On 20 Apr 2014, at 17:47, O. Hartmann <ohart...@zedat.fu-berlin.de> wrote:
> On Sat, 19 Apr 2014 20:25:51 +0200
> Dimitry Andric <d...@freebsd.org> wrote:
>> On 19 Apr 2014, at 19:30, O. Hartmann <ohart...@zedat.fu-berlin.de> wrote:
>>> On Sat, 19 Apr 2014 18:44:51 +0200
>>> Dimitry Andric <d...@freebsd.org> wrote:
>> ...
>>>>> Stop in /usr/ports/x11/kdelibs4/work/.build.
>>>>> *** [khtml/CMakeFiles/khtml.dir/all] Error code 1
>>>>> <parser-b6e15c.cpp><parser-b6e15c.sh>
>>>> 
>>>> I cannot reproduce the problem on my copy of stable/9 r264647.  Using 
>>>> /usr/bin/time
>>>> -l on it results in:
>>>> 
>>>>       3.36 real         1.01 user         0.59 sys
>>>>    71684  maximum resident set size
>>>>    30951  average shared memory size
>>>>     3349  average unshared data size
>>>>      126  average unshared stack size
>>>>    13539  page reclaims
>>>>      354  page faults
>>>>        0  swaps
>>>>        2  block input operations
>>>>        3  block output operations
>>>>       66  messages sent
>>>>        0  messages received
>>>>        0  signals received
>>>>      370  voluntary context switches
>>>>       10  involuntary context switches
>>>> 
>>>> So it compiles in 3 seconds, and uses approximately 70MB of memory.
>>> what compiler does your copy use as default? I use CLANG 3.3/3.4
>> 
>> Just the version of clang that comes with stable/9 r264647, e.g.:
>> 
>> FreeBSD clang version 3.4 (tags/RELEASE_34/final 197956) 20140216
>> 
>> Are you using a port version of clang to build kdelibs4, by any chance?
>> 
>> -Dimitry
>> 
> 
> I successfully built, as a workaround, the port x11/kdelibs4 with gcc 4.7 
> (port lang/gcc).

So it didn't work with lang/clang34 or the base version of clang either?
And with a similar type of error?


> The box in question is a Dell Latitude E6510 notebook with only 4 GB of RAM, 
> could this
> be the issue? The system very often starts swapping. Even my oldstyle E8400 
> workstation
> with only 8 GB (most recent 11.0-CURRENT) starts swapping very often and 
> recently, I saw
> musterious compiler erros and stopping compiling processes never seen bevor. 
> Restarting
> the failed portbuild most often finish successfully.

There were some postings recently, about newer versions of FreeBSD being
supposedly more "swappy", see e.g.:

http://lists.freebsd.org/pipermail/freebsd-stable/2014-April/thread.html#78361

I have no idea if this is really substantiated with evidence, or just a
feeling, though. :)

In any case, when you are experiencing mysterious compiler errors, and
your system is heavily exercising RAM and swap, it is always a good idea
to do a full hardware diagnostics test.

For your RAM, you can use memtest86+, and since you have a Dell, you can
use their diagnostics program to test other parts of the machine.

-Dimitry

Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail

Reply via email to