TODO revision 1.19 1 1.19 tsutsui $NetBSD: TODO,v 1.19 2003/05/25 15:46:15 tsutsui Exp $
2 1.2 thorpej
3 1.17 tsutsui To do list (in some particular order)
4 1.17 tsutsui XXX some entries might be obsolete.
5 1.12 soda
6 1.17 tsutsui o Boot. Standalone boot program to load ELF kernels instead of
7 1.17 tsutsui booting ECOFF kernels directly.
8 1.17 tsutsui (maybe sgimips/stand would help, but annoying buggy ARC BIOS)
9 1.17 tsutsui
10 1.17 tsutsui o sysinst
11 1.17 tsutsui maybe MD fdisk partitioning support is required to load files
12 1.17 tsutsui from FAT partition on ARC BIOS prompt.
13 1.17 tsutsui
14 1.17 tsutsui o install notes
15 1.17 tsutsui
16 1.17 tsutsui
17 1.17 tsutsui o use MI driver
18 1.17 tsutsui
19 1.17 tsutsui - make fd driver MI, and share it with other ports
20 1.17 tsutsui (contact christos about MI fd driver)
21 1.17 tsutsui
22 1.17 tsutsui - use MI bha driver instead of home grown btl
23 1.17 tsutsui XXX needs fixes of DESKstation support
24 1.14 soda
25 1.17 tsutsui o VXL framebuffer support (Magnum, RISCstation 2200)
26 1.14 soda
27 1.17 tsutsui o com_jazzio.c
28 1.17 tsutsui - clock handling clean up (obtain from ARC BIOS)
29 1.17 tsutsui - fifo disabling may be only needed on some Magnum?
30 1.1 jonathan
31 1.17 tsutsui o remove pccons and switch to wscons completely
32 1.17 tsutsui (XXX what's the problem to remove pccons?)
33 1.9 soda
34 1.17 tsutsui o AD1848 audio support
35 1.1 jonathan
36 1.17 tsutsui o missing MI devices
37 1.17 tsutsui ses?, vcoda, ...
38 1.1 jonathan
39 1.4 soda
40 1.17 tsutsui o Xserver
41 1.4 soda
42 1.17 tsutsui - VXL Magnum, RISCstation 2200
43 1.17 tsutsui - vga/S3 PICA, Image RISCstation - OpenBSD's?
44 1.17 tsutsui - vga/cirrus RISCserver 2200, Express5800/240 R4400 EISA
45 1.17 tsutsui - vga/??? DESKstation Tyne, rPC44
46 1.17 tsutsui - TGA RISCstation 2250, Express5800/230 R4400 PCI
47 1.4 soda
48 1.17 tsutsui o Find out why bitmap load to S3-928 flashes screen. (X server)
49 1.17 tsutsui Know why (enable linear mode). Need S3 info.
50 1.5 soda
51 1.5 soda
52 1.15 soda o repair DESKstation support
53 1.11 soda - requires bounce buffer bus_dma for Tyne
54 1.17 tsutsui XXX - too small bounce buffer size (128KB)
55 1.11 soda
56 1.15 soda o Olivetti M700 support
57 1.11 soda
58 1.19 tsutsui o NEC Express5800/230 R10000 PCI (NEC-J95) support
59 1.17 tsutsui (needs MI R10000 support)
60 1.11 soda
61 1.17 tsutsui o SNI RM200PCI/RM300/RM400/RM600 support
62 1.11 soda
63 1.11 soda
64 1.17 tsutsui o parse ARC BIOS configuration information and use it
65 1.11 soda
66 1.17 tsutsui o increase MAXPHYS to 64KB
67 1.17 tsutsui (XXX why is it limited to 32KB?)
68 1.11 soda
69 1.17 tsutsui o fix kernel start address
70 1.17 tsutsui (maybe requires bootloader support)
71 1.11 soda
72 1.17 tsutsui o allocate PICA_TL_BASE dynamically
73 1.5 soda
74 1.15 soda o remove inb/outb
75 1.8 nisimura
76 1.15 soda o remove UADDR
77 1.9 soda
78 1.17 tsutsui o fix mem_clusters[] usage.
79 1.17 tsutsui
80 1.17 tsutsui o intrcnt[] name cleanup, use MI evcnt(9)
81 1.17 tsutsui
82 1.17 tsutsui o test and merge soren's clean up about proc0.p_addr.
