Update to libjpeg_turbo 1.4.90

(Duplicate of https://codereview.chromium.org/1939823002/ for landing.)

TBR=noel@chromium.org,thakis@chromium.org
BUG=608347, 398235, 591927

Review URL: https://codereview.chromium.org/1953443002 .

1 files changed, 202 insertions, 124 deletions

diff --git a/jcdctmgr.c b/jcdctmgr.c
index 3234a01..aef8517 100644
--- a/jcdctmgr.c
+++ b/jcdctmgr.c
@@ -6,8 +6,9 @@
  * libjpeg-turbo Modifications:
  * Copyright (C) 1999-2006, MIYASAKA Masaru.
  * Copyright 2009 Pierre Ossman <ossman@cendio.se> for Cendio AB
- * Copyright (C) 2011 D. R. Commander
- * For conditions of distribution and use, see the accompanying README file.
+ * Copyright (C) 2011, 2014-2015, D. R. Commander.
+ * For conditions of distribution and use, see the accompanying README.ijg
+ * file.
  *
  * This file contains the forward-DCT management logic.
  * This code selects a particular DCT implementation to be used,
@@ -18,33 +19,32 @@
 #define JPEG_INTERNALS
 #include "jinclude.h"
 #include "jpeglib.h"
-#include "jdct.h"		/* Private declarations for DCT subsystem */
+#include "jdct.h"               /* Private declarations for DCT subsystem */
 #include "jsimddct.h"
 
 
 /* Private subobject for this module */
 
-typedef JMETHOD(void, forward_DCT_method_ptr, (DCTELEM * data));
-typedef JMETHOD(void, float_DCT_method_ptr, (FAST_FLOAT * data));
+typedef void (*forward_DCT_method_ptr) (DCTELEM *data);
+typedef void (*float_DCT_method_ptr) (FAST_FLOAT *data);
 
-typedef JMETHOD(void, convsamp_method_ptr,
-                (JSAMPARRAY sample_data, JDIMENSION start_col,
-                 DCTELEM * workspace));
-typedef JMETHOD(void, float_convsamp_method_ptr,
-                (JSAMPARRAY sample_data, JDIMENSION start_col,
-                 FAST_FLOAT *workspace));
+typedef void (*convsamp_method_ptr) (JSAMPARRAY sample_data,
+                                     JDIMENSION start_col,
+                                     DCTELEM *workspace);
+typedef void (*float_convsamp_method_ptr) (JSAMPARRAY sample_data,
+                                           JDIMENSION start_col,
+                                           FAST_FLOAT *workspace);
 
-typedef JMETHOD(void, quantize_method_ptr,
-                (JCOEFPTR coef_block, DCTELEM * divisors,
-                 DCTELEM * workspace));
-typedef JMETHOD(void, float_quantize_method_ptr,
-                (JCOEFPTR coef_block, FAST_FLOAT * divisors,
-                 FAST_FLOAT * workspace));
+typedef void (*quantize_method_ptr) (JCOEFPTR coef_block, DCTELEM *divisors,
+                                     DCTELEM *workspace);
+typedef void (*float_quantize_method_ptr) (JCOEFPTR coef_block,
+                                           FAST_FLOAT *divisors,
+                                           FAST_FLOAT *workspace);
 
 METHODDEF(void) quantize (JCOEFPTR, DCTELEM *, DCTELEM *);
 
 typedef struct {
-  struct jpeg_forward_dct pub;	/* public fields */
+  struct jpeg_forward_dct pub;  /* public fields */
 
   /* Pointer to the DCT routine actually in use */
   forward_DCT_method_ptr dct;
@@ -55,27 +55,30 @@ typedef struct {
    * entries, because of scaling (especially for an unnormalized DCT).
    * Each table is given in normal array order.
    */
-  DCTELEM * divisors[NUM_QUANT_TBLS];
+  DCTELEM *divisors[NUM_QUANT_TBLS];
 
   /* work area for FDCT subroutine */
-  DCTELEM * workspace;
+  DCTELEM *workspace;
 
 #ifdef DCT_FLOAT_SUPPORTED
   /* Same as above for the floating-point case. */
   float_DCT_method_ptr float_dct;
   float_convsamp_method_ptr float_convsamp;
   float_quantize_method_ptr float_quantize;
-  FAST_FLOAT * float_divisors[NUM_QUANT_TBLS];
-  FAST_FLOAT * float_workspace;
+  FAST_FLOAT *float_divisors[NUM_QUANT_TBLS];
+  FAST_FLOAT *float_workspace;
 #endif
 } my_fdct_controller;
 
