On Wed, Sep 10, 2014 at 10:01:43AM +0200, Arnd Bergmann wrote:
On Wednesday 10 September 2014 06:22:05 Build bot for Mark Brown wrote:
defconfigs with issues (other than build errors): 2 warnings 0 mismatches : arm64-allnoconfig 18 warnings 0 mismatches : arm-allmodconfig 1 warnings 0 mismatches : arm-multi_v7_defconfig 7 warnings 0 mismatches : arm64-allmodconfig
Seriously, is that all?
Are you filtering out old warnings to just see regressions or is that all that is left to fix now?
No filtering, we're actually doing pretty well (unless you turn on sparse for the builds at which point it gets miserable). Or the compiler is terrible at catching errors though I didn't change that recently. I imagine it will get worse on at least ARMv8 when PCI is enabled.
arm64-allnoconfig : PASS, 0 errors, 2 warnings, 0 section mismatches
Warnings: ../drivers/irqchip/irq-gic-v3.c:132:13: warning: 'gic_write_sgi1r' defined but not used [-Wunused-function] ../drivers/irqchip/irq-gic-v3.c:203:12: warning: 'gic_peek_irq' defined but not used [-Wunused-function]
I haven't seen this one before, should be easy to fix.
It's fixed in -next, I sent the patches just before the merge window so they didn't make it into mainline yet.
arm-allmodconfig : PASS, 0 errors, 18 warnings, 0 section mismatches
Warnings: ../arch/arm/mach-cns3xxx/pcie.c:311:1: warning: the frame size of 1072 bytes is larger than 1024 bytes [-Wframe-larger-than=]
I guess we should try to find a new cns3xxx maintainer, Anton no longer has access to the hardware as far as I know, and the patch I made for this needs to be tested.
Right, or my hack is at least no less safe than the current code.