* i386.c (i386_preferred_stack_boundary_string): New global variable.
(i386_preferred_stack_boundary): New global variable.
(override_functions): Set it. Tidy option setting code.
* i386.h (TARGET_OPTIONS): New command line option.
(i386_preferred_stack_boundary_string): Declare it.
(i386_preferred_stack_boundary): Likewise.
(PREFERRED_STACK_BOUNDARY): Use i386_preferred_stack_boundary.
From-SVN: r26527
+Sat Apr 17 21:10:10 1999 Jan Hubicka <hubicka@freesoft.cz>
+
+ * i386.c (i386_preferred_stack_boundary_string): New global variable.
+ (i386_preferred_stack_boundary): New global variable.
+ (override_functions): Set it. Tidy option setting code.
+ * i386.h (TARGET_OPTIONS): New command line option.
+ (i386_preferred_stack_boundary_string): Declare it.
+ (i386_preferred_stack_boundary): Likewise.
+ (PREFERRED_STACK_BOUNDARY): Use i386_preferred_stack_boundary.
+
Sat Apr 17 19:22:38 1999 Jan Hubicka <hubicka@freesoft.cz>
* i386.c (k6_cost): Take into account the decoding time.
/* Power of two alignment for non-loop jumps. */
char *i386_align_jumps_string;
+/* Power of two alignment for stack boundary in bytes. */
+char *i386_preferred_stack_boundary_string;
+
+/* Preferred alignment for stack boundary in bits. */
+int i386_preferred_stack_boundary;
+
/* Values 1-5: see jump.c */
int i386_branch_cost;
char *i386_branch_cost_string;
struct processor_costs *cost; /* Processor costs */
int target_enable; /* Target flags to enable. */
int target_disable; /* Target flags to disable. */
- } processor_target_table[]
- = {{PROCESSOR_I386_STRING, PROCESSOR_I386, &i386_cost, 0, 0},
- {PROCESSOR_I486_STRING, PROCESSOR_I486, &i486_cost, 0, 0},
- {PROCESSOR_I586_STRING, PROCESSOR_PENTIUM, &pentium_cost, 0, 0},
- {PROCESSOR_PENTIUM_STRING, PROCESSOR_PENTIUM, &pentium_cost, 0, 0},
- {PROCESSOR_I686_STRING, PROCESSOR_PENTIUMPRO, &pentiumpro_cost,
- 0, 0},
- {PROCESSOR_PENTIUMPRO_STRING, PROCESSOR_PENTIUMPRO,
- &pentiumpro_cost, 0, 0},
- {PROCESSOR_K6_STRING, PROCESSOR_K6, &k6_cost, 0, 0}};
+ } processor_target_table[] = {
+ {PROCESSOR_I386_STRING, PROCESSOR_I386, &i386_cost, 0, 0},
+ {PROCESSOR_I486_STRING, PROCESSOR_I486, &i486_cost, 0, 0},
+ {PROCESSOR_I586_STRING, PROCESSOR_PENTIUM, &pentium_cost, 0, 0},
+ {PROCESSOR_PENTIUM_STRING, PROCESSOR_PENTIUM, &pentium_cost, 0, 0},
+ {PROCESSOR_I686_STRING, PROCESSOR_PENTIUMPRO, &pentiumpro_cost, 0, 0},
+ {PROCESSOR_PENTIUMPRO_STRING, PROCESSOR_PENTIUMPRO,
+ &pentiumpro_cost, 0, 0},
+ {PROCESSOR_K6_STRING, PROCESSOR_K6, &k6_cost, 0, 0}
+ };
int ptt_size = sizeof (processor_target_table) / sizeof (struct ptt);
def_align = (TARGET_486) ? 4 : 2;
/* Validate -malign-loops= value, or provide default. */
+#ifdef ASM_OUTPUT_MAX_SKIP_ALIGN
+ i386_align_loops = 4;
+#else
+ i386_align_loops = 2;
+#endif
if (i386_align_loops_string)
{
i386_align_loops = atoi (i386_align_loops_string);
fatal ("-malign-loops=%d is not between 0 and %d",
i386_align_loops, MAX_CODE_ALIGN);
}
- else
+
+ /* Validate -malign-jumps= value, or provide default. */
#ifdef ASM_OUTPUT_MAX_SKIP_ALIGN
- i386_align_loops = 4;
+ i386_align_jumps = 4;
#else
- i386_align_loops = 2;
+ i386_align_jumps = def_align;
#endif
-
