mainline vs. QorIQ SDK for production with FSL SoCs...
ppc at mindchasers.com
Fri Jun 12 13:20:32 AEST 2015
Please correct me if I'm mistaken and / or provide your own insight on
I'm under the impression that it's better to develop with the mainline
(eventually freezing for production on a stable release / branch ) than
use the QorIQ SDK if the needed code (e.g., DPAA drivers) has been
upstreamed and is included in the mainline.
It's my perception that it is preferable to develop with the mainline
because there are more active open-source developers working with this
code and actively reviewing / testing it than with the SDK and
backported or stale drivers (e.g., WiFi)
I'm sure that there are at least dozens of FSL developers working with
the SDK on a daily basis, but this is behind the FSL firewall, and open
source developers, such as myself, don't have access to FSL patches
except for every six months when a new SDK is released.
It seems to me that there will be a much better chance of hearing about
new bugs from others and getting help debugging problems if we're using
the mainline. But again, the caveat is that the necessary kernel
technology for the product is included in the mainline ( SDK contains
some source for drivers and libraries only available under an FSL EULA ).
FYI - at this point, we're testing & developing with both the SDK 3.12
kernel and the mainline kernel, but it would be nice to drop the former.
Lastly, for something like the p1010, we have found it to stable with
the mainline for quite some time.
More information about the Linuxppc-dev