x86, um: get rid of system.h -> system.h include

Long-term we want to split system.h and include barriers part from
underlying target; for now copy that part to sysdep.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>

7 files changed, 266 insertions, 55 deletions

diff --git a/arch/um/Makefile b/arch/um/Makefile
index 909d16b07a44..11cb7baa0fee 100644
--- a/arch/um/Makefile
+++ b/arch/um/Makefile
@@ -19,7 +19,7 @@ core-y			+= $(ARCH_DIR)/kernel/		\
                            $(ARCH_DIR)/os-$(OS)/
 
 # Have to precede the include because the included Makefiles reference them.
-SYMLINK_HEADERS := archparam.h system.h processor.h ptrace.h \
+SYMLINK_HEADERS := archparam.h processor.h ptrace.h \
         module.h vm-flags.h elf.h
 SYMLINK_HEADERS := $(foreach header,$(SYMLINK_HEADERS),$(ARCH_DIR)/include/asm/$(header))
 
diff --git a/arch/um/include/asm/system-i386.h b/arch/um/include/asm/system-i386.h
deleted file mode 100644
index c436263e67ba..000000000000
--- a/arch/um/include/asm/system-i386.h
+++ /dev/null
@@ -1,6 +0,0 @@
-#ifndef __UM_SYSTEM_I386_H
-#define __UM_SYSTEM_I386_H
-
-#include "asm/system-generic.h"
-    
-#endif
diff --git a/arch/um/include/asm/system-ppc.h b/arch/um/include/asm/system-ppc.h
deleted file mode 100644
index 17cde6640bf5..000000000000
--- a/arch/um/include/asm/system-ppc.h
+++ /dev/null
@@ -1,12 +0,0 @@
-#ifndef __UM_SYSTEM_PPC_H
-#define __UM_SYSTEM_PPC_H
-
-#define _switch_to _ppc_switch_to
-
-#include "asm/arch/system.h"
-
-#undef _switch_to
- 
-#include "asm/system-generic.h"
-
-#endif
diff --git a/arch/um/include/asm/system-x86_64.h b/arch/um/include/asm/system-x86_64.h
deleted file mode 100644
index e1b61b580734..000000000000
--- a/arch/um/include/asm/system-x86_64.h
+++ /dev/null
@@ -1,23 +0,0 @@
-/*
- * Copyright 2003 PathScale, Inc.
- *
- * Licensed under the GPL
- */
-
-#ifndef __UM_SYSTEM_X86_64_H
-#define __UM_SYSTEM_X86_64_H
-
-#include "asm/system-generic.h"
-
-#endif
-
-/*
- * Overrides for Emacs so that we follow Linus's tabbing style.
- * Emacs will notice this stuff at the end of the file and automatically
- * adjust the settings for this buffer only.  This must remain at the end
- * of the file.
- * ---------------------------------------------------------------------------
- * Local variables:
- * c-file-style: "linux"
- * End:
- */
diff --git a/arch/um/include/asm/system-generic.h b/arch/um/include/asm/system.h
index 5bcfa35e7a22..753346e2cdfd 100644
--- a/arch/um/include/asm/system-generic.h
+++ b/arch/um/include/asm/system.h
@@ -1,19 +1,7 @@
 #ifndef __UM_SYSTEM_GENERIC_H
 #define __UM_SYSTEM_GENERIC_H
 
-#include "asm/arch/system.h"
-
-#undef switch_to
-#undef local_irq_save
-#undef local_irq_restore
-#undef local_irq_disable
-#undef local_irq_enable
-#undef local_save_flags
-#undef local_irq_restore
-#undef local_irq_enable
-#undef local_irq_disable
-#undef local_irq_save
-#undef irqs_disabled
+#include "sysdep/system.h"
 
 extern void *switch_to(void *prev, void *next, void *last);
 
