From a6c3ce6ec03516d850e6b987fcbe95aba192b76b Mon Sep 17 00:00:00 2001
From: Thomas Preud'homme <robotux@celest.fr>
Date: Wed, 18 Feb 2015 04:22:25 +0000
Subject: [PATCH] The "open a whisky and cut your finger open" patch

Make integer constant parsing C99 compliant
---
 tccpp.c                              | 75 +++++++++++++++-------------
 tests/tests2/72_long_long_constant.c | 17 +++++++
 tests/tests2/Makefile                |  7 +--
 3 files changed, 62 insertions(+), 37 deletions(-)
 create mode 100644 tests/tests2/72_long_long_constant.c

diff --git a/tccpp.c b/tccpp.c
index bc2d4a2..42f34aa 100644
--- a/tccpp.c
+++ b/tccpp.c
@@ -2012,7 +2012,8 @@ static void parse_number(const char *p)
         }
     } else {
         unsigned long long n, n1;
-        int lcount, ucount;
+        int lcount, ucount, must_64bit;
+        const char *p1;
 
         /* integer number */
         *q = '\0';
@@ -2025,17 +2026,16 @@ static void parse_number(const char *p)
         while(1) {
             t = *q++;
             /* no need for checks except for base 10 / 8 errors */
-            if (t == '\0') {
+            if (t == '\0')
                 break;
-            } else if (t >= 'a') {
+            else if (t >= 'a')
                 t = t - 'a' + 10;
-            } else if (t >= 'A') {
+            else if (t >= 'A')
                 t = t - 'A' + 10;
-            } else {
+            else
                 t = t - '0';
-                if (t >= b)
-                    tcc_error("invalid digit");
-            }
+            if (t >= b)
+                tcc_error("invalid digit");
             n1 = n;
             n = n * b + t;
             /* detect overflow */
@@ -2043,50 +2043,57 @@ static void parse_number(const char *p)
             if (n < n1)
                 tcc_error("integer constant overflow");
         }
-        
-        /* XXX: not exactly ANSI compliant */
-        if ((n & 0xffffffff00000000LL) != 0) {
-            if ((n >> 63) != 0)
-                tok = TOK_CULLONG;
-            else
-                tok = TOK_CLLONG;
-        } else if (n > 0x7fffffff) {
-            tok = TOK_CUINT;
-        } else {
-            tok = TOK_CINT;
-        }
-        lcount = 0;
-        ucount = 0;
+
+        /* Determine the characteristics (unsigned and/or 64bit) the type of
+           the constant must have according to the constant suffix(es) */
+        lcount = ucount = must_64bit = 0;
+        p1 = p;
         for(;;) {
             t = toup(ch);
             if (t == 'L') {
                 if (lcount >= 2)
                     tcc_error("three 'l's in integer constant");
+                if (lcount && *(p - 1) != ch)
+                    tcc_error("incorrect integer suffix: %s", p1);
                 lcount++;
 #if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
-                if (lcount == 2) {
-#endif
-                    if (tok == TOK_CINT)
-                        tok = TOK_CLLONG;
-                    else if (tok == TOK_CUINT)
-                        tok = TOK_CULLONG;
-#if !defined TCC_TARGET_X86_64 || defined TCC_TARGET_PE
-                }
+                if (lcount == 2)
 #endif
+                    must_64bit = 1;
                 ch = *p++;
             } else if (t == 'U') {
                 if (ucount >= 1)
                     tcc_error("two 'u's in integer constant");
                 ucount++;
-                if (tok == TOK_CINT)
-                    tok = TOK_CUINT;
-                else if (tok == TOK_CLLONG)
-                    tok = TOK_CULLONG;
                 ch = *p++;
             } else {
                 break;
             }
         }
+
+        /* Whether 64 bits are needed to hold the constant's value */
+        if (n & 0xffffffff00000000LL || must_64bit) {
+            tok = TOK_CLLONG;
+            n1 = n >> 32;
+	} else {
+            tok = TOK_CINT;
+            n1 = n;
+        }
+
+        /* Whether type must be unsigned to hold the constant's value */
+        if (ucount || ((n1 >> 31) && (b != 10))) {
+            if (tok == TOK_CLLONG)
+                tok = TOK_CULLONG;
+            else
+                tok = TOK_CUINT;
+        /* If decimal and no unsigned suffix, bump to 64 bits or throw error */
+        } else if (n1 >> 31) {
+            if (tok == TOK_CINT)
+                tok = TOK_CLLONG;
+            else
+                tcc_error("integer constant overflow");
+        }
+
         if (tok == TOK_CINT || tok == TOK_CUINT)
             tokc.ui = n;
         else
diff --git a/tests/tests2/72_long_long_constant.c b/tests/tests2/72_long_long_constant.c
new file mode 100644
index 0000000..a698fd6
--- /dev/null
+++ b/tests/tests2/72_long_long_constant.c
@@ -0,0 +1,17 @@
+#include <stdio.h>
+
+int main()
+{
+    long long int res = 0;
+
+    if (res < -2147483648LL) {
+        printf("Error: 0 < -2147483648\n");
+        return 1;
+    }
+    else
+    if (2147483647LL < res) {
+        printf("Error: 2147483647 < 0\n");
+        return 2;
+    }
+    return 0;
+}
diff --git a/tests/tests2/Makefile b/tests/tests2/Makefile
index ec636ab..8d8a5e7 100644
--- a/tests/tests2/Makefile
+++ b/tests/tests2/Makefile
@@ -91,6 +91,7 @@ TESTS =	\
  69_macro_param_list_err_2.test \
  70_floating_point_literals.test \
  71_macro_empty_arg.test \
+ 72_long_long_constant.test \
 
 
 # 34_array_assignment.test -- array assignment is not in C standard
@@ -113,14 +114,14 @@ ARGS =
 
 all test: $(filter-out $(SKIP),$(TESTS))
 
-%.test: %.c %.expect
+%.test: %.c
 	@echo Test: $*...
 
 	@$(TCC) -run $< $(ARGS) 2>&1 | grep -v 'warning: soft float ABI currently not supported: default to softfp' >$*.output || true
-	@diff -bu $*.expect $*.output && rm -f $*.output
+	@diff -Nbu $*.expect $*.output && rm -f $*.output
 
 	@($(TCC) $< -o $*.exe && ./$*.exe $(ARGS)) 2>&1 | grep -v 'warning: soft float ABI currently not supported: default to softfp' >$*.output2 || true
-	@diff -bu $*.expect $*.output2 && rm -f $*.output2 $*.exe
+	@diff -Nbu $*.expect $*.output2 && rm -f $*.output2 $*.exe
 
 clean:
 	rm -vf fred.txt *.output* *.exe