Sundrous fixes (#862)

- align.c: Remove null pointer arithmetic.
- man pages: Fix mandoc warnings pertaining to lines exceeding 80 bytes.
- Increase the SEARCHSIZE and LOOKAHEAD definitions from 80 to 500.
  (vide #460 (comment))
- tminit(): Fixed a small memory leak ASan was complaining about.
- Replaced alignof macro in nvtype.c by a version from Martijn that
  uses offsetof instead of an outdated hack with null pointer
  dereferences, fixing the following UBSan error:
  runtime error: member access within null pointer of type '_Align_'
- Fixed a compiler error that broke SHOPT_KIA by including shlex.h
  in defs.c.
- Added pragmas to ignore some superfluous compiler warnings.
- Added undefs to fix compiler warnings on illumos.
- Amended the ASan test to check the binary (in case no ASAN_OPTIONS
  flag is present). Also apply a modified form of that test to another
  regression test incompatible with ASan and profiling.

Co-authored-by: Martijn Dekker <martijn@inlv.org>

Author: JohnoKing

NEWS

@@ -2,6 +2,11 @@ This documents significant changes in the 1.0 branch of ksh 93u+m.
 For full details, see the git log at: https://github.com/ksh93/ksh/tree/1.0
 Uppercase BUG_* IDs are shell bug IDs as used by the Modernish shell library.
 
+2025-07-30:
+
+- The editor typeahead buffer and the limit for the ${.sh.edchar} variable
+  (used with the KEYBD trap) has been increased from 80 to 500 bytes.
+
 2025-07-29:
 
 - Fixed a serious but rarely occurring regression that corrupts variable

src/cmd/ksh93/Mamfile

@@ -586,6 +586,7 @@ make install virtual
 		make sh/defs.c
 			prev include/jobs.h
 			prev include/defs.h
+			prev include/shlex.h
 			prev shopt.h
 		done
 

src/cmd/ksh93/include/edit.h

@@ -25,14 +25,14 @@
  *
  */
 
-#define SEARCHSIZE	80
+#define SEARCHSIZE	500
 
 #include	"FEATURE/cmds"
 #include        "FEATURE/locale"
 #include	"terminal.h"
 
 #define STRIP		0377
-#define LOOKAHEAD	80
+#define LOOKAHEAD	500
 
 #if SHOPT_MULTIBYTE
 #   include	"national.h"

src/cmd/ksh93/include/shtable.h

@@ -2,7 +2,7 @@
 *                                                                      *
 *               This software is part of the ast package               *
 *          Copyright (c) 1982-2012 AT&T Intellectual Property          *
-*          Copyright (c) 2020-2022 Contributors to ksh 93u+m           *
+*          Copyright (c) 2020-2025 Contributors to ksh 93u+m           *
 *                      and is licensed under the                       *
 *                 Eclipse Public License, Version 2.0                  *
 *                                                                      *
@@ -60,4 +60,4 @@ extern const Shtable_t		*sh_locate(const char*, const Shtable_t*, int);
 extern int			sh_lookopt(const char*, int*);
 extern Dt_t			*sh_inittree(const struct shtable2*);
 