diff --git a/arch/um/include/shared/sysdep-i386/system.h b/arch/um/include/shared/sysdep-i386/system.h
new file mode 100644
index 000000000000..d1b93c436200
--- /dev/null
+++ b/arch/um/include/shared/sysdep-i386/system.h
@@ -0,0 +1,132 @@
+#ifndef _ASM_X86_SYSTEM_H_
+#define _ASM_X86_SYSTEM_H_
+
+#include <asm/asm.h>
+#include <asm/segment.h>
+#include <asm/cpufeature.h>
+#include <asm/cmpxchg.h>
+#include <asm/nops.h>
+
+#include <linux/kernel.h>
+#include <linux/irqflags.h>
+
+/* entries in ARCH_DLINFO: */
+#ifdef CONFIG_IA32_EMULATION
+# define AT_VECTOR_SIZE_ARCH 2
+#else
+# define AT_VECTOR_SIZE_ARCH 1
+#endif
+
+extern unsigned long arch_align_stack(unsigned long sp);
+
+void default_idle(void);
+
+/*
+ * Force strict CPU ordering.
+ * And yes, this is required on UP too when we're talking
+ * to devices.
+ */
+#ifdef CONFIG_X86_32
+/*
+ * Some non-Intel clones support out of order store. wmb() ceases to be a
+ * nop for these.
+ */
+#define mb() alternative("lock; addl $0,0(%%esp)", "mfence", X86_FEATURE_XMM2)
+#define rmb() alternative("lock; addl $0,0(%%esp)", "lfence", X86_FEATURE_XMM2)
+#define wmb() alternative("lock; addl $0,0(%%esp)", "sfence", X86_FEATURE_XMM)
+#else
+#define mb()    asm volatile("mfence":::"memory")
+#define rmb()   asm volatile("lfence":::"memory")
+#define wmb()   asm volatile("sfence" ::: "memory")
+#endif
+
+/**
+ * read_barrier_depends - Flush all pending reads that subsequents reads
+ * depend on.
+ *
+ * No data-dependent reads from memory-like regions are ever reordered
+ * over this barrier.  All reads preceding this primitive are guaranteed
+ * to access memory (but not necessarily other CPUs' caches) before any
+ * reads following this primitive that depend on the data return by
+ * any of the preceding reads.  This primitive is much lighter weight than
+ * rmb() on most CPUs, and is never heavier weight than is
+ * rmb().
+ *
+ * These ordering constraints are respected by both the local CPU
+ * and the compiler.
+ *
+ * Ordering is not guaranteed by anything other than these primitives,
+ * not even by data dependencies.  See the documentation for
+ * memory_barrier() for examples and URLs to more information.
+ *
+ * For example, the following code would force ordering (the initial
+ * value of "a" is zero, "b" is one, and "p" is "&a"):
+ *
+ * <programlisting>
+ *      CPU 0                           CPU 1
+ *
+ *      b = 2;
+ *      memory_barrier();
+ *      p = &b;                         q = p;
+ *                                      read_barrier_depends();
+ *                                      d = *q;
+ * </programlisting>
+ *
+ * because the read of "*q" depends on the read of "p" and these
+ * two reads are separated by a read_barrier_depends().  However,
+ * the following code, with the same initial values for "a" and "b":
+ *
+ * <programlisting>
+ *      CPU 0                           CPU 1
+ *
+ *      a = 2;
+ *      memory_barrier();
+ *      b = 3;                          y = b;
+ *                                      read_barrier_depends();
+ *                                      x = a;
+ * </programlisting>
+ *
+ * does not enforce ordering, since there is no data dependency between
+ * the read of "a" and the read of "b".  Therefore, on some CPUs, such
+ * as Alpha, "y" could be set to 3 and "x" to 0.  Use rmb()
+ * in cases like this where there are no data dependencies.
+ **/
+
+#define read_barrier_depends()  do { } while (0)
+
+#ifdef CONFIG_SMP
+#define smp_mb()        mb()
+#ifdef CONFIG_X86_PPRO_FENCE
+# define smp_rmb()      rmb()
+#else
+# define smp_rmb()      barrier()
+#endif
+#ifdef CONFIG_X86_OOSTORE
+# define smp_wmb()      wmb()
+#else
+# define smp_wmb()      barrier()
+#endif
+#define smp_read_barrier_depends()      read_barrier_depends()
+#define set_mb(var, value) do { (void)xchg(&var, value); } while (0)
+#else
+#define smp_mb()        barrier()
+#define smp_rmb()       barrier()
+#define smp_wmb()       barrier()
+#define smp_read_barrier_depends()      do { } while (0)
+#define set_mb(var, value) do { var = value; barrier(); } while (0)
+#endif
+
+/*
+ * Stop RDTSC speculation. This is needed when you need to use RDTSC
+ * (or get_cycles or vread that possibly accesses the TSC) in a defined
+ * code region.
+ *
+ * (Could use an alternative three way for this if there was one.)
+ */
+static inline void rdtsc_barrier(void)
+{
+        alternative(ASM_NOP3, "mfence", X86_FEATURE_MFENCE_RDTSC);
+        alternative(ASM_NOP3, "lfence", X86_FEATURE_LFENCE_RDTSC);
+}
+
+#endif
