Re: [PATCH v4 5/6] add listmount(2) syscall

[Guenter Roeck]

On Thu, Jan 11, 2024 at 07:40:32PM -0800, Linus Torvalds wrote:
> On Thu, 11 Jan 2024 at 15:57, Guenter Roeck <linux@xxxxxxxxxxxx> wrote:
> >
> > I wonder if something may be wrong with the definition and use of __m
> > for u64 accesses. The code below also fixes the build problem.
> 
> Ok, that looks like the right workaround.
> 

I think I'll go with the code below. It doesn't touch the MMU code
(which for some reason doesn't result in a build failure), and
it generates

! 5071 "fs/namespace.c" 1
        mov.l   r6,@r1
        mov.l   r7,@(4,r1)

which seems to be correct. It also matches the pattern used
for __put_user_asm(). Does that make sense ?

Thanks,
Guenter

---
diff --git a/arch/sh/include/asm/uaccess_32.h b/arch/sh/include/asm/uaccess_32.h
index 5d7ddc092afd..5ce46b2a95b6 100644
--- a/arch/sh/include/asm/uaccess_32.h
+++ b/arch/sh/include/asm/uaccess_32.h
@@ -176,6 +176,7 @@ do {							\
 } while (0)
 #endif /* CONFIG_MMU */
 
+#ifdef CONFIG_MMU
 #if defined(CONFIG_CPU_LITTLE_ENDIAN)
 #define __put_user_u64(val,addr,retval) \
 ({ \
@@ -221,6 +222,26 @@ __asm__ __volatile__( \
 	: "r" (val), "m" (__m(addr)), "i" (-EFAULT), "0" (retval) \
         : "memory"); })
 #endif
+#else
+#if defined(CONFIG_CPU_LITTLE_ENDIAN)
+#define __put_user_u64(val,addr,retval) \
+({ \
+__asm__ __volatile__( \
+	"mov.l	%R0,%1\n\t" \
+	"mov.l	%S0,%T1\n\t" \
+	: /* no outputs */ \
+	: "r" (val), "m" (*(u64 *)(addr)) \
+	: "memory"); })
+#else
+({ \
+__asm__ __volatile__( \
+	"mov.l	%S0,%1\n\t" \
+	"mov.l	%R0,%T1\n\t" \
+	: /* no outputs */ \
+	: "r" (val), "m" (*(u64 *)(addr)) \
+	: "memory"); })
+#endif
+#endif /* CONFIG_MMU */
 
 extern void __put_user_unknown(void);