CRISv10 usercopy library add lineendings to asm
authorJesper Nilsson <jesper.nilsson@axis.com>
Thu, 15 Nov 2007 01:01:30 +0000 (17:01 -0800)
committerLinus Torvalds <torvalds@woody.linux-foundation.org>
Thu, 15 Nov 2007 02:45:47 +0000 (18:45 -0800)
Removes warning when compiling arch/cris/arch-v10/lib/usercopy.c

No change except adding \n\ on the end of the lines has been done.
Removes warning about multi-line string literals.

Signed-off-by: Jesper Nilsson <jesper.nilsson@axis.com>
Cc: Mikael Starvik <mikael.starvik@axis.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/cris/arch-v10/lib/usercopy.c

index a12c708..b8e6c04 100644 (file)
@@ -92,58 +92,58 @@ __copy_user (void __user *pdst, const void *psrc, unsigned long pn)
        .ifnc %0%1%2%3,$r13$r11$r12$r10                                 \n\
        .err                                                            \n\
        .endif                                                          \n\
-
-       ;; Save the registers we'll use in the movem process
-       ;; on the stack.
-       subq    11*4,$sp
-       movem   $r10,[$sp]
-
-       ;; Now we've got this:
-       ;; r11 - src
-       ;; r13 - dst
-       ;; r12 - n
-
-       ;; Update n for the first loop
-       subq    44,$r12
-
-; Since the noted PC of a faulting instruction in a delay-slot of a taken
-; branch, is that of the branch target, we actually point at the from-movem
-; for this case.  There is no ambiguity here; if there was a fault in that
-; instruction (meaning a kernel oops), the faulted PC would be the address
-; after *that* movem.
-
-0:
-       movem   [$r11+],$r10
-       subq   44,$r12
-       bge     0b
-       movem   $r10,[$r13+]
-1:
-       addq   44,$r12  ;; compensate for last loop underflowing n
-
-       ;; Restore registers from stack
-       movem [$sp+],$r10
-2:
-       .section .fixup,\"ax\"
-
-; To provide a correct count in r10 of bytes that failed to be copied,
-; we jump back into the loop if the loop-branch was taken.  There is no
-; performance penalty for sany use; the program will segfault soon enough.
-
-3:
-       move.d [$sp],$r10
-       addq 44,$r10
-       move.d $r10,[$sp]
-       jump 0b
-4:
-       movem [$sp+],$r10
-       addq 44,$r10
-       addq 44,$r12
-       jump 2b
-
-       .previous
-       .section __ex_table,\"a\"
-       .dword 0b,3b
-       .dword 1b,4b
+                                                                       \n\
+       ;; Save the registers we'll use in the movem process            \n\
+       ;; on the stack.                                                \n\
+       subq    11*4,$sp                                                \n\
+       movem   $r10,[$sp]                                              \n\
+                                                                       \n\
+       ;; Now we've got this:                                          \n\
+       ;; r11 - src                                                    \n\
+       ;; r13 - dst                                                    \n\
+       ;; r12 - n                                                      \n\
+                                                                       \n\
+       ;; Update n for the first loop                                  \n\
+       subq    44,$r12                                                 \n\
+                                                                       \n\
+; Since the noted PC of a faulting instruction in a delay-slot of a taken \n\
+; branch, is that of the branch target, we actually point at the from-movem \n\
+; for this case.  There is no ambiguity here; if there was a fault in that \n\
+; instruction (meaning a kernel oops), the faulted PC would be the address \n\
+; after *that* movem.                                                  \n\
+                                                                       \n\
+0:                                                                     \n\
+       movem   [$r11+],$r10                                            \n\
+       subq   44,$r12                                                  \n\
+       bge     0b                                                      \n\
+       movem   $r10,[$r13+]                                            \n\
+1:                                                                     \n\
+       addq   44,$r12  ;; compensate for last loop underflowing n      \n\
+                                                                       \n\
+       ;; Restore registers from stack                                 \n\
+       movem [$sp+],$r10                                               \n\
+2:                                                                     \n\
+       .section .fixup,\"ax\"                                          \n\
+                                                                       \n\
+; To provide a correct count in r10 of bytes that failed to be copied, \n\
+; we jump back into the loop if the loop-branch was taken.  There is no        \n\
+; performance penalty for sany use; the program will segfault soon enough.\n\
+                                                                       \n\
+3:                                                                     \n\
+       move.d [$sp],$r10                                               \n\
+       addq 44,$r10                                                    \n\
+       move.d $r10,[$sp]                                               \n\
+       jump 0b                                                         \n\
+4:                                                                     \n\
+       movem [$sp+],$r10                                               \n\
+       addq 44,$r10                                                    \n\
+       addq 44,$r12                                                    \n\
+       jump 2b                                                         \n\
+                                                                       \n\
+       .previous                                                       \n\
+       .section __ex_table,\"a\"                                       \n\
+       .dword 0b,3b                                                    \n\
+       .dword 1b,4b                                                    \n\
        .previous"
 