diff --git a/arch/um/include/shared/sysdep-x86_64/system.h b/arch/um/include/shared/sysdep-x86_64/system.h
new file mode 100644
index 000000000000..d1b93c436200
--- /dev/null
+++ b/arch/um/include/shared/sysdep-x86_64/system.h
@@ -0,0 +1,132 @@
+#ifndef _ASM_X86_SYSTEM_H_
+#define _ASM_X86_SYSTEM_H_
+
+#include <asm/asm.h>
+#include <asm/segment.h>
+#include <asm/cpufeature.h>
+#include <asm/cmpxchg.h>
+#include <asm/nops.h>
+
+#include <linux/kernel.h>
+#include <linux/irqflags.h>
+
+/* entries in ARCH_DLINFO: */
+#ifdef CONFIG_IA32_EMULATION
+# define AT_VECTOR_SIZE_ARCH 2
+#else
+# define AT_VECTOR_SIZE_ARCH 1
+#endif
+
+extern unsigned long arch_align_stack(unsigned long sp);
+
+void default_idle(void);
+
+/*
+ * Force strict CPU ordering.
+ * And yes, this is required on UP too when we're talking
+ * to devices.
+ */
+#ifdef CONFIG_X86_32
+/*
+ * Some non-Intel clones support out of order store. wmb() ceases to be a
+ * nop for these.
+ */
+#define mb() alternative("lock; addl $0,0(%%esp)", "mfence", X86_FEATURE_XMM2)
+#define rmb() alternative("lock; addl $0,0(%%esp)", "lfence", X86_FEATURE_XMM2)
+#define wmb() alternative("lock; addl $0,0(%%esp)", "sfence", X86_FEATURE_XMM)
+#else
+#define mb()    asm volatile("mfence":::"memory")
+#define rmb()   asm volatile("lfence":::"memory")
+#define wmb()   asm volatile("sfence" ::: "memory")
+#endif
+
+/**
+ * read_barrier_depends - Flush all pending reads that subsequents reads
+ * depend on.
+ *
+ * No data-dependent reads from memory-like regions are ever reordered
+ * over this barrier.  All reads preceding this primitive are guaranteed
+ * to access memory (but not necessarily other CPUs' caches) before any
+ * reads following this primitive that depend on the data return by
+ * any of the preceding reads.  This primitive is much lighter weight than
+ * rmb() on most CPUs, and is never heavier weight than is
+ * rmb().
+ *
+ * These ordering constraints are respected by both the local CPU
+ * and the compiler.
+ *
+ * Ordering is not guaranteed by anything other than these primitives,
+ * not even by data dependencies.  See the documentation for
+ * memory_barrier() for examples and URLs to more information.
+ *
+ * For example, the following code would force ordering (the initial
+ * value of "a" is zero, "b" is one, and "p" is "&a"):
+ *
+ * <programlisting>
+ *      CPU 0                           CPU 1
+ *
+ *      b = 2;
+ *      memory_barrier();
+ *      p = &b;                         q = p;
+ *                                      read_barrier_depends();
+ *                                      d = *q;
+ * </programlisting>
+ *
+ * because the read of "*q" depends on the read of "p" and these
+ * two reads are separated by a read_barrier_depends().  However,
+ * the following code, with the same initial values for "a" and "b":
+ *
+ * <programlisting>
+ *      CPU 0                           CPU 1
+ *
+ *      a = 2;
+ *      memory_barrier();
+ *      b = 3;                          y = b;
+ *                                      read_barrier_depends();
+ *                                      x = a;
+ * </programlisting>
+ *
+ * does not enforce ordering, since there is no data dependency between
+ * the read of "a" and the read of "b".  Therefore, on some CPUs, such
+ * as Alpha, "y" could be set to 3 and "x" to 0.  Use rmb()
+ * in cases like this where there are no data dependencies.
+ **/
+
+#define read_barrier_depends()  do { } while (0)
+
+#ifdef CONFIG_SMP
+#define smp_mb()        mb()
+#ifdef CONFIG_X86_PPRO_FENCE
+# define smp_rmb()      rmb()
+#else
+# define smp_rmb()      barrier()
+#endif
+#ifdef CONFIG_X86_OOSTORE
+# define smp_wmb()      wmb()
+#else
+# define smp_wmb()      barrier()
+#endif
+#define smp_read_barrier_depends()      read_barrier_depends()
+#define set_mb(var, value) do { (void)xchg(&var, value); } while (0)
+#else
+#define smp_mb()        barrier()
+#define smp_rmb()       barrier()
+#define smp_wmb()       barrier()
+#define smp_read_barrier_depends()      do { } while (0)
+#define set_mb(var, value) do { var = value; barrier(); } while (0)
+#endif
+
+/*
+ * Stop RDTSC speculation. This is needed when you need to use RDTSC
+ * (or get_cycles or vread that possibly accesses the TSC) in a defined
+ * code region.
+ *
+ * (Could use an alternative three way for this if there was one.)
+ */
+static inline void rdtsc_barrier(void)
+{
+        alternative(ASM_NOP3, "mfence", X86_FEATURE_MFENCE_RDTSC);
+        alternative(ASM_NOP3, "lfence", X86_FEATURE_LFENCE_RDTSC);
+}
+
+#endif