Change Semaphore -> ThreadBlocker
This was not a semaphore, I just named it one because I didn't know
what semaphore was. I have meant to change this sooner, but it was in
no way urgent :D
Implement SMP events. Processors can now be sent SMP events through
IPIs. SMP events can be sent either to a single processor or broadcasted
to every processor.
PageTable::{map_page,map_range,unmap_page,unmap_range}() now send SMP
event to invalidate TLB caches for the changed pages.
Scheduler no longer uses a global run queue. Each processor has its own
scheduler that keeps track of the load on the processor. Once every
second schedulers do load balancing. Schedulers have no access to other
processors' schedulers, they just see approximate loads. If scheduler
decides that it has too much load, it will send a thread to another
processor through a SMP event.
Schedulers are currently run using the timer interrupt on BSB. This
should be not the case, and each processor should use its LAPIC timer
for interrupts. There is no reason to broadcast SMP event to all
processors when BSB gets timer interrupt.
Old scheduler only achieved 20% idle load on qemu. That was probably a
very inefficient implementation. This new scheduler seems to average
around 1% idle load. This is much closer to what I would expect. On my
own laptop idle load seems to be only around 0.5% on each processor.
Old StorageDevice::read_sectors() read each sector separately if the
underlying disk had a disk cache. This patch allows multiple sectors to
be read even if the disk cache exists and contains some of the sectors.
Only sectors that could not be found from the disk cache are actually
read from the disk. This optimization is not done for writing, which
still will write each sector separately, if disk cache has no memory to
store new sectors. It would feel kind of unnecessary optimization as you
have greater problems if disk cache cannot allocate a single page.
Current context saving was very hacky and dependant on compiler
behaviour that was not consistent. Now we always use iret for
context saving. This makes everything more clean.
There is techically a race condition on thread sleep and checking
done mask. This patch allows read to success even if this race
condition is hit, although the full timeout has to be waited.
This can be fixed in future with some sort of wait queues that
can properly handle this race condition.
Signal handling code was way too complex. Now everything is
simplified and there is no need for ThreadBlockers.
Only complication that this patch includes is that blocking syscalls
have to manually be made interruptable by signal. There might be some
clever solution to combat this is make this happen automatically.
Add "fast page" to KERNEL_OFFSET. This is always present in page
tables and only requires changing the page table entry to map. This
requires no interrupts since it should only be for very operations
like memcpy.
I used to map all temporary mappings to vaddr 0, but this is much
better. C++ standard always says that nullptr access is undefined
and this gets rid of it.
Fixed some bugs I found along the way
ATADevice now stores its name instead of using static buffer. Old
static buffer was changing on every name query. I just hadn't noticed
since virtual machine disks were always sda.
SATA drives can now be used with banan-os. This allows much faster
disk access (writing 10 MiB from 30s to 1.5s). This can definitely
be optimized but the main slow down is probably the whole disk
structure in the os.
AHCI drive is now the default when running qemu.
I have used two weeks in locating a bug in my ext2 implementation
while the bug was actually in disk write. If you called write_sectors
on disk it would write the first sector_size bytes repeatedly to all
asked sectors and this corrupted the disk...
Serial monitors can now be used as a output. This requires editing
init code for the stdio opening. Serial input is not supported, so qemu
still needs graphical window for ps/2 keyboard.
Bananymous
Bananymous