Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 13 Jul 2012 08:48:59 -0400
From:      Eric McCorkle <eric@shadowsun.net>
To:        "Andrey V. Elsukov" <ae@FreeBSD.org>, soc-status@FreeBSD.org
Subject:   Status report
Message-ID:  <5000193B.8060308@shadowsun.net>

next in thread | raw e-mail | index | archive | help
I reported earlier this week, but to recap, I've solved all the issues 
that were blocking loader.efi from working.  More importantly, I will be 
able to diagnose errors much more efficiently going forward.  Up to this 
point, I've only known that there was an error when the loader hangs, 
and most of the time, attempting to use the EFI OutputString function 
would cause a hang.

At this point, I'm gathering information to address the noncontiguous 
kernel issue.  The following are specific points of information I need 
to nail down:

* What is the startup state of the kernel with respect to paging (ie. is 
paging enabled, how is the page table initialized, etc)

* Does the EFI memory allocation function set up page table mappings, or 
just allocate physical memory?

* Does the kernel need to be contiguous in physical or just virtual memory?

* Are there any special considerations for the kernel's virtual memory 
mapping (ie, does it need to be at a specific physical address, is it 
mapped using superpages, etc)

Some of these are almost surely documented somewhere, and others can be 
discovered with experimentation.  Finally, I am in the process of 
modifying my test environment to more closely resemble an actual system 
that is being booted.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5000193B.8060308>