-#endif /* SH_TABLE_H */
+#endif /* _SHTABLE_H */

src/cmd/ksh93/sh.1

@@ -2082,7 +2082,8 @@ The first (default: \f3!\fP) signals the start of a history expansion.
 The second (default: \f3^\fP) is used for short-form substitutions.
 The third (default: \f3#\fP), when found as the first character of a word,
 causes history expansion to be skipped for the rest of the words on the line.
-Multi-byte characters (e.g. UTF-8) are not supported and produce undefined results.
+Multi-byte characters (e.g. UTF-8) are not supported and produce
+undefined results.
 .TP
 .B
 .SM HISTCMD
@@ -4734,8 +4735,10 @@ Normally, this means that a command line with an erroneous history expansion
 is lost and needs to be retyped from scratch,
 but if the \f3histreedit\fP shell option is turned on
 and a line editor is active (see \f2In-line Editing Options\fP below),
-the erroneous line is pre-filled into the next prompt's input buffer for correcting.
-The \f3histverify\fP option causes the same to be done for the results of successful
+the erroneous line is pre-filled into the next prompt's input buffer
+for correcting.
+The \f3histverify\fP option causes the same to be done for the results
+of successful
 history expansions, allowing verification and editing before execution.
 .PP
 A history expansion may have an \f2event specification\fP, which indicates
@@ -4783,12 +4786,14 @@ output by the \f3hist -l\fP command:
 The commands are shown with their event numbers.
 The current event, which we haven't typed in yet, is event 13.
 \f3!11\fP and \f3!\-2\fP refer to event 11.
-\f3!!\fP refers to the previous event, 12.  \f3!!\fP can be abbreviated \f3!\fP if it is
+\f3!!\fP refers to the previous event, 12.  \f3!!\fP can be
+abbreviated \f3!\fP if it is
 followed by \f3:\fP (see below).
 \f3!n\fP refers to event 9, which begins with \f3n\fP.
 \f3!?old?\fP also refers to event 12, which contains \f3old\fP.
-Without word designators or modifiers, history references simply expand to the
-entire event, so we might type \f3!cp\fP to redo the copy command or \f3!!|more\fP
+Without word designators or modifiers, history references simply
+expand to the entire event, so we might type \f3!cp\fP to redo the
+copy command or \f3!!|more\fP
 if the \f3diff\fP output scrolled off the top of the screen.
 .PP
 To select words from an event,
@@ -4842,16 +4847,17 @@ arguments from the \f3cp\fP command.  If we didn't care about the order of the
 \f3diff\fP, we might have said \f3diff !\-2:1\-2\fP or simply \f3diff !\-2:*\fP.
 The \f3cp\fP command might have been written \f3cp wumpus.man !#:1.old\fP,
 using \f3#\fP to refer to the current event.
-\f3!n:\- hurkle.man\fP would reuse the first two words from the \f3nroff\fP command
-to say \f3nroff \-man hurkle.man\fP.
+\f3!n:\- hurkle.man\fP would reuse the first two words from
+the \f3nroff\fP command to say \f3nroff \-man hurkle.man\fP.
 .PP
 The \f3:\fP separating the event specification from the word designator
 can be omitted if the argument selector begins with a
 \f3^\fP, \f3$\fP, \f3*\fP, \f3%\fP or \f3\-\fP.
 For example, our \f3diff\fP command might have been \f3diff !!^.old !!^\fP
 or, equivalently, \f3diff !!$.old !!$\fP.
 However, if \f3!!\fP is abbreviated \f3!\fP,
-an argument selector beginning with \f3\-\fP will be interpreted as an event specification.
+an argument selector beginning with \f3\-\fP will be interpreted
+as an event specification.
 .PP
 The word(s) in a history reference can be edited
 by following them with one or more modifiers,
@@ -4886,7 +4892,8 @@ If \f2l\fP is empty,
 the \f2l\fP from the previous substitution is used,
 or if there is none,
 the \f2s\fP from the most recent \f3?\fP\f2s\fP\f3?\fP search.
-The trailing delimiter may be omitted if it is immediately followed by a newline.
+The trailing delimiter may be omitted if it is immediately followed
+by a newline.
 .TP 8
 .B &
 Repeat the previous substitution.
@@ -4921,16 +4928,19 @@ We might follow \f3mail \-s "I forgot my password" rot\fP
 with \f3!:s/rot/root\fP to correct the spelling of \f3root\fP.
 .PP
 History expansions also occur when an input line begins with \f3^\fP.
-When it is the first character on an input line, it is an abbreviation of \f3!:s^\fP.
-Thus we might have said \f3^rot^root\fP to make the spelling correction in the previous example.
+When it is the first character on an input line,
+it is an abbreviation of \f3!:s^\fP.
+Thus we might have said \f3^rot^root\fP to make the spelling correction
+in the previous example.
 This is the only history expansion that does not explicitly begin with \f3!\fP.
 .PP
 If a word on a command line begins with the history comment character \f3#\fP,
 history expansion is ignored for the rest of that line.
 This usually causes the shell parser
 (which uses the same character to signal a comment)
 to treat the rest of the line as a comment as well,
-but as history expansion is parsed separately from the shell grammar and with different rules,
+but as history expansion is parsed separately from the shell grammar
+and with different rules,
 this cannot be guaranteed in all cases.
 If the history comment character is changed,
 the shell grammar comment character does not change along with it.
@@ -6310,9 +6320,11 @@ is given, the name
 and value of the alias is printed.
 The obsolete
 .B \-x
-option has no effect in most contexts, although if it's used with
+option has no effect in most contexts, but when it's used with
 .B \-t
-it will suppress all output.
+it will cause
+.B alias
+to do nothing.
 The exit status is non-zero if a
 .I name\^
 is given, but no value, and no alias has been defined for the

src/cmd/ksh93/sh/defs.c

@@ -24,6 +24,7 @@
 #include	"shopt.h"
 #include	"defs.h"
 #include	"jobs.h"
+#include	"shlex.h"
 
 Shell_t			sh = {0};
 

src/cmd/ksh93/sh/nvtype.c

@@ -115,7 +115,7 @@ typedef struct
 	char		*_dpointer;
 } _Align_;
 
