On Wed, Sep 04, 2019 at 01:30:02PM -0500, Joe Hershberger wrote:
On Wed, Sep 4, 2019 at 1:01 PM Tom Rini trini@konsulko.com wrote:
On Tue, Sep 03, 2019 at 10:04:42AM +0200, Wolfgang Denk wrote:
Dear Tom,
In message a78f0b04-c3f7-45d5-e9ac-90522dbefc2e@denx.de Heiko Schocher wrote:
I am just testing U-Boot Environment flags and they do not work anymore with current mainline U-Boot ...
...
reason is your commit:
commit 7d4776545b0f8a8827e5d061206faf61c9ba6ea9 Author: Patrick Delaunay patrick.delaunay@st.com Date: Thu Apr 18 17:32:49 2019 +0200
env: solve compilation error in SPLLooking into the history of this, I wonder if we could / should have prevented this.
Looking over my scripts, yes, I overlooked the problem. The 'edison' platform shows the same issue that Heiko's platform does but I overlooked the size change. I'm modifying my script currently so it will show more details and this should jump out more rather than the size noise of "changes in a general area". Now interesting enough, sandbox didn't blow up here but does also enable the env flags options.
As far as I can see, Patrick's patch series has not been reviewed by others, probably because general intetest in STM32 is not that big at the moment. I can see no Acked-by:, Reviewed-by: nor Tested-by: tags - nothing.
The whole patch series was then pulled from the u-boot-stm repository.
However, there was not only STM related code in there. There were changes to common code like the environment handling. common code was changed without review and without testing.
To be clear, it was tested, but sadly the environment flags code is not heavily used / enabled. More in a moment.
Are there ways to prevent this?
Yes, we can appeal to the custodians to be more careful, but I assume they are already doing their best.
It might have even been better if this had been a sub-system with a clear maintainer, but there is no such person for the environment code.
How can we prevent this in the future?
Should we define "interested developers" for such areas that have no custodian (the "Designated reviewer") entry in the MAINTAINERS file could be used for this, for example)?
This, along with some other environment related patches were things I was keeping an eye on to see if perhaps Joe would have had time to look at before it went in (as the env flag stuff came from him). I also try
I wasn't aware of it as I wasn't Cc'ed on this series. I generally don't have time to troll the list in general, which is a bit of a problem since I also missed the discussions on the UEFI env changes, some of which are already in, and are not how I would have implemented it. I only found out that it was in work from Grant Likely at his talk in San Diego.
and make use of the "Needs Review / ACK" flag in patchwork for things that stand out. Looking over the merge contents again, that particular one would not have.
So, things that would help in the future:
- An explicit environment maintainer
I would gladly volunteer for this role if Wolfgang would co-maintain to keep me in line. He seems to have an uncanny ability to keep all the cases in his head.
Wolfgang, what do you say? It's certainly an area we could use a custodian in.