On Mon, Jun 21, 2010 at 5:59 AM, Myles Watson mylesgw@gmail.com wrote:
On Sun, Jun 20, 2010 at 8:11 PM, Ward Vandewege ward@gnu.org wrote:
Hi Myles,
Everything seems fine with either patch - but there are some differences in the boot output.
I also ran the 'sensors' command.
Output here:
http://ward.vandewege.net/coreboot/s2881/20100617-myles/
I ran 4 tests: stock r5635 (head), stock r5632 (revision prior to this changeset), r5635 + patch 1 and r5635 + patch 2.
Thanks for testing.
It looks like only 5632 has the "ADT7463 properly initialized" message. One problem is that patch 2 was meant to be applied after patch 1, so the device didn't end up in the tree for that run. I'll have to think about why the only message from the new device with patch 1 is "I2C: 00:d0 missing read_resources" For some reason it doesn't look like it got the correct ops.
I wonder why the temperature values look right in all cases. Does it need to be cold booted in order for the initialization to be needed?
The ADM1027 doesn't expect to have children, so it has no scan_bus method. I had thought that the ADM1027 was some kind of a controller for the ADT4763, but it looks like the same type of device. Is there really an ADM1027 on your board? I don't see it in your sensors output.
So... the first two patches are the same as before. The third patch adds a scan_bus method to the ADM1027 so that the ADT4763 can be initialized, and the fourth patch replaces the ADM1027 with the ADT4763 in the device tree, and removes the third patch.
I'd be interested in head + 1 + 2 + 3, and head + 1 + 2 + 3 + 4.
There may be some initialization that was done by the ADM1027 code that needs to be moved to the ADT7463 code, since both were being run.
So 1 + 2 + 3 should give behavior that is closer to the original code, but if there is no ADM1027, the right thing to do is 1 + 2 + 3 + 4 + some ADM1027 code if necessary.
Signed-off-by: Myles Watson mylesgw@gmail.com
Thanks, Myles