[PATCH] add readX_relaxed() interface

From: jbarnes@sgi.com (Jesse Barnes)

Here's the patch to add the new _relaxed variants for PIO read accesses.
It's been ack'd by gregkh and Grant Grundler, and I think it's ready to
go.

PCI ordering rules also guarantee that PIO read responses arrive after any
outstanding DMA writes on that bus, since for some devices the result of a
readb() call may signal to the driver that a DMA transaction is complete.  In
many cases, however, the driver may want to indicate that the next readb()
call has no relation to any previous DMA writes performed by the device.  The
driver can use the readb_relaxed() for these cases, although only some
platforms will honor the relaxed semantics.

Documentation/DocBook/deviceiobook.tmpl

@@ -126,7 +126,9 @@
       <para>
 	The functions are named <function>readb</function>,
 	<function>readw</function>, <function>readl</function>,
-	<function>readq</function>, <function>writeb</function>,
+	<function>readq</function>, <function>readb_relaxed</function>,
+	<function>readw_relaxed</function>, <function>readl_relaxed</function>,
+	<function>readq_relaxed</function>, <function>writeb</function>,
 	<function>writew</function>, <function>writel</function> and
 	<function>writeq</function>.
       </para>
@@ -159,6 +161,18 @@
 	author cares. This kind of property cannot be hidden from driver
 	writers in the API.
       </para>
+
+      <para>
+	PCI ordering rules also guarantee that PIO read responses arrive
+	after any outstanding DMA writes on that bus, since for some devices
+	the result of a <function>readb</function> call may signal to the
+	driver that a DMA transaction is complete.  In many cases, however,
+	the driver may want to indicate that the next
+	<function>readb</function> call has no relation to any previous DMA
+	writes performed by the device.  The driver can use
+	<function>readb_relaxed</function> for these cases, although only
+	some platforms will honor the relaxed semantics.
+      </para>
     </sect1>
 
     <sect1>

arch/ia64/sn/kernel/sn2/io.c

@@ -23,6 +23,10 @@
 #undef __sn_readw
 #undef __sn_readl
 #undef __sn_readq
+#undef __sn_readb_relaxed
+#undef __sn_readw_relaxed
+#undef __sn_readl_relaxed
+#undef __sn_readq_relaxed
 
 unsigned int
 __sn_inb (unsigned long port)
@@ -84,4 +88,28 @@ __sn_readq (void *addr)
 	return ___sn_readq (addr);
 }
 
+unsigned char
+__sn_readb_relaxed (void *addr)
+{
+	return ___sn_readb_relaxed (addr);
+}
+
+unsigned short
+__sn_readw_relaxed (void *addr)
+{
+	return ___sn_readw_relaxed (addr);
+}
+
+unsigned int
+__sn_readl_relaxed (void *addr)
+{
+	return ___sn_readl_relaxed (addr);
+}
+
+unsigned long
+__sn_readq_relaxed (void *addr)
+{
+	return ___sn_readq_relaxed (addr);
+}
+
 #endif

include/asm-alpha/io.h

@@ -412,6 +412,11 @@ extern void		___raw_writeq(u64 b, unsigned long addr);
 # define readq(a)	_readq((unsigned long)(a))
 #endif
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
+#define readq_relaxed(addr) readq(addr)
+
 #ifndef writeb
 # define writeb(v,a)	_writeb((v),(unsigned long)(a))
 #endif

include/asm-arm/arch-ebsa110/io.h

@@ -38,6 +38,9 @@ u32 __readl(void *addr);
 #define readb(b)		__readb(b)
 #define readw(b)		__readw(b)
 #define readl(b)		__readl(b)
+#define readb_relaxed(addr)	readb(addr)
+#define readw_relaxed(addr)	readw(addr)
+#define readl_relaxed(addr)	readl(addr)
 
 void __writeb(u8  val, void *addr);
 void __writew(u16 val, void *addr);

include/asm-arm/io.h

@@ -149,6 +149,9 @@ extern void _memset_io(unsigned long, int, size_t);
 #define readb(c) ({ unsigned int __v = __raw_readb(__mem_pci(c)); __v; })
 #define readw(c) ({ unsigned int __v = le16_to_cpu(__raw_readw(__mem_pci(c))); __v; })
 #define readl(c) ({ unsigned int __v = le32_to_cpu(__raw_readl(__mem_pci(c))); __v; })
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
 
