Hi Guys,
I am checking in the attached patch to fix some regressions in the RX
port with regard to alignment and addressing modes. With this patch
applied I have 18 fewer failures in the GCC testsuite.
Cheers
Nick
gcc/ChangeLog
2011-03-16 Nick Clifton <ni...@redhat.com>
* config/rx/rx.h (JUMP_ALIGN): Define.
(JUMP_ALIGN_MAX_SKIP, LABEL_ALIGN_AFTER_BARRIER, LOOP_ALIGN,
(LABEL_ALIGN, LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP,
(LOOP_ALIGN_MAX_SKIP, LABEL_ALIGN_MAX_SKIP): Define.
* config/rx/rx-protos.h (rx_align_for_label): Prototype.
(rx_max_skip_for_label): Prototype.
* config/rx/rx.md (abssi3): Use CC_ZSmode.
(andis3): Fix timings.
(peephole): Add peephole to combine zero- and sign- extending
loads with arithmetic instructions.
(bset): Fix timing.
(bclr): Fix timing.
* config/rx/rx.c (rx_is_legitimate_address): Add checks for QImode
and HImode reg+int address.
(rx_print_operand): Add support for %R. Fix generation of .B and
.W addressing modes.
(rx_align_for_label): New function.
(rx_max_skip_for_label): New function.
Index: gcc/config/rx/rx.h
===================================================================
--- gcc/config/rx/rx.h (revision 171032)
+++ gcc/config/rx/rx.h (working copy)
@@ -663,3 +663,28 @@
#define SELECT_CC_MODE(OP,X,Y) rx_select_cc_mode ((OP), (X), (Y))
+#define JUMP_ALIGN(x) rx_align_for_label (x)
+#define JUMP_ALIGN_MAX_SKIP rx_max_skip_for_label (label)
+#define LABEL_ALIGN_AFTER_BARRIER(x) rx_align_for_label (x)
+#define LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP rx_max_skip_for_label (label)
+#define LOOP_ALIGN(x) rx_align_for_label (x)
+#define LOOP_ALIGN_MAX_SKIP rx_max_skip_for_label (label)
+#define LABEL_ALIGN(x) rx_align_for_label (x)
+#define LABEL_ALIGN_MAX_SKIP rx_max_skip_for_label (NULL_RTX)
+
+#define ASM_OUTPUT_MAX_SKIP_ALIGN(STREAM, LOG, MAX_SKIP) \
+ do \
+ { \
+ if ((LOG) == 0 || (MAX_SKIP) == 0) \
+ break; \
+ if (TARGET_AS100_SYNTAX) \
+ { \
+ if ((LOG) >= 2) \
+ fprintf (STREAM, "\t.ALIGN 4\t; %d alignment actually requested\n", 1 << (LOG)); \
+ else \
+ fprintf (STREAM, "\t.ALIGN 2\n"); \
+ } \
+ else \
+ fprintf (STREAM, "\t.balign %d,3,%d\n", 1 << (LOG), (MAX_SKIP)); \
+ } \
+ while (0)
Index: gcc/config/rx/rx-protos.h
===================================================================
--- gcc/config/rx/rx-protos.h (revision 171032)
+++ gcc/config/rx/rx-protos.h (working copy)
@@ -31,6 +31,7 @@
extern void rx_set_optimization_options (void);
#ifdef RTX_CODE
+extern int rx_align_for_label (rtx);
extern bool rx_compare_redundant (rtx);
extern void rx_emit_stack_popm (rtx *, bool);
extern void rx_emit_stack_pushm (rtx *);
@@ -41,6 +42,7 @@
extern bool rx_is_mode_dependent_addr (rtx);
extern bool rx_is_restricted_memory_address (rtx, Mmode);
extern bool rx_match_ccmode (rtx, Mmode);
+extern int rx_max_skip_for_label (rtx);
extern void rx_notice_update_cc (rtx body, rtx insn);
extern void rx_print_operand (FILE *, rtx, int);
extern void rx_print_operand_address (FILE *, rtx);
Index: gcc/config/rx/rx.md
===================================================================
--- gcc/config/rx/rx.md (revision 171032)
+++ gcc/config/rx/rx.md (working copy)
@@ -798,7 +798,10 @@
(set (reg CC_REG)
(compare (abs:SI (match_dup 1))
(const_int 0)))]
- "reload_completed && rx_match_ccmode (insn, CC_ZSOmode)"
+ ;; Note - although the ABS instruction does set the O bit in the processor
+ ;; status word, it does not do so in a way that is comparable with the CMP
+ ;; instruction. Hence we use CC_ZSmode rather than CC_ZSOmode.
