| 1 |
> Point #1 |
| 2 |
-------- |
| 3 |
|
| 4 |
> For your diff of sys.c, that is definitely not good. My source tree still |
| 5 |
> shows set_user() as static and it should be static. |
| 6 |
The change was a proposal. Since the beginnings of the O1-sched patch |
| 7 |
there was an "extern int set_user" change to sched.h. I have checked |
| 8 |
at that time the -ac patches and, to get rid of the warning on sched.c |
| 9 |
static int was changed there(ac) to int (the diff I have sent). I have |
| 10 |
taken it as OK (we wrote 2.4.18 kernel-time). Sorry for it, now I checked |
| 11 |
the whole source tree and indeed, set_user is only used in sched.c, so the |
| 12 |
correct solution to the warning is to remove extern int set_user from |
| 13 |
sched.h (see attached patch). I've checked the -aa patches, he does it the |
| 14 |
same way. |
| 15 |
|
| 16 |
Thanks for it. |
| 17 |
|
| 18 |
> It will make applying both the grsec and selinux patches fail horribly, but |
| 19 |
> after having gone through both 2.4.21 and 2.4.22, they both fail horribly |
| 20 |
> already with where they are now in the patch chain. I don't think it will be |
| 21 |
> that much more work to apply them last and that will, hopefully, let us keep |
| 22 |
> nearly all the other patches in their original form. Feedback on this |
| 23 |
> approach is welcome. |
| 24 |
What I am doing (it is an RPM build, from 1 spec-file, different |
| 25 |
settings): |
| 26 |
I have built many different kernels, using debian, ac, aa, ck patches |
| 27 |
(these apply first, or after some pre-kernel patch). Then I do minimal |
| 28 |
corrections (whitespace, includes order change) to get the common parts |
| 29 |
the same (ex. ac, aa, ck apply O1 differently). These "main" patches |
| 30 |
provide some "features", like |
| 31 |
debian->ipsec |
| 32 |
ac->scheduler |
| 33 |
aa->scheduler,tux |
| 34 |
ck->scheduler,preempt,lowlat |
| 35 |
These get defined and used later in the spec-file. I copy the required |
| 36 |
patch (ex. grsecurity-1.9.12-2.4.22.patch to some location like |
| 37 |
build_dir/patches/grsec.patch). This patch applies cleanly only to |
| 38 |
pristine 2.4.22, so I patch it according to the features |
| 39 |
defined by main patches (or pre-kernels) (ex. %if %sched ... %endif) to |
| 40 |
get it patched after any of the earlier patches. Some changes need to be |
| 41 |
done afterwards (O1 sched needs corrections to grsecurity/* files, or some |
| 42 |
part, where the patch has to be applied was moved from sched.c to exit.c), |
| 43 |
these are separate patches depending again on the "features" the main |
| 44 |
patches defined earlier. |
| 45 |
With this procedure I can track pre-kernels, also cvs updates to grsec/pax |
| 46 |
(the cvs updates are applied later as interdiffs), so I do not have to |
| 47 |
make changes to the original patches any more. |
| 48 |
|
| 49 |
Peter |
| 50 |
|
| 51 |
-- |
| 52 |
Peter S. Mazinger <ps.m@×××.net> ID: 0xA5F059F2 NIC: IXUYHSKQLI |
| 53 |
Key fingerprint = 92A4 31E1 56BC 3D5A 2D08 BB6E C389 975E A5F0 59F2 |
| 54 |
|
| 55 |
____________________________________________________________________ |
| 56 |
Miert fizetsz az internetert? Korlatlan, ingyenes internet hozzaferes a FreeStarttol. |
| 57 |
Probald ki most! http://www.freestart.hu |
Archives