On 2011-02-19 1:49 PM, Peter Lebbing wrote: > Hello Felix, > > You wrote the adm6996.c driver. The driver "triggered" on the Chip Identifier > of > my ADM6996M chip, so I wrote VLAN support for that chip, supposing that all > chips identified by that same Chip Identifier would probably be similar. > > The detection code masked the "Version Number" and only matched on "Product > Code", by the way (terminology from ADM6996M datasheet). > > It turned out the driver also detected the ADM6996FC as supported, but if I > look > up a datasheet for the F model (seems FC is low-power version of the F), the > chip identifier is quite different (other register address, other product > code). > > There is something odd going on here. If I look at the datasheets, it seems > the > detection code you wrote would match on the M model, and not on the F model. > But > the Kconfig says it "currently supports the ADM6996F switch". Could you tell > me > what you based the detection code on, if you still remember? It's based on information from the FC datasheet, as well as IDs read out from a device that had an FC chip on it.
> PS: By the way, I could resubmit the patch in a series of 3, if that is > preferable. I only realised it could be split after my initial submission. > > Patch 1 could reorder the current code > Patch 2 could implement the actual chip support > Patch 3 could implement the swconfig interface I think doing everything in one patch is OK as well, considering that my original driver was unfinished and didn't really do anything useful. I would like to see FC vs M detection sorted out though, so that it can let incompatible devices fall back to Generic PHY if necessary. - Felix _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
