linux-kernel - [PATCH 3/5] ihex: Simplify next record offset calculation

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-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20181221072841.30199-4-andrew.smirnov@gmail.com>
Date:   Thu, 20 Dec 2018 23:28:39 -0800
From:   Andrey Smirnov <andrew.smirnov@...il.com>
To:     linux-kernel <linux-kernel@...r.kernel.org>
Cc:     Andrey Smirnov <andrew.smirnov@...il.com>,
        Chris Healy <cphealy@...il.com>,
        Kyle McMartin <kyle@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Masahiro Yamada <yamada.masahiro@...ionext.com>,
        David Woodhouse <dwmw2@...radead.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: [PATCH 3/5] ihex: Simplify next record offset calculation

Next record calucaltion can be reduced to a much more tivial ALIGN
operation as follows:

1. Splitting 5 into 2 + 3 we get

   next = ((be16_to_cpu(rec->len) + 2 + 3) & ~3) - 2            (1)

2. Using ALIGN macro we reduce (1) to:

   ALIGN(be16_to_cpu(rec->len) + 2, 4) - 2                      (2)

3. Subsituting 'next' in original next record calucation we get:

   (void *)&rec->data[ALIGN(be16_to_cpu(rec->len) + 2, 4) - 2]  (3)

4. Converting array index to pointer arithmetic we convert (3) into:

   (void *)rec + sizeof(*rec) +
   	 ALIGN(be16_to_cpu(rec->len) + 2, 4) - 2		(4)

5. Subsituting sizeof(*rec) with its value, 6, and substracting 2,
   in (4) we get:

   (void *)rec + ALIGN(be16_to_cpu(rec->len) + 2, 4) + 4        (5)

6. Since ALIGN(X, 4) + 4 == ALIGN(X + 4, 4), (5) can be converted to:

   (void *)rec + ALIGN(be16_to_cpu(rec->len) + 6, 4)            (6)

5. Subsituting 6 in (6) to sizeof(*rec) we get:

   (void *)rec + ALIGN(be16_to_cpu(rec->len) + sizeof(*rec), 4) (7)

Using expression (7) should make it more clear that next record is
located by adding full size of the current record (payload + auxiliary
data) aligned to 4 bytes, to the location of the current one. No
functional change intended.

Cc: Chris Healy <cphealy@...il.com>
Cc: Kyle McMartin <kyle@...nel.org>
Cc: Andrew Morton <akpm@...ux-foundation.org>
Cc: Masahiro Yamada <yamada.masahiro@...ionext.com>
Cc: David Woodhouse <dwmw2@...radead.org>
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: linux-kernel <linux-kernel@...r.kernel.org>
Signed-off-by: Andrey Smirnov <andrew.smirnov@...il.com>
---
 include/linux/ihex.h | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/include/linux/ihex.h b/include/linux/ihex.h
index 9130f307a420..98cb5ce0b0a0 100644
--- a/include/linux/ihex.h
+++ b/include/linux/ihex.h
@@ -21,14 +21,18 @@ struct ihex_binrec {
 	uint8_t data[0];
 } __attribute__((packed));
 
+static inline uint16_t ihex_binrec_size(const struct ihex_binrec *p)
+{
+	return be16_to_cpu(p->len) + sizeof(*p);
+}
+
 /* Find the next record, taking into account the 4-byte alignment */
 static inline const struct ihex_binrec *
 __ihex_next_binrec(const struct ihex_binrec *rec)
 {
-	int next = ((be16_to_cpu(rec->len) + 5) & ~3) - 2;
-	rec = (void *)&rec->data[next];
+	const void *p = rec;
 
-	return rec;
+	return p + ALIGN(ihex_binrec_size(rec), 4);
 }
 
 static inline const struct ihex_binrec *
-- 
2.19.1