<div dir="ltr"><br><div class="gmail_extra"><br><br><div class="gmail_quote">2013/1/18 Adam Jackson <span dir="ltr"><<a href="mailto:ajax@redhat.com" target="_blank">ajax@redhat.com</a>></span><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">On Fri, 2013-01-18 at 20:49 +0200, cornel panceac wrote:<br>
<br>
<br>
> 2013/1/18 Justin M. Forbes <<a href="mailto:jmforbes@linuxtx.org">jmforbes@linuxtx.org</a>><br>
<br>
</div><div class="im">> There are a lof of reasons that this isn't feasible on a<br>
> running system.<br>
> You just don't have access to all of the memory for a full<br>
> check.<br>
><br>
> Justin<br>
><br>
> why not? is there something that can not be relocated in RAM?<br>
<br>
</div>The kernel image itself. DMA buffers. Memory made inaccessible by PCI<br>
BARs being in the way. Probably some other stuff I'm forgetting. Even<br>
if you fixed all that you'd still be Heisenberg-uncertain because you'd<br>
no longer be the only thing exerting cache pressure.<br>
<div class="im"><br>
> on the other hand, wasn't there supposed to be a way to exclude a<br>
> damaged RAM area by passing some arguments to the kernel?<br>
<br>
</div>See linux/Documentation/bad_memory.txt<br>
<br>
- ajax<br>
<div class="HOEnZb"><div class="h5"><br>
<br><br></div></div></blockquote><div>ok, thank you very much, ajax. <br></div></div><br></div></div>