iBook G4 chipset failure causing segfaults?

[Bruce Cran]

I've recently found that my G4 iBook is causing programs to segfault
when they start going into swap: the effect is worse on -current but
I'm also seeing it on 7.2. For example, if I run a program to allocate 500MB the
system will swap out most programs since I only have 512MB RAM. After a
while dhclient and sendmail segfault while the test program keeps
running; I've also seen tcsh and login segfault too as I logout. I
originally found the problem when building clangbsd, where 'as' would
segfault when it started building the large C++ files that come with
clang. I can run a normal buildworld so I guess this is more of a
problem with the chipset (caused by swapping?) than a memory problem?

I've run the Apple Hardware Test utility over 30 loops without any
errors - has anyone else seen this problem, and is it likely that the
hardware is simply starting to fail?

--
Bruce Cran
_______________________________________________
freebsd-ppc@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ppc
To unsubscribe, send any mail to "freebsd-ppc-unsubscribe@..."

[Nathan Whitehorn-7]

Bruce Cran wrote:

> I've recently found that my G4 iBook is causing programs to segfault
> when they start going into swap: the effect is worse on -current but
> I'm also seeing it on 7.2. For example, if I run a program to allocate 500MB the
> system will swap out most programs since I only have 512MB RAM. After a
> while dhclient and sendmail segfault while the test program keeps
> running; I've also seen tcsh and login segfault too as I logout. I
> originally found the problem when building clangbsd, where 'as' would
> segfault when it started building the large C++ files that come with
> clang. I can run a normal buildworld so I guess this is more of a
> problem with the chipset (caused by swapping?) than a memory problem?
>
> I've run the Apple Hardware Test utility over 30 loops without any
> errors - has anyone else seen this problem, and is it likely that the
> hardware is simply starting to fail?
>
>  

More likely there's a bug in the ATA driver, or the hard disk is
failing, and so swap is being corrupted somehow. Is there a way you
could try swapping to a different, external drive (firewire, for
instance), and see if the problem still exists?
-Nathan
_______________________________________________
freebsd-ppc@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ppc
To unsubscribe, send any mail to "freebsd-ppc-unsubscribe@..."

[Bruce Cran]

Nathan Whitehorn wrote:

> Bruce Cran wrote:
>> I've recently found that my G4 iBook is causing programs to segfault
>> when they start going into swap: the effect is worse on -current but
>> I'm also seeing it on 7.2. For example, if I run a program to
>> allocate 500MB the
>> system will swap out most programs since I only have 512MB RAM. After a
>> while dhclient and sendmail segfault while the test program keeps
>> running; I've also seen tcsh and login segfault too as I logout. I
>> originally found the problem when building clangbsd, where 'as' would
>> segfault when it started building the large C++ files that come with
>> clang. I can run a normal buildworld so I guess this is more of a
>> problem with the chipset (caused by swapping?) than a memory problem?
>>
>> I've run the Apple Hardware Test utility over 30 loops without any
>> errors - has anyone else seen this problem, and is it likely that the
>> hardware is simply starting to fail?
>>
>>  
> More likely there's a bug in the ATA driver, or the hard disk is
> failing, and so swap is being corrupted somehow. Is there a way you
> could try swapping to a different, external drive (firewire, for
> instance), and see if the problem still exists?

I moved swap onto an external 500GB USB drive and also tried disabling
ATA DMA on the internal HDD, but I kept seeing the segfaults.

--
Bruce
_______________________________________________
freebsd-ppc@... mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ppc
To unsubscribe, send any mail to "freebsd-ppc-unsubscribe@..."