[Patch Darwin] minor fixups in section name reading.

[Patch Darwin] minor fixups in section name reading.

[Iain Sandoe]

[-- Attachment #1: Type: text/plain, Size: 436 bytes --]

this corrects a couple of trivial typos in  
bfd_mach_o_read_section_32/64.
It also caters for the fact that segment and section names may fill  
the on-disk field which is not zero-terminated.
So one cannot use them directly as strings.
OK?,
Iain


bfd:

	* mach-o.c (bfd_mach_o_make_bfd_section): Make strings from segname,  
sectname.
	(bfd_mach_o_read_section_32): Null-terminate sectname.
	(bfd_mach_o_read_section_64): Likewise.


[-- Attachment #2: 111207-macho-minor-secname-fixes-diff.txt --]
[-- Type: text/plain, Size: 1878 bytes --]

Index: bfd/mach-o.c
===================================================================
RCS file: /cvs/src/src/bfd/mach-o.c,v
retrieving revision 1.72
diff -u -p -r1.72 mach-o.c
--- bfd/mach-o.c	7 Dec 2011 10:09:22 -0000	1.72
+++ bfd/mach-o.c	8 Dec 2011 14:12:55 -0000
@@ -1667,9 +1667,16 @@ bfd_mach_o_make_bfd_section (bfd *abfd,
 {
   const char *sname;
   flagword flags;
+  char seg[BFD_MACH_O_SEGNAME_SIZE+1];
+  char sec[BFD_MACH_O_SECTNAME_SIZE+1];
+
+  memset (seg, 0, sizeof(seg));
+  strncpy (seg, (char *)segname, BFD_MACH_O_SEGNAME_SIZE);
+  memset (sec, 0, sizeof(sec));
+  strncpy (sec, (char *)sectname, BFD_MACH_O_SECTNAME_SIZE);
 
   bfd_mach_o_convert_section_name_to_bfd
-    (abfd, (const char *)segname, (const char *)sectname, &sname, &flags);
+    (abfd, (const char *)seg, (const char *)sec, &sname, &flags);
   if (sname == NULL)
     return NULL;
 
@@ -1698,7 +1705,7 @@ bfd_mach_o_read_section_32 (bfd *abfd,
   memcpy (section->segname, raw.segname, sizeof (raw.segname));
   section->segname[BFD_MACH_O_SEGNAME_SIZE] = 0;
   memcpy (section->sectname, raw.sectname, sizeof (raw.sectname));
-  section->segname[BFD_MACH_O_SECTNAME_SIZE] = 0;
+  section->sectname[BFD_MACH_O_SECTNAME_SIZE] = 0;
   section->addr = bfd_h_get_32 (abfd, raw.addr);
   section->size = bfd_h_get_32 (abfd, raw.size);
   section->offset = bfd_h_get_32 (abfd, raw.offset);
@@ -1737,7 +1744,7 @@ bfd_mach_o_read_section_64 (bfd *abfd,
   memcpy (section->segname, raw.segname, sizeof (raw.segname));
   section->segname[BFD_MACH_O_SEGNAME_SIZE] = 0;
   memcpy (section->sectname, raw.sectname, sizeof (raw.sectname));
-  section->segname[BFD_MACH_O_SECTNAME_SIZE] = 0;
+  section->sectname[BFD_MACH_O_SECTNAME_SIZE] = 0;
   section->addr = bfd_h_get_64 (abfd, raw.addr);
   section->size = bfd_h_get_64 (abfd, raw.size);
   section->offset = bfd_h_get_32 (abfd, raw.offset);

[-- Attachment #3: Type: text/plain, Size: 1 bytes --]

Re: [Patch Darwin] minor fixups in section name reading.

[Tristan Gingold]

On Dec 8, 2011, at 3:18 PM, Iain Sandoe wrote:

> this corrects a couple of trivial typos in bfd_mach_o_read_section_32/64.
> It also caters for the fact that segment and section names may fill the on-disk field which is not zero-terminated.
> So one cannot use them directly as strings.

Thank you for working on Mach-O.

See my comments.