-typedef my_fdct_controller * my_fdct_ptr;
+typedef my_fdct_controller *my_fdct_ptr;
 
 
+#if BITS_IN_JSAMPLE == 8
+
 /*
  * Find the highest bit in an integer through binary search.
  */
+
 LOCAL(int)
 flss (UINT16 val)
 {
@@ -106,6 +109,7 @@ flss (UINT16 val)
   return bit;
 }
 
+
 /*
  * Compute values to do a division using reciprocal.
  *
@@ -147,7 +151,7 @@ flss (UINT16 val)
  *
  * In order to allow SIMD implementations we also tweak the values to
  * allow the same calculation to be made at all times:
- * 
+ *
  *   dctbl[0] = f rounded to nearest integer
  *   dctbl[1] = divisor / 2 (+ 1 if fractional part of f < 0.5)
  *   dctbl[2] = 1 << ((word size) * 2 - r)
@@ -164,13 +168,27 @@ flss (UINT16 val)
  * of in a consecutive manner, yet again in order to allow SIMD
  * routines.
  */
+
 LOCAL(int)
-compute_reciprocal (UINT16 divisor, DCTELEM * dtbl)
+compute_reciprocal (UINT16 divisor, DCTELEM *dtbl)
 {
   UDCTELEM2 fq, fr;
   UDCTELEM c;
   int b, r;
 
+  if (divisor == 1) {
+    /* divisor == 1 means unquantized, so these reciprocal/correction/shift
+     * values will cause the C quantization algorithm to act like the
+     * identity function.  Since only the C quantization algorithm is used in
+     * these cases, the scale value is irrelevant.
+     */
+    dtbl[DCTSIZE2 * 0] = (DCTELEM) 1;                       /* reciprocal */
+    dtbl[DCTSIZE2 * 1] = (DCTELEM) 0;                       /* correction */
+    dtbl[DCTSIZE2 * 2] = (DCTELEM) 1;                       /* scale */
+    dtbl[DCTSIZE2 * 3] = -(DCTELEM) (sizeof(DCTELEM) * 8);  /* shift */
+    return 0;
+  }
+
   b = flss(divisor) - 1;
   r  = sizeof(DCTELEM) * 8 + b;
 
@@ -191,13 +209,20 @@ compute_reciprocal (UINT16 divisor, DCTELEM * dtbl)
 
   dtbl[DCTSIZE2 * 0] = (DCTELEM) fq;      /* reciprocal */
   dtbl[DCTSIZE2 * 1] = (DCTELEM) c;       /* correction + roundfactor */
+#ifdef WITH_SIMD
   dtbl[DCTSIZE2 * 2] = (DCTELEM) (1 << (sizeof(DCTELEM)*8*2 - r));  /* scale */
+#else
+  dtbl[DCTSIZE2 * 2] = 1;
+#endif
   dtbl[DCTSIZE2 * 3] = (DCTELEM) r - sizeof(DCTELEM)*8; /* shift */
 
   if(r <= 16) return 0;
   else return 1;
 }
 
+#endif
+
+
 /*
  * Initialize for a processing pass.
  * Verify that all referenced Q-tables are present, and set up
@@ -213,15 +238,15 @@ start_pass_fdctmgr (j_compress_ptr cinfo)
   my_fdct_ptr fdct = (my_fdct_ptr) cinfo->fdct;
   int ci, qtblno, i;
   jpeg_component_info *compptr;
-  JQUANT_TBL * qtbl;
-  DCTELEM * dtbl;
+  JQUANT_TBL *qtbl;
+  DCTELEM *dtbl;
 