 #define readsb(p,d,l)		__raw_readsb((unsigned int)__mem_pci(p),d,l)
 #define readsw(p,d,l)		__raw_readsw((unsigned int)__mem_pci(p),d,l)

include/asm-arm26/io.h

@@ -308,6 +308,9 @@ DECLARE_IO(int,l,"")
 #define readb(c)                        (__readwrite_bug("readb"),0)
 #define readw(c)                        (__readwrite_bug("readw"),0)
 #define readl(c)                        (__readwrite_bug("readl"),0)
+#define readb_relaxed(addr)		readb(addr)
+#define readw_relaxed(addr)		readw(addr)
+#define readl_relaxed(addr)		readl(addr)
 #define writeb(v,c)                     __readwrite_bug("writeb")
 #define writew(v,c)                     __readwrite_bug("writew")
 #define writel(v,c)                     __readwrite_bug("writel")

include/asm-cris/io.h

@@ -42,6 +42,9 @@ extern void iounmap(void *addr);
 #define readb(addr) (*(volatile unsigned char *) (addr))
 #define readw(addr) (*(volatile unsigned short *) (addr))
 #define readl(addr) (*(volatile unsigned int *) (addr))
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
 #define __raw_readb readb
 #define __raw_readw readw
 #define __raw_readl readl

include/asm-h8300/io.h

@@ -47,6 +47,10 @@ static inline unsigned int _swapl(volatile unsigned long v)
 #define readl(addr) \
     ({ unsigned int __v = (*(volatile unsigned int *) (addr & 0x00ffffff)); __v; })
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
+
 #define writeb(b,addr) (void)((*(volatile unsigned char *) (addr & 0x00ffffff)) = (b))
 #define writew(b,addr) (void)((*(volatile unsigned short *) (addr & 0x00ffffff)) = (b))
 #define writel(b,addr) (void)((*(volatile unsigned int *) (addr & 0x00ffffff)) = (b))

include/asm-i386/io.h

@@ -153,6 +153,9 @@ extern void bt_iounmap(void *addr, unsigned long size);
 #define readb(addr) (*(volatile unsigned char *) __io_virt(addr))
 #define readw(addr) (*(volatile unsigned short *) __io_virt(addr))
 #define readl(addr) (*(volatile unsigned int *) __io_virt(addr))
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
 #define __raw_readb readb
 #define __raw_readw readw
 #define __raw_readl readl

include/asm-ia64/io.h

@@ -125,6 +125,10 @@ __ia64_mk_io_addr (unsigned long port)
 #define __ia64_readw	___ia64_readw
 #define __ia64_readl	___ia64_readl
 #define __ia64_readq	___ia64_readq
+#define __ia64_readb_relaxed	___ia64_readb
+#define __ia64_readw_relaxed	___ia64_readw
+#define __ia64_readl_relaxed	___ia64_readl
+#define __ia64_readq_relaxed	___ia64_readq
 #define __ia64_writeb	___ia64_writeb
 #define __ia64_writew	___ia64_writew
 #define __ia64_writel	___ia64_writel
@@ -337,15 +341,27 @@ __writeq (unsigned long val, void *addr)
 #define __readw		platform_readw
 #define __readl		platform_readl
 #define __readq		platform_readq
+#define __readb_relaxed	platform_readb_relaxed
+#define __readw_relaxed	platform_readw_relaxed
+#define __readl_relaxed	platform_readl_relaxed
+#define __readq_relaxed	platform_readq_relaxed
 
 #define readb(a)	__readb((void *)(a))
 #define readw(a)	__readw((void *)(a))
 #define readl(a)	__readl((void *)(a))
 #define readq(a)	__readq((void *)(a))
+#define readb_relaxed(a)	__readb_relaxed((void *)(a))
+#define readw_relaxed(a)	__readw_relaxed((void *)(a))
+#define readl_relaxed(a)	__readl_relaxed((void *)(a))
+#define readq_relaxed(a)	__readq_relaxed((void *)(a))
 #define __raw_readb	readb
 #define __raw_readw	readw
 #define __raw_readl	readl
 #define __raw_readq	readq
+#define __raw_readb_relaxed	readb_relaxed
+#define __raw_readw_relaxed	readw_relaxed
+#define __raw_readl_relaxed	readl_relaxed
+#define __raw_readq_relaxed	readq_relaxed
 #define writeb(v,a)	__writeb((v), (void *) (a))
 #define writew(v,a)	__writew((v), (void *) (a))
 #define writel(v,a)	__writel((v), (void *) (a))

