On 28/05/26 17:50, Oleg Endo wrote:
External email: Use caution opening links or attachments


On Thu, 2026-05-28 at 17:17 +0530, [email protected] wrote:
From: Dhruv Chawla <[email protected]>

Signed-off-by: Dhruv Chawla <[email protected]>

gcc/ChangeLog:

       * config/sh/sh-mem.cc (sh_expand_cmpstr): Fix typos.
       * config/sh/sh.cc (sh_option_override): Likewise.
       (sh_rtx_costs): Likewise.
       (fixup_addr_diff_vecs): Likewise.
       (sh_expand_epilogue): Likewise.
       (sh_legitimize_address_displacement): Likewise.
       (sh_can_use_simple_return_p): Likewise.
       (sh_find_equiv_gbr_addr): Likewise.
       (sh_check_add_incdec_notes): Likewise.
       * config/sh/sh.md: Likewise.
       * config/sh/sh4-300.md: Likewise.
       * config/sh/sh_treg_combine.cc 
(sh_treg_combine::try_combine_comparisons): Likewise.
       (sh_treg_combine::try_eliminate_cstores): Likewise.
       * config/sh/sync.md: Likewise.
---
  gcc/config/sh/sh-mem.cc          |  2 +-
  gcc/config/sh/sh.cc              | 20 ++++++++++----------
  gcc/config/sh/sh.md              | 14 +++++++-------
  gcc/config/sh/sh4-300.md         |  2 +-
  gcc/config/sh/sh_treg_combine.cc |  6 +++---
  gcc/config/sh/sync.md            |  4 ++--
  6 files changed, 24 insertions(+), 24 deletions(-)

OK, please apply.

Hi Oleg,

Thanks for the review! Could you please test the patch if possible
to make sure it doesn't break any tests? There is a change to sh.md
which modifies console output.

--
Regards,
Dhruv


Best regards,
Oleg Endo



diff --git a/gcc/config/sh/sh-mem.cc b/gcc/config/sh/sh-mem.cc
index 980b55a3a6c..591bdabad6a 100644
--- a/gcc/config/sh/sh-mem.cc
+++ b/gcc/config/sh/sh-mem.cc
@@ -273,7 +273,7 @@ sh_expand_cmpstr (rtx *operands)
    add_int_reg_note (jump, REG_BR_PROB, prob_likely);
    /* end loop.  */

-  /* Fallthu, substract words.  */
+  /* Fallthu, subtract words.  */
    if (TARGET_LITTLE_ENDIAN)
      {
        rtx low_1 = gen_lowpart (HImode, tmp1);
diff --git a/gcc/config/sh/sh.cc b/gcc/config/sh/sh.cc
index 3fe22d389b1..1dbc63f177e 100644
--- a/gcc/config/sh/sh.cc
+++ b/gcc/config/sh/sh.cc
@@ -847,7 +847,7 @@ sh_option_override (void)
        sh_cpu = PROCESSOR_SH4A;
      }

-  /* User/priviledged mode is supported only on SH3* and SH4*.
+  /* User/privileged mode is supported only on SH3* and SH4*.
       Disable it for everything else.  */
    if (!TARGET_SH3 && TARGET_USERMODE)
      TARGET_USERMODE = false;
@@ -3291,7 +3291,7 @@ sh_rtx_costs (rtx x, machine_mode mode ATTRIBUTE_UNUSED, 
int outer_code,
         && XEXP (x, 1) == const1_rtx
         && CONST_INT_P (XEXP (x, 2))
         && CONST_INT_P (XEXP (XEXP (x, 0), 1))
-       /* Check that the xor constaint overlaps with the extracted bit.  */
+       /* Check that the xor constraint overlaps with the extracted bit.  */
         && (INTVAL (XEXP (XEXP (x, 0), 1)) & (1LL << INTVAL (XEXP (x, 2)))))
       {
         *total = 1; //COSTS_N_INSNS (1);
@@ -5728,7 +5728,7 @@ fixup_addr_diff_vecs (rtx_insn *first)
           break;
       }
        /* FIXME: This is a bug in the optimizer, but it seems harmless
-      to just avoid panicing.  */
+      to just avoid panicking.  */
        if (!prev)
       continue;

@@ -7298,7 +7298,7 @@ sh_expand_epilogue (bool sibcall_p)
               else
                 break;

-           /* Use movml when all banked register are poped.  */
+           /* Use movml when all banked register are popped.  */
             if (count == LAST_BANKED_REG - FIRST_BANKED_REG + 1)
               use_movml = true;
           }
@@ -11451,7 +11451,7 @@ sh_legitimize_address_displacement (rtx *offset1, rtx 
*offset2,
    return false;
  }

