Please revert libc Symbol.map / Versions.def changes

[Matthew Dillon]

I think I've fixed the lockup issue, 97c8c48c09.  We're still working on
the symbol issues. please test the lockup issue by patching in 97c8c48c09
to the version where you backed out the symbol changes.

-Matt

On Sun, Feb 5, 2017 at 8:51 AM, John Marino <dragonflybsd at marino.st> wrote:

> On 2/5/2017 01:46, David Shao wrote:
>
>> Please revert recent changes:
>>
>> commit 626c9d706f1bffc498cc5ffda1baefba2677d693
>> Date:   Fri Feb 3 19:06:39 2017 +0100
>>
>>     libc/sys: Clean up Symbol.map for the recent get/setaffinity work.
>>
>> commit 61d257f12c46320896dfabdce3ba0591d62e06ff
>> Date:   Fri Feb 3 19:14:40 2017 +0100
>>
>>     libc: Adjust Versions.def.
>>
>> These somehow break the most basic compilation in userland
>> such as compiling pkgsrc devel/cmake which suddenly
>> claims it can't find unordered_map in std::.
>>
>> After a reversion back to
>>
>> commit 22f4a817164aa428518102732730fadabc37b8eb
>> Date:   Tue Jan 31 18:33:33 2017 -0800
>>
>>     kernel - Avoid p->p_token when sending a signal to a particular lwp
>>
>> all seems well.
>>
>> Using the kernel post the previous changes doesn't exactly
>> seem a bed of roses either with constant lockups on
>> an Intel Ivybridge machine, even trying to do a simple
>> make -j7 buildworld
>>
>>
> Yeah, you can't just redefine symbol names.
> I think rather than reverting the commit, weak symbols should be defined
> instead so it doesn't existing applications.
>
> In other words, the symbol map is now correct but we should cover the
> consequences of the earlier bad symbol map.
>
> John
>
>
>
> ---
> This email has been checked for viruses by Avast antivirus software.
> https://www.avast.com/antivirus
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.dragonflybsd.org/pipermail/users/attachments/20170205/b351c8ba/attachment-0003.htm>