+ "reload_completed && rx_match_ccmode (insn, CC_ZSmode)"
"@
abs\t%0
abs\t%1, %0"
@@ -1005,7 +1008,7 @@
and\t%1, %0
and\t%2, %1, %0
and\t%Q2, %0"
- [(set_attr "timings" "11,11,11,11,11,11,11,33,33")
+ [(set_attr "timings" "11,11,11,11,11,11,11,11,33")
(set_attr "length" "2,2,3,4,5,6,2,5,5")]
)
@@ -1027,7 +1030,7 @@
and\t%1, %0
and\t%2, %1, %0
and\t%Q2, %0"
- [(set_attr "timings" "11,11,11,11,11,11,11,33,33")
+ [(set_attr "timings" "11,11,11,11,11,11,11,11,33")
(set_attr "length" "2,2,3,4,5,6,2,5,5")]
)
@@ -1543,6 +1546,139 @@
(set_attr "length" "3,4,5,6,7,6")]
)
+;; A set of peepholes to catch extending loads followed by arithmetic operations.
+;; We use iterators where possible to reduce the amount of typing and hence the
+;; possibilities for typos.
+
+(define_code_iterator extend_types [(zero_extend "") (sign_extend "")])
+(define_code_attr letter [(zero_extend "R") (sign_extend "Q")])
+
+(define_code_iterator memex_commutative [(plus "") (and "") (ior "") (xor "")])
+(define_code_iterator memex_noncomm [(div "") (udiv "") (minus "")])
+(define_code_iterator memex_nocc [(smax "") (smin "") (mult "")])
+
+(define_code_attr op [(plus "add") (and "and") (div "div") (udiv "divu") (smax "max") (smin "min") (mult "mul") (ior "or") (minus "sub") (xor "xor")])
+
+(define_peephole2
+ [(set (match_operand:SI 0 "register_operand")
+ (extend_types:SI (match_operand:small_int_modes 1 "rx_restricted_mem_operand")))
+ (parallel [(set (match_operand:SI 2 "register_operand")
+ (memex_commutative:SI (match_dup 0)
+ (match_dup 2)))
+ (clobber (reg:CC CC_REG))])]
+ "peep2_regno_dead_p (2, REGNO (operands[0]))"
+ [(parallel [(set:SI (match_dup 2)
+ (memex_commutative:SI (match_dup 2)
+ (extend_types:SI (match_dup 1))))
+ (clobber (reg:CC CC_REG))])]
+)
+
+(define_peephole2
+ [(set (match_operand:SI 0 "register_operand")
+ (extend_types:SI (match_operand:small_int_modes 1 "rx_restricted_mem_operand")))
+ (parallel [(set (match_operand:SI 2 "register_operand")
+ (memex_commutative:SI (match_dup 2)
+ (match_dup 0)))
+ (clobber (reg:CC CC_REG))])]
+ "peep2_regno_dead_p (2, REGNO (operands[0]))"
+ [(parallel [(set:SI (match_dup 2)
+ (memex_commutative:SI (match_dup 2)
+ (extend_types:SI (match_dup 1))))
+ (clobber (reg:CC CC_REG))])]
+)
+
+(define_peephole2
+ [(set (match_operand:SI 0 "register_operand")
+ (extend_types:SI (match_operand:small_int_modes 1 "rx_restricted_mem_operand")))
+ (parallel [(set (match_operand:SI 2 "register_operand")
+ (memex_noncomm:SI (match_dup 2)
+ (match_dup 0)))
+ (clobber (reg:CC CC_REG))])]
+ "peep2_regno_dead_p (2, REGNO (operands[0]))"
+ [(parallel [(set:SI (match_dup 2)
+ (memex_noncomm:SI (match_dup 2)
+ (extend_types:SI (match_dup 1))))
+ (clobber (reg:CC CC_REG))])]
+)
+
+(define_peephole2
+ [(set (match_operand:SI 0 "register_operand")
+ (extend_types:SI (match_operand:small_int_modes 1 "rx_restricted_mem_operand")))
+ (set (match_operand:SI 2 "register_operand")
+ (memex_nocc:SI (match_dup 0)
+ (match_dup 2)))]
+ "peep2_regno_dead_p (2, REGNO (operands[0]))"
