[PATCH] x86/cpufeature: explicit comments for duplicate macro
From: Cao jin
Date: Wed Aug 28 2019 - 02:10:13 EST
Help people to understand the author's intent of apparent duplication of
BUILD_BUG_ON_ZERO(NCAPINTS != n), which is hard to detect by eyes.
CC: Dave Hansen <dave.hansen@xxxxxxxxx>
Suggested-by: Borislav Petkov <bp@xxxxxxxxx>
Signed-off-by: Cao jin <caoj.fnst@xxxxxxxxxxxxxx>
Tried my best to describe it accurately, in case of any inaccuracy, feel
free to rephrase.
arch/x86/include/asm/cpufeature.h | 11 +++++++++++
1 file changed, 11 insertions(+)
diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h
index 58acda503817..e943174abf1e 100644
@@ -61,6 +61,17 @@ extern const char * const x86_bug_flags[NBUGINTS*32];
#define CHECK_BIT_IN_MASK_WORD(maskname, word, bit) \
(((bit)>>5)==(word) && (1UL<<((bit)&31) & maskname##word ))
+ * REQUIRED_MASK_CHECK may seems duplicate, but actually has its reason to
+ * live here.
+ * New CPUID leaf added or feature bit adjustment would/may result in increase
+ * in NCAPINTS. When it does, two required-features.h and here need to be
+ * modified correspondingly. BUILD_BUG_ON_ZERO assures the modification to be
+ * carried out, checking NCAPINTS also reminds the additional lines for new
+ * word. But, required-features.h as a single header file, can't be compiled
+ * directly, that is why a wrapper is defined there and called here.
+ * Totally the same case for DISABLED_MASK_BIT_SET.
#define REQUIRED_MASK_BIT_SET(feature_bit) \
( CHECK_BIT_IN_MASK_WORD(REQUIRED_MASK, 0, feature_bit) || \
CHECK_BIT_IN_MASK_WORD(REQUIRED_MASK, 1, feature_bit) || \