8aabdd2043
around clone area, I think. Still, much improved... git-svn-id: file:///srv/svn/repos/haiku/trunk/current@1048 a95241bf-73f2-0310-859d-f6bbb57e9c96
22 lines
1.1 KiB
Plaintext
22 lines
1.1 KiB
Plaintext
1) Need to make the paging daemon load the pages in, using semaphores.
|
|
-- Looks like we *don't* have to do this, based on what is done in NewOS. Need to consult...
|
|
2) Make the caching and the VM (mmap) work together so that an mmap'ed file and a cached file will not race.
|
|
-- Maybe a vnodeManager or a lookup in a hash table or something so that we can tell if someone is already caching this vnode.
|
|
This could get a little tricky wrt permissions and all...
|
|
3) Change the architecture so that instead of multiple threads of tests, we have multiple apps of tests...
|
|
4) Tests are not done.
|
|
Test:
|
|
getNextAreaInfo
|
|
cloneArea
|
|
resizeArea
|
|
setAreaProtection
|
|
mmap / munmap
|
|
paging
|
|
disk caching
|
|
|
|
6) See how unresolvable page faults and permissions errors are handled - need to properly handle
|
|
- This may have to come at HW integration time, since they will be app signals, etc...
|
|
7) There is no arch-level integration. This is to be tested (to death) in user land first. * == can not be done in user land.
|
|
8) Add statistics counting for copy in/out
|
|
9) Must lock original area when clone()ing. Sigh.
|