[PATCH v2 0/7] ASCII output section command

[PATCH v2 0/7] ASCII output section command

[binutils]

Implement the ASCII command for output sections

Introduce a command which adds a fixed size string.

   ASCII <size>  ',' <string>

Size specified in command. 
If the string is shorter than the allocated area, it is zero filled
If the string is longer than the allocated area, it is truncated
and a NUL character is added at the end.

Note: Will be merged into fewer patches after acceptance in final version

[PATCH v2 1/7] ldlex.l: Add ASCII token
[PATCH v2 2/7] ldgram.y: Add ASCII parsing
[PATCH v2 3/7] ldlang.*: process ASCII command
[PATCH v2 4/7] ld.texi: Add ASCII to info file
[PATCH v2 5/7] Add testsuite for ASCII command
[PATCH v2 6/7] NEWS: Add ASCII command
[PATCH v2 7/7] ChangeLog: Add ASCII command

[PATCH v2 1/7] ldlex.l: Add ASCII token

[binutils]

From: Ulf Samuelsson <ulf@emagii.com>

Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
---
 ld/ldlex.l | 1 +
 1 file changed, 1 insertion(+)

diff --git a/ld/ldlex.l b/ld/ldlex.l
index aa613100db0..c67c83689d7 100644
--- a/ld/ldlex.l
+++ b/ld/ldlex.l
@@ -311,6 +311,7 @@ V_IDENTIFIER [*?.$_a-zA-Z\[\]\-\!\^\\]([*?.$_a-zA-Z0-9\[\]\-\!\^\\]|::)*
 <WILD>"LONG"				{ RTOKEN(LONG); }
 <WILD>"SHORT"				{ RTOKEN(SHORT); }
 <WILD>"BYTE"				{ RTOKEN(BYTE); }
+<WILD>"ASCII"				{ RTOKEN(ASCII); }
 <WILD>"ASCIZ"				{ RTOKEN(ASCIZ); }
 <WILD>"LINKER_VERSION"			{ RTOKEN(LINKER_VERSION); }
 <SCRIPT>"NOFLOAT"			{ RTOKEN(NOFLOAT); }
-- 
2.17.1

[PATCH v2 2/7] ldgram.y: Add ASCII parsing

[binutils]

From: Ulf Samuelsson <ulf@emagii.com>

Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
---
 ld/ldgram.y | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/ld/ldgram.y b/ld/ldgram.y
index 7c04025f526..1d6965ff49d 100644
--- a/ld/ldgram.y
+++ b/ld/ldgram.y
@@ -128,7 +128,7 @@ static void yyerror (const char *);
 %right UNARY
 %token END
 %left <token> '('
-%token <token> ALIGN_K BLOCK BIND QUAD SQUAD LONG SHORT BYTE ASCIZ
+%token <token> ALIGN_K BLOCK BIND QUAD SQUAD LONG SHORT BYTE ASCII ASCIZ
 %token SECTIONS PHDRS INSERT_K AFTER BEFORE LINKER_VERSION
 %token DATA_SEGMENT_ALIGN DATA_SEGMENT_RELRO_END DATA_SEGMENT_END
 %token SORT_BY_NAME SORT_BY_ALIGNMENT SORT_NONE
@@ -703,9 +703,18 @@ statement:
 		{
 		  lang_add_data ((int) $1, $3);
 		}
+        | ASCII mustbe_exp ',' NAME
+		{
+		  /* 'value' is a memory leak, do we care?  */
+		  etree_type *value = $2;
+		  if (value->type.node_code == INT)
+		    lang_add_string (value->value.value, $4);
+		  else
+		    einfo (_("%X%P:%pS: ASCII expression must be an integer\n"), NULL);
+		}
 	| ASCIZ NAME
 		{
-		  lang_add_string ($2);
+		  lang_add_string (0, $2);
 		}
 	| FILL '(' fill_exp ')'
 		{
-- 
2.17.1

[PATCH v2 3/7] ldlang.*: process ASCII command

[binutils]

