Hi,

some may recall the "micro-optimisation" thread I once started. Here is
now a simple approach to deal with the yet suboptimal mux-code
calculation in user-space. Code saving:

Before:
   text    data     bss     dec     hex filename
  18004     476       8   18488    4838 native/.libs/libnative.so
  27445     696       4   28145    6df1 posix/.libs/libpthread_rt.so

After:
   text    data     bss     dec     hex filename
  17172     476       8   17656    44f8 native/.libs/libnative.so
  26805     696       4   27505    6b71 posix/.libs/libpthread_rt.so

Of course, this also results in a few ops less being executed on each
Xenomai syscall invocation (not many cycles, though).

Tested without problems on x86 so far.

Jan
---
 include/asm-arm/syscall.h      |   25 +++++++++++++------------
 include/asm-blackfin/syscall.h |    4 +++-
 include/asm-i386/syscall.h     |    3 ++-
 include/asm-ia64/syscall.h     |   23 ++++++++++++-----------
 include/asm-powerpc/syscall.h  |    3 ++-
 ksrc/nucleus/shadow.c          |    3 ++-
 6 files changed, 34 insertions(+), 27 deletions(-)

Index: xenomai/include/asm-arm/syscall.h
===================================================================
--- xenomai.orig/include/asm-arm/syscall.h
+++ xenomai/include/asm-arm/syscall.h
@@ -26,6 +26,7 @@
 #include <asm-generic/xenomai/syscall.h>
 
 #define __xn_mux_code(id,op)    ((op << 24)|((id << 16) & 
0xff0000)|(__xn_sys_mux & 0xffff))
+#define __xn_mux_code_shft(shifted_id,op) ((op << 24)|shifted_id|(__xn_sys_mux 
& 0xffff))
 
 #define XENO_ARM_SYSCALL        0x009F0042     /* carefully chosen... */
 
@@ -131,19 +132,19 @@ static inline int __xn_interrupted_p(str
 #define __sys1(x)      __sys2(x)
 
 #define XENOMAI_DO_SYSCALL(nr, id, op, args...)                        \
-  ({                                                                           
    \
-        unsigned long __res;                                           \
+  ({                                                           \
+        unsigned long __res;                                   \
        register unsigned long __res_r0 __asm__ ("r0");         \
-       ASM_INDECL_##nr;                                                        
\
-                                                                               
        \
-       LOADARGS_##nr(__xn_mux_code(id,op), args);                  \
-       __asm__ __volatile__ (                                              \
-"       swi " __sys1(XENO_ARM_SYSCALL)                         \
-               : "=r" (__res_r0)                                               
\
-               : ASM_INPUT_##nr                                                
\
-               : "memory");                                                    
\
-       __res = __res_r0;                                                       
\
-       (int) __res;                                                            
\
+       ASM_INDECL_##nr;                                        \
+                                                               \
+       LOADARGS_##nr(__xn_mux_code_shft(id,op), args);         \
+       __asm__ __volatile__ (                                  \
+"       swi " __sys1(XENO_ARM_SYSCALL)                         \
+               : "=r" (__res_r0)                               \
+               : ASM_INPUT_##nr                                \
+               : "memory");                                    \
+       __res = __res_r0;                                       \
+       (int) __res;                                            \
   })
 
 #define XENOMAI_SYSCALL0(op)                XENOMAI_DO_SYSCALL(0,0,op)
Index: xenomai/include/asm-blackfin/syscall.h
===================================================================
--- xenomai.orig/include/asm-blackfin/syscall.h
+++ xenomai/include/asm-blackfin/syscall.h
@@ -27,6 +27,7 @@
    (i.e. negative syscall number in orig_p0 meaning "non-syscall
    entry"). */
 #define __xn_mux_code(id,op)   ((id << 24)|((op << 16) & 
0xff0000)|(__xn_sys_mux & 0xffff))
+#define __xn_mux_code_fast(shifted_id,op) ((op << 24)|shifted_id|(__xn_sys_mux 
& 0xffff))
 
 /* Local syscalls -- the braindamage thing about this arch is the
    absence of atomic ops usable from user-space; so we export what
@@ -213,7 +214,8 @@ static inline int __xn_interrupted_p(str
   __res;                                                               \
 })
 
-#define XENOMAI_DO_SYSCALL(nr, id, op, args...)        
__emit_syscall##nr(__xn_mux_code(id,op), ##args)
+#define XENOMAI_DO_SYSCALL(nr, id, op, args...) \
+    __emit_syscall##nr(__xn_mux_code_shft(id,op), ##args)
 
 #define XENOMAI_SYSCALL0(op)                XENOMAI_DO_SYSCALL(0,0,op)
 #define XENOMAI_SYSCALL1(op,a1)             XENOMAI_DO_SYSCALL(1,0,op,a1)
Index: xenomai/include/asm-i386/syscall.h
===================================================================
--- xenomai.orig/include/asm-i386/syscall.h
+++ xenomai/include/asm-i386/syscall.h
@@ -23,6 +23,7 @@
 #include <asm-generic/xenomai/syscall.h>
 
