| 1 |
I did already exactly that, and glibc failed in first step! |
| 2 |
What else should I check?? |
| 3 |
|
| 4 |
On 10/25/08, Mike Edenfield <kutulu@××××××.org> wrote: |
| 5 |
> Jan Klod wrote: |
| 6 |
>> Hello. |
| 7 |
>> |
| 8 |
>> I was trying to make a switch form normal, freshly installed gentoo to |
| 9 |
>> hardened like described in PaX quickstart. |
| 10 |
>> http://www.gentoo.org/proj/en/hardened/pax-quickstart.xml Though, that |
| 11 |
>> guide is missing to inform about when should I boot hardened-sources. |
| 12 |
>> Not sure if after "emerge binutils gcc virtual/libc" or "emerge -e |
| 13 |
>> world", but I got this error, which persists: |
| 14 |
>> |
| 15 |
>> ============================================================================ |
| 16 |
>> *** stack smashing detected ***: cc1 - terminated |
| 17 |
>> cc1: stack smashing attack in function ix86_split_to_parts - terminated |
| 18 |
>> Report to http://bugs.gentoo.org/ |
| 19 |
>> i686-pc-linux-gnu-gcc: Internal error: Killed (program cc1) |
| 20 |
>> Please submit a full bug report. |
| 21 |
>> See <URL:http://bugs.gentoo.org/> for instructions. |
| 22 |
>> make[2]: *** |
| 23 |
>> [/var/tmp/portage/sys-libs/glibc-2.6.1/work/build-default-i686-pc-linux-gnu-nptl/math/s_catanl.o] |
| 24 |
>> Error 1 |
| 25 |
>> make[2]: Leaving directory |
| 26 |
>> `/var/tmp/portage/sys-libs/glibc-2.6.1/work/glibc-2.6.1/math' |
| 27 |
>> make[1]: *** [math/others] Error 2 |
| 28 |
>> make[1]: Leaving directory |
| 29 |
>> `/var/tmp/portage/sys-libs/glibc-2.6.1/work/glibc-2.6.1' |
| 30 |
>> make: *** [all] Error 2 |
| 31 |
>> * |
| 32 |
>> * ERROR: sys-libs/glibc-2.6.1 failed. |
| 33 |
>> * Call stack: |
| 34 |
>> * ebuild.sh, line 49: Called src_compile |
| 35 |
>> * environment, line 3350: Called eblit-run 'src_compile' |
| 36 |
>> * environment, line 1075: Called eblit-glibc-src_compile |
| 37 |
>> * src_compile.eblit, line 181: Called toolchain-glibc_src_compile |
| 38 |
>> * src_compile.eblit, line 122: Called die |
| 39 |
>> * The specific snippet of code: |
| 40 |
>> * make PARALLELMFLAGS="${MAKEOPTS}" || die "make for |
| 41 |
>> ${ABI} failed" |
| 42 |
>> * The die message: |
| 43 |
>> * make for default failed |
| 44 |
>> * |
| 45 |
>> * If you need support, post the topmost build error, and the call |
| 46 |
>> stack if relevant. |
| 47 |
>> * A complete build log is located at |
| 48 |
>> '/var/tmp/portage/sys-libs/glibc-2.6.1/temp/build.log'. |
| 49 |
>> * The ebuild environment file is located at |
| 50 |
>> '/var/tmp/portage/sys-libs/glibc-2.6.1/temp/environment'. |
| 51 |
>> ============================================================================ |
| 52 |
>> |
| 53 |
>> Should I really discard my work with normal install (lost |
| 54 |
>> configurations + some hustle) and use hardened stage3? (somehow I |
| 55 |
>> don't believe, I discovered something really "bug") |
| 56 |
>> And one more question: is hardened toolchain built userland going to |
| 57 |
>> work with gentoo-sources? |
| 58 |
> |
| 59 |
> You should not boot into the PaX-enabled kernel until you |
| 60 |
> have rebuilt everything with the new binutils/gcc |
| 61 |
> combination. Otherwise you risk a critical application |
| 62 |
> still doing something bad, which is happening here. |
| 63 |
> |
| 64 |
> To fix this, you need to boot into a non-PaX kernel but |
| 65 |
> hardened profile, and re-emerge gcc. I'd suggest, to be |
| 66 |
> absolutely sure you got everything, you just rerun the two |
| 67 |
> emerge steps again from the beginning: |
| 68 |
> |
| 69 |
> emerge -1 bintils gcc virtual/libc |
| 70 |
> emerge -e world |
| 71 |
> |
| 72 |
> |
| 73 |
> |
Archives