Re-fix buf img mode decompr err w/short prog JPEGs

This commit reverts 4dbc293 and
9f8f683 (the previous two commits) and
fixes libjpeg-turbo#613 the correct way.  The crux of the issue wasn't the size of
the whole_image virtual array but rather that, since last_iMCU_row is
unsigned, (last_iMCU_row - 1) wrapped around to 0xFFFFFFFF when
last_iMCU_row was 0.  This caused the interblock smoothing algorithm
introduced in 6d91e95 to erroneously
try to access the next two iMCU rows, neither of which existed.  The
first attempt at a fix (4dbc293)
exposed a NULL dereference, detected by OSS-Fuzz, that occurred when
attempting to decompress a specially-crafted malformed JPEG image to a
YUV buffer using tjDecompressToYUV*() with 1/4 IDCT scaling.

Fixes libjpeg-turbo#613 (again)
Also fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=49898

ChangeLog.md

@@ -18,9 +18,10 @@ This allows both AltiVec-equipped (PowerPC G4 and G5) and non-AltiVec-equipped
 (PowerPC G3) CPUs to be supported using the same build of libjpeg-turbo.
 
 4. Fixed an error ("Bogus virtual array access") that occurred when attempting
-to decompress a progressive JPEG image with a height less than or equal to
-(32 * the vertical sampling factor) using buffered image mode.  This was a
-regression introduced by 2.1 beta1[6(b)].
+to decompress a progressive JPEG image with a height less than or equal to one
+iMCU (8 * the vertical sampling factor) using buffered-image mode with
+interblock smoothing enabled.  This was a regression introduced by
+2.1 beta1[6(b)].
 
 
 2.1.3

jdcoefct.c

@@ -475,7 +475,7 @@ decompress_smooth_data(j_decompress_ptr cinfo, JSAMPIMAGE output_buf)
     if (!compptr->component_needed)
       continue;
     /* Count non-dummy DCT block rows in this iMCU row. */
-    if (cinfo->output_iMCU_row < last_iMCU_row - 1) {
+    if (cinfo->output_iMCU_row + 1 < last_iMCU_row) {
       block_rows = compptr->v_samp_factor;
       access_rows = block_rows * 3; /* this and next two iMCU rows */
     } else if (cinfo->output_iMCU_row < last_iMCU_row) {
@@ -560,7 +560,7 @@ decompress_smooth_data(j_decompress_ptr cinfo, JSAMPIMAGE output_buf)
         next_block_row = buffer_ptr;
 
       if (block_row < block_rows - 2 ||
-          cinfo->output_iMCU_row < last_iMCU_row - 1)
+          cinfo->output_iMCU_row + 1 < last_iMCU_row)
         next_next_block_row =
           buffer[block_row + 2] + cinfo->master->first_MCU_col[ci];
       else
@@ -835,21 +835,18 @@ jinit_d_coef_controller(j_decompress_ptr cinfo, boolean need_full_buffer)
 
     for (ci = 0, compptr = cinfo->comp_info; ci < cinfo->num_components;
          ci++, compptr++) {
-      JDIMENSION num_rows =
-        (JDIMENSION)jround_up((long)compptr->height_in_blocks,
-                              (long)compptr->v_samp_factor);
       access_rows = compptr->v_samp_factor;
 #ifdef BLOCK_SMOOTHING_SUPPORTED
       /* If block smoothing could be used, need a bigger window */
-      if (cinfo->progressive_mode) {
+      if (cinfo->progressive_mode)
         access_rows *= 5;
-        num_rows = MAX(num_rows, (JDIMENSION)access_rows);
-      }
 #endif
       coef->whole_image[ci] = (*cinfo->mem->request_virt_barray)
         ((j_common_ptr)cinfo, JPOOL_IMAGE, TRUE,
          (JDIMENSION)jround_up((long)compptr->width_in_blocks,
-                               (long)compptr->h_samp_factor), num_rows,
+                               (long)compptr->h_samp_factor),
+         (JDIMENSION)jround_up((long)compptr->height_in_blocks,
+                               (long)compptr->v_samp_factor),
          (JDIMENSION)access_rows);
     }
     coef->pub.consume_data = consume_data;