lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241031063707.795842-1-aardelean@baylibre.com>
Date: Thu, 31 Oct 2024 08:37:06 +0200
From: Alexandru Ardelean <aardelean@...libre.com>
To: linux-iio@...r.kernel.org,
	linux-kernel@...r.kernel.org
Cc: jic23@...nel.org,
	bartosz.golaszewski@...aro.org,
	gregkh@...uxfoundation.org,
	akpm@...ux-foundation.org,
	Alexandru Ardelean <aardelean@...libre.com>
Subject: [PATCH 1/2] util_macros.h: fix/rework find_closest() macros

A bug was found in the find_closest() (find_closest_descending() is also
affected after some testing), where for certain values with small
progressions, the rounding (done by averaging 2 values) causes an incorrect
index to be returned.
The rounding issues occur for progressions of 1, 2 and 3. It goes away when
the progression/interval between two values is 4 or larger.

It's particularly bad for progressions of 1. For example if there's an
array of 'a = { 1, 2, 3 }', using 'find_closest(2, a ...)' would return 0
(the index of '1'), rather than returning 1 (the index of '2').
This means that for exact values (with a progression of 1), find_closest()
will misbehave and return the index of the value smaller than the one we're
searching for.
For progressions of 2 and 3, the exact values are obtained correctly; but
values aren't approximated correctly (as one would expect). Starting with
progressions of 4, all seems to be good.

This change reworks the find_closest(x,) macros to also check the
difference between the left and right elements when 'x'. If the distance to
the right is smaller (than the distance to the left), the index is
incremented by 1. This also makes redundant the need for using the
DIV_ROUND_CLOSEST() macro.

For find_closest_descending(), the operator was changed from '>=' to '>'.
Since the iteration is happening from the highest-to-lowest values, the
'>=' comparison would (for small progressions) prefer higher values (as
closer to the given values).
For example:
  Given array 'a[] = { 10, 7, 4, 1 };'
     find_closest_descending(2, a,...) returns the index[2] for 4
     find_closest_descending(5, a,...) returns the index[1] for 7
     find_closest_descending(8, a,...) returns the index[0] for 10

Signed-off-by: Alexandru Ardelean <aardelean@...libre.com>
---
 include/linux/util_macros.h | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/include/linux/util_macros.h b/include/linux/util_macros.h
index 6bb460c3e818..60c74770b703 100644
--- a/include/linux/util_macros.h
+++ b/include/linux/util_macros.h
@@ -7,12 +7,18 @@
 #define __find_closest(x, a, as, op)					\
 ({									\
 	typeof(as) __fc_i, __fc_as = (as) - 1;				\
-	typeof(x) __fc_x = (x);						\
+	typeof(x) __fc_mid_x, __fc_x = (x);				\
+	typeof(x) __fc_left, __fc_right;				\
 	typeof(*a) const *__fc_a = (a);					\
 	for (__fc_i = 0; __fc_i < __fc_as; __fc_i++) {			\
-		if (__fc_x op DIV_ROUND_CLOSEST(__fc_a[__fc_i] +	\
-						__fc_a[__fc_i + 1], 2))	\
+		__fc_mid_x = (__fc_a[__fc_i] + __fc_a[__fc_i + 1]) / 2;	\
+		if (__fc_x op __fc_mid_x) {				\
+			__fc_left = __fc_mid_x - __fc_a[__fc_i];	\
+			__fc_right = __fc_a[__fc_i + 1] - __fc_mid_x;	\
+			if (__fc_right < __fc_left)			\
+				__fc_i++;				\
 			break;						\
+		}							\
 	}								\
 	(__fc_i);							\
 })
@@ -38,7 +44,7 @@
  * Similar to find_closest() but 'a' is expected to be sorted in descending
  * order.
  */
-#define find_closest_descending(x, a, as) __find_closest(x, a, as, >=)
+#define find_closest_descending(x, a, as) __find_closest(x, a, as, >)
 
 /**
  * is_insidevar - check if the @ptr points inside the @var memory range.
-- 
2.46.1


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