 #define __xn_mux_code(id,op)  ((op << 24)|((id << 16) & 
0xff0000)|(__xn_sys_mux & 0x7fff))
+#define __xn_mux_code_shft(shifted_id,op) ((op << 24)|shifted_id|(__xn_sys_mux 
& 0x7fff))
 
 #ifdef __KERNEL__
 
@@ -167,7 +168,7 @@ asm (".L__X'%ebx = 1\n\t"
 
 #define XENOMAI_SKIN_MUX(nr, id, op, args...) \
   ({                                                                 \
-    int muxcode = __xn_mux_code(id,op);                               \
+    int muxcode = __xn_mux_code_shft(id,op);                         \
     unsigned resultvar;                                                      \
     asm volatile (                                                   \
     LOADARGS_##nr                                                    \
Index: xenomai/include/asm-ia64/syscall.h
===================================================================
--- xenomai.orig/include/asm-ia64/syscall.h
+++ xenomai/include/asm-ia64/syscall.h
@@ -24,6 +24,7 @@
 #include <asm-generic/xenomai/syscall.h>
 
 #define __xn_mux_code(id,op)        ((op << 24)|(((id << 16) & 
0xff0000UL)|(__xn_sys_mux & 0xffffUL)))
+#define __xn_mux_code_shft(shifted_id,op) ((op << 24)|shifted_id|(__xn_sys_mux 
& 0xffffUL))
 
 #ifdef __KERNEL__
 
@@ -151,17 +152,17 @@ static inline int __xn_interrupted_p(str
   /* Branch registers.  */                                             \
   "b6", "b7"
 
-#define XENOMAI_SKIN_MUX(nr, id, op, args...)                   \
-  ({                                                            \
-    register long _r15 asm ("r15") = (__xn_mux_code(id,op));    \
-    register long _retval asm ("r8");                           \
-    register long err asm ("r10");                              \
-    LOAD_ARGS_##nr (args);                                      \
-    __asm __volatile ("break %3;;\n\t"                          \
-                      : "=r" (_retval), "=r" (_r15), "=r" (err) \
-                      : "i" (__BREAK_SYSCALL), "1" (_r15)       \
-                       ASM_ARGS_##nr                           \
-                      : "memory" ASM_CLOBBERS_##nr);            \
+#define XENOMAI_SKIN_MUX(nr, id, op, args...)                          \
+  ({                                                                   \
+    register long _r15 asm ("r15") = (__xn_mux_code_shft(id,op));      \
+    register long _retval asm ("r8");                                  \
+    register long err asm ("r10");                                     \
+    LOAD_ARGS_##nr (args);                                             \
+    __asm __volatile ("break %3;;\n\t"                                 \
+                     : "=r" (_retval), "=r" (_r15), "=r" (err)         \
+                     : "i" (__BREAK_SYSCALL), "1" (_r15)               \
+                       ASM_ARGS_##nr                                   \
+                     : "memory" ASM_CLOBBERS_##nr);                    \
     err < 0 ? -_retval : _retval; })
 
 #define XENOMAI_SYS_MUX(nr, op, args...) XENOMAI_SKIN_MUX(nr, 0, op , ##args)
Index: xenomai/include/asm-powerpc/syscall.h
===================================================================
--- xenomai.orig/include/asm-powerpc/syscall.h
+++ xenomai/include/asm-powerpc/syscall.h
@@ -26,6 +26,7 @@
 #include <asm-generic/xenomai/syscall.h>
 
 #define __xn_mux_code(id,op)        ((op << 24)|((id << 16) & 
0xff0000)|(__xn_sys_mux & 0xffff))
+#define __xn_mux_code_shft(shifted_id,op) ((op << 24)|shifted_id|(__xn_sys_mux 
& 0xffff))
 
 #ifdef __KERNEL__
 
@@ -132,7 +133,7 @@ static inline int __xn_interrupted_p(str
        register unsigned long __sc_6  __asm__ ("r6");          \
        register unsigned long __sc_7  __asm__ ("r7");          \
                                                                \
-       LOADARGS_##nr(__xn_mux_code(id,op), args);              \
+       LOADARGS_##nr(__xn_mux_code_shft(id,op), args);         \
        __asm__ __volatile__                                    \
                ("sc           \n\t"                            \
                 "mfcr %0      "                                \
Index: xenomai/ksrc/nucleus/shadow.c
===================================================================
--- xenomai.orig/ksrc/nucleus/shadow.c
+++ xenomai/ksrc/nucleus/shadow.c
@@ -1187,7 +1187,8 @@ static int xnshadow_sys_bind(struct task
                return err;
        }
 
-       return muxid;
+       /* return pre-shifted muxid - saves some cycles in user-space */
+       return __xn_mux_code(muxid,0);
 }
 
 static int xnshadow_sys_info(struct task_struct *curr, struct pt_regs *regs)

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
Xenomai-core mailing list
Xenomai-core@gna.org
https://mail.gna.org/listinfo/xenomai-core

Reply via email to