cpuid/x86: avoid trailing nulls in output and storage

cpuid/x86.cpp

@@ -45,7 +45,6 @@ x86::x86 ()
     memcpy (vendor_name.data () + 0, &vendor0, sizeof (vendor0));
     memcpy (vendor_name.data () + 4, &vendor1, sizeof (vendor1));
     memcpy (vendor_name.data () + 8, &vendor2, sizeof (vendor2));
-    vendor_name.back () = '\0';
 
     const bool is_amd = vendor0 == 0x68747541 && vendor1 == 0x69746e65 && vendor2 == 0x444d4163;
 
@@ -72,7 +71,6 @@ x86::x86 ()
         memcpy (&product_name[0x00], &product0, sizeof (product0));
         memcpy (&product_name[0x10], &product1, sizeof (product1));
         memcpy (&product_name[0x20], &product2, sizeof (product2));
-        product_name.back () = '\0';
     }
 
     // Function 4 isn't implemented by AMD. Intel uses it for cache
@@ -93,13 +91,20 @@ x86::x86 ()
 std::ostream&
 cruft::cpu::operator<< (std::ostream &os, const x86 &val)
 {
+    auto const to_string = [] (auto const &obj) {
+        return cruft::view {
+            obj.begin (),
+            std::find (obj.begin (), obj.end (), '\0')
+        };
+    };
+
     return os << cruft::format::printf (
         "{ name: { vendor: '%!', product: '%!' }"
         ", cores: { logical: %!, physical: %!, hyper_threading: %! }"
         ", simd: { sse: %!, sse2: %!, sse3: %!, ssse3: %!, sse41: %!, sse42: %! }"
         " }",
-        cruft::view {val.vendor_name},
-        cruft::view {val.product_name},
+        to_string (val.vendor_name),
+        to_string (val.product_name),
         val.cores.logical,
         val.cores.physical,
         (val.cores.hyper_threading ? "true" : "false"),

cpuid/x86.hpp

@@ -41,8 +41,8 @@ namespace cruft::cpu {
             bool avx;
         } simd;
 
-        std::array<char,12+1> vendor_name;
-        std::array<char,48+1> product_name;
+        std::array<char,12> vendor_name;
+        std::array<char,48> product_name;
     };
 
     std::ostream&