   for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
        ci++, compptr++) {
     qtblno = compptr->quant_tbl_no;
     /* Make sure specified quantization table is present */
     if (qtblno < 0 || qtblno >= NUM_QUANT_TBLS ||
-	cinfo->quant_tbl_ptrs[qtblno] == NULL)
+        cinfo->quant_tbl_ptrs[qtblno] == NULL)
       ERREXIT1(cinfo, JERR_NO_QUANT_TABLE, qtblno);
     qtbl = cinfo->quant_tbl_ptrs[qtblno];
     /* Compute divisors for this quant table */
@@ -233,91 +258,102 @@ start_pass_fdctmgr (j_compress_ptr cinfo)
        * coefficients multiplied by 8 (to counteract scaling).
        */
       if (fdct->divisors[qtblno] == NULL) {
-	fdct->divisors[qtblno] = (DCTELEM *)
-	  (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				      (DCTSIZE2 * 4) * SIZEOF(DCTELEM));
+        fdct->divisors[qtblno] = (DCTELEM *)
+          (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
+                                      (DCTSIZE2 * 4) * sizeof(DCTELEM));
       }
       dtbl = fdct->divisors[qtblno];
       for (i = 0; i < DCTSIZE2; i++) {
-	if(!compute_reciprocal(qtbl->quantval[i] << 3, &dtbl[i])
-	  && fdct->quantize == jsimd_quantize)
-	  fdct->quantize = quantize;
+#if BITS_IN_JSAMPLE == 8
+        if (!compute_reciprocal(qtbl->quantval[i] << 3, &dtbl[i]) &&
+            fdct->quantize == jsimd_quantize)
+          fdct->quantize = quantize;
+#else
+        dtbl[i] = ((DCTELEM) qtbl->quantval[i]) << 3;
+#endif
       }
       break;
 #endif
 #ifdef DCT_IFAST_SUPPORTED
     case JDCT_IFAST:
       {
-	/* For AA&N IDCT method, divisors are equal to quantization
-	 * coefficients scaled by scalefactor[row]*scalefactor[col], where
-	 *   scalefactor[0] = 1
-	 *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
-	 * We apply a further scale factor of 8.
-	 */
+        /* For AA&N IDCT method, divisors are equal to quantization
+         * coefficients scaled by scalefactor[row]*scalefactor[col], where
+         *   scalefactor[0] = 1
+         *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
+         * We apply a further scale factor of 8.
+         */
 #define CONST_BITS 14
-	static const INT16 aanscales[DCTSIZE2] = {
-	  /* precomputed values scaled up by 14 bits */
-	  16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
-	  22725, 31521, 29692, 26722, 22725, 17855, 12299,  6270,
-	  21407, 29692, 27969, 25172, 21407, 16819, 11585,  5906,
-	  19266, 26722, 25172, 22654, 19266, 15137, 10426,  5315,
-	  16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
-	  12873, 17855, 16819, 15137, 12873, 10114,  6967,  3552,
-	   8867, 12299, 11585, 10426,  8867,  6967,  4799,  2446,
-	   4520,  6270,  5906,  5315,  4520,  3552,  2446,  1247
-	};
-	SHIFT_TEMPS
-
-	if (fdct->divisors[qtblno] == NULL) {
-	  fdct->divisors[qtblno] = (DCTELEM *)
-	    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-					(DCTSIZE2 * 4) * SIZEOF(DCTELEM));
-	}
-	dtbl = fdct->divisors[qtblno];
-	for (i = 0; i < DCTSIZE2; i++) {
-	  if(!compute_reciprocal(
-	    DESCALE(MULTIPLY16V16((INT32) qtbl->quantval[i],
-				  (INT32) aanscales[i]),
-		    CONST_BITS-3), &dtbl[i])
-	    && fdct->quantize == jsimd_quantize)
-	    fdct->quantize = quantize;
-	}
+        static const INT16 aanscales[DCTSIZE2] = {
+          /* precomputed values scaled up by 14 bits */
+          16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
+          22725, 31521, 29692, 26722, 22725, 17855, 12299,  6270,
+          21407, 29692, 27969, 25172, 21407, 16819, 11585,  5906,
+          19266, 26722, 25172, 22654, 19266, 15137, 10426,  5315,
+          16384, 22725, 21407, 19266, 16384, 12873,  8867,  4520,