From: Ulf Samuelsson <ulf@emagii.com>

Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
---
 ld/ldlang.c | 65 +++++++++++++++++++++++++++++++++++++++++------------
 ld/ldlang.h |  2 +-
 2 files changed, 52 insertions(+), 15 deletions(-)

diff --git a/ld/ldlang.c b/ld/ldlang.c
index 9e8cc224f4d..6e592ee5318 100644
--- a/ld/ldlang.c
+++ b/ld/ldlang.c
@@ -8663,15 +8663,20 @@ lang_add_data (int type, union etree_union *exp)
   new_stmt->type = type;
 }
 
-void
-lang_add_string (const char *s)
+/* Convert escape codes in S.
+   Supports \n, \r, \t and \NNN octals.
+   Returns a copy of S in a malloc'ed buffer.  */
+
+static char *
+convert_string (const char * s)
 {
-  bfd_vma  len = strlen (s);
-  bfd_vma  i;
-  bool     escape = false;
+  size_t  len = strlen (s);
+  size_t  i;
+  bool    escape = false;
+  char *  buffer = xmalloc (len + 1);
+  char *  b;
 
-  /* Add byte expressions until end of string.  */
-  for (i = 0 ; i < len; i++)
+  for (i = 0, b = buffer; i < len; i++)
     {
       char c = *s++;
 
@@ -8729,21 +8734,53 @@ lang_add_string (const char *s)
 	      }
 	      break;
 	    }
-
-	  lang_add_data (BYTE, exp_intop (c));
 	  escape = false;
 	}
       else
 	{
 	  if (c == '\\')
-	    escape = true;
-	  else
-	    lang_add_data (BYTE, exp_intop (c));
+	    {
+	      escape = true;
+	      continue;
+	    }
 	}
+
+      * b ++ = c;
+    }
+
+  * b = 0;
+  return buffer;
+}
+
+void
+lang_add_string (size_t size, const char *s)
+{
+  size_t  len;
+  size_t  i;
+  char *  string;
+
+  string = convert_string (s);
+  len = strlen (string);
+
+  /* Check if it is ASCIZ command (len == 0) */
+  if (size == 0)
+    /* Make sure that we include the terminating nul byte.  */
+    size = len + 1;
+  else if (len >= size)
+    {
+      len = size - 1;
+
+      einfo (_("%P:%pS: warning: ASCII string does not fit in allocated space,"
+		" truncated\n"), NULL);
     }
 
-  /* Remeber to terminate the string.  */
-  lang_add_data (BYTE, exp_intop (0));
+  for (i = 0; i < len; i++)
+    lang_add_data (BYTE, exp_intop (string[i]));
+
+  while (i++ < size)
+    lang_add_data (BYTE, exp_intop ('\0'));
+
+  free (string);
 }
 
 /* Create a new reloc statement.  RELOC is the BFD relocation type to
diff --git a/ld/ldlang.h b/ld/ldlang.h
index f36e04c586a..bf1b602799c 100644
--- a/ld/ldlang.h
+++ b/ld/ldlang.h
@@ -656,7 +656,7 @@ extern void pop_stat_ptr
 extern void lang_add_data
   (int, union etree_union *);
 extern void lang_add_string
-  (const char *);
+  (size_t, const char *s);
 extern void lang_add_reloc
   (bfd_reloc_code_real_type, reloc_howto_type *, asection *, const char *,
    union etree_union *);
-- 
2.17.1

[PATCH v2 4/7] ld.texi: Add ASCII to info file

[binutils]

From: Ulf Samuelsson <ulf@emagii.com>

Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
---
 ld/ld.texi | 25 ++++++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)

diff --git a/ld/ld.texi b/ld/ld.texi
index 89e3913317a..7f86de5452d 100644
--- a/ld/ld.texi
+++ b/ld/ld.texi
@@ -5643,6 +5643,7 @@ C identifiers because they contain a @samp{.} character.
 @cindex section data
 @cindex output section data
 @kindex ASCIZ ``@var{string}''
+@kindex ASCII (@var{expression}) ``@var{string}''
 @kindex BYTE(@var{expression})
 @kindex SHORT(@var{expression})
 @kindex LONG(@var{expression})
@@ -5681,7 +5682,7 @@ endianness of the first input object file.
 