      /* Outputs */ : "=r" (dst), "=r" (src), "=r" (n), "=r" (retn)
@@ -253,59 +253,59 @@ __copy_user_zeroing (void __user *pdst, const void *psrc, unsigned long pn)
        If you want to check that the allocation was right; then
        check the equalities in the first comment.  It should say
        "r13=r13, r11=r11, r12=r12" */
-    __asm__ volatile ("
+    __asm__ volatile ("\n\
        .ifnc %0%1%2%3,$r13$r11$r12$r10                                 \n\
        .err                                                            \n\
        .endif                                                          \n\
-
-       ;; Save the registers we'll use in the movem process
-       ;; on the stack.
-       subq    11*4,$sp
-       movem   $r10,[$sp]
-
-       ;; Now we've got this:
-       ;; r11 - src
-       ;; r13 - dst
-       ;; r12 - n
-
-       ;; Update n for the first loop
-       subq    44,$r12
-0:
-       movem   [$r11+],$r10
-1:
-       subq   44,$r12
-       bge     0b
-       movem   $r10,[$r13+]
-
-       addq   44,$r12  ;; compensate for last loop underflowing n
-
-       ;; Restore registers from stack
-       movem [$sp+],$r10
-4:
-       .section .fixup,\"ax\"
-
-;; Do not jump back into the loop if we fail.  For some uses, we get a
-;; page fault somewhere on the line.  Without checking for page limits,
-;; we don't know where, but we need to copy accurately and keep an
-;; accurate count; not just clear the whole line.  To do that, we fall
-;; down in the code below, proceeding with smaller amounts.  It should
-;; be kept in mind that we have to cater to code like what at one time
-;; was in fs/super.c:
-;;  i = size - copy_from_user((void *)page, data, size);
-;; which would cause repeated faults while clearing the remainder of
-;; the SIZE bytes at PAGE after the first fault.
-;; A caveat here is that we must not fall through from a failing page
-;; to a valid page.
-
-3:
-       movem  [$sp+],$r10
-       addq    44,$r12 ;; Get back count before faulting point.
-       subq    44,$r11 ;; Get back pointer to faulting movem-line.
-       jump    4b      ;; Fall through, pretending the fault didn't happen.
-
-       .previous
-       .section __ex_table,\"a\"
-       .dword 1b,3b
+                                                                       \n\
+       ;; Save the registers we'll use in the movem process            \n\
+       ;; on the stack.                                                \n\
+       subq    11*4,$sp                                                \n\
+       movem   $r10,[$sp]                                              \n\
+                                                                       \n\
+       ;; Now we've got this:                                          \n\
+       ;; r11 - src                                                    \n\
+       ;; r13 - dst                                                    \n\
+       ;; r12 - n                                                      \n\
+                                                                       \n\
+       ;; Update n for the first loop                                  \n\
+       subq    44,$r12                                                 \n\
+0:                                                                     \n\
+       movem   [$r11+],$r10                                            \n\
+1:                                                                     \n\
+       subq   44,$r12                                                  \n\
+       bge     0b                                                      \n\
+       movem   $r10,[$r13+]                                            \n\
+                                                                       \n\
+       addq   44,$r12  ;; compensate for last loop underflowing n      \n\
+                                                                       \n\
+       ;; Restore registers from stack                                 \n\
+       movem [$sp+],$r10                                               \n\
+4:                                                                     \n\
+       .section .fixup,\"ax\"                                          \n\
+                                                                       \n\
+;; Do not jump back into the loop if we fail.  For some uses, we get a \n\
+;; page fault somewhere on the line.  Without checking for page limits,        \n\
+;; we don't know where, but we need to copy accurately and keep an     \n\
+;; accurate count; not just clear the whole line.  To do that, we fall \n\
+;; down in the code below, proceeding with smaller amounts.  It should \n\
+;; be kept in mind that we have to cater to code like what at one time \n\
+;; was in fs/super.c:                                                  \n\
+;;  i = size - copy_from_user((void *)page, data, size);               \n\
+;; which would cause repeated faults while clearing the remainder of   \n\
+;; the SIZE bytes at PAGE after the first fault.                       \n\
+;; A caveat here is that we must not fall through from a failing page  \n\
+;; to a valid page.                                                    \n\
+                                                                       \n\
+3:                                                                     \n\
+       movem  [$sp+],$r10                                              \n\
+       addq    44,$r12 ;; Get back count before faulting point.        \n\
+       subq    44,$r11 ;; Get back pointer to faulting movem-line.     \n\
+       jump    4b      ;; Fall through, pretending the fault didn't happen.\n\
+                                                                       \n\
+       .previous                                                       \n\
+       .section __ex_table,\"a\"                                       \n\
+       .dword 1b,3b                                                    \n\
        .previous"
 
