[git commit nptl] i386/bits/syscalls.h: more compact, but stack-hungry syscall code. disabled so far

[Denys Vlasenko]

commit: http://git.uclibc.org/uClibc/commit/?id=cd50d9a1c93bf1e02f4d4e2df4ece6c881f24719
branch: http://git.uclibc.org/uClibc/commit/?id=refs/heads/nptl

Signed-off-by: Denys Vlasenko <vda.linux at googlemail.com>
---
 libc/sysdeps/linux/i386/bits/syscalls.h |   96 +++++++++++++++++++++++++------
 1 files changed, 78 insertions(+), 18 deletions(-)

diff --git a/libc/sysdeps/linux/i386/bits/syscalls.h b/libc/sysdeps/linux/i386/bits/syscalls.h
index 09d48a2..0427d91 100644
--- a/libc/sysdeps/linux/i386/bits/syscalls.h
+++ b/libc/sysdeps/linux/i386/bits/syscalls.h
@@ -14,13 +14,36 @@
 
 #include <errno.h>
 
+#define INTERNAL_SYSCALL_NCS(name, err, nr, args...) \
+({ \
+	register unsigned int resultvar; \
+	__asm__ __volatile__ ( \
+		LOADARGS_##nr                                   \
+		"movl	%1, %%eax\n\t"                          \
+		"int	$0x80\n\t"                              \
+		RESTOREARGS_##nr                                \
+		: "=a" (resultvar)                              \
+		: "i" (name) ASMFMT_##nr(args) : "memory", "cc" \
+	); \
+	(int) resultvar; \
+})
+
+
+#if 1 /* defined __PIC__ || defined __pic__ */
+
+/* This code avoids pushing/popping ebx as much as possible.
+ * I think the main reason was that older GCCs had problems
+ * with proper saving/restoring of ebx if "b" constraint was used,
+ * which was breaking -fPIC code really badly.
+ * At least gcc 4.2.x seems to not need these tricks anymore,
+ * but this code is still useful because it often avoids
+ * using stack for saving ebx.
+ * Keeping it unconditionally enabled for now.
+ */
+
 /* We need some help from the assembler to generate optimal code.
  * We define some macros here which later will be used.  */
 
-#if defined __SUPPORT_LD_DEBUG__ && defined __DOMULTI__
-#error LD debugging and DOMULTI are incompatible
-#endif
-
 __asm__ (
 #ifdef __DOMULTI__
 	/* Protect against asm macro redefinition (happens in __DOMULTI__ mode).
@@ -74,20 +97,6 @@ __asm__ (
 #endif
 );
 
-#define INTERNAL_SYSCALL_NCS(name, err, nr, args...) \
-({ \
-	register unsigned int resultvar; \
-	__asm__ __volatile__ ( \
-		LOADARGS_##nr                                   \
-		"movl	%1, %%eax\n\t"                          \
-		"int	$0x80\n\t"                              \
-		RESTOREARGS_##nr                                \
-		: "=a" (resultvar)                              \
-		: "i" (name) ASMFMT_##nr(args) : "memory", "cc" \
-	); \
-	(int) resultvar; \
-})
-
 #define LOADARGS_0
 #define LOADARGS_1  "bpushl .L__X'%k2, %k2\n\t" "bmovl .L__X'%k2, %k2\n\t"
 #define LOADARGS_2  LOADARGS_1
@@ -125,5 +134,56 @@ __asm__ (
 #define ASMFMT_6(arg1, arg2, arg3, arg4, arg5, arg6) \
 	, "a" (arg1), "c" (arg2), "d" (arg3), "S" (arg4), "D" (arg5), "m" (arg6)
 
+#else /* !PIC */
+
+/* Simpler code which just uses "b" constraint to load ebx.
+ * Seems to work with gc 4.2.x, and generates slightly smaller,
+ * but slightly slower code. Example (time syscall):
+ *
+ * -	8b 4c 24 04            mov    0x4(%esp),%ecx
+ * -	87 cb                  xchg   %ecx,%ebx
+ * +	53                     push   %ebx
+ * +	8b 5c 24 08            mov    0x8(%esp),%ebx
+ *	b8 0d 00 00 00         mov    $0xd,%eax
+ *	cd 80                  int    $0x80
+ * -	87 cb                  xchg   %ecx,%ebx
+ * +	5b                     pop    %ebx
+ *	c3                     ret
+ *
+ * 2 bytes smaller, but uses stack via "push/pop ebx"
+ */
+
+#define LOADARGS_0
+#define LOADARGS_1
+#define LOADARGS_2
+#define LOADARGS_3
+#define LOADARGS_4
+#define LOADARGS_5
+#define LOADARGS_6  "push %%ebp\n\t" "movl %7, %%ebp\n\t"
+
+#define RESTOREARGS_0
+#define RESTOREARGS_1
+#define RESTOREARGS_2
+#define RESTOREARGS_3
+#define RESTOREARGS_4
+#define RESTOREARGS_5
+#define RESTOREARGS_6  "pop %%ebp\n\t"
+
+#define ASMFMT_0()
+#define ASMFMT_1(arg1) \
+	, "b" (arg1)
+#define ASMFMT_2(arg1, arg2) \
+	, "b" (arg1), "c" (arg2)
+#define ASMFMT_3(arg1, arg2, arg3) \
+	, "b" (arg1), "c" (arg2), "d" (arg3)
+#define ASMFMT_4(arg1, arg2, arg3, arg4) \
+	, "b" (arg1), "c" (arg2), "d" (arg3), "S" (arg4)
+#define ASMFMT_5(arg1, arg2, arg3, arg4, arg5) \
+	, "b" (arg1), "c" (arg2), "d" (arg3), "S" (arg4), "D" (arg5)
+#define ASMFMT_6(arg1, arg2, arg3, arg4, arg5, arg6) \
+	, "b" (arg1), "c" (arg2), "d" (arg3), "S" (arg4), "D" (arg5), "m" (arg6)
+
+#endif /* !PIC */
+
 #endif /* __ASSEMBLER__ */
 #endif /* _BITS_SYSCALLS_H */
-- 
1.6.3.3