diff --git a/internal/rfmt/fmtsort/sort.go b/internal/rfmt/fmtsort/sort.go
index 9d2d580..a64d5cd 100644
--- a/internal/rfmt/fmtsort/sort.go
+++ b/internal/rfmt/fmtsort/sort.go
@@ -132,7 +132,7 @@ func compare(aVal, bVal reflect.Value) int {
 		default:
 			return -1
 		}
-	case reflect.Ptr:
+	case reflect.Ptr, reflect.UnsafePointer:
 		a, b := aVal.Pointer(), bVal.Pointer()
 		switch {
 		case a < b:
diff --git a/internal/rfmt/format.go b/internal/rfmt/format.go
index 49296f4..7d953cd 100644
--- a/internal/rfmt/format.go
+++ b/internal/rfmt/format.go
@@ -15,6 +15,8 @@ import (
 const (
 	ldigits = "0123456789abcdefx"
 	udigits = "0123456789ABCDEFX"
+
+	ellipsisS = "…"
 )
 
 const (
@@ -32,6 +34,9 @@ type fmtFlags struct {
 	space       bool
 	zero        bool
 
+	// ellipsis tells printf to insert an ellipsis after a truncated value.
+	ellipsis bool
+
 	// For the formats %+v %#v, we set the plusV/sharpV flags
 	// and clear the plus/sharp flags since %+v and %#v are in effect
 	// different, flagless formats set at the top level.
@@ -315,14 +320,18 @@ func (f *fmt) fmtInteger(u uint64, base int, isSigned bool, verb rune, digits st
 	f.zero = oldZero
 }
 
-// truncate truncates the string s to the specified precision, if present.
+// truncateString truncates the string s to the specified precision, if present.
 func (f *fmt) truncateString(s string) string {
 	if f.precPresent {
 		n := f.prec
 		for i := range s {
 			n--
 			if n < 0 {
-				return s[:i]
+				res := s[:i]
+				if f.ellipsis {
+					res += ellipsisS
+				}
+				return res
 			}
 		}
 	}
@@ -336,7 +345,11 @@ func (f *fmt) truncate(b []byte) []byte {
 		for i := 0; i < len(b); {
 			n--
 			if n < 0 {
-				return b[:i]
+				res := b[:i]
+				if f.ellipsis {
+					res = append(res, ellipsisS...)
+				}
+				return res
 			}
 			wid := 1
 			if b[i] >= utf8.RuneSelf {
@@ -368,8 +381,12 @@ func (f *fmt) fmtSbx(s string, b []byte, digits string) {
 		length = len(s)
 	}
 	// Set length to not process more bytes than the precision demands.
+	addEllipsis := false
 	if f.precPresent && f.prec < length {
 		length = f.prec
+		if f.ellipsis {
+			addEllipsis = true
+		}
 	}
 	// Compute width of the encoding taking into account the f.sharp and f.space flag.
 	width := 2 * length
@@ -421,6 +438,11 @@ func (f *fmt) fmtSbx(s string, b []byte, digits string) {
 		f.buf.WriteByte(digits[c>>4])
 		f.buf.WriteByte(digits[c&0xF])
 	}
+	// If an ellipsis was requested, add it.
+	if addEllipsis {
+		f.buf.writeString(ellipsisS)
+	}
+
 	// Handle padding to the right.
 	if f.widPresent && f.wid > width && f.minus {
 		f.writePadding(f.wid - width)
diff --git a/internal/rfmt/format.go.diff b/internal/rfmt/format.go.diff
index 7bf918c..762e11b 100644
--- a/internal/rfmt/format.go.diff
+++ b/internal/rfmt/format.go.diff
@@ -1,13 +1,38 @@
---- format.go.orig	2021-06-18 16:45:51.065116000 +0200
-+++ format.go	2021-06-18 17:30:58.628179000 +0200
-@@ -1,3 +1,6 @@
+--- format.go.orig	2021-09-03 11:01:24.164634000 +0200
++++ format.go	2021-09-03 11:06:51.632979000 +0200
+@@ -1,8 +1,11 @@
 +// Code generated from the Go standard library. DO NOT EDIT
 +// GENERATED FILE DO NOT EDIT
 +//
  // Copyright 2009 The Go Authors. All rights reserved.
  // Use of this source code is governed by a BSD-style
  // license that can be found in the LICENSE file.
-@@ -65,25 +68,17 @@
+ 
+-package fmt
++package rfmt
+ 
+ import (
+ 	"strconv"
+@@ -12,6 +15,8 @@
+ const (
+ 	ldigits = "0123456789abcdefx"
+ 	udigits = "0123456789ABCDEFX"
++
++	ellipsisS = "…"
+ )
+ 
+ const (
+@@ -29,6 +34,9 @@
+ 	space       bool
+ 	zero        bool
+ 
++	// ellipsis tells printf to insert an ellipsis after a truncated value.
++	ellipsis bool
++
+ 	// For the formats %+v %#v, we set the plusV/sharpV flags
+ 	// and clear the plus/sharp flags since %+v and %#v are in effect
+ 	// different, flagless formats set at the top level.
+@@ -65,25 +73,17 @@
  	if n <= 0 { // No padding bytes needed.
  		return
  	}
@@ -36,7 +61,46 @@
  }
  
