Re: U-Boot Digest, Vol 158, Issue 63
On Thu, Aug 05, 2021 at 04:31:19PM -0400, Tom Rini wrote: > On Tue, Aug 03, 2021 at 09:21:57AM -0400, Oleksandr G Zhadan wrote: > > > Hi Tom, > > > > Thanks for the quick notes. > > > > As I already mentioned, unfortunately, I did not notice any "warnings" in > > time - we haven’t been building the bootloader recently, but our customer is > > using it. > > > > My only concern is the removal of support for the uCP1020 board. This > > product is not dead and we would like to leave it on the u-boot host. > > > > Please give us a couple of weeks to fix this (after I get back from > > vacation). > > I think this, along with taking most of Simon's series (such tat uCP1020 > still builds) is the best path forward right now. I would rather see a > series you've tested to update the platform soon than to attempt it > myself now. So it's been a little over a month. Where do things stand on this? Thanks! -- Tom signature.asc Description: PGP signature
Re: U-Boot Digest, Vol 158, Issue 63
On Tue, Aug 03, 2021 at 09:21:57AM -0400, Oleksandr G Zhadan wrote: > Hi Tom, > > Thanks for the quick notes. > > As I already mentioned, unfortunately, I did not notice any "warnings" in > time - we haven’t been building the bootloader recently, but our customer is > using it. > > My only concern is the removal of support for the uCP1020 board. This > product is not dead and we would like to leave it on the u-boot host. > > Please give us a couple of weeks to fix this (after I get back from > vacation). I think this, along with taking most of Simon's series (such tat uCP1020 still builds) is the best path forward right now. I would rather see a series you've tested to update the platform soon than to attempt it myself now. -- Tom signature.asc Description: PGP signature
Re: U-Boot Digest, Vol 158, Issue 63
Hi Tom,
Thanks for the quick notes.
As I already mentioned, unfortunately, I did not notice any "warnings"
in time - we haven’t been building the bootloader recently, but our
customer is using it.
My only concern is the removal of support for the uCP1020 board. This
product is not dead and we would like to leave it on the u-boot host.
Please give us a couple of weeks to fix this (after I get back from
vacation).
Thank you,
Oleksandr
On 7/28/21 12:43 PM, Tom Rini wrote:
On Wed, Jul 28, 2021 at 12:22:57PM -0400, Oleksandr G Zhadan wrote:
Hello,
Please see inline.
On 7/26/21 9:40 AM, u-boot-requ...@lists.denx.de wrote:
Message: 13
Date: Mon, 26 Jul 2021 07:34:06 -0600
From: Simon Glass
To: U-Boot Mailing List
Cc: Tom Rini, Simon Glass,
Albert Aribaud, Andy Fleming
, Joe Hershberger, Marek
Vasut, Mario Six, Oleksandr
Zhadan and Michael Durrant, Pavel
Herrmann, Priyanka Jain
, Rob Herring, Stefan
Roese, Stefano Babic, Wolfgang Denk
Subject: [PATCH 00/33] pci: Drop all pre-driver model code
Message-ID:<20210726133440.634682-1-...@chromium.org>
The hard work to actually enable DM_PCI everywhere was done recently. This
series attempts to drop most of the code that it no-longer needed now that
PCI has been converted to driver model.
It also drops the UCP1020 board since it has various unique build issues.
What issues are we talking about?Please name it.
Today's u-boot clone builds and runs fine in our module without any warnings
except:
git clone https://source.denx.de/u-boot/u-boot.git u-boot-20210728
cd u-boot-20210728
make UCP1020_defconfig
CROSS_COMPILE=powerpc-linux- make 2>222
> cat 222
= WARNING ==
This board does not use CONFIG_DM. CONFIG_DM will be
compulsory starting with the v2020.01 release.
Failure to update may result in board removal.
See doc/driver-model/migration.rst for more info.
= WARNING ==
This board does not use CONFIG_DM_ETH (Driver Model
for Ethernet drivers). Please update the board to use
CONFIG_DM_ETH before the v2020.07 release. Failure to
update by the deadline may result in board removal.
See doc/driver-model/migration.rst for more info.
===
It doesn't even support driver model so it seems reasonable to just remove
it.