- /* Validate -malign-jumps= value, or provide default. */
if (i386_align_jumps_string)
{
i386_align_jumps = atoi (i386_align_jumps_string);
fatal ("-malign-jumps=%d is not between 0 and %d",
i386_align_jumps, MAX_CODE_ALIGN);
}
- else
-#ifdef ASM_OUTPUT_MAX_SKIP_ALIGN
- i386_align_jumps = 4;
-#else
- i386_align_jumps = def_align;
-#endif
/* Validate -malign-functions= value, or provide default. */
+ i386_align_funcs = def_align;
if (i386_align_funcs_string)
{
i386_align_funcs = atoi (i386_align_funcs_string);
fatal ("-malign-functions=%d is not between 0 and %d",
i386_align_funcs, MAX_CODE_ALIGN);
}
- else
- i386_align_funcs = def_align;
+
+ /* Validate -mpreferred_stack_boundary= value, or provide default.
+ The default of 128 bits is for Pentium III's SSE __m128. */
+ i386_preferred_stack_boundary = 128;
+ if (i386_preferred_stack_boundary_string)
+ {
+ i = atoi (i386_preferred_stack_boundary_string);
+ if (i < 2 || i > 31)
+ fatal ("-mpreferred_stack_boundary=%d is not between 2 and 31", i);
+ i386_preferred_stack_boundary = (1 << i) * BITS_PER_UNIT;
+ }
/* Validate -mbranch-cost= value, or provide default. */
+ i386_branch_cost = 1;
if (i386_branch_cost_string)
{
i386_branch_cost = atoi (i386_branch_cost_string);
if (i386_branch_cost < 0 || i386_branch_cost > 5)
- fatal ("-mbranch-cost=%d is not between 0 and 5",
- i386_branch_cost);
+ fatal ("-mbranch-cost=%d is not between 0 and 5", i386_branch_cost);
}
- else
- i386_branch_cost = 1;
/* Keep nonleaf frame pointers. */
if (TARGET_OMIT_LEAF_FRAME_POINTER)
{ "hard-float", MASK_80387, "Use hardware fp" }, \
{ "soft-float", -MASK_80387, "Do not use hardware fp" },\
{ "no-soft-float", MASK_80387, "Use hardware fp" }, \
- { "386", 0, "Optimize for i80386" }, \
- { "486", 0, "Optimize for i80486" }, \
- { "pentium", 0, "Optimize for Pentium" }, \
- { "pentiumpro", 0, "Optimize for Pentium Pro, Pentium II" },\
+ { "386", 0, "Same as -mcpu=i386" }, \
+ { "486", 0, "Same as -mcpu=i486" }, \
+ { "pentium", 0, "Same as -mcpu=pentium" }, \
+ { "pentiumpro", 0, "Same as -mcpu=pentiumpro" }, \
{ "rtd", MASK_RTD, "Alternate calling convention" },\
{ "no-rtd", -MASK_RTD, "Use normal calling convention" },\
{ "align-double", MASK_ALIGN_DOUBLE, "Align some doubles on dword boundary" },\
{ "align-loops=", &i386_align_loops_string, "Loop code aligned to this power of 2" }, \
{ "align-jumps=", &i386_align_jumps_string, "Jump targets are aligned to this power of 2" }, \
{ "align-functions=", &i386_align_funcs_string, "Function starts are aligned to this power of 2" }, \
+ { "preferred-stack-boundary=", &i386_preferred_stack_boundary_string, "Attempt to keep stack aligned to this power of 2" }, \
{ "branch-cost=", &i386_branch_cost_string, "Branches are this expensive (1-5, arbitrary units)" }, \
SUBTARGET_OPTIONS \
}
/* Allocation boundary (in *bits*) for storing arguments in argument list. */
#define PARM_BOUNDARY 32
-/* Boundary (in *bits*) on which stack pointer should be aligned. */
+/* Boundary (in *bits*) on which the stack pointer must be aligned. */
#define STACK_BOUNDARY 32
-/* We want to keep the stack aligned to 128 bits when possible, for the