  // pad appends b to f.buf, padded on left (!f.minus) or right (f.minus).
-@@ -401,19 +396,20 @@
+@@ -327,7 +327,11 @@
+ 		for i := range s {
+ 			n--
+ 			if n < 0 {
+-				return s[:i]
++				res := s[:i]
++				if f.ellipsis {
++					res += ellipsisS
++				}
++				return res
+ 			}
+ 		}
+ 	}
+@@ -341,7 +345,11 @@
+ 		for i := 0; i < len(b); {
+ 			n--
+ 			if n < 0 {
+-				return b[:i]
++				res := b[:i]
++				if f.ellipsis {
++					res = append(res, ellipsisS...)
++				}
++				return res
+ 			}
+ 			wid := 1
+ 			if b[i] >= utf8.RuneSelf {
+@@ -373,8 +381,12 @@
+ 		length = len(s)
+ 	}
+ 	// Set length to not process more bytes than the precision demands.
++	addEllipsis := false
+ 	if f.precPresent && f.prec < length {
+ 		length = f.prec
++		if f.ellipsis {
++			addEllipsis = true
++		}
+ 	}
+ 	// Compute width of the encoding taking into account the f.sharp and f.space flag.
+ 	width := 2 * length
+@@ -401,19 +413,20 @@
  		f.writePadding(f.wid - width)
  	}
  	// Write the encoding directly into the output buffer.
@@ -61,7 +125,7 @@
  			}
  		}
  		if b != nil {
-@@ -422,9 +418,9 @@
+@@ -422,9 +435,14 @@
  			c = s[i] // Take a byte from the input string.
  		}
  		// Encode each byte as two hexadecimal digits.
@@ -70,6 +134,11 @@
 +		f.buf.WriteByte(digits[c&0xF])
  	}
 -	*f.buf = buf
++	// If an ellipsis was requested, add it.
++	if addEllipsis {
++		f.buf.writeString(ellipsisS)
++	}
++
  	// Handle padding to the right.
  	if f.widPresent && f.wid > width && f.minus {
  		f.writePadding(f.wid - width)
diff --git a/internal/rfmt/print.go b/internal/rfmt/print.go
index 16ba3ce..b105bd0 100644
--- a/internal/rfmt/print.go
+++ b/internal/rfmt/print.go
@@ -1129,6 +1129,8 @@ formatLoop:
 				p.fmt.zero = false // Do not pad with zeros to the right.
 			case ' ':
 				p.fmt.space = true
+			case '!':
+				p.fmt.ellipsis = true
 			default:
 				// Fast path for common case of ascii lower case simple verbs
 				// without precision or width or argument indices.
diff --git a/internal/rfmt/print.go.diff b/internal/rfmt/print.go.diff
index 997e138..b63f0fa 100644
--- a/internal/rfmt/print.go.diff
+++ b/internal/rfmt/print.go.diff
@@ -1,5 +1,5 @@
---- print.go.orig	2021-06-18 16:45:56.623967000 +0200
-+++ print.go	2021-06-19 16:26:32.751061000 +0200
+--- print.go.orig	2021-09-03 11:01:24.167095000 +0200
++++ print.go	2021-09-03 11:06:51.631705000 +0200
 @@ -1,16 +1,27 @@
 +// Code generated from print.go.orig. DO NOT EDIT
 +// GENERATED FILE DO NOT EDIT
@@ -471,7 +471,16 @@
  	end := len(format)
  	argNum := 0         // we process one argument per non-trivial format
  	afterIndex := false // previous item in format was an index like [3].
-@@ -1147,6 +1265,7 @@
+@@ -1011,6 +1129,8 @@
+ 				p.fmt.zero = false // Do not pad with zeros to the right.
+ 			case ' ':
+ 				p.fmt.space = true
++			case '!':
++				p.fmt.ellipsis = true
+ 			default:
+ 				// Fast path for common case of ascii lower case simple verbs
+ 				// without precision or width or argument indices.
+@@ -1147,6 +1267,7 @@
  }
  
