Extension enhancements and tests.

15 files changed, 243 insertions, 21 deletions

diff --git a/ChangeLog b/ChangeLog
index ec3279ea..16febe5e 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,7 @@
+2012-04-07         Andrew J. Schorr     <aschorr@telemetry-investments.com>
+
+	* TODO.xgawk: Update to reflect progress.
+
 2012-04-01         Andrew J. Schorr     <aschorr@telemetry-investments.com>
 
 	* TODO.xgawk: Move valgrind-noleak item into "done" section.
diff --git a/TODO.xgawk b/TODO.xgawk
index 7a06b0fb..421eae24 100644
--- a/TODO.xgawk
+++ b/TODO.xgawk
@@ -21,9 +21,21 @@ Done:
 
 - Add valgrind-noleak target.
 
+- Fix minor bug in fork extension, and add wait function.
+
+- Patch filefuncs extension to read symbolic links more robustly.
+
+- Add shared library tests.
+
 
 To do (not necessarily in this order):
 
+- Enhance extension/fork.c waitpid to allow the caller to specify the options.
+  And add an optional array argument to wait and waitpid in which to return
+  exit status information.
+
+- Maybe add more shared library tests.
+
 - Figure out how to support xgawk on platforms such as Cygwin where a DLL
   cannot be linked with unresolved references.  There are currently 3
   possible solutions:
@@ -50,11 +62,6 @@ To do (not necessarily in this order):
   defined but not used.  So there should probably be an enhancement to func_use
   and ftable to indicate if it's a shared library function.
 
-- Patch standard extensions to work properly.  For example, fork is broken.
-  And I think we made a few improvements in the xgawk versions.
-
-- Add tests for standard extensions.
-
 - Develop a libgawk shared library for use by extensions.  In particular,
   a few existing extensions use a hash API for mapping string handles to
   structures.  In xgawk, we had this API inside array.c, but it probably
diff --git a/extension/ChangeLog b/extension/ChangeLog
index 287ce309..994b3c92 100644
--- a/extension/ChangeLog
+++ b/extension/ChangeLog
@@ -1,3 +1,11 @@
+2012-04-07         Andrew J. Schorr     <aschorr@telemetry-investments.com>
+
+	* filefuncs.c: Remove unnecessary #include <sys/sysmacros.h>.
+	(read_symlink): New function to read symbolic links more robustly.
+	(do_stat): Use read_symlink instead of readlink.
+	* fork.c (do_wait): new function.
+	(dlload): Call make_builtin to add "wait" function.
+
 2012-04-02         Andrew J. Schorr     <aschorr@telemetry-investments.com>
 
 	* fork.c (do_fork): Test whether PROCINFO_node exists before updating
diff --git a/extension/filefuncs.c b/extension/filefuncs.c
index 63010c35..6d46c5e5 100644
--- a/extension/filefuncs.c
+++ b/extension/filefuncs.c
@@ -29,8 +29,6 @@
 
 #include "awk.h"
 
-#include <sys/sysmacros.h>
-
 int plugin_is_GPL_compatible;
 
 /*  do_chdir --- provide dynamically loaded chdir() builtin for gawk */
@@ -157,6 +155,60 @@ format_mode(unsigned long fmode)
 	return outbuf;
 }
 
+/* read_symlink -- read a symbolic link into an allocated buffer.
+   This is based on xreadlink; the basic problem is that lstat cannot be relied
+   upon to return the proper size for a symbolic link.  This happens,
+   for example, on linux in the /proc filesystem, where the symbolic link
+   sizes are often 0. */
+
+#ifndef SIZE_MAX
+# define SIZE_MAX ((size_t) -1)
+#endif
+#ifndef SSIZE_MAX
+# define SSIZE_MAX ((ssize_t) (SIZE_MAX / 2))
+#endif
+
+#define MAXSIZE (SIZE_MAX < SSIZE_MAX ? SIZE_MAX : SSIZE_MAX)
+
+static char *
+read_symlink(const char *fname, size_t bufsize, ssize_t *linksize)
+{
+	if (bufsize)
+		bufsize += 2;
+	else
+		bufsize = BUFSIZ*2;
+	/* Make sure that bufsize >= 2 and within range */
+	if ((bufsize > MAXSIZE) || (bufsize < 2))
+		bufsize = MAXSIZE;
+	while (1) {
+		char *buf;
+
+		emalloc(buf, char *, bufsize, "read_symlink");
+		if ((*linksize = readlink(fname, buf, bufsize)) < 0) {
+			/* On AIX 5L v5.3 and HP-UX 11i v2 04/09, readlink
+			   returns -1 with errno == ERANGE if the buffer is
+			   too small.  */
+			if (errno != ERANGE) {
+				free(buf);
+				return NULL;
+			}
+		}
+		/* N.B. This test is safe because bufsize must be >= 2 */
+		else if ((size_t)*linksize <= bufsize-2) {
+			buf[*linksize] = '\0';
+			return buf;
+		}
+		free(buf);
+		if (bufsize <= MAXSIZE/2)
+			bufsize *= 2;
+		else if (bufsize < MAXSIZE)
+			bufsize = MAXSIZE;
+		else
+			return NULL;
+	}
+	return NULL;
+}
+
 /* do_stat --- provide a stat() function for gawk */
 