-/* Return true if movsf insn should be splited with an additional
+/* Return true if movsf insn should be split with an additional
     register.  */
  bool
  sh_movsf_ie_ra_split_p (rtx op0, rtx op1, rtx op2)
@@ -11552,7 +11552,7 @@ sh_can_use_simple_return_p (void)
    if (! reload_completed || frame_pointer_needed)
      return false;

-  /* Moving prologue around does't reduce the size.  */
+  /* Moving prologue around doesn't reduce the size.  */
    if (optimize_function_for_size_p (cfun))
      return false;

@@ -11719,7 +11719,7 @@ sh_find_equiv_gbr_addr (rtx_insn* insn, rtx mem)
      {
        /* If GBR is marked as call clobbered we bail out if we see a call.
        FIXME: Actually should check if this mem refers to the gbr value
-      before or after the call.  If there is a store_gbr preceeding this
+      before or after the call.  If there is a store_gbr preceding this
        mem, it's safe to use GBR for this mem.

        If GBR is not marked as call clobbered, but there is some other
@@ -11884,7 +11884,7 @@ sh_check_add_incdec_notes (rtx_insn* i)
    return i;
  }

-/* Given a move insn destiation and a source, make sure that the move source
+/* Given a move insn destination and a source, make sure that the move source
     operand is not a post-inc mem load with the same address reg as the
     destination.  Returns the modified source operand with the post-inc removed
     if necessary.  */
@@ -12251,7 +12251,7 @@ private:
  };

  /* Given an rtx x, determine whether the expression can be used to create
-   an insn that calulates x and stores the result in the T bit.
+   an insn that calculates x and stores the result in the T bit.
     This is used by the 'treg_set_expr' predicate to construct insns sequences
     where T bit results are fed into other insns, such as addc, subc, negc
     insns.
@@ -12260,7 +12260,7 @@ private:
     distinguish between 'positive' and 'negative' forms.  For now this has to
     be done in the preparation code.  We could also introduce
     'pos_treg_set_expr' and 'neg_treg_set_expr' predicates for that and write
-   two different patterns for the 'postive' and 'negative' forms.  However,
+   two different patterns for the 'positive' and 'negative' forms.  However,
     the total amount of lines of code seems to be about the same and the
     '{pos|neg}_treg_set_expr' predicates would be more expensive, because the
     recog function would need to look inside the expression by temporarily
diff --git a/gcc/config/sh/sh.md b/gcc/config/sh/sh.md
index 07f8968b0f5..98790f6f7fe 100644
--- a/gcc/config/sh/sh.md
+++ b/gcc/config/sh/sh.md
@@ -823,7 +823,7 @@

  ;; Sometimes combine fails to form the (eq (and (op) (op)) 0) tst insn.
  ;; Try to fix that in the split1 pass by looking for the previous set
-;; of the tested op.  Also see if there is a preceeding sign/zero
+;; of the tested op.  Also see if there is a preceding sign/zero
  ;; extension that can be avoided.
  (define_split
    [(set (reg:SI T_REG)
@@ -854,7 +854,7 @@
        && !sh_insn_operands_modified_between_p (op.insn, op.insn, curr_insn))
      {
        if (dump_file)
-     fprintf (dump_file, "cmpeqsi_t: found preceeding and in insn %d\n",
+     fprintf (dump_file, "cmpeqsi_t: found preceding and in insn %d\n",
                INSN_UID (op.insn));

        if (!(arith_reg_operand (XEXP (op.set_src, 0), SImode)
@@ -1653,7 +1653,7 @@
     (set (match_dup 0) (plus:SI (match_dup 0) (const_int 1)))])


-;; The tree optimiziers canonicalize
+;; The tree optimizers canonicalize
  ;;    reg + (reg & 1)
  ;; into
  ;;    (reg + 1) & -2
@@ -5318,7 +5318,7 @@
  })

