TODO revision 1.40 
1/* $NetBSD: TODO,v 1.40 2021/03/30 20:23:45 palle Exp $ */
2
3Things to be done:
4
5common:
6- make %g6 point to curcpu
7- make %g7 point to curlwp
8- change run-time checks for cpu type to function pointers
9
10sun4u:
11- GENERIC.UP kernel hangs on v445 (missing interrupt?)
12
13sun4v:
14 - current status
15     T5 ldom with 2 VCPU and 4GB:
16       - kernel boots from miniroot.fs via ldom fisk (vdsk)
17	   - ldom virtual network interface (vnet) is working
18	     (verified by exiting sysinst and issuing a ping command)
19	   - the sysinst tool starts, disk setup is working,
20	     but the process crashes when selecting network installation method (trap 0x34 ALIGN is received)
21	 T2000 ldom with 8 VCPU and 4GB:
22	   - crashes in /sbin/init doing an access() call where %o0 is corrupted (zero)
23	 S7 ldom with 8 VCPU and 16GB (primary ldom is Solaris 11.4 SRU30):
24	   - during autoconfiguration of vnet devices the call to hv_vintr_setenabled()
25	     returns error code 7 (EBADTRAP?)
26- idle handling - call hypervisor trap (openbsd code...??)	   
27- 64-bit kernel support
28- 32-bit kernel support
29- libkvm
30- ofwboot: tlb_init_sun4v() hardcodes number of slots to 64
31- locore.s: sun4v_datatrap missing implementation for trap level 1
32- check build without SUN4V defined
33- replace relevant references to %ver with GET_MAXCWP
34- pmap_mp_init(): sun4v missing handling
35- replace constructs like "wrpr %g0, PSTATE_KERN, %pstate" with NORMAL_GLOBALS
36- replace constructs line "wrpr %g0, PSTATE_INTR, %pstate" with ALTERNATE_GOBALS
37- sun4v tsb no need to lock... per cpu... anyway...
38- ci_tsb_desc->td_ctxidx: -1 or 1?
39- MP support - currently bypassed in pmap_bootstrap() for sun4v
40- vpci.c/vpcivar.h: cleanup FIXMEs
41- interrupts not handled properly (com at ebus only...)
42- mpt(4) complains: mpt0: Phy 0: Link Status Unknown
43- man pages for drivers imported from OpenBSD lke vpci, vbus, cbus, vdsk, ldc etc.
44- vdsk and ldc drivers: code maked with OPENBSD_BUSDMA - make the bus_dma stuff work properly
45- vbus.c: handle prom_getprop() memory leaks
46- locore.s: rft_user (sun4v specific manaul fill) - seems to work, but is it good enough (compared to openbsds rft_user?
47- openfirmware interface
48    bounce handlng could need rework
49      - only #if __arch64__ independent for SUN4V, as OF for all ultrasparcs is
50        plain 32bit code too (and this #if only because the point below becomes
51        always false otherwise)
52      - only bounce if the address is >= 1UL<<32
53      - perhaps apply the lock to OF calls in general, not only the bounce'd ones
54        (in other words: let's protect the OF context with the lock, not
55        only the bounce buffer contents).
56	  - Are there any other bounce-worthy OF calls? If so, apply handling 
57	    to all of them now
58      - consider method used by Solaris/OpenSolaris/Illumos (look for PROM_32BIT_ADDRS in
59	    the Illumos codebase)
60 - platforms tested so far:
61 -- qemu sun4v/niagara - reaches the point where the kernel needs to mount a root fs
62 -- T2000 - mmu fault somewhere
63 -- T5 - stickcmpr is not properly initialized ???
64