<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Am 10.03.2012 11:49, schrieb Jaspal Dhillon:<br>
<blockquote
cite="mid:CAHgtvXNC5R45tc7ACAcbVr5OaoKvTgb8ktoQKd-GjmxC9RmCkQ@mail.gmail.com"
type="cite">I was going through the <a moz-do-not-send="true"
href="http://coreboot.org/" target="_blank">coreboot.org</a> and
I am definitely interested in it.</blockquote>
Welcome to coreboot, then!<br>
<br>
<blockquote
cite="mid:CAHgtvXNC5R45tc7ACAcbVr5OaoKvTgb8ktoQKd-GjmxC9RmCkQ@mail.gmail.com"
type="cite"> So wanted to ask couple of things. <br>
I have not read the source code yet but tried it out in this way :<br>
1.Checked out latest version of coreboot on 32bit Fedora 15. make
config with settings as : <br>
MainBoard Vendor : Emulation , Model : Qemu , ROM Size : 256KB ,
Payload : SeaBIOS<br>
make was successful . <br>
</blockquote>
<br>
<blockquote
cite="mid:CAHgtvXNC5R45tc7ACAcbVr5OaoKvTgb8ktoQKd-GjmxC9RmCkQ@mail.gmail.com"
type="cite">Since my laptop is a old one and does not have
hardware virtualization support , I use a lab pc ( 64 bit Fedora
15 ) for some of my experiments.<br>
</blockquote>
qemu requires no hardware virtualization extensions (only kvm does,
which is an extended version), and for basic tests it's usually fast
enough.<br>
<br>
<blockquote
cite="mid:CAHgtvXNC5R45tc7ACAcbVr5OaoKvTgb8ktoQKd-GjmxC9RmCkQ@mail.gmail.com"
type="cite">
<div id=":2m">Installation of CentOS ( Minimum installation ) was
successful and at the end , it asked me to reboot. When I
clicked on reboot option , coreboot got stuck in an infinite
loop and I was getting this (in a loop) on the stdio : <br>
<br>
Changing serial settings was 0/0 now 3/0<br>
In resume (status=0)<br>
In 32bit resume<br>
Attempting a hard reboot<br>
</div>
</blockquote>
These are seabios messages, so its reboot handler doesn't seem to
work correctly. Unfortunately, there are ~6 different ways of
rebooting an x86 machine (not all of which are available on any
given system), so it's possible that CentOS triggered the "wrong"
one in seabios, which in turn tries to use a method not supported by
qemu (or something like that).<br>
<blockquote
cite="mid:CAHgtvXNC5R45tc7ACAcbVr5OaoKvTgb8ktoQKd-GjmxC9RmCkQ@mail.gmail.com"
type="cite">
<div id=":2m">Could you give me some pointers as to where to begin
for GSoC or some little hacks that I should attempt out ? I am
trying different payloads in the meantime. <br>
</div>
</blockquote>
Different payloads is a good start. Ultimately, there so much you
could do, that it's better if you specify what area you're most
interested in.<br>
<blockquote
cite="mid:CAHgtvXNC5R45tc7ACAcbVr5OaoKvTgb8ktoQKd-GjmxC9RmCkQ@mail.gmail.com"
type="cite">
<div id=":2m">I read about mkelfImage command and also tried it
out but the resultant size of linux.elf is ~5MB which exceeds
the size of ROM chip , declared in Qemu. Can I use a kernel
payload ( any other ) without hacking Qemu ? ( I think not but
want to know , if yes , how ) .</div>
</blockquote>
Only by reducing the kernel size - kernel images are usually already
compressed, so there's not much we can do.<br>
Fortunately, when putting the kernel in flash, you usually know
rather well what kind of hardware you're running it on - so you can
reduce kernel/initrd to just the drivers required to access the
disk, and go from there.<br>
<br>
I think current qemu versions are more tolerant with flash sizes -
you probably can't go above 4MB, but that should be enough to
contain an optimized kernel.<br>
<br>
<br>
Regards,<br>
Patrick<br>
</body>
</html>