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
88
89
90
|
[[!meta copyright="Copyright © 2013 Free Software Foundation, Inc."]]
[[!meta license="""[[!toggle id="license" text="GFDL 1.2+"]][[!toggleable
id="license" text="Permission is granted to copy, distribute and/or modify this
document under the terms of the GNU Free Documentation License, Version 1.2 or
any later version published by the Free Software Foundation; with no Invariant
Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license
is included in the section entitled [[GNU Free Documentation
License|/fdl]]."]]"""]]
# IRC, freenode, #hurd, 2013-01-18
<braunr> youpi: what would you think of adding a debug-related syscall to
gnumach so that we have a printf-like tool even in situations where the
code can't perform an rpc (i.e. glibc)
<youpi> could be useful indeed
<youpi> I've found myself having a hard time making some printfs from odd
places of glibc :)
<braunr> i also suggest we make it unprivileged
<youpi> not enabled by default then
<youpi> otherwise it's an easy DoS
<braunr> well, for now, we don't care much about DoS, but we do care about
debugging
<braunr> at least until the very core issues we have are understood and
resolved
<youpi> I usually frown on debugging feature being too open
<braunr> me too
<youpi> you would always forget dropping one
<braunr> that's why i didn't suggest it earlier
<braunr> but i'm wasting too much time finding other decently effective
ways
## IRC, freenode, #hurd, 2013-01-19
<braunr> youpi: how about we build this debugging system call in debugging
versions only ?
<braunr> i suppose you already use such versions for buildds anyway
<braunr> MACH_DEBUG is always enabled
<braunr> the debugging version only adds --enable-kdb if i'm right
<pinotree> check debian/rules
<braunr> that, and -fno-omit-frame-pointer
## IRC, freenode, #hurd, 2013-01-21
<braunr> youpi: concerning gnumach, i've added a mach_print system call,
with one argument (a null terminated string) to -dbg kernels
(--enable-kbd)
<youpi> k
<braunr> if it's fine with you, i'll commit it too
<youpi> I'm fine
## IRC, freenode, #hurd, 2013-04-07
<braunr> see http://www.sceen.net/~rbraun/mach_print/ for an example to use
it
[[Makefile]], [[mach_print.S]], [[main.c]].
## IRC, freenode, #hurd, 2013-07-01
<youpi> braunr: btw, we are missing the symbol in mach/Versions
<braunr> youpi: what symbol ?
<youpi> so the libc-provided RPC stub is not available
<youpi> mach_printf
<youpi> -f
<braunr> it's a system calll
<braunr> not exported
<youpi> s/RPC/system call/
<braunr> that's expected
<youpi> libc does provide stubs for system calls too
<braunr> yes but not for this one
<youpi> I don't see why we wouldn't want to include it
<youpi> ?! it does
<braunr> it's temporary
<braunr> oh
<braunr> there must be automatic parsing during build
<youpi> sure
<braunr> nice
<braunr> youpi: if we're going to make this system call exported by glibc,
i should change its interface first
<braunr> it was meant as a very quick-and-dirty hack and directly accesses
the caller's address space without going through a special copy-from-user
function
<braunr> not very portable
|