 static NODE *
@@ -265,22 +317,17 @@ do_stat(int nargs)
 	/* for symbolic links, add a linkval field */
 	if (S_ISLNK(sbuf.st_mode)) {
 		char *buf;
-		int linksize;
-
-		emalloc(buf, char *, sbuf.st_size + 2, "do_stat");
-		if (((linksize = readlink(file->stptr, buf,
-					  sbuf.st_size + 2)) >= 0) &&
-		    (linksize <= sbuf.st_size)) {
-			/*
-			 * set the linkval field only if we are able to
-			 * retrieve the entire link value successfully.
-			 */
-			buf[linksize] = '\0';
+		ssize_t linksize;
 
+		if ((buf = read_symlink(file->stptr, sbuf.st_size,
+					&linksize)) != NULL) {
 			aptr = assoc_lookup(array, tmp = make_string("linkval", 7));
 			*aptr = make_str_node(buf, linksize, ALREADY_MALLOCED);
 			unref(tmp);
 		}
+		else
+			warning(_("unable to read symbolic link `%s'"),
+				file->stptr);
 	}
 
 	/* add a type field */
diff --git a/extension/fork.c b/extension/fork.c
index 7d6ab362..5a6e96d5 100644
--- a/extension/fork.c
+++ b/extension/fork.c
@@ -93,6 +93,25 @@ do_waitpid(int nargs)
 	return make_number((AWKNUM) ret);
 }
 
+
+/*  do_wait --- provide dynamically loaded wait() builtin for gawk */
+
+static NODE *
+do_wait(int nargs)
+{
+	int ret;
+
+	if  (do_lint && nargs > 0)
+		lintwarn("wait: called with too many arguments");
+
+	ret = wait(NULL);
+	if (ret < 0)
+		update_ERRNO_int(errno);
+
+	/* Set the return value */
+	return make_number((AWKNUM) ret);
+}
+
 /* dlload --- load new builtins in this library */
 
 NODE *
@@ -102,5 +121,6 @@ void *dl;
 {
 	make_builtin("fork", do_fork, 0);
 	make_builtin("waitpid", do_waitpid, 1);
+	make_builtin("wait", do_wait, 0);
 	return make_number((AWKNUM) 0);
 }
diff --git a/test/ChangeLog b/test/ChangeLog
index 236288e2..5993ca23 100644
--- a/test/ChangeLog
+++ b/test/ChangeLog
@@ -1,3 +1,11 @@
+2012-04-07         Andrew J. Schorr     <aschorr@telemetry-investments.com>
+
+	* Makefile.am (check): Add new shlib-tests target.
+	(SHLIB_TESTS): Add tests ordchr, ordchr2, fork, fork2, readfile and
+	filefuncs.
+	* ordchr.awk, ordchr.ok, fork.awk, fork.ok, fork2.awk, fork2.ok,
+	filefuncs.awk, filefuncs.ok: New files.
+
 2012-04-01         Andrew J. Schorr     <aschorr@telemetry-investments.com>
 
 	* Makefile.am (valgrind-scan): Update to match modern valgrind output.
diff --git a/test/Makefile.am b/test/Makefile.am
index afd98299..854a8f7c 100644
--- a/test/Makefile.am
+++ b/test/Makefile.am
@@ -201,6 +201,8 @@ EXTRA_DIST = \
 	fieldwdth.awk \
 	fieldwdth.in \
 	fieldwdth.ok \
+	filefuncs.awk \
+	filefuncs.ok \
 	fldchg.awk \
 	fldchg.in \
 	fldchg.ok \
@@ -230,6 +232,8 @@ EXTRA_DIST = \
 	fnmisc.ok \
 	fnparydl.awk \
 	fnparydl.ok \
+	fork.awk \
+	fork2.awk \
 	fpat1.awk \
 	fpat1.in \
 	fpat1.ok \
@@ -522,6 +526,8 @@ EXTRA_DIST = \
 	opasnidx.ok \
 	opasnslf.awk \
 	opasnslf.ok \
+	ordchr.awk \
+	ordchr.ok \
 	out1.ok \
 	out2.ok \
 	out3.ok \
@@ -846,6 +852,8 @@ LOCALE_CHARSET_TESTS = \
 	asort asorti fmttest fnarydel fnparydl lc_num1 mbfw1 \
 	mbprintf1 mbprintf2 mbprintf3 rebt8b2 rtlenmb sort1 sprintfc
 