-#define alignof(t)	((char*)&((_Align_*)0)->_d##t-(char*)&((_Align_*)0)->_c##t)
+#define alignof(t)	( offsetof(_Align_, _d##t) - offsetof(_Align_, _c##t) )
 
 static void put_type(Namval_t*, const char*, int, Namfun_t*);
 static Namval_t* create_type(Namval_t*, const char*, int, Namfun_t*);

src/cmd/ksh93/tests/basic.sh

@@ -693,19 +693,28 @@ getCmdline() {
 		getPsOutput "$1"
 	fi
 }
-if	[[ ! $(uname -s) =~ ^SunOS$ ]] &&
-	getPsOutput "$$" &&
-	[[ "$SHELL $0" == "$actual"* ]]  # "$SHELL $0" is how shtests invokes this script
-then	expect='./atest 1 2'
-	echo 'sleep 10; exit 0' >atest
-	chmod 755 atest
-	./atest 1 2 &
-	getCmdline "$!"
-	kill "$!"
-	[[ $actual == "$expect" ]] || err_exit "ksh didn't rewrite argv correctly" \
-		"(expected $(printf %q "$expect"), got $(printf %q "$actual"))"
+check_profiling_or_asan() {
+	# Skip test when the binary was compiled with ASAN or is gathering profiling data
+	[[ -v ASAN_OPTIONS || -v TSAN_OPTIONS || -v MSAN_OPTIONS || -v LSAN_OPTIONS ]] && return 0
+	! whence -q readelf && return 1
+	[[ -n $(readelf -s "$SHELL" | grep -E "_asan_|__gcov") ]] && return 0
+	return 1
+}
+if ! check_profiling_or_asan
+then	if	[[ ! $(uname -s) =~ ^SunOS$ ]] &&
+		getPsOutput "$$" &&
+		[[ "$SHELL $0" == "$actual"* ]]  # "$SHELL $0" is how shtests invokes this script
+	then	expect='./atest 1 2'
+		echo 'sleep 10; exit 0' >atest
+		chmod 755 atest
+		./atest 1 2 &
+		getCmdline "$!"
+		kill "$!"
+		[[ $actual == "$expect" ]] || err_exit "ksh didn't rewrite argv correctly" \
+			"(expected $(printf %q "$expect"), got $(printf %q "$actual"))"
+	fi
+	unset -f getPsOutput getCmdline
 fi
-unset -f getPsOutput getCmdline
 
 # ======
 # https://bugzilla.redhat.com/1241013

src/cmd/ksh93/tests/signal.sh

@@ -397,7 +397,14 @@ yes() for ((;;)); do print y; done
 # The test for SIGBUS trap handling below is incompatible with ASan because ASan
 # implements its own SIGBUS handler independently of ksh.
 # Also, Android does not allow ignoring SIGBUS.
-if ! [[ -v ASAN_OPTIONS || -v TSAN_OPTIONS || -v MSAN_OPTIONS || -v LSAN_OPTIONS || $HOSTTYPE == android.* ]]; then
+check_asan() {
+	# Skip test when the binary was compiled with ASAN or is gathering profiling data
+	[[ -v ASAN_OPTIONS || -v TSAN_OPTIONS || -v MSAN_OPTIONS || -v LSAN_OPTIONS ]] && return 0
+	! whence -q readelf && return 1
+	[[ -n $(readelf -s "$SHELL" | grep -E "_asan_") ]] && return 0
+	return 1
+}
+if ! check_asan && ! [[ $HOSTTYPE == android.* ]]; then
 	trap '' SIGBUS
 	got=$("$SHELL" -c 'trap date SIGBUS; trap -p SIGBUS')
 	[[ "$got" ]] && err_exit 'SIGBUS should not have a trap' \

src/lib/libast/features/align.c

@@ -110,7 +110,7 @@ main(void)
 	printf("\n");
 	printf("#define ALIGN_CHUNK		%d\n", sizeof(char*) >= 4 ? 8192 : 1024);
 	printf("#define ALIGN_INTEGRAL		uintptr_t\n");
-	printf("#define ALIGN_INTEGER(x)	((char*)(x)-(char*)0)\n");
+	printf("#define ALIGN_INTEGER(x)	((intptr_t)(x))\n");
 	printf("#define ALIGN_POINTER(x)	((char*)(x))\n");
 	if (bits2 == (align2 - 1))
 		printf("#define ALIGN_ROUND(x,y)	ALIGN_POINTER(ALIGN_INTEGER((x)+(y)-1)&~((y)-1))\n");