1 |
On 09/12/2013 11:03 AM, Richard Yao wrote: |
2 |
> On 09/10/2013 09:17 PM, Rich Freeman wrote: |
3 |
>> On Tue, Sep 10, 2013 at 6:41 PM, Richard Yao <ryao@g.o> wrote: |
4 |
>>> 1. The kernel expects -fno-stack-protector to be the default. What will |
5 |
>>> the effect be on kernel configuration once -fstack-protector is the default? |
6 |
>> |
7 |
>> Nothing, since the kernel build system doesn't source make.conf. If |
8 |
>> somebody creates an ebuild that actually installs a kernel then it |
9 |
>> might be an issue, though it could be filtered if it is a problem. |
10 |
> |
11 |
> My understanding is that this change would be made to GCC's spec files, |
12 |
> which affects everything that uses GCC. |
13 |
> |
14 |
>>> |
15 |
>>> 2. We should make sure that -fno-stack-protector is a supported CFLAG. |
16 |
>>> This will make it easier to handle complaints from the vocal minority of |
17 |
>>> our user base that want every last percentage point of performance. |
18 |
>> |
19 |
>> No reason that it would be any less supported than -fstack-protector |
20 |
>> already is. |
21 |
> |
22 |
> Just making certain. |
23 |
> |
24 |
>>> |
25 |
>>> 3. I would like to point out that we are talking about deviating from |
26 |
>>> upstream behavior and everyone is okay with it. Anyone who thinks we |
27 |
>>> should stick to upstream when it is not good for us should speak now or |
28 |
>>> risk being asked "where were you when..." whenever they try to use |
29 |
>>> upstream as an excuse to hold back progress. ;) |
30 |
>> |
31 |
>> I don't really see this as an issue - in general maintainers are |
32 |
>> expected to accommodate reasonable CFLAGS. This doesn't mean that |
33 |
>> arbitrary CFLAGS are "supported" so much as bugs should be taken |
34 |
>> seriously if they're really bugs. If -fstack-protector causes serious |
35 |
>> problems and this is inherent to the nature of the package/etc then |
36 |
>> just filter it. If it causes problems because upstream isn't doing |
37 |
>> things right, then this is no different than how things were 10 years |
38 |
>> ago when we were stomping out amd64 issues left and right (not working |
39 |
>> on amd64 wasn't a reason to mask a package for x86, but we didn't |
40 |
>> accept "upstream doesn't care about amd64" as an excuse). |
41 |
>> |
42 |
>> Staying close to upstream is a good philosophy in general. I don't |
43 |
>> think that means that we can't have reasonable CFLAGS. Otherwise we |
44 |
>> wouldn't set CFLAGS at all and would just use whatever defaults the |
45 |
>> upstream build system applies. We're a distro - doing integration |
46 |
>> work is a value-add, not a sin. If we aren't doing integration, then |
47 |
>> users might as well just do Linux-from-scratch. |
48 |
>> |
49 |
>> Rich |
50 |
>> |
51 |
> |
52 |
> Past events have led me to think that we are sometimes too dependent on |
53 |
> upstream for guidance. I have certainly deviated from upstream whenever |
54 |
> it made sense and the results have been fairly positive. I am not saying |
55 |
> that acting for the best interests of Gentoo is mutually exclusive with |
56 |
> collaboration with upstream, but I am saying that the two sometimes |
57 |
> conflict. This being one of them. I am taking this opportunity to point |
58 |
> out that what is best for Gentoo should come first and that it is okay |
59 |
> to make decisions on our own. |
60 |
> |
61 |
|
62 |
I just checked the kernel's makefiles. It does pass -fno-stack-protector |
63 |
when the option is disabled. Therefore a GCC spec file change will not |
64 |
have cause GCC to generate kernels that differ from their .config files. |