TODO revision 1.28 1 /* $NetBSD: TODO,v 1.28 2018/12/17 20:13:46 palle Exp $ */
2
3 Things to be done:
4
5 common:
6 - make %g6 point to curcpu
7 - make %g7 point to curlwp
8 - change run-time checks for cpu type to function pointers
9
10 sun4u:
11 - GENERIC.UP kernel hangs on v445 (missing interrupt?)
12
13 sun4v:
14 - current status: The kernel boots and starts the init process (pid 1)
15 which spawns /sbin/sh (pid 2). This is forked as a new /sbin/sh (pid 3)
16 which crashes, probably due to MMU DPROT issues.
17 Looks like the user process tries to write to the stack as part of a save
18 instruction since cansave is zero. This causes a DMMU PROT trap (0x6c) which
19 calls sun4v_tl0_dtsb_prot which calls sun4v_datatrap which calls TRAP_SETUP().
20 This macro has code that does a save instruction, again to the user stack,
21 which again causes a MMU DPROT trap...
22 - 64-bit kernel support
23 - 32-bit kernel support
24 - libkvm
25 - ofwboot: tlb_init_sun4v() hardcodes number of slots to 64
26 - locore.s: sun4v_datatrap missing implementation for trap level 1
27 - check build without SUN4V defined
28 - replace relevant references to %ver with GET_MAXCWP
29 - pmap_mp_init(): sun4v missing handling
30 - replace constructs like "wrpr %g0, PSTATE_KERN, %pstate" with NORMAL_GLOBALS
31 - replace constructs line "wrpr %g0, PSTATE_INTR, %pstate" with ALTERNATE_GOBALS
32 - sun4v tsb no need to lock... per cpu... anyway...
33 - ci_tsb_desc->td_ctxidx: -1 or 1?
34 - MP support - currently bypassed in pmap_bootstrap() for sun4v
35 - vpci.c/vpcivar.h: cleanup FIXMEs
36 - interrups not handled properly (com at ebus only...)
37 - mpt(4) complains: mpt0: Phy 0: Link Status Unknown
38 - man pages for drivers imported from OpenBSD lke vpci, vbus, cbus, vdsk, ldc etc.
39 - vdsk and ldc drivers: code maked with OPENBSD_BUSDMA - make the bus_dma stuff work properly
40 - vbus.c: handle prom_getprop() memory leaks
41 - locore.s: rft_user (sun4v specific manaul fill) - seems to work, but is it good enough (compared to openbsds rft_user?
42 - platforms tested so far:
43 -- qemu sun4v/niagara - crash in init process relaed to fork (stack?)
44 -- T2000 - mmu fault somewhere
45 -- T5 - stickcmpr is not properly initialized ???
46