include/asm-ia64/machvec.h

@@ -65,6 +65,10 @@ typedef unsigned char ia64_mv_readb_t (void *);
 typedef unsigned short ia64_mv_readw_t (void *);
 typedef unsigned int ia64_mv_readl_t (void *);
 typedef unsigned long ia64_mv_readq_t (void *);
+typedef unsigned char ia64_mv_readb_relaxed_t (void *);
+typedef unsigned short ia64_mv_readw_relaxed_t (void *);
+typedef unsigned int ia64_mv_readl_relaxed_t (void *);
+typedef unsigned long ia64_mv_readq_relaxed_t (void *);
 
 extern void machvec_noop (void);
 extern void machvec_memory_fence (void);
@@ -116,6 +120,10 @@ extern void machvec_memory_fence (void);
 #  define platform_readw        ia64_mv.readw
 #  define platform_readl        ia64_mv.readl
 #  define platform_readq        ia64_mv.readq
+#  define platform_readb_relaxed        ia64_mv.readb_relaxed
+#  define platform_readw_relaxed        ia64_mv.readw_relaxed
+#  define platform_readl_relaxed        ia64_mv.readl_relaxed
+#  define platform_readq_relaxed        ia64_mv.readq_relaxed
 # endif
 
 /* __attribute__((__aligned__(16))) is required to make size of the
@@ -158,6 +166,10 @@ struct ia64_machine_vector {
 	ia64_mv_readw_t *readw;
 	ia64_mv_readl_t *readl;
 	ia64_mv_readq_t *readq;
+	ia64_mv_readb_relaxed_t *readb_relaxed;
+	ia64_mv_readw_relaxed_t *readw_relaxed;
+	ia64_mv_readl_relaxed_t *readl_relaxed;
+	ia64_mv_readq_relaxed_t *readq_relaxed;
 } __attribute__((__aligned__(16))); /* align attrib? see above comment */
 
 #define MACHVEC_INIT(name)			\
@@ -196,6 +208,10 @@ struct ia64_machine_vector {
 	platform_readw,				\
 	platform_readl,				\
 	platform_readq,				\
+	platform_readb_relaxed,			\
+	platform_readw_relaxed,			\
+	platform_readl_relaxed,			\
+	platform_readq_relaxed,			\
 }
 
 extern struct ia64_machine_vector ia64_mv;
@@ -322,5 +338,17 @@ extern ia64_mv_dma_supported		swiotlb_dma_supported;
 #ifndef platform_readq
 # define platform_readq		__ia64_readq
 #endif
+#ifndef platform_readb_relaxed
+# define platform_readb_relaxed	__ia64_readb_relaxed
+#endif
+#ifndef platform_readw_relaxed
+# define platform_readw_relaxed	__ia64_readw_relaxed
+#endif
+#ifndef platform_readl_relaxed
+# define platform_readl_relaxed	__ia64_readl_relaxed
+#endif
+#ifndef platform_readq_relaxed
+# define platform_readq_relaxed	__ia64_readq_relaxed
+#endif
 
 #endif /* _ASM_IA64_MACHVEC_H */

include/asm-ia64/machvec_sn2.h

@@ -52,6 +52,10 @@ extern ia64_mv_readb_t __sn_readb;
 extern ia64_mv_readw_t __sn_readw;
 extern ia64_mv_readl_t __sn_readl;
 extern ia64_mv_readq_t __sn_readq;
+extern ia64_mv_readb_t __sn_readb_relaxed;
+extern ia64_mv_readw_t __sn_readw_relaxed;
+extern ia64_mv_readl_t __sn_readl_relaxed;
+extern ia64_mv_readq_t __sn_readq_relaxed;
 extern ia64_mv_dma_alloc_coherent	sn_dma_alloc_coherent;
 extern ia64_mv_dma_free_coherent	sn_dma_free_coherent;
 extern ia64_mv_dma_map_single		sn_dma_map_single;
@@ -87,6 +91,10 @@ extern ia64_mv_dma_supported		sn_dma_supported;
 #define platform_readw			__sn_readw
 #define platform_readl			__sn_readl
 #define platform_readq			__sn_readq
+#define platform_readb_relaxed		__sn_readb_relaxed
+#define platform_readw_relaxed		__sn_readw_relaxed
+#define platform_readl_relaxed		__sn_readl_relaxed
+#define platform_readq_relaxed		__sn_readq_relaxed
 #define platform_irq_desc		sn_irq_desc
 #define platform_irq_to_vector		sn_irq_to_vector
 #define platform_local_vector_to_irq	sn_local_vector_to_irq