+          12873, 17855, 16819, 15137, 12873, 10114,  6967,  3552,
+           8867, 12299, 11585, 10426,  8867,  6967,  4799,  2446,
+           4520,  6270,  5906,  5315,  4520,  3552,  2446,  1247
+        };
+        SHIFT_TEMPS
+
+        if (fdct->divisors[qtblno] == NULL) {
+          fdct->divisors[qtblno] = (DCTELEM *)
+            (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
+                                        (DCTSIZE2 * 4) * sizeof(DCTELEM));
+        }
+        dtbl = fdct->divisors[qtblno];
+        for (i = 0; i < DCTSIZE2; i++) {
+#if BITS_IN_JSAMPLE == 8
+          if (!compute_reciprocal(
+                DESCALE(MULTIPLY16V16((JLONG) qtbl->quantval[i],
+                                      (JLONG) aanscales[i]),
+                        CONST_BITS-3), &dtbl[i]) &&
+              fdct->quantize == jsimd_quantize)
+            fdct->quantize = quantize;
+#else
+           dtbl[i] = (DCTELEM)
+             DESCALE(MULTIPLY16V16((JLONG) qtbl->quantval[i],
+                                   (JLONG) aanscales[i]),
+                     CONST_BITS-3);
+#endif
+        }
       }
       break;
 #endif
 #ifdef DCT_FLOAT_SUPPORTED
     case JDCT_FLOAT:
       {
-	/* For float AA&N IDCT method, divisors are equal to quantization
-	 * coefficients scaled by scalefactor[row]*scalefactor[col], where
-	 *   scalefactor[0] = 1
-	 *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
-	 * We apply a further scale factor of 8.
-	 * What's actually stored is 1/divisor so that the inner loop can
-	 * use a multiplication rather than a division.
-	 */
-	FAST_FLOAT * fdtbl;
-	int row, col;
-	static const double aanscalefactor[DCTSIZE] = {
-	  1.0, 1.387039845, 1.306562965, 1.175875602,
-	  1.0, 0.785694958, 0.541196100, 0.275899379
-	};
-
-	if (fdct->float_divisors[qtblno] == NULL) {
-	  fdct->float_divisors[qtblno] = (FAST_FLOAT *)
-	    (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-					DCTSIZE2 * SIZEOF(FAST_FLOAT));
-	}
-	fdtbl = fdct->float_divisors[qtblno];
-	i = 0;
-	for (row = 0; row < DCTSIZE; row++) {
-	  for (col = 0; col < DCTSIZE; col++) {
-	    fdtbl[i] = (FAST_FLOAT)
-	      (1.0 / (((double) qtbl->quantval[i] *
-		       aanscalefactor[row] * aanscalefactor[col] * 8.0)));
-	    i++;
-	  }
-	}
+        /* For float AA&N IDCT method, divisors are equal to quantization
+         * coefficients scaled by scalefactor[row]*scalefactor[col], where
+         *   scalefactor[0] = 1
+         *   scalefactor[k] = cos(k*PI/16) * sqrt(2)    for k=1..7
+         * We apply a further scale factor of 8.
+         * What's actually stored is 1/divisor so that the inner loop can
+         * use a multiplication rather than a division.
+         */
+        FAST_FLOAT *fdtbl;
+        int row, col;
+        static const double aanscalefactor[DCTSIZE] = {
+          1.0, 1.387039845, 1.306562965, 1.175875602,
+          1.0, 0.785694958, 0.541196100, 0.275899379
+        };
+
+        if (fdct->float_divisors[qtblno] == NULL) {
+          fdct->float_divisors[qtblno] = (FAST_FLOAT *)
+            (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
+                                        DCTSIZE2 * sizeof(FAST_FLOAT));
+        }
+        fdtbl = fdct->float_divisors[qtblno];
+        i = 0;
+        for (row = 0; row < DCTSIZE; row++) {
+          for (col = 0; col < DCTSIZE; col++) {
+            fdtbl[i] = (FAST_FLOAT)
+              (1.0 / (((double) qtbl->quantval[i] *
+                       aanscalefactor[row] * aanscalefactor[col] * 8.0)));
+            i++;
+          }
+        }
       }
       break;
 #endif
@@ -334,7 +370,7 @@ start_pass_fdctmgr (j_compress_ptr cinfo)
  */
 
