-2007-06-12 Seongbae Park <seongbae.park@gmail.com>
+2007-06-13 Kazu Hirata <kazu@codesourcery.com>
+
+ * auto-inc-dec.c, c-incpath.c, config/c4x/libgcc.S,
+ config/sh/divcost-analysis, dbgcnt.def, df-core.c,
+ df-problems.c, df-scan.c, df.h, dominance.c, dse.c, regstat.c,
+ tree-data-ref.c, tree-ssa-loop-im.c, tree-ssa-loop-prefetch.c,
+ tree-vect-transform.c: Fix comment typos. Follow spelling
+ conventions.
+
+2007-06-12 Seongbae Park <seongbae.park@gmail.com>
* df-scan.c (df_get_exit-block_use_set): Always add the stack pointer
to the exit block use set.
HAVE_PRE_INCREMENT, HAVE_POST_INCREMENT, HAVE_PRE_DECREMENT or
HAVE_POST_DECREMENT defined.
- 2) c is a contant not equal to the width of the value being accessed
+ 2) c is a constant not equal to the width of the value being accessed
by the pointer. This is useful for machines that have
HAVE_PRE_MODIFY_DISP, HAVE_POST_MODIFY_DISP defined.
/* Move dead note that match PATTERN to TO_INSN from FROM_INSN. We do
not really care about moving any other notes from the inc or add
insn. Moving the REG_EQUAL and REG_EQUIV is clearly wrong and it
- does not appear that there are any other kinds of relavant notes. */
+ does not appear that there are any other kinds of relevant notes. */
static void
move_dead_notes (rtx to_insn, rtx from_insn, rtx pattern)
return false;
}
- /* Need to check that there are no assignemnts to b
+ /* Need to check that there are no assignments to b
before the add insn. */
other_insn
= get_next_ref (REGNO (inc_insn.reg1), bb, reg_next_def);
#if defined (HAVE_DOS_BASED_FILE_SYSTEM)
/* Remove unnecessary trailing slashes. On some versions of MS
Windows, trailing _forward_ slashes cause no problems for stat().
- On newer versions, stat() does not recognise a directory that ends
+ On newer versions, stat() does not recognize a directory that ends
in a '\\' or '/', unless it is a drive root dir, such as "c:/",
where it is obligatory. */
int pathlen = strlen (path);
; Now divisor and dividend are aligned. Do first SUBC by hand, save
; of the forst quotient digit. Then, shift divisor right rather
; than shifting dividend left. This leaves a zero in the top bit of
-; the divident
+; the dividend
;
ldi 1, ar0 ; Initizialize MSB of quotient
lsh rc, ar0 ; create a mask for MSBs
div_le128_neg -> rts 18
sh4-200 absolute divisor range:
- 1 [2..128] [129..64K) [64K..|divident|/256] >=64K,>|divident/256|
+ 1 [2..128] [129..64K) [64K..|dividend|/256] >=64K,>|dividend/256|
udiv 18 22 38 32 30
sdiv pos: 20 24 41 35 32
sdiv neg: 15 25 42 36 33
How does it work ?
- Everytime dbg_cnt(named-counter) is called,
+ Every time dbg_cnt(named-counter) is called,
the counter is incremented for the named-counter.
And the incremented value is compared against the threshold (limit)
specified by the option.
some instruction within the block has changed.
The top layer is the dataflow solution itself. The dataflow solution
-is computed by using an efficient iterative solver and the trasfer
+is computed by using an efficient iterative solver and the transfer
functions. The dataflow solution must be recomputed whenever the
control changes or if one of the transfer function changes.
DF_ANALYZE is completes, the IN and OUT sets for each basic block
contain the computer information. The DF_*_BB_INFO macros can be used
to access these bitvectors. All deferred rescannings are down before
-the transfer functions are recompited.
+the transfer functions are recomputed.
DF_DUMP can then be called to dump the information produce to some
file. This calls DF_DUMP_START, to print the information that is not
rescanned may be impractical. Cse and regrename fall into this
category.
