[RFA] gdbserver, errno handling in my_waitpid

[RFA] gdbserver, errno handling in my_waitpid

[Michael Snyder]

[-- Attachment #1: Type: text/plain, Size: 233 bytes --]

This is not really my most familiar bit of the code, so
you gdbserver guys check me, ok?

It seems to me that we need to set errno to zero before calling waitpid,
if we're going to check it right after.  Especially the second time.


[-- Attachment #2: errno.txt --]
[-- Type: text/plain, Size: 825 bytes --]

2010-03-10  Michael Snyder  <msnyder@vmware.com>

	* linux-low.c (my_waitpid): Set errno to 0 before waitpid.

Index: linux-low.c
===================================================================
RCS file: /cvs/src/src/gdb/gdbserver/linux-low.c,v
retrieving revision 1.130
diff -u -p -r1.130 linux-low.c
--- linux-low.c	6 Mar 2010 00:11:42 -0000	1.130
+++ linux-low.c	10 Mar 2010 21:20:14 -0000
@@ -309,6 +309,7 @@ my_waitpid (int pid, int *status, int fl
 	{
 	  /* Since all signals are blocked, there's no need to check
 	     for EINTR here.  */
+	  errno = 0;
 	  ret = waitpid (pid, status, flags);
 	  out_errno = errno;
 
@@ -338,6 +339,7 @@ my_waitpid (int pid, int *status, int fl
     }
   else
     {
+      errno = 0;
       do
 	ret = waitpid (pid, status, flags);
       while (ret == -1 && errno == EINTR);

Re: [RFA] gdbserver, errno handling in my_waitpid

[Pedro Alves]

On Wednesday 10 March 2010 21:24:57, Michael Snyder wrote:
> This is not really my most familiar bit of the code, so
> you gdbserver guys check me, ok?
> 
> It seems to me that we need to set errno to zero before calling waitpid,
> if we're going to check it right after.  Especially the second time.

I don't think do; we only check it if waitpid returns -1, and
in that case, waitpid should have changed errno itself.  Or
have you found an actual problem here?

-- 
Pedro Alves

Re: [RFA] gdbserver, errno handling in my_waitpid

[Michael Snyder]

Pedro Alves wrote:
> On Wednesday 10 March 2010 21:24:57, Michael Snyder wrote:
>> This is not really my most familiar bit of the code, so
>> you gdbserver guys check me, ok?
>>
>> It seems to me that we need to set errno to zero before calling waitpid,
>> if we're going to check it right after.  Especially the second time.
> 
> I don't think do; we only check it if waitpid returns -1, and
> in that case, waitpid should have changed errno itself.  Or
> have you found an actual problem here?
> 

No, I guess you're right -- it just looked suspicious to me.