 You can include a zero-terminated string in an output section by using
 @code{ASCIZ}.  The keyword is followed by a string which is stored at
-the current value of the location counter adding a zero byte at the
+the current value of the location counter adding a NUL byte at the
 end.  If the string includes spaces it must be enclosed in double
 quotes.  The string may contain '\n', '\r', '\t' and octal numbers.
 Hex numbers are not supported.
@@ -5691,6 +5692,28 @@ For example, this string of 16 characters will create a 17 byte area
   ASCIZ "This is 16 bytes"
 @end smallexample
 
+You can include a fixed size string in an output section by using
+@code{ASCII}.  The keyword is followed by a size and a string
+separated by a comma.  The string is stored at the current
+value of the location counter. If the string is shorted than the allocated
+size, NUL character are added at the end to fill up to the specified size.
+Note the fill value is ignored for this padding.
+If the string includes spaces it must be enclosed in double
+quotes.  The string may contain '\n', '\r', '\t' and octal numbers.
+Hex numbers are not supported.
+
+If the string is too long, a warning is issued and the string is
+truncated.  The string will still be zero-terminated in this case.
+
+If the expression evaluates to zero then the directive will be treated
+as if it were @code{ASCIZ} instead.
+
+Example: This is string of 16 characters and will create a 16 byte
+area, not zero-terminated.
+@smallexample
+  ASCII 16, "This is 16 bytes"
+@end smallexample
+
 Note---these commands only work inside a section description and not
 between them, so the following will produce an error from the linker:
 @smallexample
-- 
2.17.1

Re: [PATCH v2 4/7] ld.texi: Add ASCII to info file

[Jan Beulich]

On 05.07.2024 18:22, binutils@emagii.com wrote:
> --- a/ld/ld.texi
> +++ b/ld/ld.texi
> @@ -5643,6 +5643,7 @@ C identifiers because they contain a @samp{.} character.
>  @cindex section data
>  @cindex output section data
>  @kindex ASCIZ ``@var{string}''
> +@kindex ASCII (@var{expression}) ``@var{string}''

I couldn't really spot anywhere you would say what has changed from the
earlier version, so I'm unsure: Are, as per above, parentheses still
required here, or ...

> @@ -5691,6 +5692,28 @@ For example, this string of 16 characters will create a 17 byte area
>    ASCIZ "This is 16 bytes"
>  @end smallexample
>  
> +You can include a fixed size string in an output section by using
> +@code{ASCII}.  The keyword is followed by a size and a string
> +separated by a comma.  The string is stored at the current
> +value of the location counter. If the string is shorted than the allocated
> +size, NUL character are added at the end to fill up to the specified size.
> +Note the fill value is ignored for this padding.
> +If the string includes spaces it must be enclosed in double
> +quotes.  The string may contain '\n', '\r', '\t' and octal numbers.
> +Hex numbers are not supported.
> +
> +If the string is too long, a warning is issued and the string is
> +truncated.  The string will still be zero-terminated in this case.
> +
> +If the expression evaluates to zero then the directive will be treated
> +as if it were @code{ASCIZ} instead.
> +
> +Example: This is string of 16 characters and will create a 16 byte
> +area, not zero-terminated.
> +@smallexample
> +  ASCII 16, "This is 16 bytes"
> +@end smallexample

... are they not anymore, as per this and the cover letter?

Jan

Re: [PATCH v2 4/7] ld.texi: Add ASCII to info file

[Ulf Samuelsson]

