Hi,
On 16 November 2015 at 01:57, Heiko Schocher hs@denx.de wrote:
Hello Andreas,
Am 16.11.2015 um 09:48 schrieb Andreas Bießmann:
Hi Heiko,
On 11/16/2015 07:29 AM, Heiko Schocher wrote:
Hello Simon,
I see an compile error for avr32 on travis.org for current mainline U-Boot, see [1].
git bisecting it, with the information that last working avr32 build on travis.org is [2], shows:
pollux:u-boot hs [(kein Branch, binäre Suche begonnen bei 20151116)] $ git bisect log git bisect start # good: [a5325cd5e91f77a2214e80198ae31c1d8b7e7c3c] configs: Remove CONFIG_SERIAL_MULTI git bisect good a5325cd5e91f77a2214e80198ae31c1d8b7e7c3c # bad: [ae27120c31d58b8bb694d9155bcffdcfae8552a6] Merge git://git.denx.de/u-boot-dm git bisect bad ae27120c31d58b8bb694d9155bcffdcfae8552a6 # bad: [bae34335e8aef46d7103907580c40f18f1876bb0] exynos: Remove unneeded device tree control #ifdefs git bisect bad bae34335e8aef46d7103907580c40f18f1876bb0 # bad: [f615e6a64d23952d3c68b9ac92b4ef5fa5ebdf8a] dm: power: Add support for S5M8767 regulators git bisect bad f615e6a64d23952d3c68b9ac92b4ef5fa5ebdf8a # bad: [f1ac35b7a6c6bc878d473b5955370b1cc49dba54] exynos: dts: Sync up I2C ports with the kernel git bisect bad f1ac35b7a6c6bc878d473b5955370b1cc49dba54 # good: [7fc65bcf8a0a5463db86efbb273a40448c845efc] dm: i2c: Move definitions to the top of the header file git bisect good 7fc65bcf8a0a5463db86efbb273a40448c845efc # bad: [b725dc458f9db44957d4f541d05defe178dda7b5] i2c: Add a mux for GPIO-based I2C bus arbitration git bisect bad b725dc458f9db44957d4f541d05defe178dda7b5 # bad: [3d1957f0ea0133ec06f9c6fd85dc1acdf66ad29c] dm: i2c: Add support for multiplexed I2C buses git bisect bad 3d1957f0ea0133ec06f9c6fd85dc1acdf66ad29c # good: [df358c6beca8f7cc2f2b67c54f0bf255ebd5cc22] dm: i2c: Add a function to transfer messages git bisect good df358c6beca8f7cc2f2b67c54f0bf255ebd5cc22 # first bad commit: [3d1957f0ea0133ec06f9c6fd85dc1acdf66ad29c] dm: i2c: Add support for multiplexed I2C buses pollux:u-boot hs [(kein Branch, binäre Suche begonnen bei 20151116)] $
Hmm... it seems, that introducing the subdir "muxes" leads into this error for avr32 ... I just tried the following patch:
I know about this but had no time to investigate it deeper. It seems this is a toolchain problem ... and there is no way to get another toolchain.
diff --git a/drivers/i2c/Makefile b/drivers/i2c/Makefile index 32198bd..811ad9b 100644 --- a/drivers/i2c/Makefile +++ b/drivers/i2c/Makefile @@ -40,4 +40,4 @@ obj-$(CONFIG_SYS_I2C_UNIPHIER) += i2c-uniphier.o obj-$(CONFIG_SYS_I2C_UNIPHIER_F) += i2c-uniphier-f.o obj-$(CONFIG_SYS_I2C_ZYNQ) += zynq_i2c.o
-obj-y += muxes/ +obj-$(CONFIG_I2C_MUX) += muxes/ pollux:u-boot hs [20151116] $
and the error "drivers/i2c/built-in.o: warning: input is not relaxable" is gone ...
Do you have an idea, why this breaks? Is above patch a valid fix? (I checked, that all boards using CONFIG_I2C_ARB_GPIO_CHALLENGE have also set CONFIG_I2C_MUX)
For me this is a valid fix. Would you send a proper patch?
Yes, of course, but I want to wait for Simons comment, thanks!
Yes that fix seems good to me. I'm not quite sure what is going on though.
Regards, Simon