Patch: Fix the segment missing in gcore maps parser.

Patch: Fix the segment missing in gcore maps parser.

[Nora Pan]

A bug in gdb /proc/<pid>/maps parser causes gcore can't capture the
segments following the one with a file name and zero inode entry in the
maps file. For example:

30028000-3002f000 rw-s 00000000 00:05 0 	 /SYSV0000bdad (deleted)

Okay for mainline?

-- 
Nora Pan
MontaVista Software




2005-03-18  Nora Pan  <qpan@mvista.com>

	* linux-nat.c (read_mapping): Capturing the segments after the
one with file name and zero inode.


diff -urNp src/gdb.orig/linux-nat.c src/gdb/linux-nat.c
--- src/gdb.orig/linux-nat.c	2005-03-06 08:42:20.000000000 -0800
+++ src/gdb/linux-nat.c	2005-03-18 17:48:05.000000000 -0800
@@ -2453,7 +2453,8 @@ read_mapping (FILE *mapfile,
   int ret = fscanf (mapfile, "%llx-%llx %s %llx %s %llx",
 		    addr, endaddr, permissions, offset, device, inode);
 
-  if (ret > 0 && ret != EOF && *inode != 0)
+  filename[0] = '\0';	
+  if (ret > 0 && ret != EOF)
     {
       /* Eat everything up to EOL for the filename.  This will prevent
          weird filenames (such as one with embedded whitespace) from

@@ -2464,11 +2465,7 @@ read_mapping (FILE *mapfile,
          only.  */
       ret += fscanf (mapfile, "%[^\n]\n", filename);
     }
-  else
-    {
-      filename[0] = '\0';	/* no filename */
-      fscanf (mapfile, "\n");
-    }
+
   return (ret != 0 && ret != EOF);
 }

Re: Patch: Fix the segment missing in gcore maps parser.

[Michael Snyder]

Nora Pan wrote:
> A bug in gdb /proc/<pid>/maps parser causes gcore can't capture the
> segments following the one with a file name and zero inode entry in the
> maps file. For example:
> 
> 30028000-3002f000 rw-s 00000000 00:05 0 	 /SYSV0000bdad (deleted)
> 
> Okay for mainline?

Well, as the gcore author, it looks fine to me.
Any linux native maintainers object?

Re: Patch: Fix the segment missing in gcore maps parser.

[Daniel Jacobowitz]

On Mon, Mar 21, 2005 at 11:23:07AM -0800, Michael Snyder wrote:
> Nora Pan wrote:
> >A bug in gdb /proc/<pid>/maps parser causes gcore can't capture the
> >segments following the one with a file name and zero inode entry in the
> >maps file. For example:
> >
> >30028000-3002f000 rw-s 00000000 00:05 0 	 /SYSV0000bdad (deleted)
> >
> >Okay for mainline?
> 
> Well, as the gcore author, it looks fine to me.
> Any linux native maintainers object?

Looks fine to me too.

-- 
Daniel Jacobowitz
CodeSourcery, LLC

RE: Patch: Fix the segment missing in gcore maps parser.

[Nora Pan]

The patch is still not in the community tree. Would you check in? 

Thanks,

Nora



>On Mon, Mar 21, 2005 at 11:23:07AM -0800, Michael Snyder wrote:
>> Nora Pan wrote:
>> >A bug in gdb /proc/<pid>/maps parser causes gcore can't capture the
>> >segments following the one with a file name and zero inode entry in
the
>> >maps file. For example:
>> >
>> >30028000-3002f000 rw-s 00000000 00:05 0 	 /SYSV0000bdad (deleted)
>> >
>> >Okay for mainline?
>> 
>> Well, as the gcore author, it looks fine to me.
>> Any linux native maintainers object?

>Looks fine to me too.

>-- 
>Daniel Jacobowitz
>CodeSourcery, LLC

Re: Patch: Fix the segment missing in gcore maps parser.

[Michael Snyder]

Nora Pan wrote:
> The patch is still not in the community tree. Would you check in? 

Committed.
Thanks for your contribution.

Michael

>>On Mon, Mar 21, 2005 at 11:23:07AM -0800, Michael Snyder wrote:
>>
>>>Nora Pan wrote:
>>>
>>>>A bug in gdb /proc/<pid>/maps parser causes gcore can't capture the
>>>>segments following the one with a file name and zero inode entry in
> 
> the
> 
>>>>maps file. For example:
>>>>
>>>>30028000-3002f000 rw-s 00000000 00:05 0 	 /SYSV0000bdad (deleted)
>>>>
>>>>Okay for mainline?
>>>
>>>Well, as the gcore author, it looks fine to me.
>>>Any linux native maintainers object?
> 
> 
>>Looks fine to me too.
> 
> 
>>-- 
>>Daniel Jacobowitz
>>CodeSourcery, LLC
> 
> 
>