Anyway, tetris can be played. |
|
| 4068 |
5546 d 0 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/src/trap/sun4v/ |
Fixed a stupid bug. |
|
| 4065 |
5547 d 0 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/src/ |
minor cleanup |
|
| 4064 |
5547 d 0 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/src/trap/sun4v/ |
Workaround for the bug which usually caused rewriting saved register window data on the uspace stack. More investigation needed. |
|
| 4063 |
5547 d 0 h |
rimsky |
/branches/sparc/uspace/lib/libc/arch/sparc64/include/ |
Fixed the bug which caused unexpected MMU misses. The fix breaks sun4u ports - this will be solved once the sun4v is integrated with the new configuration mechanism, which allows to define common options for both kernel and uspace. |
|
| 3993 |
5553 d 23 h |
rimsky |
/branches/sparc/kernel/ |
Implemented preemptible trap handler for userspace (including syscalls - did not forget enabling interrupts for them) and the ralated stuff. Implemented handlers of instruction/data MMU miss/protection. Now some userspace tasks are run; there are, however, still some bugs causing unexpected data MMU misses. |
|
| 3864 |
5575 d 0 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/src/sun4v/ |
Allocated bigger memory block for the Machine Description. This was the problem which prevented running HelenOS on real Niagara. Now all kernel tests except for 'slab2' pass on real Niagara. |
|
| 3863 |
5575 d 23 h |
rimsky |
/branches/sparc/ |
Simics machine: modifications needed to run userspace (the next step will be to implement the trap handler for uspace); real machine: small fixes, after them the machine passes some parts of the initialization, but fails to determine CPU frequency (probably due to a different format of Machine Description). |
|
| 3862 |
5579 d 0 h |
rimsky |
/branches/sparc/kernel/ |
Changed the structure of header files, which have sun4u and sun4v versions. Implemented some sun4v TSB functions. |
|
| 3835 |
5583 d 23 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/ |
Niagara: Implemented (and debugged) installing identity mapping for kernel. Now all the kernel tests pass on Simics. |
|
| 3817 |
5591 d 23 h |
rimsky |
/branches/sparc/kernel/ |
Progress on Niagara: Commented out portions of mm, which are not required for kernel threads (temporarily). Scheduler is working. Implemented a driver of hypervisor's standard input. Kernel console is usable. |
|
| 3801 |
5596 d 1 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/src/trap/sun4v/ |
Minor fixes in the preemptible trap handler. |
|
| 3798 |
5597 d 0 h |
rimsky |
/branches/sparc/kernel/ |
Implemented preemptible trap handler for sun4v for trapping from kernel (not userspace). Not properly tested/debugged yet. HelenOS now reaches the as_install_arch function (where it, of course, fails). |
|
| 3783 |
5609 d 0 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/ |
Machine description traversal implemented. Now used to detect CPU frequency. |
|
| 3771 |
5612 d 1 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/ |
Forgotten files comitted to repository. |
|
| 3770 |
5612 d 1 h |
rimsky |
/branches/sparc/ |
Working on Niagara port - TLB initialization, MMU fault status area initialization, hypercall enhancement, modified tick.c so that the (hyperprivileged) TICK register does not have to be accessed. Now the initialization phase proceeds, but some parts of it are omitted for now (e.g. CPU initialization). |
|
| 3748 |
5619 d 2 h |
rimsky |
/branches/sparc/kernel/arch/sparc64/ |
Be careful to preserve the output registers when processing the fast data access MMU miss. |
|