patch for rcp to work around sh4 math bug in gcc <= 3.0.2

[Dan Kegel]

I noticed that rcp'ing from an sh4-linux system always
fails for me, and tracked down the problem to a line in
inetutils-1.4.2/rcp/util.c:
         size = ((((((stb).st_blksize > 0 ? (stb).st_blksize : 512))+((blksize)-1))/(blksize))*(blksize));
When st_blksize was 4096, and blksize was 8192, this was
setting size to 2^31 or so.  It was quite flaky;
executing printf("") just prior to the function containing
the bad code, or adding print statements into the function
containing the bad code (even if after the bad code)
made the problem go away.

The problem occurs with both a gcc-2.97 snapshot,
and with vanilla gcc-3.0.2, but doesn't seem to happen
anymore with gcc-3.2.3.

Just thought I'd mention it here in case anyone else was
trying to bring up rcp on an sh4 system with an old compiler :-)
The following patch seems to work around the problem for me.
It should probably be rewritten to use and's and or's rather
than div and mult, but I just did a quick fix.
- Dan

--- inetutils-1.4.2/rcp/util.c.old	Tue Jul  8 11:38:47 2003
+++ inetutils-1.4.2/rcp/util.c	Tue Jul  8 11:40:03 2003
@@ -130,6 +130,11 @@
  	return (status);
  }

+static size_t froundup(size_t x, size_t y)
+{
+	return (((x+(y-1))/y)*y);
+}
+
  BUF *
  allocbuf(BUF *bp, int fd, int blksize)
  {
@@ -140,10 +145,7 @@
  		run_err("fstat: %s", strerror(errno));
  		return (0);
  	}
-#ifndef roundup
-#  define roundup(x, y)   ((((x)+((y)-1))/(y))*(y))
-#endif
-	size = roundup(ST_BLKSIZE(stb), blksize);
+	size = froundup(ST_BLKSIZE(stb), blksize);
  	if (size == 0)
  		size = blksize;
  	if (bp->cnt >= size)


-- 
Dan Kegel
http://www.kegel.com
http://counter.li.org/cgi-bin/runscript/display-person.cgi?user=78045


------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com