'Giant' list of current issues

[winocm]

Currently, the kernel seems to suck and not suck a lot.

Current issues are:

• pmap is bad (I'm working on fixing this.)
• etimer doesn't have a real sense of time, it runs too fast.
• nothing is synchronized properly.
• the abort handlers need to handle more cases (like the real iOS kernel) including user faults. The same thing goes for sleh_undef.
• forking causes the kernel to die.
• weird weird race conditions need to be fixed.
• 4.3.5 dyld has issues with vfs, random files appear to be read as null bytes.

If you find anything else, put it in this list and hopefully we can get it fixed.

[winocm]

The new pmap is about 50% done.

[winocm]

The following issues are fixed:

• pmap is fixed (for the most part.)
• etimer is less stupid, but still stupid
• synchronization is now present
• abort handlers are now fixed (for the most part.)
• forking does not cause the kernel to die.
• 4.3.5 dyld works, and so does bash.

[winocm]

The following issues are fixed:

• pmap is nearly feature complete (some tiny bugs remain.)
• event timer works properly, matches up to bus/timer Hz rate, see rtclock_bus_configure. Dynamic decrementer set is yet to be implemented properly however.
• 4.3.5 root filesystem works great, including storage drivers such as AppleOMAPHSMMC+IOStorageFamily.

The following issues remain:

• The kernel randomly deadlocks after launching a new process.
• Processes that use 100% of CPU time cause extreme unresponsiveness.
• UNIX signal handling proper (jumping to __sigtramp and sigreturn).
• Rooting off a storage device causes a lock up.

The following enhancements remain:

• Fix inbuilt kxld so that mkexts/kexts can work (if passed from booter in /chosen/memory-map), means that we won't need to use kextcache to build kernelcaches anymore.
• Power management drivers.
• SMP support.
• Various security 'enhancements'.
• Dtrace/kdebug probes for machine dependent code.
• Dtrace itself.