installworld fails on 7.2-RELEASE/amd64

[Alex R]

Boris Samorodov wrote:
> On Tue, 11 Aug 2009 05:17:09 +1000 Alex R wrote:
>   
>> Boris Samorodov wrote:
>>     
>>> On Mon, 10 Aug 2009 18:30:22 +1000 Alex R wrote:
>>>       
>
>   
>>>> /usr/obj/usr/src/sys/boot/i386/boot2/../btx/btx/btx -l boot2.ldr  -o
>>>> boot2.ld -P 1 boot2.bin
>>>> btxld:No such file or directory
>>>> *** Error code 1
>>>>         
>
>   
>>> This error (not only with btxld but with some random file) often
>>> occures when the system timer has been changed (imho stepped back)
>>> while the system is building/installing world. World rebuilding
>>> helps in that case. 
>>>       
>
>   
>> Why might the system timer do this? I am confused. The thing that
>>     
>
> Well, there are too many possibilities here. Like some run an
> ntpdate command. If you have logs you may check them up.
>
>   
>> ended up fixing it was completely rebuilding /usr/src (deleting the
>> dir and installing the system sources via csup again)
>>     
>
> Seems like the case I supposed.
>
>   
>> It's a new computer so perhaps there is some compatibility problem or
>> fault with the machine? During a couple of port builds, I noticed a
>> few processes relating to the build of a port had died with signal 10
>> in dmesg (bus error i think this means), and during a build of apache,
>> something called confcheck had died with signal 12.
>>     
>
> Hm, that is not good imho. Smells like hardware fault.
>
>   
>> I ran memtest86 on this system for about 6 hours and after about 20
>> passes, no errors reported.
>>     
>
> Memory is only one system component. A processor/disk may be overheated,
> coolers stopped, etc. A very good test is make world (one after another
> several times).
>
>   
Shouldn't be a heating issue, the case has fans galore in it, the 
thermal side of things look ok from what I can see :) It's one of the 
recent gigabyte motherboards that uses DDR3 memory. I did find a setting 
in the BIOS that had a title of DRAM performance enhance, it was set to 
turbo by default, I have set it back to standard in case that was 
causing stability issues (the machine is not overclocked). I have also 
gone back to the i386 release instead of amd64. Done a build world and 
have built several ports, no core dumps or unexplained phenomena as of 
yet (fingers crossed). though if the system starts to act up again, I 
will be sending the motherboard back!

Off topic, it wouldn't be a first time that the amd64 release has 
presented odd issues. I have a machine with an Intel desktop board with 
a core duo cpu in it (EMT64 capable) with 4GB of DDR3 memory, freebsd 
7/amd64 or freebsd 8/amd64 refuse to boot. On that machine, it just page 
faults during the kernel init (had a PR open for over a year now, going 
nowhere), however the i386 release of freebsd boots ok. 64 bit linux 
works perfectly on that board.

It's a bit of a hit and miss thing these days with motherboards and open 
source operating systems. I've generally had a good run with FreeBSD on 
Gigabyte hardware.

Thanks for your suggestions though.