Re: random panics: vm_phys_free_pages: page 0xc161f2a0 has unexpected order 10

From: 韓家標 Bill Hacker <askbill_at_conducive.net>
Date: Wed, 24 Oct 2007 12:56:17 -0400
Brad Allen wrote:
> 
>> --- askbill_at_conducive.net wrote:
*snip*

>> Might try replicating the problem here on other hardware & several
>> different 'vintages' of builds.
>> 
>> IF I had a few more details. I've got two servers sitting within arm's
>> length that don't go into a rack for another 12 hours or so... Both Core-2
>> Quad, otherwise 'enough' different it might be helpful.
>> 
>> I don't code "C" - but I can test...
>> 
>> Bill
>> 
> Currently I'm using the October snapshot, I've also tried the August & June
> snapshots with the same result. I'm running an Athlon XP 2400+.
> 
> I can _generally_ get away with compiling the smaller ports (bash, nano,
> etc), however xservers and kernels, to name a couple, are guaranteed to fail.

OK - I may have 'tested' some of it already.

Other than NOT using 'portsnap', and having different CPU & RAM, specifically:

ASUS P5K, 4 GB DDR 800, Core-2 Quad 2.4 GHz,

GigaByte G33-DS3R, 4 GB DDR 800, Core-2 Quad 2.4 GHz

But we DO have overlap with the FreeBSD versions you are trying, ex:
i386 & AMD-64 Aug, Sep snapshots. cvsup from October onward 'til about 2 days ago.

- two instances of make build~ failing to complete. OK with new cvsuping 12 
hours later. Might have been OK within minutes if re-pulled. 'not known'

- a few instances of pkg_add being unable to find a consenting tarball, but a 
make install from the ports tree working OK for the same port.

- a few instances of a 'common' port refusing to build on AMD-64 at all (Opera 
comes to mind). Not to mention the odd few ports simply marked as 'broken'.

No occcurence of the sort of error you cited.

JMHO, but...

"SWAG" - most likely to be stale debris in your build environment. We either 
nuke /usr/src, or rely on the 'delete' tag to clean it up for us.

Second choice is RAM complement. Third suspect is RAM quality.

Meanwhile;

- Give me the *exact* commands you are using, I'll try them, too.

- See if you can lay hands on another 1 GB of RAM and if it at least delays the 
issue.

Standing by...

Bill
Received on Wed Oct 24 2007 - 14:56:40 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:20 UTC