Den 2024-07-08 kl. 08:23, skrev Jan Beulich:
> On 05.07.2024 18:22, binutils@emagii.com wrote:
>> --- a/ld/ld.texi
>> +++ b/ld/ld.texi
>> @@ -5643,6 +5643,7 @@ C identifiers because they contain a @samp{.} character.
>>   @cindex section data
>>   @cindex output section data
>>   @kindex ASCIZ ``@var{string}''
>> +@kindex ASCII (@var{expression}) ``@var{string}''
> I couldn't really spot anywhere you would say what has changed from the
> earlier version, so I'm unsure: Are, as per above, parentheses still
> required here, or ...
>
>> @@ -5691,6 +5692,28 @@ For example, this string of 16 characters will create a 17 byte area
>>     ASCIZ "This is 16 bytes"
>>   @end smallexample
>>   
>> +You can include a fixed size string in an output section by using
>> +@code{ASCII}.  The keyword is followed by a size and a string
>> +separated by a comma.  The string is stored at the current
>> +value of the location counter. If the string is shorted than the allocated
>> +size, NUL character are added at the end to fill up to the specified size.
>> +Note the fill value is ignored for this padding.
>> +If the string includes spaces it must be enclosed in double
>> +quotes.  The string may contain '\n', '\r', '\t' and octal numbers.
>> +Hex numbers are not supported.
>> +
>> +If the string is too long, a warning is issued and the string is
>> +truncated.  The string will still be zero-terminated in this case.
>> +
>> +If the expression evaluates to zero then the directive will be treated
>> +as if it were @code{ASCIZ} instead.
>> +
>> +Example: This is string of 16 characters and will create a 16 byte
>> +area, not zero-terminated.
>> +@smallexample
>> +  ASCII 16, "This is 16 bytes"
>> +@end smallexample
> ... are they not anymore, as per this and the cover letter?
>
> Jan

I changed it from

ASCII '(' <size> ')' <string>

to

ASCII <size> ',' <string>


-- 
Best Regards
Ulf Samuelsson

[PATCH v2 5/7] Add testsuite for ASCII command

[binutils]

From: Ulf Samuelsson <ulf@emagii.com>

Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
---
 ld/testsuite/ld-scripts/ascii.d    | 25 ++++++++++++++++++++
 ld/testsuite/ld-scripts/ascii.s    | 11 +++++++++
 ld/testsuite/ld-scripts/ascii.t    | 38 ++++++++++++++++++++++++++++++
 ld/testsuite/ld-scripts/script.exp |  1 +
 4 files changed, 75 insertions(+)
 create mode 100644 ld/testsuite/ld-scripts/ascii.d
 create mode 100644 ld/testsuite/ld-scripts/ascii.s
 create mode 100644 ld/testsuite/ld-scripts/ascii.t

diff --git a/ld/testsuite/ld-scripts/ascii.d b/ld/testsuite/ld-scripts/ascii.d
new file mode 100644
index 00000000000..ff3bd328c15
--- /dev/null
+++ b/ld/testsuite/ld-scripts/ascii.d
@@ -0,0 +1,25 @@
+#source: ascii.s
+#ld: -T ascii.t
+#objdump: -s -j .header
+#target: [is_elf_format] [is_coff_format]
+#notarget: tic4x-*-* tic54x-*-*
+
+.*:     file format .*
+
+Contents of section .header:
+ .... 70726f67 72616d20 6e616d65 00000000  program name....
+ .... 656d7074 79000000 00000000 00000000  empty...........
+ .... 00000000 00000000 00000000 00000000  ................
+ .... 00000000 00000000 00000000 00000000  ................
+ .... 00000000 00000000 00000000 00000000  ................
+ .... 636f6d6d 656e7420 310a0000 00000000  comment 1.......
+ .... 00000000 00000000 00000000 00000000  ................
+ .... 636f6d6d 656e7420 320a0000 00000000  comment 2.......
+ .... 00000000 00000000 00000000 00000000  ................
+ .... 636f6d6d 656e7420 330a0000 00000000  comment 3.......
+ .... 00000000 00000000 00000000 00000000  ................
+ .... 636f6d6d 656e7420 340a0000 00000000  comment 4.......
+ .... 00000000 00000000 49206d65 616e7420  ........I meant 
+ .... 746f2073 61793a20 54686973 20697320  to say: This is 
+ .... 77617920 746f6f20 6c6f6e67 00000000  way too long....
+#pass
diff --git a/ld/testsuite/ld-scripts/ascii.s b/ld/testsuite/ld-scripts/ascii.s
new file mode 100644
index 00000000000..a1b6148dc79
--- /dev/null
+++ b/ld/testsuite/ld-scripts/ascii.s
@@ -0,0 +1,11 @@
+        .extern ecc_start
+	.section .text
+main:
+	.long 0x45444F43
+	.long 0x12345678
+	
+	.section .data
+	.long 0x9abcdef0
+	
+	.section .bss
+	.long 0
diff --git a/ld/testsuite/ld-scripts/ascii.t b/ld/testsuite/ld-scripts/ascii.t
new file mode 100644
index 00000000000..dd7249ff8a2
--- /dev/null
+++ b/ld/testsuite/ld-scripts/ascii.t
@@ -0,0 +1,38 @@
+_start = 0x000000;
+
+SECTIONS
+{
+  . = 0x1000 + SIZEOF_HEADERS;
+
+  .header ALIGN (0x100) (READONLY) :
+    {
+      ASCII 16, "program name"
+      ASCII 64, "empty"
+      ASCII 4 * 8, "comment 1\n"
+      ASCII 32, "comment 2\n"
+      ASCII 32, "comment 3\n"
+      ASCII 24, "comment 4\n"
+      ASCII 64, "I meant to say: This is way too long"
+    }
+
+  .text ALIGN (0x100) :
+  {
+      entry = .;
+      *(.text)
+  }
+
+  .data : AT (0x400000)
+  {
+	*(.data)
+  }
+  
+  . = ALIGN(0x20);
+  
+  .bss :
+  {
+	*(.bss)
+  }
+
+  /DISCARD/ : { *(*) }
+}
+
diff --git a/ld/testsuite/ld-scripts/script.exp b/ld/testsuite/ld-scripts/script.exp
index 213383de962..bc3fffe92d8 100644
--- a/ld/testsuite/ld-scripts/script.exp
+++ b/ld/testsuite/ld-scripts/script.exp
@@ -228,6 +228,7 @@ foreach test_script $test_script_list {
 }
 
