hl: fix incorrect hvl_t comments in H5DS.c and CHANGELOG

VL strings use char* in memory, not hvl_t (which is only for VL
sequences). Correct two comments in H5DS.c and the CHANGELOG entry
per jhendersonHDF review on PR HDFGroup#6371.

Author: brtnfld

hl/src/H5DS.c

@@ -2262,8 +2262,9 @@ H5DSis_scale(hid_t did)
             is_ds = 0;
             goto out;
         }
-        /* reject VLEN strings; H5Aread into a char* buffer would write an hvl_t
-           instead of character data, so this cannot be a dimension scale */
+        /* reject VLEN strings; H5Aread into a char* buffer would overwrite it
+           with a heap-allocated char* pointer rather than the string content,
+           so this cannot be a dimension scale */
         if ((isvl = H5Tis_variable_str(tid)) < 0)
             goto out;
         if (isvl > 0) {
@@ -2485,9 +2486,10 @@ H5DS_is_reserved(hid_t did, bool *is_reserved)
     if (H5T_STRING != H5Tget_class(tid))
         goto error;
 
-    /* Reject VLEN strings; H5Aread into a char* buffer would write an hvl_t,
-     * not character data. Treat as an error for consistency with the
-     * existing non-NULLTERM rejection below. */
+    /* Reject VLEN strings; H5Aread into a char* buffer would overwrite it
+     * with a heap-allocated char* pointer rather than the string content.
+     * Treat as an error for consistency with the existing non-NULLTERM
+     * rejection below. */
     if (H5Tis_variable_str(tid) > 0)
         goto error;
 

release_docs/CHANGELOG.md

@@ -215,8 +215,9 @@ We would like to thank the many HDF5 community members who contributed to this r
    Additional fixes applied to all four routines:
 
    - **VLEN-string CLASS attributes are now explicitly rejected.** Previously, reading a VLEN-typed
-     attribute into a fixed `char *` buffer would write an `hvl_t` struct over the character data,
-     which is undefined behaviour and could corrupt memory or produce garbage comparison results.
+     attribute into a fixed `char *` buffer would overwrite it with a heap-allocated `char *` pointer
+     rather than the string content, which is undefined behaviour and could corrupt memory or produce
+     garbage comparison results.
      `H5DSis_scale()`, `H5IMis_image()`, and `H5IMis_palette()` now return 0 (not a scale/image/palette)
      for a VLEN CLASS attribute. `H5DS_is_reserved()` (an internal routine called by `H5DSattach_scale()`)
      returns FAIL, so `H5DSattach_scale()` will return FAIL for a dataset whose CLASS attribute uses a