1. Sure it's my fault that I didn't notice the warnings that CONFIG_DM is
the only choice to build and run u-boot.
2. Why is CONFIG_DM still present in `make menuconfig` options?
3. We only use u-boot as the first stage bootloader to start the kernel. We
need minimal device controllers support. We don't need anything like PCI,
USB, etc.
4. At least you need to add doc/driver-model/migration.rst to the source
tree from https://source.denx.de/u-boot/u-boot.git if you point to it in
your warnings.
And
5. I have seen NO reason to remove ANY support board without REAL reasons to
do so.
So, unfortunately this was the only board that was not triggering the
PCI migration warning as well because of how it was using the legacy PCI
subsystem. For today, if you want to continue to be in mainline, just
removing the PCI and other options that you don't use, and submitting
that patch is sufficient.
SUMMARY:
1. UCP1020_defconfig does not have any compiler warnings on build (at least
2021.07.28 clone).
2. If now the only choice is CONFIG_DM, then it should be removed as an
option from "make menuconfig", and we will definitely fix the related
warnings/errors.
3. IMPO board support can be removed if it breaks when built and/or breaks
any other builds.
Note that CONFIG_DM is still an option because it won't be until after
the v2022.01 window (when that migration warning you noted above) will
have been present for about 3 years. It's still an option as we're down
to 15 boards (UCP included) that need migration still. Having just
counted that, I'm quite likely to fire off a separate email now and see
if there's interest in these boards being updated or not, now that we're
down to so few boards.
--
Oleksandr Zhadan
ol...@arcturusnetworks.com
416.621.0125 x.235
}|{/\|)/\|-|
Re: U-Boot Digest, Vol 158, Issue 63
On Wed, Jul 28, 2021 at 12:22:57PM -0400, Oleksandr G Zhadan wrote: > Hello, > > Please see inline. > > On 7/26/21 9:40 AM, u-boot-requ...@lists.denx.de wrote: > > Message: 13 > > Date: Mon, 26 Jul 2021 07:34:06 -0600 > > From: Simon Glass > > To: U-Boot Mailing List > > Cc: Tom Rini, Simon Glass, > > Albert Aribaud, Andy Fleming > > , Joe Hershberger, Marek > > Vasut, Mario Six, Oleksandr > > Zhadan and Michael Durrant, Pavel > > Herrmann, Priyanka Jain > > , Rob Herring, Stefan > > Roese, Stefano Babic, Wolfgang Denk > > > > Subject: [PATCH 00/33] pci: Drop all pre-driver model code > > Message-ID:<20210726133440.634682-1-...@chromium.org> > > > > The hard work to actually enable DM_PCI everywhere was done recently. This > > series attempts to drop most of the code that it no-longer needed now that > > PCI has been converted to driver model. > > > It also drops the UCP1020 board since it has various unique build issues. > > What issues are we talking about?Please name it. > > Today's u-boot clone builds and runs fine in our module without any warnings > except: > > > git clone https://source.denx.de/u-boot/u-boot.git u-boot-20210728 > > > cd u-boot-20210728 > > > make UCP1020_defconfig > > > CROSS_COMPILE=powerpc-linux- make 2>222 > > > cat 222 > = WARNING == > This board does not use CONFIG_DM. CONFIG_DM will be > compulsory starting with the v2020.01 release. > Failure to update may result in board removal. > See doc/driver-model/migration.rst for more info. > > = WARNING == > This board does not use CONFIG_DM_ETH (Driver Model > for Ethernet drivers). Please update the board to use > CONFIG_DM_ETH before the v2020.07 release. Failure to > update by the deadline may result in board removal. > See doc/driver-model/migration.rst for more info. > === > > > It doesn't even support driver model so it seems reasonable to just remove > > it. > 1. Sure it's my fault that I didn't notice the warnings that CONFIG_DM is > the only choice to build and run u-boot. > > 2. Why is CONFIG_DM still present in `make menuconfig` options? > > 3. We only use u-boot as the first stage bootloader to start the kernel. We > need minimal device controllers support. We don't need anything like PCI, > USB, etc. > > 4. At least you need to add doc/driver-model/migration.rst to the source > tree from https://source.denx.de/u-boot/u-boot.git if you point to it in > your warnings. > > And > > 5. I have seen NO reason to remove ANY support board without REAL reasons to > do so. So, unfortunately this was the only board that was not triggering the PCI migration warning as well because of how it was using the legacy PCI subsystem. For today, if you want to continue to be in mainline, just removing the PCI and other options that you don't use, and submitting that patch is sufficient. > SUMMARY: > > 1. UCP1020_defconfig does not have any compiler warnings on build (at least > 2021.07.28 clone). > > 2. If now the only choice is CONFIG_DM, then it should be removed as an > option from "make menuconfig", and we will definitely fix the related > warnings/errors. > > 3. IMPO board support can be removed if it breaks when built and/or breaks > any other builds. Note that CONFIG_DM is still an option because it won't be until after the v2022.01 window (when that migration warning you noted above) will have been present for about 3 years. It's still an option as we're down to 15 boards (UCP included) that need migration still. Having just counted that, I'm quite likely to fire off a separate email now and see if there's interest in these boards being updated or not, now that we're down to so few boards. -- Tom signature.asc Description: PGP signature
Re: U-Boot Digest, Vol 158, Issue 63
Hello,
Please see inline.
