Rev |
Age |
Author |
Path |
Log message |
Diff |
1251 |
6765 d 6 h |
jermar |
/kernel/trunk/ |
Fix BITS2BYTES macro to return 0 when passed 0 as argument.
Fix ia32 TSS segment granularity to be 0.
Fix ia32 and amd64 initial TSS limit to be 103.
Little textual changes here and there. |
|
750 |
6854 d 1 h |
jermar |
/kernel/trunk/ |
ia32 SMP updates.
Update info about supported version of Bochs (bump the counter to 2.2.6).
Add Task Priority Register type and initialize this register to a known state. |
|
672 |
6880 d 22 h |
jermar |
/kernel/trunk/arch/ia32/ |
ia32 work.
Initialize LDR and DFR registers in Local APIC so that logical destination addresses work. |
|
576 |
6904 d 20 h |
palkovsky |
/kernel/trunk/ |
Changed ia32 & amd64 to use exc_register instead of trap_register.
Fixed dependency list building. I hope you all have 'makedepend' installed,
if you don't it's time to install it, as CC -M builds the dependency
list without directory names..and it just does not work. |
|
534 |
6910 d 1 h |
jermar |
/ |
Rename HelenOS/SPARTAN to HelenOS/kernel again. |
|
532 |
6910 d 1 h |
jermar |
/ |
Revert renaming of HelenOS/SPARTAN to HelenOS/kernel because of important pending patch. |
|
515 |
6920 d 2 h |
jermar |
/kernel/trunk/ |
SMP cleanup continued.
Add nice type for IO APIC ID register.
Replace some magic numbers with more descriptive macros. |
|
514 |
6920 d 20 h |
jermar |
/kernel/trunk/arch/ia32/ |
More SMP cleanup.
Suddenly, keyboard started to work on SMP under Simics.
Still not functional on Bochs (will consult Bochs people).
Doxygen style comments for apic.c. |
|
513 |
6921 d 3 h |
jermar |
/kernel/trunk/arch/ia32/ |
APIC code cleanup. |
|
512 |
6921 d 7 h |
jermar |
/kernel/trunk/ |
SMP work.
Add madt_irq_to_pin().
Make ksmp() use virtual irq_to_pin() function, which makes better sence for ACPI configurations. |
|
501 |
6922 d 5 h |
jermar |
/ |
Rename HelenOS/SPARTAN to HelenOS/kernel. |
|
273 |
7002 d 2 h |
palkovsky |
/SPARTAN/trunk/ |
Changes to make ia32 smp code not architecture dependent.
Changes to accomodate new gdtr loading scheme. |
|
268 |
7002 d 4 h |
palkovsky |
/SPARTAN/trunk/ |
Added symbol table lookup in exceptions.
This breaks ia64 & ppc architecture compiles. |
|
128 |
7049 d 8 h |
jermar |
/SPARTAN/trunk/arch/ia32/ |
Create IA-32 interface for bringing SMP up.
This interface hides differences between ACPI MADT and MPS configuration tables.
Provide MPS implementation thereof. |
|
111 |
7091 d 1 h |
palkovsky |
/SPARTAN/trunk/ |
Code cleanup in scheduler.c thread.c - removed unnecessary spinlock.
atomic_inc, atomic_dec moved to arch/atomic.h instead of arch/smp/atomic.h,
advisable to use even in non-smp mode.
Fixed atomic_inc, atomic_dec in mips architecture. |
|
81 |
7111 d 0 h |
jermar |
/SPARTAN/trunk/ |
Add SP_DELTA macro for each architecture.
On IA-32, this macro represents two doublewords that must be on the stack when switching to a new stack.
On MIPS, there is no need to fix sp when switching to a new stack.
On IA-64, this macro is used to allocate the 16-byte scratch area.
IA-64 work.
Changes in start.S. More changes to come.
Define memcopy().
Janitorial fixes.
Add missing #include <panic.h> to rwlock4/test.c.
Remove wrong use of inline keyword from both declaration and definition of l_apic_id(). |
|
34 |
7128 d 4 h |
jermar |
/SPARTAN/trunk/ |
More ACPI work.
Initial MADT table parsing.
SMP renaming and reorganization to reflect there are more ways to bring SMP up. |
|
27 |
7133 d 23 h |
jermar |
/SPARTAN/trunk/ |
APIC changes and fixes.
Be more robust during MP configuration.
Do not use APIC ID as CPU_ID_ARCH anymore.
Changing APIC ID's is not a good idea.
Use dr0 register instead. |
|
21 |
7142 d 4 h |
jermar |
/SPARTAN/trunk/arch/ia32/ |
Minor changes and cleanups of recent modifications. |
|
16 |
7148 d 4 h |
jermar |
/SPARTAN/trunk/ |
Switch from mm-based 'the' mechanism to macro-based 'cpu_private_data[CPU_ID_ARCH]' mechanism.
Added l_apic_id() and some other minor APIC changes.
Move gdtr to K_DATA_START section.
Move K_DATA_START section immediately behind K_TEXT_START section so that real-mode addresses work even with growing size of kernel code. |
|