+SHLIB_TESTS = ordchr ordchr2 fork fork2 readfile filefuncs
+
 # List of the tests which should be run with --lint option:
 NEED_LINT = \
 	defref fmtspcl lintwarn noeffect nofmtch shadow \
@@ -870,8 +878,10 @@ PGAWKPROG = ../pgawk$(EXEEXT)
 #
 # This can also be done in individual tests where we wish to
 # check things specifically not in the C locale.
-AWK = LC_ALL=$${GAWKLOCALE:-C} LANG=$${GAWKLOCALE:-C} $(AWKPROG)
-PGAWK = LC_ALL=$${GAWKLOCALE:-C} LANG=$${GAWKLOCALE:-C} $(PGAWKPROG)
+#
+# And we set AWKLIBPATH to find the extension libraries we built.
+AWK = LC_ALL=$${GAWKLOCALE:-C} LANG=$${GAWKLOCALE:-C} AWKLIBPATH=../extension/.libs $(AWKPROG)
+PGAWK = LC_ALL=$${GAWKLOCALE:-C} LANG=$${GAWKLOCALE:-C} AWKLIBPATH=../extension/.libs $(PGAWKPROG)
 
 # Message stuff is to make it a little easier to follow.
 # Make the pass-fail last and dependent on others to avoid
@@ -882,7 +892,8 @@ check:	msg \
 	unix-msg-start   unix-tests      unix-msg-end \
 	extend-msg-start gawk-extensions extend-msg-end \
 	machine-msg-start machine-tests machine-msg-end \
-	charset-msg-start charset-tests charset-msg-end
+	charset-msg-start charset-tests charset-msg-end \
+	shlib-msg-start  shlib-tests     shlib-msg-end
 	@$(MAKE) pass-fail
 
 basic:	$(BASIC_TESTS)
@@ -899,6 +910,8 @@ inet:	inetmesg $(INET_TESTS)
 
 machine-tests: $(MACHINE_TESTS)
 
+shlib-tests: $(SHLIB_TESTS)
+
 msg::
 	@echo ''
 	@echo 'Any output from "cmp" is bad news, although some differences'
@@ -939,6 +952,12 @@ charset-msg-start:
 charset-msg-end:
 	@echo "======== Done with tests that can vary based on character set or locale support ========"
 
+shlib-msg-start:
+	@echo "======== Starting shared library tests ========"
+
+shlib-msg-end:
+	@echo "======== Done with shared library tests ========"
+
 
 lc_num1:
 	@echo $@
@@ -1444,6 +1463,16 @@ rri1::
 	AWKPATH=$(srcdir) $(AWK) -f $@.awk  < $(srcdir)/$@.in >_$@ 2>&1 || echo EXIT CODE: $$? >>_$@
 	@-$(CMP) $(srcdir)/$@.ok _$@ && rm -f _$@
 
+ordchr2::
+	@echo $@
+	@$(AWK) -l ordchr 'BEGIN {print chr(ord("A"))}' >_$@ 2>&1
+	@-$(CMP) $(srcdir)/ordchr.ok _$@ && rm -f _$@
+
+readfile::
+	@echo $@
+	@$(AWK) -l readfile 'BEGIN {printf "%s", readfile("Makefile")}' 2>&1 \
+	| $(CMP) Makefile -
+
 # Targets generated for other tests:
 include Maketests
 
