Skip to content

RFC kconfig: warn if an unknown symbol is selected

Paul Bolle linux-kbuild lkml 1 version New

Get series with b4
b4 am 1411769296.15241.36.camel@x220

Versions newest first

  1. v1 RFC Latest0 of 1 patch 0 repliesNew

Patches in v1 No applies recorded yet

The patch the cover letter announces is not in this archive.

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.

Read the v1 thread

Cover letter [PATCH 0/1] [RFC] kconfig: warn if an unknown symbol is selected

This is an attempt to make a recurring kconfig issue more visible:
select statements for unknown, or unreachable, Kconfig symbols. See, I've submitted quite a few patches that remove statements like
select BOGUS

One of the annoying properties of kconfig is that select statements like that are silently treated as a nop. That makes them surprisingly easy to stay unnoticed. Some of them are caught by my local "800 line perl monster", but not all. But perhaps silently skipping these statements is actually a feature. I think it is not.

Anyhow, I've tweaked kconfig to warn about them. Then I ran kconfig on all (500+) in tree defconfig files as of next-20140926. These are the select statements warns about when doing that.

1) arch/arm/Kconfig:429:warning: 'ARCH_EFM32' selects unknown symbol 'NO_DMA'

Unknown on arm. Patch submitted.

2) arch/arm/mach-meson/Kconfig:11:warning: 'MACH_MESON6' selects unknown symbol 'MESON6_TIMER'

Clearly bogus select. But Kconfig symbol MESON6_TIMER is apparently queued.

3) arch/arm/mach-mvebu/Kconfig:40:warning: 'MACH_ARMADA_375' selects unknown symbol 'ARM_ERRATA_753970' arch/arm/mach-mvebu/Kconfig:55:warning: 'MACH_ARMADA_38X' selects unknown symbol 'ARM_ERRATA_753970'

Typo. Patch is being discussed for some time now.

4) arch/arm/mach-shmobile/Kconfig:39:warning: 'ARCH_SHMOBILE_MULTI' selects unknown symbol 'ARCH_HAS_OPP'

Unfinished Kconfig cleanup. Patches submitted and/or queued.

5) arch/hexagon/Kconfig:22:warning: 'HEXAGON' selects unknown symbol 'NO_IOPORT_MAP'

Unknown on hexagon. Patch submitted.

6) arch/powerpc/platforms/44x/Kconfig:217:warning: 'AKEBONO' selects unknown symbol 'IBM_EMAC_RGMII_WOL' arch/powerpc/platforms/44x/Kconfig:223:warning: 'AKEBONO' selects unknown symbol 'MMC_SDHCI_OF_476GTR'

Clearly bogus selects. One symbol might be queued. Not sure why the other select hasn't been removed.

7) arch/score/Kconfig:24:warning: 'ARCH_SCORE7' selects unknown symbol 'SYS_SUPPORTS_32BIT_KERNEL' arch/score/Kconfig:28:warning: 'MACH_SPCT6600' selects unknown symbol 'SYS_SUPPORTS_32BIT_KERNEL' arch/score/Kconfig:32:warning: 'SCORE_SIM' selects unknown symbol 'SYS_SUPPORTS_32BIT_KERNEL'

Bogus. Submitted a patch long ago. Nothing happened.

8) drivers/acpi/Kconfig:165:warning: 'ACPI_PROCESSOR' selects unknown symbol 'CPU_IDLE'

Unreachable for eg ia64. Perhaps fixed by:

drivers/acpi/Kconfig +1 -1
@@ -162,7 +162,7 @@ config ACPI_DOCK
 config ACPI_PROCESSOR
        tristate "Processor"
        select THERMAL
-       select CPU_IDLE
+       select CPU_IDLE if (ARM || ARM64 || MIPS || PPC || X86 || SUPERH)
        default y
        help
          This driver installs ACPI as the idle handler for Linux and uses

9) drivers/irqchip/Kconfig:16:warning: 'ARM_GIC_V3' selects unknown symbol 'MULTI_IRQ_HANDLER' drivers/irqchip/Kconfig:8:warning: 'ARM_GIC' selects unknown symbol 'MULTI_IRQ_HANDLER'

MULTI_IRQ_HANDLER is arm specific. So arm64 builds trigger this. Suggested fix:

drivers/irqchip/Kconfig +1 -2
@@ -5,7 +5,7 @@ config IRQCHIP
 config ARM_GIC
        bool
        select IRQ_DOMAIN
-       select MULTI_IRQ_HANDLER
+       select MULTI_IRQ_HANDLER if ARM
 
 config GIC_NON_BANKED
        bool
@@ -13,7 +13,6 @@ config GIC_NON_BANKED
 config ARM_GIC_V3
        bool
        select IRQ_DOMAIN
-       select MULTI_IRQ_HANDLER
 
 config ARM_NVIC
        bool

10) lib/Kconfig.debug:1008:warning: 'DEBUG_ATOMIC_SLEEP' selects unknown symbol 'PREEMPT_COUNT'

Not all architectures have PREEMPT_COUNT. Maybe fixed by:

lib/Kconfig.debug +1 -1
@@ -1005,7 +1005,7 @@ config DEBUG_LOCKDEP
 
 config DEBUG_ATOMIC_SLEEP
        bool "Sleep inside atomic section checking"
-       select PREEMPT_COUNT
+       select PREEMPT_COUNT if !(ALPHA || FRV || HEXAGON || UML)
        depends on DEBUG_KERNEL
        help
          If you say Y here, various routines which may sleep will become very

Not sure, however.

11) sound/pci/Kconfig:158:warning: 'SND_AZT3328' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:28:warning: 'SND_ALS300' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:466:warning: 'SND_EMU10K1' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:482:warning: 'SND_EMU10K1X' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:516:warning: 'SND_ES1938' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:528:warning: 'SND_ES1968' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:53:warning: 'SND_ALI5451' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:615:warning: 'SND_ICE1712' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:703:warning: 'SND_MAESTRO3' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:812:warning: 'SND_SONICVIBES' selects unknown symbol 'ZONE_DMA' sound/pci/Kconfig:824:warning: 'SND_TRIDENT' selects unknown symbol 'ZONE_DMA'

Tricky. See commit 80ab8eae70e5 ("ALSA: Enable CONFIG_ZONE_DMA for smaller PCI DMA masks"). Nice quote from Andrew Morton about adding these selects (instead of fixing the underlying problem):

    But then we'll just forget about it :(

(https://bugzilla.kernel.org/show_bug.cgi?id=68221#c19 ).

Comments welcome!

Paul Bolle (1):
kconfig: warn if an unknown symbol is selected

 scripts/kconfig/conf.c | 41 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 41 insertions(+)
Signature · 1 line
1.9.3