1 |
On 16/09/2017 22:31, allan gottlieb wrote: |
2 |
> I am one of the users experiencing the |
3 |
> infinite rebuild of binutils |
4 |
> bug. Today it took a turn I find worrisome |
5 |
|
6 |
|
7 |
Replying here so it's at the top. |
8 |
|
9 |
Several things are happening, but I suppose the primary one is (and I |
10 |
sorts need to be direct here): |
11 |
|
12 |
Stop being scared and stop being confused by the big words. It's |
13 |
blinding you to what is right there is plain sight right in front of you. |
14 |
|
15 |
binutils has a history of getting itself and portage confused as to what |
16 |
it needs and does not need with respect to preserved-rebuild. A few |
17 |
other packages do this as well over the years. The most direct solution |
18 |
is to simply unmerge the package, letting portage clean out the slate |
19 |
and all the cruft, then merge the package back. The junk goes away, |
20 |
portage sorts itself out, and stops it's whining. |
21 |
|
22 |
The wrinkle as you point out is that you can't merge binutils without |
23 |
bintils as you need ld. But never fear! this is Gentoo and you have all |
24 |
the tools. Just quickpkg it, unmerge the damn thing and untar it back |
25 |
like all the advise you've read tells you to do. Now you have an ld and |
26 |
can properly merge binutils, so do so. portage will complain about file |
27 |
collisions as expected, but because you are doing this in "yes, I DO |
28 |
know what I am doing" mode, say yes. |
29 |
|
30 |
That should fix your preserved-rebuild. |
31 |
|
32 |
|
33 |
Now, for the masking notice. This is a completely different thing and |
34 |
the only similarlity is that this sequence of letters "b,i,n,u,t,i,l,s" |
35 |
shows up in both. |
36 |
|
37 |
binutils is slotted. You have 2.28.1 and the mask applies to 2.25.1-r1 |
38 |
Portage is really saying "Yo dude, so this binutils you have? I've been |
39 |
told to mask it and not just use it. You need to tell me if you want to |
40 |
use it anyway, or if I can kill it with fire, or what." |
41 |
|
42 |
It's a simple problem of deciding what you want to do. 2.25 is ancient, |
43 |
so if you have it installed for some reason, you should find out why and |
44 |
deal with that. It's probably safe, but you do need to find why you have |
45 |
it first. Then proceed as a routine package removal (which is actually |
46 |
all this is). |
47 |
|
48 |
|
49 |
> |
50 |
> To summarize for months now after every emerge I get |
51 |
> |
52 |
> !!! existing preserved libs: |
53 |
> >>> package: sys-libs/binutils-libs-2.28.1 |
54 |
> * - /usr/lib64/libbfd-2.25.1.so |
55 |
> * used by |
56 |
> /usr/lib64/binutils/x86_64-pc-linux-gnu/2.25.1/libopcodes-2.25.1.so |
57 |
> (sys-devel/binutils-2.25.1-r1) |
58 |
> Use emerge @preserved-rebuild to rebuild packages using these libraries |
59 |
> |
60 |
> If if then |
61 |
> emerge @preserved-rebuild |
62 |
> it rebuilds binutils but I still get the above message. I have rebuilt |
63 |
> it at least a dozen times in the past months. |
64 |
> |
65 |
> A solution was posted that involved making a quickpkg (I think), |
66 |
> unmerging binutils, and then manually untaring (after the unmerge of |
67 |
> binutils you can't simply remerge since ld is gone). |
68 |
> |
69 |
> This solution frightened me and I am living with always being told to run |
70 |
> emerge @preserved-rebuild |
71 |
> and actually doing that fruitless emerge about once a month. |
72 |
> |
73 |
> Today was different, I guess because binutils-lib was involved |
74 |
> |
75 |
> My "twice-weekly" |
76 |
> emerge --update --changed-use --with-bdeps=y @world |
77 |
> resulted in |
78 |
> Calculating dependencies... done! |
79 |
> [ebuild NS ] sys-kernel/gentoo-sources-4.12.12 [3.10.3-r1, 3.11.0, 3.12.13, 3.18.11, 3.18.12, 4.0.5, 4.0.9, 4.1.12, 4.1.15-r1, 4.4.6, 4.4.21, 4.4.26, 4.4.39, 4.9.6-r1, 4.9.16, 4.9.34, 4.12.5] USE="-build -experimental -symlink" |
80 |
> [ebuild U ] www-plugins/adobe-flash-27.0.0.130-r1 [27.0.0.130] |
81 |
> [ebuild rR ] x11-libs/cairo-1.14.8 |
82 |
> [ebuild r U ] sys-libs/binutils-libs-2.28.1 [2.28-r1] |
83 |
> |
84 |
> The following packages are causing rebuilds: |
85 |
> |
86 |
> (sys-libs/binutils-libs-2.28.1:0/2.28.1::gentoo, ebuild scheduled for merge) causes rebuilds for: |
87 |
> (x11-libs/cairo-1.14.8:0/0::gentoo, ebuild scheduled for merge) |
88 |
> |
89 |
> When the emerge finished I received the usual |
90 |
> |
91 |
> !!! existing preserved libs: |
92 |
>>>> package: sys-libs/binutils-libs-2.28.1 |
93 |
> * - /usr/lib64/libbfd-2.25.1.so |
94 |
> * used by /usr/lib64/binutils/x86_64-pc-linux-gnu/2.25.1/libopcodes-2.25.1.so (sys-devel/binutils-2.25.1-r1) |
95 |
> Use emerge @preserved-rebuild to rebuild packages using these libraries |
96 |
> |
97 |
> But when I ran |
98 |
> emerge @preserved-rebuild |
99 |
> out came the following |
100 |
> |
101 |
> The following mask changes are necessary to proceed: |
102 |
> (see "package.unmask" in the portage(5) man page for more details) |
103 |
> # required by @preserved-rebuild (argument) |
104 |
> # /var/portage/profiles/package.mask: |
105 |
> # Michał Górny <mgorny@g.o>, Andreas K. Hüttel <dilfridge@g.o>, |
106 |
> # Matthias Maier <tamiko@g.o> (21 May 2017) |
107 |
> # These old versions of toolchain packages (binutils, gcc, glibc) are no |
108 |
> # longer officially supported and are not suitable for general use. Using |
109 |
> # these packages can result in build failures (and possible breakage) for |
110 |
> # many packages, and may leave your system vulnerable to known security |
111 |
> # exploits. |
112 |
> # If you still use one of these old toolchain packages, please upgrade (and |
113 |
> # switch the compiler / the binutils) ASAP. If you need them for a specific |
114 |
> # (isolated) use case, feel free to unmask them on your system. |
115 |
> =sys-devel/binutils-2.25.1-r1 |
116 |
> |
117 |
> NOTE: The --autounmask-keep-masks option will prevent emerge |
118 |
> from creating package.unmask or ** keyword changes. |
119 |
> |
120 |
> Would you like to add these changes to your config files? [Yes/No] |
121 |
> |
122 |
> I said no since I don't want to unmask, but am not sure how to proceed |
123 |
> |
124 |
> Thanks in advance for any help. |
125 |
> |
126 |
> allan |
127 |
> |
128 |
> |
129 |
|
130 |
|
131 |
-- |
132 |
Alan McKinnon |
133 |
alan.mckinnon@×××××.com |