83 1.17 tsutsui
84 1.15 soda o redesign interrupt handler framework.
85 1.15 soda i/o bus devices should have sane IPL, but currently doesn't.
86 1.13 soda
87 1.15 soda also, current MIPS interrupt handler has overblocking and
88 1.15 soda other problems as follows:
89 1.9 soda
90 1.9 soda - SR_INT_IE should be enabled before calling hardclock().
91 1.9 soda Since this is not done currently, spllowersoftclock()
92 1.9 soda on hardclock() doesn't have effect, and softclock() is
93 1.9 soda handled with all interrupt disabled in this case.
94 1.9 soda -> overblocking, possibly causes missing hardclock()
95 1.9 soda
96 1.9 soda - MIPS3_CLKF_BASEPRI() doesn't work correctly,
97 1.9 soda when MIPS_INT_MASK_5 (== MIPS_INT_MASK_CLOCK) is disabled.
98 1.9 soda -> micro optimization on hardclock() doesn't work.
99 1.9 soda but currently this may make hardclock() latency better
100 1.9 soda due to above SR_INT_IE problem.
101 1.9 soda s/MIPS_INT_MASK/MIPS3_INT_MASK/ makes this work, although tricky.
102 1.9 soda
103 1.9 soda - if (ipending & INT_MASK_REAL_DEV) == 0,
104 1.9 soda softnet() and softclock() are handled with all interrupt disabled.
105 1.9 soda -> overblocking, possibly causes missing hardclock()
106 1.9 soda
107 1.9 soda - `netisr' handling in netintr() implies potential race condition.
108 1.15 soda The access to `netisr' should be protected by splnet().
109 1.15 soda Currently this is not real problem due to above overblocking.
110 1.9 soda
111 1.10 soda - INT_MASK_REAL_DEV should be removed
112 1.10 soda
113 1.11 soda - make CLKF_INTR() work.
114 1.11 soda
115 1.9 soda o it is better to always disable MIPS_INT_MASK_CLOCK.
116 1.9 soda those are the points which should be fixed:
117 1.9 soda mips_idle: li t0, (MIPS_INT_MASK | MIPS_SR_INT_IE)
118 1.9 soda machdep.c: curpcb->pcb_context[11] = MIPS_INT_MASK | MIPS_SR_INT_IE;
119 1.9 soda spl0()
120 1.9 soda splnone()
121 1.10 soda
122 1.10 soda - MIPS_INT_MASK_CLOCK should be removed in someway
123 1.1 jonathan
124 1.17 tsutsui o XXX at least 2000/06/07 version is already quite unstable
125 1.17 tsutsui on PICA and NEC Image RISCstation. (but almost OK on Magnum)
126 1.17 tsutsui Userland commands dumps core randomly.
127 1.17 tsutsui This version is before _MIPS_PADDR_T_64BIT changes
128 1.17 tsutsui and MIPS3_TLB_WIRED_UPAGES changes.
129 1.1 jonathan
130 1.17 tsutsui "vm_page_zero_enable = FALSE" makes this problem disappeared.
131 1.17 tsutsui (vm_page_zero_enable = FALSE by default on all archs w/ UBC, now)
132 1.11 soda
133 1.17 tsutsui currently, page zero in the idle loop is also disabled on
134 1.17 tsutsui untested platforms like DESKstation rPC44/Tyne and SNI for safety.
135 1.11 soda
136 1.17 tsutsui XXX what's the current status of uvm_pageidlezero()?
137 1.11 soda
138 1.4 soda
139 1.17 tsutsui o resolve "XXX"
140 1.5 soda
141 1.4 soda
142 1.17 tsutsui (following entries might be MI MIPS items)
143 1.11 soda
144 1.17 tsutsui o Move the RO and WIRED attribute from the pte to the pv table.
145 1.17 tsutsui This saves four instructions in the tlb miss handler.
146 1.15 soda
147 1.17 tsutsui o Can we have 32 double registers?
148 1.15 soda
149 1.17 tsutsui o 64bit kernel/userland
150 1.15 soda
151 1.17 tsutsui o fix implementation of DELAY(), clean up clock implementation
152 1.4 soda
153 1.17 tsutsui o omit __SWAP_BROKEN in <mips/types.h>
154 1.11 soda
155 1.11 soda o clean up ALEAF/NLEAF/NON_LEAF/NNON_LEAF in userland.
156 1.1 jonathan
157 1.1 jonathan Lots of other things.....
158