On 7/26/21 9:40 AM, u-boot-requ...@lists.denx.de wrote:
Message: 13
Date: Mon, 26 Jul 2021 07:34:06 -0600
From: Simon Glass
To: U-Boot Mailing List
Cc: Tom Rini, Simon Glass,
Albert Aribaud, Andy Fleming
, Joe Hershberger, Marek
Vasut, Mario Six, Oleksandr
Zhadan and Michael Durrant, Pavel
Herrmann, Priyanka Jain
, Rob Herring, Stefan
Roese, Stefano Babic, Wolfgang Denk
Subject: [PATCH 00/33] pci: Drop all pre-driver model code
Message-ID:<20210726133440.634682-1-...@chromium.org>
The hard work to actually enable DM_PCI everywhere was done recently. This
series attempts to drop most of the code that it no-longer needed now that
PCI has been converted to driver model.
It also drops the UCP1020 board since it has various unique build issues.
What issues are we talking about?Please name it.
Today's u-boot clone builds and runs fine in our module without any
warnings except:
> git clone https://source.denx.de/u-boot/u-boot.git u-boot-20210728
> cd u-boot-20210728
> make UCP1020_defconfig
> CROSS_COMPILE=powerpc-linux- make 2>222
> cat 222
= WARNING ==
This board does not use CONFIG_DM. CONFIG_DM will be
compulsory starting with the v2020.01 release.
Failure to update may result in board removal.
See doc/driver-model/migration.rst for more info.
= WARNING ==
This board does not use CONFIG_DM_ETH (Driver Model
for Ethernet drivers). Please update the board to use
CONFIG_DM_ETH before the v2020.07 release. Failure to
update by the deadline may result in board removal.
See doc/driver-model/migration.rst for more info.
===
It doesn't even support driver model so it seems reasonable to just remove
it.
1. Sure it's my fault that I didn't notice the warnings that CONFIG_DM
is the only choice to build and run u-boot.
2. Why is CONFIG_DM still present in `make menuconfig` options?
3. We only use u-boot as the first stage bootloader to start the kernel.
We need minimal device controllers support. We don't need anything like
PCI, USB, etc.
4. At least you need to add doc/driver-model/migration.rst to the source
tree from https://source.denx.de/u-boot/u-boot.git if you point to it in
your warnings.
And
5. I have seen NO reason to remove ANY support board without REAL
reasons to do so.
SUMMARY:
1. UCP1020_defconfig does not have any compiler warnings on build (at
least 2021.07.28 clone).
2. If now the only choice is CONFIG_DM, then it should be removed as an
option from "make menuconfig", and we will definitely fix the related
warnings/errors.
3. IMPO board support can be removed if it breaks when built and/or
breaks any other builds.
Thank you,
Oleks
The DM_PCI option disappears and only PCI is left.
--
Oleksandr Zhadan
ol...@arcturusnetworks.com
416.621.0125 x.235
}|{/\|)/\|-|