 METHODDEF(void)
-convsamp (JSAMPARRAY sample_data, JDIMENSION start_col, DCTELEM * workspace)
+convsamp (JSAMPARRAY sample_data, JDIMENSION start_col, DCTELEM *workspace)
 {
   register DCTELEM *workspaceptr;
   register JSAMPROW elemptr;
@@ -344,7 +380,7 @@ convsamp (JSAMPARRAY sample_data, JDIMENSION start_col, DCTELEM * workspace)
   for (elemr = 0; elemr < DCTSIZE; elemr++) {
     elemptr = sample_data[elemr] + start_col;
 
-#if DCTSIZE == 8		/* unroll the inner loop */
+#if DCTSIZE == 8                /* unroll the inner loop */
     *workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
     *workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
     *workspaceptr++ = GETJSAMPLE(*elemptr++) - CENTERJSAMPLE;
@@ -369,14 +405,18 @@ convsamp (JSAMPARRAY sample_data, JDIMENSION start_col, DCTELEM * workspace)
  */
 
 METHODDEF(void)
-quantize (JCOEFPTR coef_block, DCTELEM * divisors, DCTELEM * workspace)
+quantize (JCOEFPTR coef_block, DCTELEM *divisors, DCTELEM *workspace)
 {
   int i;
   DCTELEM temp;
-  UDCTELEM recip, corr, shift;
-  UDCTELEM2 product;
   JCOEFPTR output_ptr = coef_block;
 
+#if BITS_IN_JSAMPLE == 8
+
+  UDCTELEM recip, corr;
+  int shift;
+  UDCTELEM2 product;
+
   for (i = 0; i < DCTSIZE2; i++) {
     temp = workspace[i];
     recip = divisors[i + DCTSIZE2 * 0];
@@ -387,16 +427,54 @@ quantize (JCOEFPTR coef_block, DCTELEM * divisors, DCTELEM * workspace)
       temp = -temp;
       product = (UDCTELEM2)(temp + corr) * recip;
       product >>= shift + sizeof(DCTELEM)*8;
-      temp = product;
+      temp = (DCTELEM)product;
       temp = -temp;
     } else {
       product = (UDCTELEM2)(temp + corr) * recip;
       product >>= shift + sizeof(DCTELEM)*8;
-      temp = product;
+      temp = (DCTELEM)product;
     }
+    output_ptr[i] = (JCOEF) temp;
+  }
+
+#else
+
+  register DCTELEM qval;
 
+  for (i = 0; i < DCTSIZE2; i++) {
+    qval = divisors[i];
+    temp = workspace[i];
+    /* Divide the coefficient value by qval, ensuring proper rounding.
+     * Since C does not specify the direction of rounding for negative
+     * quotients, we have to force the dividend positive for portability.
+     *
+     * In most files, at least half of the output values will be zero
+     * (at default quantization settings, more like three-quarters...)
+     * so we should ensure that this case is fast.  On many machines,
+     * a comparison is enough cheaper than a divide to make a special test
+     * a win.  Since both inputs will be nonnegative, we need only test
+     * for a < b to discover whether a/b is 0.
+     * If your machine's division is fast enough, define FAST_DIVIDE.
+     */
+#ifdef FAST_DIVIDE
+#define DIVIDE_BY(a,b)  a /= b
+#else
+#define DIVIDE_BY(a,b)  if (a >= b) a /= b; else a = 0
+#endif
+    if (temp < 0) {
+      temp = -temp;
+      temp += qval>>1;  /* for rounding */
+      DIVIDE_BY(temp, qval);
+      temp = -temp;
+    } else {
+      temp += qval>>1;  /* for rounding */
+      DIVIDE_BY(temp, qval);
+    }
     output_ptr[i] = (JCOEF) temp;
   }
+
+#endif
+
 }
 
 
@@ -409,16 +487,16 @@ quantize (JCOEFPTR coef_block, DCTELEM * divisors, DCTELEM * workspace)
  */
 
 METHODDEF(void)
-forward_DCT (j_compress_ptr cinfo, jpeg_component_info * compptr,
-	     JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
-	     JDIMENSION start_row, JDIMENSION start_col,
-	     JDIMENSION num_blocks)
+forward_DCT (j_compress_ptr cinfo, jpeg_component_info *compptr,
+             JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
+             JDIMENSION start_row, JDIMENSION start_col,
+             JDIMENSION num_blocks)
 /* This version is used for integer DCT implementations. */
 {
   /* This routine is heavily used, so it's worth coding it tightly. */
   my_fdct_ptr fdct = (my_fdct_ptr) cinfo->fdct;
-  DCTELEM * divisors = fdct->divisors[compptr->quant_tbl_no];
-  DCTELEM * workspace;
+  DCTELEM *divisors = fdct->divisors[compptr->quant_tbl_no];
+  DCTELEM *workspace;
   JDIMENSION bi;
 
   /* Make sure the compiler doesn't look up these every pass */
@@ -427,7 +505,7 @@ forward_DCT (j_compress_ptr cinfo, jpeg_component_info * compptr,
   quantize_method_ptr do_quantize = fdct->quantize;
   workspace = fdct->workspace;
 
-  sample_data += start_row;	/* fold in the vertical offset once */
+  sample_data += start_row;     /* fold in the vertical offset once */
 
   for (bi = 0; bi < num_blocks; bi++, start_col += DCTSIZE) {
     /* Load data into workspace, applying unsigned->signed conversion */
@@ -446,7 +524,7 @@ forward_DCT (j_compress_ptr cinfo, jpeg_component_info * compptr,
 
