1 |
On Sat, 13 May 2006 23:04:10 -0700 |
2 |
Donnie Berkholz <spyderous@g.o> wrote: |
3 |
|
4 |
> Kevin F. Quinn (Gentoo) wrote: |
5 |
> |
6 |
> Oh, OK, let's argue semantics. It's suggested by a hardened user on a |
7 |
> bug the hardened team is CC'd on, but the team didn't say anything was |
8 |
> wrong with the change. |
9 |
|
10 |
That's because for the moment we don't have a better suggestion; we |
11 |
can't say "don't do it" in this case until we have a solution. Our |
12 |
silence doesn't mean we like the solution; it means we haven't got |
13 |
anything better to suggest for now. |
14 |
|
15 |
> > With regards to Duncan's (non-hardened) problem, adding: |
16 |
> > |
17 |
> > filter-ldflags -Wl,-z,now |
18 |
> > |
19 |
> > to x-modular.eclass as he suggests should be fine; his issue is |
20 |
> > different to that with the hardened compiler in as much as he has |
21 |
> > added the '-Wl,-z,now' to LDFLAGS as advised by the QA message and |
22 |
> > the above filter will just remove it again; whereas to deal with |
23 |
> > the hardened compiler we need to reliably add a flag to all the |
24 |
> > relevant link commands (the bit that takes the effort is working |
25 |
> > out which are relevant). |
26 |
> |
27 |
> Now I'm confused. Do you want this filter instead of the current |
28 |
> situation, in addition to, or what? This is exactly why I asked for a |
29 |
> patch. |
30 |
|
31 |
This is a completely separate issue, nothing to do with the hardened |
32 |
team or the hardened compiler. It causes the same problem in the end, |
33 |
but a completely different way. |
34 |
|
35 |
|
36 |
The QA checks in portage advise the user to try: |
37 |
|
38 |
LDFLAGS='-Wl,-z,now' emerge ${PN} |
39 |
|
40 |
because the X server is "suid, dyn linked and using lazy |
41 |
bindings". This warning becomes fatal if FEATURES=stricter, |
42 |
so you may want to RESTRICT it (which doesn't remove the warning, so |
43 |
you should be able to find it in your build logs for xorg-server). |
44 |
|
45 |
|
46 |
In summary, for Duncan's issue I suggest adding: |
47 |
|
48 |
# Xorg server is unaviodably suid with lazy bindings |
49 |
RESTRICT="stricter" |
50 |
|
51 |
to the xorg-server ebuild to stop it dying for people with |
52 |
FEATURES=stricter (the comment helps people who have enabled STRICTER |
53 |
to see why it's disabled, in case anything else crops up) and also to |
54 |
add: |
55 |
|
56 |
filter-ldflags -Wl,-z,now |
57 |
|
58 |
to the eclass (perhaps in x-modular_src_compile, or in both |
59 |
x-modular_src_config and x-modular_src_make). If you do it just on the |
60 |
xorg-server ebuild, and people do what Duncan did and set LDFLAGS in |
61 |
make.conf, it'll set BIND_NOW on everything which at the very least |
62 |
will cause the radeon and GL drivers to fail to load. |
63 |
|
64 |
Obviously I haven't tried it so it would be useful if Duncan could |
65 |
raise a bug with the exact change he made. |
66 |
|
67 |
-- |
68 |
Kevin F. Quinn |