-2) Defered rescanning - Calls to df_insn_rescan, df_notes_rescan, and
+2) Deferred rescanning - Calls to df_insn_rescan, df_notes_rescan, and
df_insn_delete do not immediately change the insn but instead make
a note that the insn needs to be rescanned. The next call to
df_analyze, df_finish_pass, or df_process_deferred_rescans will
/* Remove all of the problems that are not permanent. Scanning, lr,
- ur and live are permanent, the rest are removeable. Also clear all
+ ur and live are permanent, the rest are removable. Also clear all
of the changeable_flags. */
void
dataflow infomation is not being updated properly. You can just
sprinkle calls in until you find the place that is changing an
underlying structure without calling the proper updating
- rountine. */
+ routine. */
void
df_verify (void)
}
-/* Apply the artifical uses and defs at the top of BB in a forwards
+/* Apply the artificial uses and defs at the top of BB in a forwards
direction. */
void
}
-/* Apply the artifical uses and defs at the end of BB in a backwards
+/* Apply the artificial uses and defs at the end of BB in a backwards
direction. */
void
}
-/* Organinze the refs by insn into the table in REF_INFO. If
+/* Organize the refs by insn into the table in REF_INFO. If
blocks_to_analyze is defined, use that set, otherwise the entire
program. Include the defs if INCLUDE_DEFS. Include the uses if
INCLUDE_USES. Include the eq_uses if INCLUDE_EQ_USES. */
/* Return the (conservative) set of hard registers that are defined on
entry to the function.
- It uses df->entry_block_defs to determine which regster
+ It uses df->entry_block_defs to determine which register
reference to include. */
static void
#define df_chain (df->problems_by_index[DF_CHAIN])
#define df_note (df->problems_by_index[DF_NOTE])
-/* This symbol turns on checking that each modfication of the cfg has
+/* This symbol turns on checking that each modification of the cfg has
been identified to the appropriate df routines. It is not part of
verification per se because the check that the final solution has
not changed covers this. However, if the solution is not being
Then, we need to establish the dominance relation among the basic blocks
in BBS. We split the dominance tree by removing the immediate dominator
- edges from BBS, creating a forrest F. We form a graph G whose vertices
+ edges from BBS, creating a forest F. We form a graph G whose vertices
are BBS and ENTRY and X -> Y is an edge of G if there exists an edge
- X' -> Y in CFG such that X' belongs to the tree of the dominance forrest
+ X' -> Y in CFG such that X' belongs to the tree of the dominance forest
whose root is X. We then determine dominance tree of G. Note that
for X, Y in BBS, X dominates Y in CFG if and only if X dominates Y in G.
In this step, we can use arbitrary algorithm to determine dominators.
possible for each address. This pass is a forwards pass through
each basic block. From the point of view of the global technique,
the first pass could examine a block in either direction. The
- forwards ordering is to accomodate cselib.
+ forwards ordering is to accommodate cselib.
We a simplifying assumption: addresses fall into four broad
categories:
c) For any pass that may prespill, there is currently no
mechanism to tell the dse pass that the slot being used has the
special properties that reload uses. It may be that all that is
- requirred is to have those passes make the same calls that reload
+ required is to have those passes make the same calls that reload
does, assuming that the alias sets can be manipulated in the same
way. */
First step.
Scan all of the insns. Any random ordering of the blocks is fine.
- Each block is scanned in forward order to accomodate cselib which
+ Each block is scanned in forward order to accommodate cselib which
is used to remove stores with non-constant bases.
----------------------------------------------------------------------------*/
case PRE_MODIFY:
case POST_MODIFY:
{
- /* We can resuse the add because we are about to delete the
+ /* We can reuse the add because we are about to delete the
insn that contained it. */
rtx add = XEXP (x, 0);
rtx r1 = XEXP (add, 0);
read_info->next = insn_info->read_rec;
insn_info->read_rec = read_info;
- /* We ignore the clobbers in store_info. The is mildly agressive,
+ /* We ignore the clobbers in store_info. The is mildly aggressive,
but there really should not be a clobber followed by a read. */
if (spill_alias_set)
bb_info->last_insn = insn_info;
- /* Cselib clears the table for this case, so we have to essentually
+ /* Cselib clears the table for this case, so we have to essentially
do the same. */
if (NONJUMP_INSN_P (insn)
&& GET_CODE (PATTERN (insn)) == ASM_OPERANDS
algorithm must take a more conservative view of block
mode reads than the local alg does. So to get the case
where you have a store to the frame followed by a non
- overlaping block more read, we look at the active local
+ overlapping block more read, we look at the active local
stores at the end of the function and delete all of the
frame and spill based ones. */
if (stores_off_frame_dead_at_return
/*----------------------------------------------------------------------------
Fifth step.
