On Thu, Apr 21, 2016 at 04:11:25PM +0200, Heiko Schocher wrote:
Hello Tom,
Am 21.04.2016 um 15:53 schrieb Tom Rini:
On Thu, Apr 21, 2016 at 03:50:28PM +0200, Marek Vasut wrote:
On 04/21/2016 03:35 PM, Simon Glass wrote:
Hi,
On 21 April 2016 at 07:25, Marek Vasut marex@denx.de wrote:
On 04/21/2016 03:17 PM, Heiko Schocher wrote:
Hello Marek,
Am 21.04.2016 um 14:51 schrieb Marek Vasut: >On 04/21/2016 02:48 PM, Heiko Schocher wrote: >>suppress a lot of >>"reg or ranges property, but no unit name" warnings, >>through the dtc compiler flag "-Wno-unit_address_vs_reg". >> >>If all DTS are fixed, we can remove this flag again. >> >>Signed-off-by: Heiko Schocher hs@denx.de >>--- >>There is also a solution to suppress warnings from >>the dtc compiler by the "-q" flag, but that would >>suppress all warnings. Not realy what I want. > >Yep > >>With this patch and patch: >>http://patchwork.ozlabs.org/patch/609150/ >> >>travis build passes, see: >>https://travis-ci.org/hsdenx/u-boot/builds/124723016 >> >> >> arch/arc/dts/Makefile | 2 ++ >> arch/arm/dts/Makefile | 3 ++- >> arch/microblaze/dts/Makefile | 2 ++ >> arch/mips/dts/Makefile | 3 ++- >> arch/nios2/dts/Makefile | 2 ++ >> arch/powerpc/dts/Makefile | 2 ++ >> arch/sandbox/dts/Makefile | 2 ++ >> arch/x86/dts/Makefile | 2 ++ >> 8 files changed, 16 insertions(+), 2 deletions(-) > >Isn't there some common place in scripts/ or so where we can disable >this warning using an one-liner ?
I don;t know ... but I prefer to disable this per arch .. so we can enable the check back if one arch is fixed ...
In my opinion, we should stick to the same behavior Linux does. Ccing a few more people.
Wouldn't it be better to fix the problems?
My impression was that these warnings are just the result of over-eagerness of DTC, that's why Linux prints them only if you increase the W= (warning) verbosity. I might be wrong tho.
They are minor problems. For the vast majority of the dts files we have, the fixes will come in via re-syncs with the kernel and in at least some cases it's not just a simple regex but also "oh, lets give things better names". With respect to dts files that we really do own (ie x86) yes, we should fix them.
So, this patch from me could be still an option?
http://patchwork.ozlabs.org/patch/610866/
(at least for the sandbox fixes?) (I have a v2 where I worked in the comments from Bin ...)
Yes, but I want to make sure we think about the renames and aren't trying to silence the warning. I'm sorry but the MIPS "fixes" have me leery of other blind changes.