diff --git a/test/filefuncs.awk b/test/filefuncs.awk
new file mode 100644
index 00000000..aa532741
--- /dev/null
+++ b/test/filefuncs.awk
@@ -0,0 +1,25 @@
+@load "filefuncs"
+
+BEGIN {
+   if (chdir("..") < 0) {
+      printf "Error: chdir failed with ERRNO %s\n", ERRNO
+      exit 1
+   }
+
+   if (stat(ARGV[0], st) < 0) {
+      printf "Error: stat(%s) failed with ERRNO %s\n", ARGV[0], ERRNO
+      exit 1
+   }
+
+   nf = split("name dev ino mode nlink uid gid size blocks atime mtime ctime pmode type", f)
+
+   for (i = 1; i <= nf; i++) {
+      if (!(f[i] in st)) {
+	 printf "stat value for %s is missing\n",f[i]
+	 rc = 1
+      }
+      else
+	 delete st[f[i]]
+   }
+   exit rc+0
+}
diff --git a/test/filefuncs.ok b/test/filefuncs.ok
new file mode 100644
index 00000000..e69de29b
--- /dev/null
+++ b/test/filefuncs.ok
diff --git a/test/fork.awk b/test/fork.awk
new file mode 100644
index 00000000..0b29f9ff
--- /dev/null
+++ b/test/fork.awk
@@ -0,0 +1,33 @@
+@load "fork"
+
+BEGIN {
+   fn = ("fork.tmp." PROCINFO["pid"])
+   switch (pid = fork()) {
+   case -1:
+      printf "Error: fork failed with ERRNO %s\n", ERRNO
+      exit 1
+   case 0:
+      # child
+      printf "pid %s ppid %s\n", PROCINFO["pid"], PROCINFO["ppid"] > fn
+      exit 0
+   default:
+      # parent
+      erc = 1
+      if ((rc = wait()) < 0)
+	 printf "Error: wait failed with ERRNO %s\n", ERRNO
+      else if (rc != pid)
+	 printf "Error: wait returned %s instead of child pid %s\n", rc, pid
+      else if ((getline x < fn) != 1)
+	 printf "Error: getline failed on temp file %s\n", fn
+      else {
+	 expected = ("pid " pid " ppid " PROCINFO["pid"])
+	 if (x != expected)
+	    printf "Error: child data (%s) != expected (%s)\n", x, expected
+	 else if ((rc = system("rm  " fn)) != 0)
+	    printf "Error removing temp file %s with ERRNO %s\n", fn, ERRNO
+	 else
+	    erc = 0
+      }
+      exit erc
+   }
+}
diff --git a/test/fork.ok b/test/fork.ok
new file mode 100644
index 00000000..e69de29b
--- /dev/null
+++ b/test/fork.ok
diff --git a/test/fork2.awk b/test/fork2.awk
new file mode 100644
index 00000000..bd364286
--- /dev/null
+++ b/test/fork2.awk
@@ -0,0 +1,35 @@
+@load "fork"
+
+BEGIN {
+   # avoid instantiating PROCINFO prior to the fork
+   switch (pid = fork()) {
+   case -1:
+      printf "Error: fork failed with ERRNO %s\n", ERRNO
+      exit 1
+   case 0:
+      # child
+      fn = ("fork.tmp." PROCINFO["pid"])
+      printf "pid %s ppid %s\n", PROCINFO["pid"], PROCINFO["ppid"] > fn
+      exit 0
+   default:
+      # parent
+      erc = 1
+      fn = ("fork.tmp." pid)
+      if ((rc = wait()) < 0)
+	 printf "Error: wait failed with ERRNO %s\n", ERRNO
+      else if (rc != pid)
+	 printf "Error: wait returned %s instead of child pid %s\n", rc, pid
+      else if ((getline x < fn) != 1)
+	 printf "Error: getline failed on temp file %s\n", fn
+      else {
+	 expected = ("pid " pid " ppid " PROCINFO["pid"])
+	 if (x != expected)
+	    printf "Error: child data (%s) != expected (%s)\n", x, expected
+	 else if ((rc = system("rm  " fn)) != 0)
+	    printf "Error removing temp file %s with ERRNO %s\n", fn, ERRNO
+	 else
+	    erc = 0
+      }
+      exit erc
+   }
+}
diff --git a/test/fork2.ok b/test/fork2.ok
new file mode 100644
index 00000000..e69de29b
--- /dev/null
+++ b/test/fork2.ok
diff --git a/test/ordchr.awk b/test/ordchr.awk
new file mode 100644
index 00000000..abb793a0
--- /dev/null
+++ b/test/ordchr.awk
@@ -0,0 +1,5 @@
+@load "ordchr"
+
+BEGIN {
+   print chr(ord("A"))
+}
diff --git a/test/ordchr.ok b/test/ordchr.ok
new file mode 100644
index 00000000..f70f10e4
--- /dev/null
+++ b/test/ordchr.ok
@@ -0,0 +1 @@
+A