Commit-ID:  69309a05907546fb686b251d4ab041c26afe1e1d
Gitweb:     http://git.kernel.org/tip/69309a05907546fb686b251d4ab041c26afe1e1d
Author:     H. Peter Anvin <hpa@zytor.com>
AuthorDate: Tue, 27 Jul 2010 23:29:52 -0700
Committer:  H. Peter Anvin <hpa@zytor.com>
CommitDate: Tue, 27 Jul 2010 23:29:52 -0700

x86, asm: Clean up and simplify set_64bit()

Clean up and simplify set_64bit().  This code is quite old (1.3.11)
and contains a fair bit of auxilliary machinery that current versions
of gcc handle just fine automatically.  Worse, the auxilliary
machinery can actually cause an unnecessary spill to memory.

Furthermore, the loading of the old value inside the loop in the
32-bit case is unnecessary: if the value doesn't match, the CMPXCHG8B
instruction will already have loaded the "new previous" value for us.

Clean up the comment, too, and remove page references to obsolete
versions of the Intel SDM.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
LKML-Reference: <tip-*@vger.kernel.org>
---
 arch/x86/include/asm/cmpxchg_32.h |   67 +++++++++++--------------------------
 arch/x86/include/asm/cmpxchg_64.h |    4 +--
 2 files changed, 21 insertions(+), 50 deletions(-)

diff --git a/arch/x86/include/asm/cmpxchg_32.h b/arch/x86/include/asm/cmpxchg_32.h
index c1cf59d..20955ea 100644
--- a/arch/x86/include/asm/cmpxchg_32.h
+++ b/arch/x86/include/asm/cmpxchg_32.h
@@ -53,60 +53,33 @@ struct __xchg_dummy {
 	__xchg((v), (ptr), sizeof(*ptr))
 
 /*
- * The semantics of XCHGCMP8B are a bit strange, this is why
- * there is a loop and the loading of %%eax and %%edx has to
- * be inside. This inlines well in most cases, the cached
- * cost is around ~38 cycles. (in the future we might want
- * to do an SIMD/3DNOW!/MMX/FPU 64-bit store here, but that
- * might have an implicit FPU-save as a cost, so it's not
- * clear which path to go.)
+ * CMPXCHG8B only writes to the target if we had the previous
+ * value in registers, otherwise it acts as a read and gives us the
+ * "new ...
[ message continues ]

Commit-ID:  bf676945cb5bfe455321f57968967c18976f4995
Gitweb:     http://git.kernel.org/tip/bf676945cb5bfe455321f57968967c18976f4995
Author:     H. Peter Anvin <hpa@zytor.com>
AuthorDate: Tue, 3 Aug 2010 07:00:16 -0700
Committer:  H. Peter Anvin <hpa@zytor.com>
CommitDate: Tue, 3 Aug 2010 07:00:16 -0700

um, x86: Cast to (u64 *) inside set_64bit()

After tightening up the types passed to set_64bit(), the cast to
(phys_t *) triggers a warning apparently because phys_t is defined as
"unsigned long" when building on 64 bits; however, u64 is defined as
"unsigned long long".  This is, however, a explicit cast inside a
size-specific call, so just make the cast explicitly (u64 *).

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Cc: Jeff Dike <jdike@addtoit.com>
LKML-Reference: <tip-69309a05907546fb686b251d4ab041c26afe1e1d@git.kernel.org>
---
 arch/um/include/asm/pgtable-3level.h |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/um/include/asm/pgtable-3level.h b/arch/um/include/asm/pgtable-3level.h
index 084de4a..0032f92 100644
--- a/arch/um/include/asm/pgtable-3level.h
+++ b/arch/um/include/asm/pgtable-3level.h
@@ -60,7 +60,7 @@
 	set_pud(pud, __pud(_PAGE_TABLE + __pa(pmd)))
 
 #ifdef CONFIG_64BIT
-#define set_pud(pudptr, pudval) set_64bit((phys_t *) (pudptr), pud_val(pudval))
+#define set_pud(pudptr, pudval) set_64bit((u64 *) (pudptr), pud_val(pudval))
 #else
 #define set_pud(pudptr, pudval) (*(pudptr) = (pudval))
 #endif
@@ -73,7 +73,7 @@ static inline int pgd_newpage(pgd_t pgd)
 static inline void pgd_mkuptodate(pgd_t pgd) { pgd_val(pgd) &= ~_PAGE_NEWPAGE; }
 
 #ifdef CONFIG_64BIT
-#define set_pmd(pmdptr, pmdval) set_64bit((phys_t *) (pmdptr), pmd_val(pmdval))
+#define set_pmd(pmdptr, pmdval) set_64bit((u64 *) (pmdptr), pmd_val(pmdval))
 #else
 #define set_pmd(pmdptr, pmdval) (*(pmdptr) = (pmdval))
 #endif
--

[ message continues ]