- benefit of doubles and SSE __m128. But the compiler can not rely on
- the stack having this alignment.*/
-#define PREFERRED_STACK_BOUNDARY 128
+/* Boundary (in *bits*) on which the stack pointer preferrs to be
+ aligned; the compiler cannot rely on having this alignment. */
+#define PREFERRED_STACK_BOUNDARY i386_preferred_stack_boundary
/* Allocation boundary (in *bits*) for the code of a function.
For i486, we get better performance by aligning to a cache
extern char *i386_align_loops_string; /* power of two alignment for loops */
extern char *i386_align_jumps_string; /* power of two alignment for non-loop jumps */
extern char *i386_align_funcs_string; /* power of two alignment for functions */
+extern char *i386_preferred_stack_boundary_string;/* power of two alignment for stack boundary */
extern char *i386_branch_cost_string; /* values 1-5: see jump.c */
extern int i386_regparm; /* i386_regparm_string as a number */
extern int i386_align_loops; /* power of two alignment for loops */
extern int i386_align_jumps; /* power of two alignment for non-loop jumps */
extern int i386_align_funcs; /* power of two alignment for functions */
+extern int i386_preferred_stack_boundary; /* preferred stack boundary alignment in bits */
extern int i386_branch_cost; /* values 1-5: see jump.c */
extern char *hi_reg_name[]; /* names for 16 bit regs */
extern char *qi_reg_name[]; /* names for 8 bit regs (low) */
-mno-wide-multiply -mrtd -malign-double
-mreg-alloc=@var{list} -mregparm=@var{num}
-malign-jumps=@var{num} -malign-loops=@var{num}
--malign-functions=@var{num}
+-malign-functions=@var{num} -mpreferred_stack_boundary=@var{num}
@emph{HPPA Options}
-mbig-switch -mdisable-fpregs -mdisable-indexing
for a 386, and 4 if optimizing for a 486.
@end table
+@item -mpreferred-stack-boundary=@var{num}
+Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
+byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
+the default is 4 (16 bytes or 128 bits).
+
+The stack is required to be aligned on a 4 byte boundary. On Pentium
+and PentiumPro, @code{double} and @code{long double} values should be
+aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
+significant run time performance penalties. On Pentium III, the
+Streaming SIMD Extention (SSE) data type @code{__m128} suffers similar
+penalties if it is not 16 byte aligned.
+
+To ensure proper alignment of this values on the stack, the stack boundary
+must be as aligned as that required by any value stored on the stack.
+Further, every function must be generated such that it keeps the stack
+aligned. Thus calling a function compiled with a higher preferred
+stack boundary from a function compiled with a lower preferred stack
+boundary will most likely misalign the stack. It is recommended that
+libraries that use callbacks always use the default setting.
+
+This extra alignment does consume extra stack space. Code that is sensitive
+to stack space usage, such as embedded systems and operating system kernels,
+may want to reduce the preferred alignment to
+@samp{-mpreferred-stack-boundary=2}.
+@end table
+
@node HPPA Options
@subsection HPPA Options
@cindex HPPA Options
projects / gcc.git / commitdiff