+ [(set:SI (match_dup 2)
+ (memex_nocc:SI (match_dup 2)
+ (extend_types:SI (match_dup 1))))]
+)
+
+(define_peephole2
+ [(set (match_operand:SI 0 "register_operand")
+ (extend_types:SI (match_operand:small_int_modes 1 "rx_restricted_mem_operand")))
+ (set (match_operand:SI 2 "register_operand")
+ (memex_nocc:SI (match_dup 2)
+ (match_dup 0)))]
+ "peep2_regno_dead_p (2, REGNO (operands[0]))"
+ [(set:SI (match_dup 2)
+ (memex_nocc:SI (match_dup 2)
+ (extend_types:SI (match_dup 1))))]
+)
+
+(define_insn "*<memex_commutative:code>si3_<extend_types:code><small_int_modes:mode>"
+ [(set (match_operand:SI 0 "register_operand" "=r")
+ (memex_commutative:SI (match_operand:SI 1 "register_operand" "%0")
+ (extend_types:SI (match_operand:small_int_modes 2 "rx_restricted_mem_operand" "Q"))))
+ (clobber (reg:CC CC_REG))]
+ ""
+ "<memex_commutative:op>\t%<extend_types:letter>2, %0"
+ [(set_attr "timings" "33")
+ (set_attr "length" "5")] ;; Worst case sceanario. FIXME: If we defined separate patterns
+) ;; rather than using iterators we could specify exact sizes.
+
+(define_insn "*<memex_noncomm:code>si3_<extend_types:code><small_int_modes:mode>"
+ [(set (match_operand:SI 0 "register_operand" "=r")
+ (memex_noncomm:SI (match_operand:SI 1 "register_operand" "0")
+ (extend_types:SI (match_operand:small_int_modes 2 "rx_restricted_mem_operand" "Q"))))
+ (clobber (reg:CC CC_REG))]
+ ""
+ "<memex_noncomm:op>\t%<extend_types:letter>2, %0"
+ [(set_attr "timings" "33")
+ (set_attr "length" "5")] ;; Worst case sceanario. FIXME: If we defined separate patterns
+) ;; rather than using iterators we could specify exact sizes.
+
+(define_insn "*<memex_nocc:code>si3_<extend_types:code><small_int_modes:mode>"
+ [(set (match_operand:SI 0 "register_operand" "=r")
+ (memex_nocc:SI (match_operand:SI 1 "register_operand" "%0")
+ (extend_types:SI (match_operand:small_int_modes 2 "rx_restricted_mem_operand" "Q"))))]
+ ""
+ "<memex_nocc:op>\t%<extend_types:letter>2, %0"
+ [(set_attr "timings" "33")
+ (set_attr "length" "5")] ;; Worst case sceanario. FIXME: If we defined separate patterns
+) ;; rather than using iterators we could specify exact sizes.
+
+(define_peephole2
+ [(set (match_operand:SI 0 "register_operand")
+ (extend_types:SI (match_operand:small_int_modes 1 "rx_restricted_mem_operand")))
+ (set (reg:CC CC_REG)
+ (compare:CC (match_operand:SI 2 "register_operand")
+ (match_dup 0)))]
+ "peep2_regno_dead_p (2, REGNO (operands[0]))"
+ [(set (reg:CC CC_REG)
+ (compare:CC (match_dup 2)
+ (extend_types:SI (match_dup 1))))]
+)
+
+(define_insn "*comparesi3_<extend_types:code><small_int_modes:mode>"
+ [(set (reg:CC CC_REG)
+ (compare:CC (match_operand:SI 0 "register_operand" "=r")
+ (extend_types:SI (match_operand:small_int_modes 1 "rx_restricted_mem_operand" "Q"))))]
+ ""
+ "cmp\t%<extend_types:letter>1, %0"
+ [(set_attr "timings" "33")
+ (set_attr "length" "5")] ;; Worst case sceanario. FIXME: If we defined separate patterns
+) ;; rather than using iterators we could specify exact sizes.