  ;; The address %0 is assumed to be 4-aligned at least.  Thus, by ORing
-;; 0xf0000008, we get the low-oder bits *1*00 (binary), which fits
+;; 0xf0000008, we get the low-order bits *1*00 (binary), which fits
  ;; the requirement *1*00 for associative address writes.  The alignment of
  ;; %0 implies that its least significant bit is cleared,
  ;; thus we clear the V bit of a matching entry if there is one.
@@ -5438,11 +5438,11 @@
  ;; placed into delay slots.  Since there is no QImode PC relative load, the
  ;; Q constraint and general_movsrc_operand will reject it for QImode.
  ;; The Sid/Ssd alternatives should come before Sdd in order to avoid
-;; a preference of using r0 als the register operand for addressing modes
+;; a preference of using r0 as the register operand for addressing modes
  ;; other than displacement addressing.
  ;; The Sdd alternatives allow only r0 as register operand, even though on
  ;; SH2A any register could be allowed by switching to a 32 bit insn.
-;; Generally sticking to the r0 is preferrable, since it generates smaller
+;; Generally sticking to the r0 is preferable, since it generates smaller
  ;; code.  Obvious r0 reloads can then be eliminated with a peephole on SH2A.
  (define_insn "*mov<mode>"
    [(set (match_operand:QIHI 0 "general_movdst_operand"
@@ -10892,7 +10892,7 @@
  ;; to be one.  It tries to convert a sequence such as
  ;;   movt    r2      ->      movt    r2
  ;;   movt    r13             mov     r2,r13
-;; This gives the schduler a bit more freedom to hoist a following
+;; This gives the scheduler a bit more freedom to hoist a following
  ;; comparison insn.  Moreover, it the reg-reg mov insn is MT group which has
  ;; better chances for parallel execution.
  ;; We can do this with a peephole2 pattern, but then the cprop_hardreg
diff --git a/gcc/config/sh/sh4-300.md b/gcc/config/sh/sh4-300.md
index df128b7fff8..0f31ed85a4e 100644
--- a/gcc/config/sh/sh4-300.md
+++ b/gcc/config/sh/sh4-300.md
@@ -269,7 +269,7 @@
         (eq_attr "type" "arith3"))
    "sh4_300_issue,all*4")

-;; arith3b insns without brach redirection make use of the 0-offset 0-latency
+;; arith3b insns without branch redirection make use of the 0-offset 0-latency
  ;; branch feature, and thus schedule the same no matter if the branch is taken
  ;; or not.  If the branch is redirected, the taken branch might take longer,
  ;; but then, we don't have to take the next branch.
diff --git a/gcc/config/sh/sh_treg_combine.cc b/gcc/config/sh/sh_treg_combine.cc
index df820d7d39c..05225365b99 100644
--- a/gcc/config/sh/sh_treg_combine.cc
+++ b/gcc/config/sh/sh_treg_combine.cc
@@ -57,7 +57,7 @@ checking whether the ccreg is set before the conditional 
branch
  by testing another register for != 0, which was set by a ccreg store.
  This can be optimized by eliminating the redundant comparison and
  inverting the branch condition.  There can be multiple comparisons in
-different basic blocks that all end up in the redunant test insn before the
+different basic blocks that all end up in the redundant test insn before the
  conditional branch.  Some example RTL ...

  Example 1)
@@ -1085,7 +1085,7 @@ sh_treg_combine::try_combine_comparisons (cbranch_trace& 
trace,
      {
        int i_empty_count = i->setcc.empty () + i->cstore.empty ();

-      // A completly empty entry is OK (could be the BB of the cbranch).
+      // A completely empty entry is OK (could be the BB of the cbranch).
        if (i_empty_count == 2)
       continue;

@@ -1261,7 +1261,7 @@ sh_treg_combine::try_eliminate_cstores (cbranch_trace& 
trace,
    for (std::list<bb_entry>::const_iterator i = trace.bb_entries.begin ();
         i != trace.bb_entries.end (); ++i)
      {
-      // A completly empty entry is OK (could be the BB of the cbranch).
+      // A completely empty entry is OK (could be the BB of the cbranch).
        if (i->setcc.empty () && i->cstore.empty ())
       continue;

diff --git a/gcc/config/sh/sync.md b/gcc/config/sh/sync.md
index 2eca3accbc8..b0b412b3d9d 100644
--- a/gcc/config/sh/sync.md
+++ b/gcc/config/sh/sync.md
@@ -42,7 +42,7 @@
  ;; Hardware Atomics (-matomic-model=hard-llcs; SH4A only)
  ;;
  ;; Hardware atomics implement all atomic operations using the 'movli.l' and
-;; 'movco.l' instructions that are availble on SH4A.  On multi-core hardware
+;; 'movco.l' instructions that are available on SH4A.  On multi-core hardware
  ;; configurations hardware atomics is the only safe mode.
  ;; However, it can also be safely used on single-core configurations.
  ;; Since these instructions operate on SImode memory only, QImode and HImode
@@ -196,7 +196,7 @@
    [(plus "add") (minus "sub") (ior "or") (xor "xor") (and "and")])

  
;;------------------------------------------------------------------------------
-;; comapre and swap
+;; compare and swap

  ;; Only the hard_llcs SImode patterns can use an I08 for the comparison
  ;; or for the new swapped in value.
--
2.43.0

Reply via email to