Friday, March 23, 2012 11:06 PMDaniel Kahn Gillmor wrote: On Fri, 23 Mar 2012 19:51:37 -0700 (PDT), Will Set <debiandu...@yahoo.com> wrote:
>> I know of 5 bug reports that confirm using boot parameter - processor.nocst=1 >> as a workaround for kernels < 2.6.38 > >Thanks, i will try this the next time i get a chance to restart this >machine (it's currently crashed again and i don't have physical access >right now). Probably not necessary since you are using 2.6.32 (squeeze) kernels. Normally, kernels older than 2.6.39 don't need the extra boot parameter. But 4 GiB of ram is max for the board and although HP has spec for 4 GiB of ram for your model, they also have notes recommending only 2 GiB of Ram, in another doc. http://bizsupport2.austin.hp.com/bc/docs/support/SupportManual/c00072736/c00072736.pdf > >Do you have a link to a couple of these bug reports so i could read them >myself? I'd appreciate it if you do. https://bugzilla.redhat.com/show_bug.cgi?id=727865 > >> linux-image-2.6.39 through linux-image 3.3.0-rc6-686-pae - on both of >> my 865g based boxes. > >I'm not sure what this sentence means -- is it related to the sentence >above? if so, how? I have two computers with i865g chipsets and 1 GiB of ram each. I'm not sure what I was thinking when I refered to kernels your not using. > >> Also, iirc, the bigmem kernel was swallowed by the 686-pae kernel, >> which might be a reason for the instability when using 486. > >I wouldn't expect the -486 flavor to be able to fully address all 4GiB >of RAM (i.e. i doubt it would make use of the physical address >extensions). But i don't understand why this would cause instability. >My workload during these crashes is definitely not memory-intensive. >This is the first i'm hearing that the -486 flavor would cause >instability on highmem machines. > >Can you point me to some documentation so i could understand why that >might be the case? Here is a link to linux-image-2.6.32-5-686-bigmem http://packages.debian.org/squeeze/linux-image-2.6-686-bigmem. best regards, Will