Re: GMAKE 3.81 vs GMAKE 4.2

[Paul Smith]

On Tue, 2019-12-17 at 13:17 +0530, nikhil jain wrote:
> A rule was processed and then as soon as it ends, the next dependent rule
> processed. It has to work on the object file generated by the first rule.
> But the timestamp of the file was older so the next rule did not process.
> I am on NFS.

I completely understand the issue.

> I assume that this is because we are doing an open and read operation in
> safe_stat which might remove the stale old file handle and make can see a
> newer updated file.
> 
> Also, the comment mentioned in the misc.c in the code you can see it has
> to be related with NFS caching - 
> 
>     /* For safety's sake, read something to make sure NFS attribute cache
>  *      * is updated
>  *           */

I have no idea where you've found this code.  This code does not exist in
GNU make and, as far as I can tell from Git, has _never_ existed in GNU
make.  I've searched the entire history back to 1988 and the term "NFS
attribute" has never appeared anywhere in the GNU make source.

Also, the implementation of safe_stat from the early 1990's has nothing to
do with opening or reading: this is the implementation:

commit 710e8e6f6f8c43e68bc4ee067bf7c9e37ca1f4e6
Author: Roland McGrath <address@hidden>
Date:   1994-07-04 21:46:58 +0000

    (safe_stat): New function, EINTR-safe wrapper around stat.

diff --git a/misc.c b/misc.c
index 06be0b27..b9e76b9b 100644
--- a/misc.c
+++ b/misc.c
@@ -716,3 +716,23 @@ get_path_max ()
   return value;
 }
 #endif
+^L
+/* On some systems, stat can return EINTR.  */
+
+int
+safe_stat (name, buf)
+     char *name;
+     struct stat *buf;
+{
+  int ret;
+
+#ifdef EINTR
+  do
+#endif
+    ret = stat (name, buf);
+#ifdef EINTR
+  while (ret < 0 && errno == EINTR);
+#endif
+
+  return  ret;
+}

It was removed again in Dec 1995 as I mentioned.  As you can see all it
does is wrap the stat(2) call with a check for EINTR.

> To answer your first question - 
> 
> YES, it works if i do a make clean (which deletes all the object files)
> and the next make command will work. But for incremental build, It wont
> work. I can reproduce the issue 8 out of 10 times with stat and 0 out of
> 1000 times with safe_stat..

Well I certainly wasn't suggesting you would run "make clean"!

You've said you've implemented a facility that sends a compile command out
to a different system to be run there, correct?  I was suggesting that you
change this implementation so that on the local system BEFORE you send the
command to another system to be invoked, you delete the specific target
(and only that target) that will be built by that command.

> Cant share exact code snippets as I am not allowed to.. but this is what
> is happening...

I'm not sure what you mean: if you've found implementations of safe_stat
that was distributed with GNU make, somewhere, then those implementations
are covered under the GPL.