Re: fix for pthread_broadcast

[Christopher Faylor <cgf at redhat dot com>]

On Sun, May 06, 2001 at 05:13:25PM +1000, Robert Collins wrote:
>pthread_broadcast was broken. This patch fixes it for the testcase
>reported by Greg Smith. I've also introduced per-cond variable locking
>to make broadcasts atomic. There are still races present, however the
>worst case is an occasional dropped signal. (And note that to trigger
>the race, the users code must be such that the signal could be missed
>_anyway_ ).

This looks good except for two formatting problems.

1) Please submit a ChangeLog entry, not a diff of a ChangeLog entry.  This
   is pretty much a standard requirement for any GNU program submission.

2) There are some formatting problems.  This block illustrates two:

+  if ((temperr=pthread_mutex_init(&this->cond_access, NULL)))
	       ^                  ^
	       missing spaces  here 
+    {
+      system_printf("couldn't init mutex, this = %0p errno=%d\n",this,temperr);
+      /* we need the mutex for correct behaviour */
+      magic = 0;

And, just a very minor nit, which you can take or leave.  I have tried to
avoid using foo = %s in debugging output in favor of foo %s, which I
think is equally clear to a human and results in less output in an
strace log file.

It isn't a universal constant, of course.  I don't know if Corinna follows
this rule or not.  It's just an opinion that you can either take or leave.

If you make the formatting changes above, feel free to check this in.

cgf