 
 METHODDEF(void)
-convsamp_float (JSAMPARRAY sample_data, JDIMENSION start_col, FAST_FLOAT * workspace)
+convsamp_float (JSAMPARRAY sample_data, JDIMENSION start_col, FAST_FLOAT *workspace)
 {
   register FAST_FLOAT *workspaceptr;
   register JSAMPROW elemptr;
@@ -455,7 +533,7 @@ convsamp_float (JSAMPARRAY sample_data, JDIMENSION start_col, FAST_FLOAT * works
   workspaceptr = workspace;
   for (elemr = 0; elemr < DCTSIZE; elemr++) {
     elemptr = sample_data[elemr] + start_col;
-#if DCTSIZE == 8		/* unroll the inner loop */
+#if DCTSIZE == 8                /* unroll the inner loop */
     *workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
     *workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
     *workspaceptr++ = (FAST_FLOAT)(GETJSAMPLE(*elemptr++) - CENTERJSAMPLE);
@@ -477,7 +555,7 @@ convsamp_float (JSAMPARRAY sample_data, JDIMENSION start_col, FAST_FLOAT * works
 
 
 METHODDEF(void)
-quantize_float (JCOEFPTR coef_block, FAST_FLOAT * divisors, FAST_FLOAT * workspace)
+quantize_float (JCOEFPTR coef_block, FAST_FLOAT *divisors, FAST_FLOAT *workspace)
 {
   register FAST_FLOAT temp;
   register int i;
@@ -499,16 +577,16 @@ quantize_float (JCOEFPTR coef_block, FAST_FLOAT * divisors, FAST_FLOAT * workspa
 
 
 METHODDEF(void)
-forward_DCT_float (j_compress_ptr cinfo, jpeg_component_info * compptr,
-		   JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
-		   JDIMENSION start_row, JDIMENSION start_col,
-		   JDIMENSION num_blocks)
+forward_DCT_float (j_compress_ptr cinfo, jpeg_component_info *compptr,
+                   JSAMPARRAY sample_data, JBLOCKROW coef_blocks,
+                   JDIMENSION start_row, JDIMENSION start_col,
+                   JDIMENSION num_blocks)
 /* This version is used for floating-point DCT implementations. */
 {
   /* This routine is heavily used, so it's worth coding it tightly. */
   my_fdct_ptr fdct = (my_fdct_ptr) cinfo->fdct;
-  FAST_FLOAT * divisors = fdct->float_divisors[compptr->quant_tbl_no];
-  FAST_FLOAT * workspace;
+  FAST_FLOAT *divisors = fdct->float_divisors[compptr->quant_tbl_no];
+  FAST_FLOAT *workspace;
   JDIMENSION bi;
 
 
@@ -518,7 +596,7 @@ forward_DCT_float (j_compress_ptr cinfo, jpeg_component_info * compptr,
   float_quantize_method_ptr do_quantize = fdct->float_quantize;
   workspace = fdct->float_workspace;
 
-  sample_data += start_row;	/* fold in the vertical offset once */
+  sample_data += start_row;     /* fold in the vertical offset once */
 
   for (bi = 0; bi < num_blocks; bi++, start_col += DCTSIZE) {
     /* Load data into workspace, applying unsigned->signed conversion */
@@ -547,7 +625,7 @@ jinit_forward_dct (j_compress_ptr cinfo)
 
   fdct = (my_fdct_ptr)
     (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				SIZEOF(my_fdct_controller));
+                                sizeof(my_fdct_controller));
   cinfo->fdct = (struct jpeg_forward_dct *) fdct;
   fdct->pub.start_pass = start_pass_fdctmgr;
 
@@ -626,12 +704,12 @@ jinit_forward_dct (j_compress_ptr cinfo)
   if (cinfo->dct_method == JDCT_FLOAT)
     fdct->float_workspace = (FAST_FLOAT *)
       (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(FAST_FLOAT) * DCTSIZE2);
+                                  sizeof(FAST_FLOAT) * DCTSIZE2);
   else
 #endif
     fdct->workspace = (DCTELEM *)
       (*cinfo->mem->alloc_small) ((j_common_ptr) cinfo, JPOOL_IMAGE,
-				  SIZEOF(DCTELEM) * DCTSIZE2);
+                                  sizeof(DCTELEM) * DCTSIZE2);
 
   /* Mark divisor tables unallocated */
   for (i = 0; i < NUM_QUANT_TBLS; i++) {