aboutsummaryrefslogtreecommitdiffstats
path: root/arch/ia64/pci/pci.c
diff options
context:
space:
mode:
authorRussell King <rmk+kernel@arm.linux.org.uk>2009-12-12 11:20:57 -0500
committerRussell King <rmk+kernel@arm.linux.org.uk>2009-12-13 14:38:08 -0500
commit1937f5b91833e2e8e53bcc821fc7a5fbe6ccb9b5 (patch)
tree0d3094976fcec5bd82db27144224f7ec7c9e06c3 /arch/ia64/pci/pci.c
parentc7baab5d1e97437a2fca63b71d467f193dbebb02 (diff)
ARM: fix sa1100 build
Fix: arch/arm/mach-sa1100/generic.c:117: error: redefinition of 'cpufreq_get' include/linux/cpufreq.h:299: error: previous definition of 'cpufreq_get' was here cpufreq_get() is used on these platforms to tell drivers what the CPU frequency is, and therefore the bus frequency - which is critical for setting the PCMCIA and LCD timings. Adding ifdefs to these drivers to select cpufreq_get() or some other interface adds confusion. Making these drivers use some other interface for the normal paths and cpufreq stuff for the cpufreq notifier is insane as well. (Why x86 can't provide a version of cpufreq_get() which returns the CPU frequency when CPUFREQ is disabled is beyond me, rather than requiring a dummy zero-returning cpufreq_get(). Especially as they do: unsigned long khz = cpufreq_get(cpu); if (!khz) khz = tsc_khz; In other words, if CPUFREQ is disabled, get it from tsc_khz - why not provide a dummy cpufreq_get() which returns tsc_khz?) Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Diffstat (limited to 'arch/ia64/pci/pci.c')
0 files changed, 0 insertions, 0 deletions