- Delete the stores that can only be deleted using the global informantion.
+ Delete the stores that can only be deleted using the global information.
----------------------------------------------------------------------------*/
-/* Compute callse crossed for BB. Live is a scratch bitvector. */
+/* Compute calls crossed for BB. Live is a scratch bitvector. */
static void
regstat_bb_compute_calls_crossed (unsigned int bb_index, bitmap live)
/* Analyzes memory reference MEMREF accessed in STMT. The reference
is read if IS_READ is true, write otherwise. Returns the
data_reference description of MEMREF. NEST is the outermost loop of the
- loop nest in that the reference should be analysed. */
+ loop nest in that the reference should be analyzed. */
struct data_reference *
create_data_ref (struct loop *nest, tree memref, tree stmt, bool is_read)
}
/* If the base of the object is not invariant in the loop nest, we cannot
- analyse it. TODO -- in fact, it would suffice to record that there may
+ analyze it. TODO -- in fact, it would suffice to record that there may
be arbitrary dependences in the loops where the base object varies. */
if (!object_address_invariant_in_loop_p (VEC_index (loop_p, loop_nest, 0),
DR_BASE_OBJECT (a)))
/* Stores the data references in STMT to DATAREFS. If there is an unanalyzable
reference, returns false, otherwise returns true. NEST is the outermost
- loop of the loop nest in that the references should be analysed. */
+ loop of the loop nest in that the references should be analyzed. */
static bool
find_data_references_in_stmt (struct loop *nest, tree stmt,
if (TREE_CODE (stmt1) != GIMPLE_MODIFY_STMT)
return stmt;
- /* There is a conversion inbetween possibly inserted by fold. */
+ /* There is a conversion in between possibly inserted by fold. */
t = GIMPLE_STMT_OPERAND (stmt1, 1);
if (TREE_CODE (t) == NOP_EXPR
|| TREE_CODE (t) == CONVERT_EXPR)
}
/* Prepare the references in the form suitable for data dependence
- analysis. We ignore unanalysable data references (the results
+ analysis. We ignore unanalyzable data references (the results
are used just as a heuristics to estimate temporality of the
references, hence we do not need to worry about correctness). */
for (gr = refs; gr; gr = gr->next)
if (DDR_ARE_DEPENDENT (dep) == chrec_dont_know
|| DDR_NUM_DIST_VECTS (dep) == 0)
{
- /* If the dependence cannot be analysed, assume that there might be
+ /* If the dependence cannot be analyzed, assume that there might be
a reuse. */
dist = 0;
}
else
{
- /* The distance vectors are normalised to be always lexicographically
+ /* The distance vectors are normalized to be always lexicographically
positive, hence we cannot tell just from them whether DDR_A comes
before DDR_B or vice versa. However, it is not important,
anyway -- if DDR_A is close to DDR_B, then it is either reused in
"prologue peel iters set conservatively.");
/* If peeling for alignment is unknown, loop bound of main loop becomes
- unkown. */
+ unknown. */
peel_iters_epilogue = vf - 1;
if (vect_print_dump_info (REPORT_DETAILS))
fprintf (vect_dump, "cost model: "
/* TODO: Close dependency between vect_model_*_cost and vectorizable_*
- functions. Design better to avoid maintainence issues. */
+ functions. Design better to avoid maintenance issues. */
/* Function vect_model_reduction_cost.
/* Unaligned software pipeline has a load of an address, an initial
load, and possibly a mask operation to "prime" the loop. However,
if this is an access in a group of loads, which provide strided
- acccess, then the above cost should only be considered for one
+ access, then the above cost should only be considered for one
access in the group. Inside the loop, there is a load op
and a realignment op. */
git://gcc.gnu.org / gcc.git / commitdiff