I've also got the V2 release working without any major problems (a fantastic achievement, thanks Jonathan et al). I have a few observations and questions that may help others.
I did stop the NAND backup process because it was incredibly slow. "top" showed that the system wasn't using much CPU but the load was over 2.0 Using "df" also revealed that the USB stick was only having 16 bytes written to it every second or so.
The E3 keyboard is mapped as a standard desktop keyboard so most shifted characters are wrong and symbol-shift characters are ignored.
I replaced the supplied 2.6.16 uImage with the one from the 2.6.17 directory in the hope that the keyboard mappings would be fixed. It worked but it only provided a shell login for the serial connection unlike the 2.6.16 image which provides both a serial and an onboard E3 login.
The copying of the Linux image from NAND seems slow, I recall some discussion about this being fixed within u-boot. Does the copy of u-boot in the V2 release contain this fix?
My USB-to-Ethernet dongle is a Psion Dacom Gold Port using the Kawasaki LSI KL5KUSB101B chip which was detected by both the 2.6.16 and 2.6.17 kernels (it uses the kaweth device driver). Upon connection it reports some "rx-16" problem or other but it works fine.
Again, thanks to everyone involved in contributing to this project,
Alex.
-----Original Message----- From: e3-hacking-bounces@earth.li [mailto:e3-hacking-bounces@earth.li]
On
Behalf Of David Given Sent: 19 September 2006 23:40 To: Discussion of the Amstrad E3 emailer hardware/software Subject: [E3-hacking] v2 release
I've just given the v2 release a try --- impressive stuff! The
download
Just Worked. No more faffing around with serial downloads every time I want
to
boot the thing, hurrah. It's also really nice to have a real boot loader.
Plus,
thanks to the magic of jffs2 there's an astonishing amount of root
file
system space...
A few minor quibbles:
- Busybox's loadkmap is very stupid and can't cope with my ASCII
keymap
table. (It also wants to refer to the obsolete /dev/vc/0.) I can't figure out
how
to compile my keymap, or even if such a concept makes sense these days.
- Is there an ipkg feed available where I can get iwconfig? I've tried
the
same feed that I use for my NSLU2, but it's not compatible. The
statically
compiled version I built myself (I couldn't make a dynamically
compiled
version to work) is over five megabytes. Ho hum.
- OpenEmbedded wants to rebuild monotone's roster cache. Apparently
this
is going to take about four hours.
Currently I'm trying to build a working zd1201 driver for wireless networking, which is involving some entertaining messing around with cross
compilers -
-- it's suprisingly hard to find a decent cross compiler package for
Debian.
-- +- David Given --McQ-+ "Gaping from its single obling socket was | dg@cowlark.com | scintillating, many fauceted scarlet
emerald..."
| (dg@tao-group.com) | --- Jim Theis, _The Eye of Argon_ (spelling +- www.cowlark.com --+ original)
On Tue, Sep 26, 2006 at 06:13:54PM +0100, Alex Goodyear wrote:
I've also got the V2 release working without any major problems (a fantastic achievement, thanks Jonathan et al). I have a few observations and questions that may help others.
I did stop the NAND backup process because it was incredibly slow. "top" showed that the system wasn't using much CPU but the load was over 2.0 Using "df" also revealed that the USB stick was only having 16 bytes written to it every second or so.
The E3 keyboard is mapped as a standard desktop keyboard so most shifted characters are wrong and symbol-shift characters are ignored.
There was a new keymap posted to the list by David Given a while back, though I had meant to update the one in the driver itself.
I replaced the supplied 2.6.16 uImage with the one from the 2.6.17 directory in the hope that the keyboard mappings would be fixed. It worked but it only provided a shell login for the serial connection unlike the 2.6.16 image which provides both a serial and an onboard E3 login.
2.6.16 is the one that supports most things; later kernels gradually get more of the devices supported by mainline linux-omap rather than having to be patched, but both the keyboard and sound drivers weren't written by me so they tend to end up last on my list of things to forward port. I'm hoping to find some time to get a 2.6.18 kernel up and running in the next week or two, along with the diminishing patch set.
The copying of the Linux image from NAND seems slow, I recall some discussion about this being fixed within u-boot. Does the copy of u-boot in the V2 release contain this fix?
No, the version in V2 is 1.4, but there's a patch against uboot git that adds support but also speeds up the Linux image loading.
J.
No, the version in V2 is 1.4, but there's a patch against uboot git that adds support but also speeds up the Linux image loading.
Umm... Where would that patch be ? :)
I hope it's not u-boot-amsdelta-20060519.diff, because it made no difference for me :(
On Fri, Sep 29, 2006 at 11:03:05AM +0200, Piotr Michniewski wrote:
No, the version in V2 is 1.4, but there's a patch against uboot git that adds support but also speeds up the Linux image loading.
Umm... Where would that patch be ? :)
I hope it's not u-boot-amsdelta-20060519.diff, because it made no difference for me :(
http://www.earth.li/pipermail/e3-hacking/2006-June/000537.html
J.
2006/9/29, Jonathan McDowell noodles@earth.li:
http://www.earth.li/pipermail/e3-hacking/2006-June/000537.html
This one boots really faster! Good job :) I was really tired of waiting for it to boot...
Oh and by the way... I tried to modify the notldr mem layout, because there's no sense in reading 2 MB of Flash during booting, but PBL couldn't find it. I suspect that I have to adjust the CRC somehow in the MEM block, but I have no idea how to do it. Could you help me out a bit ?
On Fri, Sep 29, 2006 at 11:50:37PM +0200, Piotr Michniewski wrote:
2006/9/29, Jonathan McDowell noodles@earth.li:
http://www.earth.li/pipermail/e3-hacking/2006-June/000537.html
This one boots really faster! Good job :) I was really tired of waiting for it to boot...
Oh and by the way... I tried to modify the notldr mem layout, because there's no sense in reading 2 MB of Flash during booting, but PBL couldn't find it. I suspect that I have to adjust the CRC somehow in the MEM block, but I have no idea how to do it. Could you help me out a bit ?
I've never bothered modifying the MEM block. For the checksum look at:
http://www.earth.li/pipermail/e3-hacking/2006-April/000477.html
The 16 byte checksum ends up in offset 6 + 7 of the block IIRC.
J.