+
;; Floating Point Instructions
(define_insn "addsf3"
@@ -1641,7 +1777,7 @@
""
"bset\t%1, %0.B"
[(set_attr "length" "3")
- (set_attr "timings" "34")]
+ (set_attr "timings" "33")]
)
(define_insn "*bitinvert"
@@ -1687,7 +1823,7 @@
""
"bclr\t%1, %0.B"
[(set_attr "length" "3")
- (set_attr "timings" "34")]
+ (set_attr "timings" "33")]
)
(define_insn "*insv_imm"
Index: gcc/config/rx/rx.c
===================================================================
--- gcc/config/rx/rx.c (revision 171032)
+++ gcc/config/rx/rx.c (working copy)
@@ -80,7 +80,9 @@
/* Register Indirect. */
return true;
- if (GET_MODE_SIZE (mode) == 4
+ if ((GET_MODE_SIZE (mode) == 4
+ || GET_MODE_SIZE (mode) == 2
+ || GET_MODE_SIZE (mode) == 1)
&& (GET_CODE (x) == PRE_DEC || GET_CODE (x) == POST_INC))
/* Pre-decrement Register Indirect or
Post-increment Register Indirect. */
@@ -182,8 +184,11 @@
base = XEXP (mem, 0);
index = XEXP (mem, 1);
- return RX_REG_P (base) && CONST_INT_P (index);
+ if (! RX_REG_P (base) || ! CONST_INT_P (index))
+ return false;
+ return IN_RANGE (INTVAL (index), 0, (0x10000 * GET_MODE_SIZE (mode)) - 1);
+
case SYMBOL_REF:
/* Can happen when small data is being supported.
Assume that it will be resolved into GP+INT. */
@@ -447,11 +452,14 @@
%L Print low part of a DImode register, integer or address.
%N Print the negation of the immediate value.
%Q If the operand is a MEM, then correctly generate
- register indirect or register relative addressing. */
+ register indirect or register relative addressing.
+ %R Like %Q but for zero-extending loads. */
void
rx_print_operand (FILE * file, rtx op, int letter)
{
+ bool unsigned_load = false;
+
switch (letter)
{
case 'A':
@@ -615,10 +623,15 @@
rx_print_integer (file, - INTVAL (op));
break;
+ case 'R':
+ gcc_assert (GET_MODE_SIZE (GET_MODE (op)) < 4);
+ unsigned_load = true;
+ /* Fall through. */
case 'Q':
if (MEM_P (op))
{
HOST_WIDE_INT offset;
+ rtx mem = op;
op = XEXP (op, 0);
@@ -653,22 +666,24 @@
rx_print_operand (file, op, 0);
fprintf (file, "].");
- switch (GET_MODE_SIZE (GET_MODE (op)))
+ switch (GET_MODE_SIZE (GET_MODE (mem)))
{
case 1:
- gcc_assert (offset < 65535 * 1);
- fprintf (file, "B");
+ gcc_assert (offset <= 65535 * 1);
+ fprintf (file, unsigned_load ? "UB" : "B");
break;
case 2:
gcc_assert (offset % 2 == 0);
- gcc_assert (offset < 65535 * 2);
- fprintf (file, "W");
+ gcc_assert (offset <= 65535 * 2);
+ fprintf (file, unsigned_load ? "UW" : "W");
break;
- default:
+ case 4:
gcc_assert (offset % 4 == 0);
- gcc_assert (offset < 65535 * 4);
+ gcc_assert (offset <= 65535 * 4);
fprintf (file, "L");
break;
+ default:
+ gcc_unreachable ();
}
break;
}
@@ -2656,6 +2671,36 @@
return true;
}
+int
+rx_align_for_label (rtx lab ATTRIBUTE_UNUSED)
+{
+ return optimize_size ? 1 : 3;
+}
+
+int
+rx_max_skip_for_label (rtx lab)
+{
+ int opsize;
+ rtx op;
+
+ if (lab == NULL_RTX)
+ return 0;
+ op = lab;
+ do
+ {
+ op = next_nonnote_insn (op);
+ }
+ while (op && (LABEL_P (op)
+ || (INSN_P (op) && GET_CODE (PATTERN (op)) == USE)));
+ if (!op)
+ return 0;
+
+ opsize = get_attr_length (op);
+ if (opsize >= 0 && opsize < 8)
+ return opsize - 1;
+ return 0;
+}
+
#undef TARGET_FUNCTION_VALUE
#define TARGET_FUNCTION_VALUE rx_function_value
