On Mon, Nov 30, 2009 at 2:13 AM, Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net wrote:
On 30.11.2009 10:19, David Bartley wrote:
The way I figured this out was by picking the chip closest to the BIOS (the VT8237A southbridge) and inverting each of the GPIO lines one at a time. The GPIO stuff was gleaned from the VT8237R datasheet since I couldn't find the 'A one anywhere. All in all, a very fun and educating experience; next step, coreboot!
Thanks for your patch. Overall, it looks fine. Someone with VIA datasheets may want to verify.
Could you please add a Signed-off-by: statement to your patch? If you reply to your own mail with the Signed-off-by: on a separate line, that is sufficient. We need this to create an audit trail of who contributed which code. Details are available at http://www.coreboot.org/Development_Guidelines#Sign-off_Procedure
Done.
On a related note, reading the flash back seems slightly broken; sometimes the read gets a few bytes wrong.
This is very interesting. If you read a few times (without writing/erasing in between), do the results change? If not, this points to an error during writing and we should make sure that this is fixed. An error during reading is much more difficult to diagnose. Please note that latest flashrom (0.9.1-r790) has some improved write checks and should detect write errors better. If you see a pattern in the broken addresses (e.g. always the same, wandering by x bytes, common prefix, common suffix), please tell us about it as well.
I noticed this before I had erase/write working and can reproduce this if I reboot and try reading. I can no longer reproduce this after flashing though.
-- David