Around Sun 09 Nov 2014 22:33:53 -0800 or thereabout, Guenter Roeck wrote: > On 11/09/2014 10:20 PM, Hans-Christian Egtvedt wrote: >>Around Sun 09 Nov 2014 20:49:31 -0800 or thereabout, Guenter Roeck wrote: >>>Hi, >>> >>>v3.17-stable-queue fails to buildavr32:atngw100mkii_evklcd101_defconfig >>>with the following link error. >>> >>>vsprintf.c:(.text+0x16dc): relocation truncated to fit: R_AVR32_16N_PCREL against symbol `_ctype' defined in .text section in lib/lib.a(ctype.o) >>>make: *** [vmlinux] Error 1 >> >>Something has grown too large for a 16-bit relative jump. Are you building >>with -Os? >> > Just whatever the default build does; no special configuration. Default should set CONFIG_CC_OPTIMIZE_FOR_SIZE=y in kconfig, you could always double check. >>>Culprit is commit "random: add and use memzero_explicit() for clearing data". >>>I don't see the error upstream. I have no idea why it fails to build in stable-queue, >>>nor why the build is fine with the upstream kernel. >>> >>>avr32:defconfig and avr32:merisc_defconfig builds pass without error. >> >>I would suspect atngw100mkii_evklcd101_defconfig enables more stuff than the >>other defconfigs, which might explain why. I typically build with only the >>set of features I need. >> >>Sadly, no updated toolchain will be available to my knowledge. >> > Still, it is odd that both the upstream kernel and -next compile. > Any idea why that might be the case ? In the past we have had a couple of cases like this, but then it would typically be assembler code with a badly picked instruction. This looks to be generic C code. I will grab the stable branch and see if I can spot anything obvious. -- Hans-Christian Egtvedt -- To unsubscribe from this list: send the line "unsubscribe stable" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html