include/asm-ia64/sn/sn2/io.h

@@ -27,6 +27,10 @@ extern void sn_dma_flush(unsigned long);
 #define __sn_readw ___sn_readw
 #define __sn_readl ___sn_readl
 #define __sn_readq ___sn_readq
+#define __sn_readb_relaxed ___sn_readb_relaxed
+#define __sn_readw_relaxed ___sn_readw_relaxed
+#define __sn_readl_relaxed ___sn_readl_relaxed
+#define __sn_readq_relaxed ___sn_readq_relaxed
 
 /*
  * The following routines are SN Platform specific, called when
@@ -208,25 +212,25 @@ sn_inl_fast (unsigned long port)
 }
 
 static inline unsigned char
-sn_readb_fast (void *addr)
+___sn_readb_relaxed (void *addr)
 {
 	return *(volatile unsigned char *)addr;
 }
 
 static inline unsigned short
-sn_readw_fast (void *addr)
+___sn_readw_relaxed (void *addr)
 {
 	return *(volatile unsigned short *)addr;
 }
 
 static inline unsigned int
-sn_readl_fast (void *addr)
+___sn_readl_relaxed (void *addr)
 {
 	return *(volatile unsigned int *) addr;
 }
 
 static inline unsigned long
-sn_readq_fast (void *addr)
+___sn_readq_relaxed (void *addr)
 {
 	return *(volatile unsigned long *) addr;
 }

include/asm-m68k/io.h

@@ -261,6 +261,10 @@ static inline void isa_delay(void)
 #define writeb(val,addr)  out_8((addr),(val))
 #define writew(val,addr)  out_le16((addr),(val))
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
+
 #ifndef CONFIG_ISA
 #define inb(port)      in_8(port)
 #define outb(val,port) out_8((port),(val))

include/asm-m68knommu/io.h

@@ -39,6 +39,10 @@ static inline unsigned int _swapl(volatile unsigned long v)
 #define readl(addr) \
     ({ unsigned int __v = (*(volatile unsigned int *) (addr)); __v; })
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
+
 #define writeb(b,addr) (void)((*(volatile unsigned char *) (addr)) = (b))
 #define writew(b,addr) (void)((*(volatile unsigned short *) (addr)) = (b))
 #define writel(b,addr) (void)((*(volatile unsigned int *) (addr)) = (b))

include/asm-mips/io.h

@@ -264,6 +264,10 @@ extern void iounmap(void *addr);
 #define readw(addr)		__ioswab16(__raw_readw(addr))
 #define readl(addr)		__ioswab32(__raw_readl(addr))
 #define readq(addr)		__ioswab64(__raw_readq(addr))
+#define readb_relaxed(addr)	readb(addr)
+#define readw_relaxed(addr)	readw(addr)
+#define readl_relaxed(addr)	readl(addr)
+#define readq_relaxed(addr)	readq(addr)
 
 #define __raw_writeb(b,addr)	((*(volatile unsigned char *)(addr)) = (b))
 #define __raw_writew(w,addr)	((*(volatile unsigned short *)(addr)) = (w))

include/asm-parisc/io.h

@@ -171,6 +171,11 @@ extern __inline__ void ___raw_writeq(unsigned long long val, unsigned long addr)
 #define writeq(b,addr) __raw_writeq(cpu_to_le64(b),addr)
 #endif /* !USE_HPPA_IOREMAP */
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
+#define readq_relaxed(addr) readq(addr)
+
 extern void __memcpy_fromio(unsigned long dest, unsigned long src, int count);
 extern void __memcpy_toio(unsigned long dest, unsigned long src, int count);
 extern void __memset_io(unsigned long dest, char fill, int count);

include/asm-ppc/io.h

@@ -58,6 +58,9 @@ extern unsigned long pci_dram_offset;
 #define writel(b,addr) out_le32((volatile u32 *)(addr),(b))
 #endif /* CONFIG_APUS */
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
 