(I think it is easier to review patches if they are inline).

Tristan.

> Index: bfd/mach-o.c
> ===================================================================
> RCS file: /cvs/src/src/bfd/mach-o.c,v
> retrieving revision 1.72
> diff -u -p -r1.72 mach-o.c
> --- bfd/mach-o.c	7 Dec 2011 10:09:22 -0000	1.72
> +++ bfd/mach-o.c	8 Dec 2011 14:12:55 -0000
> @@ -1667,9 +1667,16 @@ bfd_mach_o_make_bfd_section (bfd *abfd,
>  {
>    const char *sname;
>    flagword flags;
> +  char seg[BFD_MACH_O_SEGNAME_SIZE+1];
> +  char sec[BFD_MACH_O_SECTNAME_SIZE+1];
> +
> +  memset (seg, 0, sizeof(seg));
> +  strncpy (seg, (char *)segname, BFD_MACH_O_SEGNAME_SIZE);
> +  memset (sec, 0, sizeof(sec));
> +  strncpy (sec, (char *)sectname, BFD_MACH_O_SECTNAME_SIZE);
>  
>    bfd_mach_o_convert_section_name_to_bfd
> -    (abfd, (const char *)segname, (const char *)sectname, &sname, &flags);
> +    (abfd, (const char *)seg, (const char *)sec, &sname, &flags);
>    if (sname == NULL)
>      return NULL;

This is not necessary as bfd_mach_o_convert_section_name_to_bfd is documented as taking non nul terminated strings.  If there is a bug, it is in the later function.

> @@ -1698,7 +1705,7 @@ bfd_mach_o_read_section_32 (bfd *abfd,
>    memcpy (section->segname, raw.segname, sizeof (raw.segname));
>    section->segname[BFD_MACH_O_SEGNAME_SIZE] = 0;
>    memcpy (section->sectname, raw.sectname, sizeof (raw.sectname));
> -  section->segname[BFD_MACH_O_SECTNAME_SIZE] = 0;
> +  section->sectname[BFD_MACH_O_SECTNAME_SIZE] = 0;
>    section->addr = bfd_h_get_32 (abfd, raw.addr);
>    section->size = bfd_h_get_32 (abfd, raw.size);
>    section->offset = bfd_h_get_32 (abfd, raw.offset);
> @@ -1737,7 +1744,7 @@ bfd_mach_o_read_section_64 (bfd *abfd,
>    memcpy (section->segname, raw.segname, sizeof (raw.segname));
>    section->segname[BFD_MACH_O_SEGNAME_SIZE] = 0;
>    memcpy (section->sectname, raw.sectname, sizeof (raw.sectname));
> -  section->segname[BFD_MACH_O_SECTNAME_SIZE] = 0;
> +  section->sectname[BFD_MACH_O_SECTNAME_SIZE] = 0;
>    section->addr = bfd_h_get_64 (abfd, raw.addr);
>    section->size = bfd_h_get_64 (abfd, raw.size);
>    section->offset = bfd_h_get_32 (abfd, raw.offset);

Good catch.  You can commit this part.

Thanks,
Tristan.

Re: [Patch Darwin] minor fixups in section name reading.

[Iain Sandoe]

On 9 Dec 2011, at 08:31, Tristan Gingold wrote:
>>
>>   section->size = bfd_h_get_64 (abfd, raw.size);
>>   section->offset = bfd_h_get_32 (abfd, raw.offset);
>
> Good catch.  You can commit this part.

My gcc.gnu.org user credentials appear to work for the checkout - but  
not for the commit.

Is there an additional procedure?
  (write access is not specifically mentioned on the binutils webpage).

thanks
Iain

Re: [Patch Darwin] minor fixups in section name reading.

[Ian Lance Taylor]

Iain Sandoe <developer@sandoe-acoustics.co.uk> writes:

> My gcc.gnu.org user credentials appear to work for the checkout - but
> not for the commit.
>
> Is there an additional procedure?
>  (write access is not specifically mentioned on the binutils webpage).

Having write access to gcc is not the same as having write access to the
binutils.  They are managed and granted separately.

Ian