The ctt_size of integers and floating point numbers is in bytes not bits.

OK mpi@

diff --git a/usr.bin/ctfconv/generate.c b/usr.bin/ctfconv/generate.c
index c2911f5..fdbaa0e 100644 (file)
--- a/usr.bin/ctfconv/generate.c
+++ b/usr.bin/ctfconv/generate.c
@@ -1,4 +1,4 @@
-/*     $OpenBSD: generate.c,v 1.5 2022/08/14 14:54:13 millert Exp $ */
+/*     $OpenBSD: generate.c,v 1.6 2024/02/22 13:15:17 claudio Exp $ */
 
 /*
  * Copyright (c) 2017 Martin Pieuchot
@@ -196,7 +196,20 @@ imcs_add_type(struct imcs *imcs, struct itype *it)
                ctt.ctt_type = it->it_refp->it_idx;
                ctsz = sizeof(struct ctf_stype);
        } else if (size <= CTF_MAX_SIZE) {
-               ctt.ctt_size = size;
+               if (kind == CTF_K_INTEGER || kind == CTF_K_FLOAT) {
+                       assert(size <= 64);
+                       if (size == 0)
+                               ctt.ctt_size = 0;
+                       else if (size <= 8)
+                               ctt.ctt_size = 1;
+                       else if (size <= 16)
+                               ctt.ctt_size = 2;
+                       else if (size <= 32)
+                               ctt.ctt_size = 4;
+                       else
+                               ctt.ctt_size = 8;
+               } else
+                       ctt.ctt_size = size;
                ctsz = sizeof(struct ctf_stype);
        } else {
                ctt.ctt_lsizehi = CTF_SIZE_TO_LSIZE_HI(size);