[PATCH] Fix buf inside tracepoint tfile memory read

[PATCH] Fix buf inside tracepoint tfile memory read

[Hui Zhu]

Hi,

When I try tfile with KGTP, got some error that tfine didn't give the
right value of memory.
So I checked the code and found that:
	  if (maddr <= offset && offset < (maddr + mlen))
	    {
	      amt = (maddr + mlen) - offset;
	      if (amt > len)
		amt = len;

	      tfile_read (readbuf, amt);
	      return amt;
	    }
This part code is for memroy read form tfile.  But if the offset is
not same with maddr, it will not read from the right part.  Because
this code always read from the 0 offset of this part of memory.
So I add some code to handle this issue,  and it is fixed.

Please help me review it.

Joel, I remember GDB will release a new version after 3 weeks.  I
suggest this issue be fixed before this release.

Thanks,
Hui

2012-06-15  Hui Zhu  <hui_zhu@mentor.com>

	* tracepoint.c (tfile_xfer_partial): Add a lseek.

---
 tracepoint.c |    2 ++
 1 file changed, 2 insertions(+)

--- a/tracepoint.c
+++ b/tracepoint.c
@@ -4545,6 +4545,8 @@ tfile_xfer_partial (struct target_ops *o
 	      if (amt > len)
 		amt = len;

+	      if (maddr != offset)
+	        lseek (trace_fd, offset - maddr, SEEK_CUR);
 	      tfile_read (readbuf, amt);
 	      return amt;
 	    }

Re: [PATCH] Fix buf inside tracepoint tfile memory read

[Pedro Alves]

On 06/15/2012 11:28 AM, Hui Zhu wrote:

> 2012-06-15  Hui Zhu  <hui_zhu@mentor.com>
> 
> 	* tracepoint.c (tfile_xfer_partial): Add a lseek.


Okay, thanks.

I wonder if a testcase could be constructed?

-- 
Pedro Alves

Re: [PATCH] Fix buf inside tracepoint tfile memory read

[Hui Zhu]

On Fri, Jun 15, 2012 at 11:06 PM, Pedro Alves <palves@redhat.com> wrote:
> On 06/15/2012 11:28 AM, Hui Zhu wrote:
>
>> 2012-06-15  Hui Zhu  <hui_zhu@mentor.com>
>>
>>       * tracepoint.c (tfile_xfer_partial): Add a lseek.
>
>
> Okay, thanks.
>
> I wonder if a testcase could be constructed?
>
> --
> Pedro Alves

Thanks.  Checked in.

http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/ChangeLog.diff?cvsroot=src&r1=1.14375&r2=1.14376
http://sourceware.org/cgi-bin/cvsweb.cgi/src/gdb/tracepoint.c.diff?cvsroot=src&r1=1.257&r2=1.258

I will post the test for it later.

Best,
Hui