      /* Outputs */ : "=r" (dst), "=r" (src), "=r" (n), "=r" (retn)
@@ -425,64 +425,64 @@ __do_clear_user (void __user *pto, unsigned long pn)
       If you want to check that the allocation was right; then
       check the equalities in the first comment.  It should say
       something like "r13=r13, r11=r11, r12=r12". */
-    __asm__ volatile ("
+    __asm__ volatile ("\n\
        .ifnc %0%1%2,$r13$r12$r10                                       \n\
        .err                                                            \n\
        .endif                                                          \n\
-
-       ;; Save the registers we'll clobber in the movem process
-       ;; on the stack.  Don't mention them to gcc, it will only be
-       ;; upset.
-       subq    11*4,$sp
-       movem   $r10,[$sp]
-
-       clear.d $r0
-       clear.d $r1
-       clear.d $r2
-       clear.d $r3
-       clear.d $r4
-       clear.d $r5
-       clear.d $r6
-       clear.d $r7
-       clear.d $r8
-       clear.d $r9
-       clear.d $r10
-       clear.d $r11
-
-       ;; Now we've got this:
-       ;; r13 - dst
-       ;; r12 - n
-
-       ;; Update n for the first loop
-       subq    12*4,$r12
-0:
-       subq   12*4,$r12
-       bge     0b
-       movem   $r11,[$r13+]
-1:
-       addq   12*4,$r12        ;; compensate for last loop underflowing n
-
-       ;; Restore registers from stack
-       movem [$sp+],$r10
-2:
-       .section .fixup,\"ax\"
-3:
-       move.d [$sp],$r10
-       addq 12*4,$r10
-       move.d $r10,[$sp]
-       clear.d $r10
-       jump 0b
-
-4:
-       movem [$sp+],$r10
-       addq 12*4,$r10
-       addq 12*4,$r12
-       jump 2b
-
-       .previous
-       .section __ex_table,\"a\"
-       .dword 0b,3b
-       .dword 1b,4b
+                                                                       \n\
+       ;; Save the registers we'll clobber in the movem process        \n\
+       ;; on the stack.  Don't mention them to gcc, it will only be    \n\
+       ;; upset.                                                       \n\
+       subq    11*4,$sp                                                \n\
+       movem   $r10,[$sp]                                              \n\
+                                                                       \n\
+       clear.d $r0                                                     \n\
+       clear.d $r1                                                     \n\
+       clear.d $r2                                                     \n\
+       clear.d $r3                                                     \n\
+       clear.d $r4                                                     \n\
+       clear.d $r5                                                     \n\
+       clear.d $r6                                                     \n\
+       clear.d $r7                                                     \n\
+       clear.d $r8                                                     \n\
+       clear.d $r9                                                     \n\
+       clear.d $r10                                                    \n\
+       clear.d $r11                                                    \n\
+                                                                       \n\
+       ;; Now we've got this:                                          \n\
+       ;; r13 - dst                                                    \n\
+       ;; r12 - n                                                      \n\
+                                                                       \n\
+       ;; Update n for the first loop                                  \n\
+       subq    12*4,$r12                                               \n\
+0:                                                                     \n\
+       subq   12*4,$r12                                                \n\
+       bge     0b                                                      \n\
+       movem   $r11,[$r13+]                                            \n\
+1:                                                                     \n\
+       addq   12*4,$r12        ;; compensate for last loop underflowing n\n\
+                                                                       \n\
+       ;; Restore registers from stack                                 \n\
+       movem [$sp+],$r10                                               \n\
+2:                                                                     \n\
+       .section .fixup,\"ax\"                                          \n\
+3:                                                                     \n\
+       move.d [$sp],$r10                                               \n\
+       addq 12*4,$r10                                                  \n\
+       move.d $r10,[$sp]                                               \n\
+       clear.d $r10                                                    \n\
+       jump 0b                                                         \n\
+                                                                       \n\
+4:                                                                     \n\
+       movem [$sp+],$r10                                               \n\
+       addq 12*4,$r10                                                  \n\
+       addq 12*4,$r12                                                  \n\
+       jump 2b                                                         \n\
+                                                                       \n\
+       .previous                                                       \n\
+       .section __ex_table,\"a\"                                       \n\
+       .dword 0b,3b                                                    \n\
+       .dword 1b,4b                                                    \n\
        .previous"
 
      /* Outputs */ : "=r" (dst), "=r" (n), "=r" (retn)