-
Notifications
You must be signed in to change notification settings - Fork 12
/
Copy pathBUGS
87 lines (69 loc) · 2.96 KB
/
BUGS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
# ############################################################################
# K42: (C) Copyright IBM Corp. 2000.
# All Rights Reserved
#
# This file is distributed under the GNU LGPL. You should have
# received a copy of the License along with K42; see the file LICENSE.html
# in the top-level directory for more details.
#
# $Id: BUGS,v 1.6 2002/07/30 20:47:57 jimix Exp $
# ############################################################################
This is a list of currently known but not-yet-fixed bugs in K42.
May 1/2000
- the ethernet drivers are not yet working on kitch12, kitch13, and k1 with a
single binary.
May 1/2000
- k1 does not boot with 1GB of RAM.
May 1/2000
- random hangs in err_printf()'s have been seen in Kevin's code.
May 1/2000
- random hangs on hardware during boot around initKernelProcess.
May 1/2000
- there is no way of killing a user process. If it crashes or hangs, you have
to reboot.
May 1/2000
- %lld was behaving differently on hardware than it did on the
simulator before we fixed it by adding the file
sysdeps/k42/gmp-mparam.h in glibc. Should look into why there was a
difference.
May 1/2000
- "XHandleTrans.C:182 XHandleTransTable full" error appears after
running hardware for a while -- due to token circulation not
happening. This is impacting our uptimes.
May 1/2000
- unmap range of a large range hangs in kernel (MP?).
May 1/2000
- PPC doesn't scale on hardware. Need to look and see why.
May 1/2000
- migrate test with 50 threads causes hangs. Need to look and see why.
May 1/2000
- wait for 100000 usec to reap process in
ProcessShared<ALLOC>::startExecuting() and
ProcessReplicated::startExecuting()
Slows process creation destroy tests.
May 1/2000
- FileLinux::create() crashes w/ garbage argument (should fail). Eg,
"cat" (without an argument) crashes the system.
May 1/2000
- Arguments to PPC calls are not error checked on the server end,
opening a potential security hole.
May 2/2000
- Alignment handlers are not implemented in K42 yet, meaning that
4-byte aligned loads/stores to doubles fail if they happen to cross a
page boundary on PowerPC.
May 2/2000
- In kitch-linux/apps/splash2/codes/apps/barnes/defs.Hm4 there are
some fields in structs added as a workaround to force doubles to be 8
byte aligned. Remove these as soon as the alignment handler is
implemented.
May 6/2000
- err_printf() is not migration safe, and so if it is called from a
migratable thread it causes an assert failure in the Scheduler. This
assert calls err_printf(), which sets up a nice loop...
May 6/2000
- Thread allocation does not interact well with migration. If a
thread is allocated on VP 0, migrates to VP 1, then ends, it leaves a
thread on the free list of VP 1. If you repeat this a number of times
VP 0 may run out of threads (since it is drawing on the fixed pool of
the ThreadManager) while VP 1 has a large list of free threads. Some
mechanism to address this problem needs to be introduced.