  func (p *pp) doPrint(a []interface{}) {
@@ -479,7 +488,7 @@
  	prevString := false
  	for argNum, arg := range a {
  		isString := arg != nil && reflect.TypeOf(arg).Kind() == reflect.String
-@@ -1162,6 +1281,7 @@
+@@ -1162,6 +1283,7 @@
  // doPrintln is like doPrint but always adds a space between arguments
  // and a newline after the last argument.
  func (p *pp) doPrintln(a []interface{}) {
diff --git a/markers_test.go b/markers_test.go
index 18f216e..d2f390c 100644
--- a/markers_test.go
+++ b/markers_test.go
@@ -204,6 +204,48 @@ func TestPrinter(t *testing.T) {
 		{func(w p) { w.Print(buf) }, "safe ‹unsafe›"},
 		{func(w p) { w.Printf("%v", &buf) }, "safe ‹unsafe›"},
 		{func(w p) { w.Print(&buf) }, "safe ‹unsafe›"},
+
+		// Redaction supports truncation for simple strings like printf.
+		{func(w p) { w.Printf("tr %.3s %.3s", "hello", Safe("world")) }, `tr ‹hel› wor`},
+		{func(w p) { w.Printf("tr %.3q %.3q", "hello", Safe("world")) }, `tr ‹"hel"› "wor"`},
+		{func(w p) { w.Printf("tr %.3v %.3v", "hello", Safe("world")) }, `tr ‹hel› wor`},
+		{func(w p) { w.Printf("tr %#.3q %#.3q", "hello", Safe("world")) }, "tr ‹`hel`› `wor`"},
+		{func(w p) { w.Printf("tr %.1T", 123) }, `tr i`},
+		{func(w p) { w.Printf("tr %.3v", map[string]string{"hello": "world"}) }, `tr map[‹hel›:‹wor›]`},
+		// Note that the precision affects the amount of input represented, not the size of the output.
+		// For example, with hexadecimal output.
+		{func(w p) { w.Printf("tr %.3x", "hello") }, `tr ‹68656c›`},
+		// Respect UTF-8 boundaries.
+		{func(w p) { w.Printf("tr %.1v", "☃☀") }, `tr ‹☃›`},
+		// Byte arrays are either represented as byte values or strings.
+		{func(w p) { w.Printf("tr %.1v %.1v", []byte("hello"), Safe([]byte("world"))) },
+			`tr [‹104› ‹101› ‹108› ‹108› ‹111›] [119 111 114 108 100]`},
+		{func(w p) { w.Printf("tr %.3s %.3s", []byte("hello"), Safe([]byte("world"))) }, `tr ‹hel› wor`},
+		// Special values do not get truncated.
+		{func(w p) { w.Printf("tr %.1v", nil) }, `tr <nil>`},
+		// Simple objects like booleans, integers etc do not get truncated.
+		{func(w p) { w.Printf("tr %.1v %.1v %.1v", 123, true, 11i) }, `tr ‹123› ‹true› (‹0›‹+1e+01›i)`},
+		{func(w p) { w.Printf("tr %.1v %.1v %.1v", Safe(123), Safe(true), Safe(11i)) }, `tr 123 true (0+1e+01i)`},
+		// Redactable strings do not get truncated by precision, because they have internal structure
+		// that could be broken by truncation.
+		{func(w p) { w.Printf("tr %.3v", RedactableString("‹hello› world")) }, `tr ‹hello› world`},
+		// By default, complex objects with a SafeFormat function do not get truncated by precision.
+		// This is because SafeFormat is itself responsible for implementing truncation by recognizing
+		// the precision field.
+		{func(w p) { w.Printf("tr %.3v", &safeNil{}) }, `tr hello ‹world›`},
+		{func(w p) { w.Printf("tr %.3v", buf) }, "tr safe ‹unsafe›"},
+
+		// Additionally, it can report when truncation has occurred.
+		{func(w p) { w.Printf("tre %!.3v %!.3v", "hello", Safe("world")) }, `tre ‹hel…› wor…`},
+		{func(w p) { w.Printf("tre %!.3s %!.3s", "hello", Safe("world")) }, `tre ‹hel…› wor…`},
+		{func(w p) { w.Printf("tre %!.3q %!.3q", "hello", Safe("world")) }, `tre ‹"hel…"› "wor…"`},
+		{func(w p) { w.Printf("tre %!#.3q %!#.3q", "hello", Safe("world")) }, "tre ‹`hel…`› `wor…`"},
+		{func(w p) { w.Printf("tre %!.3v", map[string]string{"hello": "world"}) }, `tre map[‹hel…›:‹wor…›]`},
+		{func(w p) { w.Printf("tre %!.1T", 123) }, `tre i…`},
+		{func(w p) { w.Printf("tre %!.3s %!.3s", []byte("hello"), Safe([]byte("world"))) }, `tre ‹hel…› wor…`},
+		{func(w p) { w.Printf("tre %!.3x", "hello") }, `tre ‹68656c…›`},
+		{func(w p) { w.Printf("tre %!.1v", "☃☀") }, `tre ‹☃…›`},
+
 	}
 
 	var methods = []struct {