 run_dump_test "asciz"
+run_dump_test "ascii"
 run_dump_test "align-with-input"
 run_dump_test "pr20302"
 run_dump_test "output-section-types"
-- 
2.17.1

[PATCH v2 6/7] NEWS: Add ASCII command

[binutils]

From: Ulf Samuelsson <ulf@emagii.com>

Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
---
 ld/NEWS | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/ld/NEWS b/ld/NEWS
index e0b9341a8ef..58144afd7a4 100644
--- a/ld/NEWS
+++ b/ld/NEWS
@@ -1,5 +1,9 @@
 -*- text -*-
 
+* The linker script syntax has a new command for output sections: ASCII (<size>) "string"
+  This will insert a string of the specified length at the current location.
+  It will be zero-filled if the string is shorter than <size> and truncated if longer.
+
 * Add -z isa-level-report=[none|all|needed|used] to the x86 ELF linker
   to report needed and used x86-64 ISA levels.
 
-- 
2.17.1

[PATCH v2 7/7] ChangeLog: Add ASCII command

[binutils]

From: Ulf Samuelsson <ulf@emagii.com>

Signed-off-by: Ulf Samuelsson <ulf@emagii.com>
---
 ld/ChangeLog | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/ld/ChangeLog b/ld/ChangeLog
index ab881707e11..0d7fc62a984 100644
--- a/ld/ChangeLog
+++ b/ld/ChangeLog
@@ -1,3 +1,18 @@
+2024-06-28  Ulf Samuelsson <binutils@emagii.com>
+	    Nick Clifton  <nickc@redhat.com>
+
+	* ldlex.l: Add ASCII token.
+	* ldgram.y: Add parsing of the ASCII command.
+	* ldlang.c (lang_add_string): Add maximum size parameter.  Move
+	escape character handling code into separate function.
+	* ldlang.h (lang_add_string): Update prototype.
+	* NEWS: Mention the new feature.
+	* ld.texi (Output Section Data): Document the new directives.
+	* testsuite/ld-scripts/ascii.d: New test driver.
+	* testsuite/ld-scripts/ascii.s: New test assembler source.
+	* testsuite/ld-scripts/ascii.t: New test script.
+	* testsuite/ld-scripts/script.exp: Run the new test.
+
 2024-01-15  Nick Clifton  <nickc@redhat.com>
 
 	* configure: Regenerate.
-- 
2.17.1