On Tue, Sep 10, 2024 at 12:41:11PM -0600, Simon Glass wrote:
Hi Tom,
On Mon, 2 Sept 2024 at 09:39, Tom Rini trini@konsulko.com wrote:
On Sun, Sep 01, 2024 at 02:09:39PM -0600, Simon Glass wrote:
Hi Tom,
On Thu, 22 Aug 2024 at 08:10, Tom Rini trini@konsulko.com wrote:
On Wed, Aug 21, 2024 at 09:00:25PM -0600, Simon Glass wrote:
Hi Tom,
On Fri, 16 Aug 2024 at 17:53, Simon Glass sjg@chromium.org wrote:
Hi Tom,
On Fri, 16 Aug 2024 at 11:22, Tom Rini trini@konsulko.com wrote: > > On Thu, Aug 15, 2024 at 01:57:45PM -0600, Simon Glass wrote: > > > For most boards, the device-tree compiler is built in-tree, ignoring the > > system version. Add a special option to skip this build. This can be > > useful when the system dtc is up-to-date, as it speeds up the build. > > > > Signed-off-by: Simon Glass sjg@chromium.org > > --- > > > > (no changes since v1) > > > > tools/buildman/builder.py | 27 ++++++++++++++++++++++++++- > > tools/buildman/builderthread.py | 4 ++-- > > tools/buildman/buildman.rst | 3 +++ > > tools/buildman/cmdline.py | 2 ++ > > tools/buildman/control.py | 3 ++- > > tools/buildman/test.py | 31 +++++++++++++++++++++++++++++++ > > 6 files changed, 66 insertions(+), 4 deletions(-) > > We should probably do this more generically, outside of buildman. We > have scripts/dtc-version.sh and if the system version isn't new enough > (and we just need to define whatever the minimum version is), then we > build our (not currently that new anymore) dtc instead.
Yes I think I did a patch for that ages ago [1], but it was rejected.
I'd be very happy for it to be applied as I think it is a better solution than this one.
I see that some poor sod tried to do this in Linux this morning.
Any thoughts on that patch?
I'm open to re-considering [1] again, but we need to handle the warning problem first. That means...
Also I do see one problem. Newer dtc version produce a lot of warnings, which causes CI to fail. So if we always use the newest version, people are going to see a ton of warnings when they run locally. Am I missing something here?
Well, it would be great to get our Kbuild logic anywhere close to in-sync again with upstream. But syncing up the disabling warning flags shouldn't be too hard.
So, coming back to this patch, the nice thing about it is that it is deterministic. So people who build U-Boot and don't want funky behaviour will be happy. It will use the internal dtc by default. To use the external one, you must provide an option.
This patch only affects buildman, but as you can see the mechanism it uses is to set the DTC variable, which people can do without buildman. It's just a convenience, but useful enough to have a flag, I believe.
Wait, that's right, we have DTC as a thing that can be set in the environment, so why do we need something for buildman at all?
It's a convenience. I have found that it speeds things up quite a bit, so I want to enable it most of the time. But we can't do it by default and need it to be optional, I believe.
I wasn't clear, sorry. Why is: $ export DTC=/usr/local/bin/dtc $ tools/buildman/buildman ...
Enough?