On Wed, Oct 31, 2018 at 9:26 PM Nadav Har'El <[email protected]> wrote:

>
>
> On Wed, Oct 31, 2018 at 9:17 PM Nadav Har'El <[email protected]> wrote:
>
>> I was able to retrieve mkfs.so from the last build on Jenkins,
>> http://jenkins.cloudius-systems.com:8080/job/osv-build/ws/build/release/tools/mkfs/
>> and replaced the one on my machine with this version, and got:
>>
>> Assertion failed: (phdr.p_flags & (PF_R | PF_W | PF_X)) == PF_R
>> (core/elf.cc: fix_permissions: 495)
>>
>> And indeed,
>>
>> $ readelf -a mkfs.so
>> ...
>>   GNU_RELRO      0x0000000000002d70 0x0000000000003d70 0x0000000000003d70
>>                  0x0000000000000290 0x0000000000000290  RW     0x8
>>
>> So it's really "RW" there instead of R as it should be! So it's probably
>> a coincidence that something changed in the build machine in the same time
>> you changed nearby code...
>> Now I need to figure out *why* there's a RW there. If it's something
>> legal, we should just drop this assert. But I don't understand the point of
>> a relro section with RW permissions...
>>
>
> It appears the problematic mkfs.so was compiled with ld.gold:
>
> Displaying notes found in: .note.gnu.gold-version
>   Owner                 Data size       Description
>   GNU                  0x00000009       NT_GNU_GOLD_VERSION (gold version)
>     Version: gold 1.14
>
> This is different from my machine, where "ld" is ld.bfd, not ld.gold.
>

Turns out we knew of this "gold" problem long ago:

https://github.com/cloudius-systems/osv/issues/891

-- 
You received this message because you are subscribed to the Google Groups "OSv 
Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to