Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757486AbXH2Mej (ORCPT ); Wed, 29 Aug 2007 08:34:39 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754691AbXH2Mea (ORCPT ); Wed, 29 Aug 2007 08:34:30 -0400 Received: from mgw1.diku.dk ([130.225.96.91]:60904 "EHLO mgw1.diku.dk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753990AbXH2Me3 (ORCPT ); Wed, 29 Aug 2007 08:34:29 -0400 Subject: [PATCH] avoid negative shifts in radix-tree.c, take 2 From: Peter Lund To: Christoph Lameter Cc: Christoph Hellwig , trivial@kernel.org, linux-kernel@vger.kernel.org, Momchil Velikov , "Maciej W. Rozycki" Content-Type: text/plain Date: Wed, 29 Aug 2007 14:34:25 +0200 Message-Id: <1188390865.7216.5.camel@localhost.localdomain> Mime-Version: 1.0 X-Mailer: Evolution 2.10.1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2382 Lines: 63 From: Peter Lund Negative shifts are not allowed in C (the result is undefined). It works on most platforms but not on the VAX with gcc 4.0.1 (it results in an "operand reserved" fault). Applies to Linux 2.6.22. Signed-off-by: Peter Lund --- Shifting by more than the width of the value on the left is also not allowed. I think the extra '>> 1' tacked on at the end in the original code was an attempt to work around that. Getting rid of that is an extra feature of this patch. Since the shift amount is what causes the trouble, I felt it was better to name that value than the return value, which in any case becames much easier to read after removal of ' - 1' and '>> 1' and naming of the shift amount. Here's the chapter and verse, taken from the final draft of the C99 standard ("6.5.7 Bitwise shift operators", paragraph 3): "The integer promotions are performed on each of the operands. The type of the result is that of the promoted left operand. If the value of the right operand is negative or is greater than or equal to the width of the promoted left operand, the behavior is undefined." Thank you to Jan-Benedict Glaw and Christoph Hellwig for review. I could change the indentation so the variables and equal signs no longer line up but I'm pretty sure that would not be an improvement. I could also remove the else and unindent the second return statement but isn't that just a matter of personal taste? --- linux-2.6.22/lib/radix-tree.c.orig 2007-08-27 15:42:37.000000000 +0200 +++ linux-2.6.22/lib/radix-tree.c 2007-08-29 13:19:19.000000000 +0200 @@ -980,12 +980,13 @@ static __init unsigned long __maxindex(unsigned int height) { - unsigned int tmp = height * RADIX_TREE_MAP_SHIFT; - unsigned long index = (~0UL >> (RADIX_TREE_INDEX_BITS - tmp - 1)) >> 1; + unsigned int width = height * RADIX_TREE_MAP_SHIFT; + int shift = RADIX_TREE_INDEX_BITS - width; - if (tmp >= RADIX_TREE_INDEX_BITS) - index = ~0UL; - return index; + if (shift < 0) + return ~0UL; + else + return ~0UL >> shift; } static __init void radix_tree_init_maxindex(void) - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/