 #define __raw_readb(addr)	(*(volatile unsigned char *)(addr))
 #define __raw_readw(addr)	(*(volatile unsigned short *)(addr))

include/asm-ppc64/io.h

@@ -79,6 +79,10 @@ extern unsigned long pci_io_base;
 #define outsl(port, buf, nl)	_outsl_ns((u32 *)((port)+pci_io_base), (buf), (nl))
 #endif
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
+
 extern void _insb(volatile u8 *port, void *buf, int ns);
 extern void _outsb(volatile u8 *port, const void *buf, int ns);
 extern void _insw(volatile u16 *port, void *buf, int ns);

include/asm-s390/io.h

@@ -87,6 +87,10 @@ extern void iounmap(void *addr);
 #define readw(addr) (*(volatile unsigned short *) __io_virt(addr))
 #define readl(addr) (*(volatile unsigned int *) __io_virt(addr))
 
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
+
 #define writeb(b,addr) (*(volatile unsigned char *) __io_virt(addr) = (b))
 #define writew(b,addr) (*(volatile unsigned short *) __io_virt(addr) = (b))
 #define writel(b,addr) (*(volatile unsigned int *) __io_virt(addr) = (b))

include/asm-sh/io.h

@@ -130,6 +130,10 @@
 # define writel(v,a)	({ __raw_writel((v),(unsigned long)(a)); mb(); })
 #endif
 
+#define readb_relaxed(a) readb(a)
+#define readw_relaxed(a) readw(a)
+#define readl_relaxed(a) readl(a)
+
 /*
  * If the platform has PC-like I/O, this function converts the offset into
  * an address.

include/asm-sparc/io.h

@@ -99,6 +99,9 @@ static inline void __writel(u32 b, unsigned long addr)
 #define readb(addr)	__readb((unsigned long)(addr))
 #define readw(addr)	__readw((unsigned long)(addr))
 #define readl(addr)	__readl((unsigned long)(addr))
+#define readb_relaxed(addr) readb(addr)
+#define readw_relaxed(addr) readw(addr)
+#define readl_relaxed(addr) readl(addr)
 
 #define writeb(b, addr)	__writeb((b),(unsigned long)(addr))
 #define writew(b, addr)	__writew((b),(unsigned long)(addr))

include/asm-sparc64/io.h

@@ -176,6 +176,10 @@ static __inline__ void _writeq(u64 q, unsigned long addr)
 #define readw(__addr)		(_readw((unsigned long)(__addr)))
 #define readl(__addr)		(_readl((unsigned long)(__addr)))
 #define readq(__addr)		(_readq((unsigned long)(__addr)))
+#define readb_relaxed(a)	readb(a)
+#define readw_relaxed(a)	readw(a)
+#define readl_relaxed(a)	readl(a)
+#define readq_relaxed(a)	readq(a)
 #define writeb(__b, __addr)	(_writeb((u8)(__b), (unsigned long)(__addr)))
 #define writew(__w, __addr)	(_writew((u16)(__w), (unsigned long)(__addr)))
 #define writel(__l, __addr)	(_writel((u32)(__l), (unsigned long)(__addr)))

include/asm-v850/io.h

@@ -23,6 +23,10 @@
 #define readl(addr) \
   ({ unsigned long __v = (*(volatile unsigned long *) (addr)); __v; })
 
+#define readb_relaxed(a) readb(a)
+#define readw_relaxed(a) readw(a)
+#define readl_relaxed(a) readl(a)
+
 #define writeb(b, addr) \
   (void)((*(volatile unsigned char *) (addr)) = (b))
 #define writew(b, addr) \

include/asm-x86_64/io.h

@@ -188,6 +188,10 @@ extern void iounmap(void *addr);
 #define readw(addr) (*(volatile unsigned short *) __io_virt(addr))
 #define readl(addr) (*(volatile unsigned int *) __io_virt(addr))
 #define readq(addr) (*(volatile unsigned long *) __io_virt(addr))
+#define readb_relaxed(a) readb(a)
+#define readw_relaxed(a) readw(a)
+#define readl_relaxed(a) readl(a)
+#define readq_relaxed(a) readq(a)
 #define __raw_readb readb
 #define __raw_readw readw
 #define __raw_readl readl