Skip to contentSkip to conversation

Use __kernel_long_t in struct mq_attr

PATCH In review H.J. Lu 2 participants 6 messages Dec 25, 2013 – Dec 26, 2013

lore
Share & tools
lore

H.J. Lu

author
#0 lore raw

PATCH 1/1 Use __kernel_long_t in struct mq_attr

Both x32 and x86-64 use the same struct mq_attr for system calls. But x32 long is 32-bit. This patch replaces long with __kernel_long_t in struct mq_attr.

---
 include/uapi/linux/mqueue.h | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
include/uapi/linux/mqueue.h +5 -5
index 8b5a796..d0a2b8e 100644
@@ -23,11 +23,11 @@
 #define MQ_BYTES_MAX	819200
 
 struct mq_attr {
-	long	mq_flags;	/* message queue flags			*/
-	long	mq_maxmsg;	/* maximum number of messages		*/
-	long	mq_msgsize;	/* maximum message size			*/
-	long	mq_curmsgs;	/* number of messages currently queued	*/
-	long	__reserved[4];	/* ignored for input, zeroed for output */
+	__kernel_long_t	mq_flags;	/* message queue flags			*/
+	__kernel_long_t	mq_maxmsg;	/* maximum number of messages		*/
+	__kernel_long_t	mq_msgsize;	/* maximum message size			*/
+	__kernel_long_t	mq_curmsgs;	/* number of messages currently queued	*/
+	__kernel_long_t	__reserved[4];	/* ignored for input, zeroed for output */
 };
 
 /*
Signature · 1 line
1.8.4.2
author
#1 lore raw
Quoted 4 lines · H.J. Lu

Both x32 and x86-64 use the same struct mq_attr for system calls. But x32 long is 32-bit. This patch replaces long with __kernel_long_t in struct mq_attr.

Here is the updated patch which uses __kernel_long_t only if __BITS_PER_LONG == 64.

H.J.

Quoted 1 line

From fc117a05dd0a4a9ec7e6ae286aac9433609ce155 Mon Sep 17 00:00:00 2001

From: "H.J. Lu" <[email protected]>
Date: Wed, 25 Dec 2013 08:56:22 -0800
Subject: [PATCH] Use __kernel_long_t in struct mq_attr

Both x32 and x86-64 use the same struct mq_attr for system calls. But x32 long is 32-bit. This patch replaces long with __kernel_long_t in struct mq_attr if __BITS_PER_LONG == 64.

---
 include/uapi/linux/mqueue.h | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)
include/uapi/linux/mqueue.h +13 -5
index 8b5a796..03bf30e 100644
@@ -23,11 +23,19 @@
 #define MQ_BYTES_MAX	819200
 
 struct mq_attr {
-	long	mq_flags;	/* message queue flags			*/
-	long	mq_maxmsg;	/* maximum number of messages		*/
-	long	mq_msgsize;	/* maximum message size			*/
-	long	mq_curmsgs;	/* number of messages currently queued	*/
-	long	__reserved[4];	/* ignored for input, zeroed for output */
+#if __BITS_PER_LONG == 64
+	__kernel_long_t	mq_flags;	/* message queue flags			*/
+	__kernel_long_t	mq_maxmsg;	/* maximum number of messages		*/
+	__kernel_long_t	mq_msgsize;	/* maximum message size			*/
+	__kernel_long_t	mq_curmsgs;	/* number of messages currently queued	*/
+	__kernel_long_t	__reserved[4];	/* ignored for input, zeroed for output */
+#else
+	long	mq_flags;		/* message queue flags			*/
+	long	mq_maxmsg;		/* maximum number of messages		*/
+	long	mq_msgsize;		/* maximum message size			*/
+	long	mq_curmsgs;		/* number of messages currently queued	*/
+	long	__reserved[4];		/* ignored for input, zeroed for output */
+#endif
 };
 
 /*
Signature · 1 line
1.8.4.2
#2 lore raw
Quoted 9 lines · H.J. Lu
Quoted 4 lines · H.J. Lu

Both x32 and x86-64 use the same struct mq_attr for system calls. But x32 long is 32-bit. This patch replaces long with __kernel_long_t in struct mq_attr.

Here is the updated patch which uses __kernel_long_t only if __BITS_PER_LONG == 64.

That seems wrong in multiple ways... not only does it add more clutter, but it would seem to give the wrong types when __BITS_PER_LONG in userspace is 32.

-hpa

author
#3 lore raw
Quoted 7 lines · H. Peter Anvin
Quoted 6 lines · H.J. Lu
Quoted 4 lines · H.J. Lu

Both x32 and x86-64 use the same struct mq_attr for system calls. But x32 long is 32-bit. This patch replaces long with __kernel_long_t in struct mq_attr.

Quoted 5 lines
Quoted 3 lines

Here is the updated patch which uses __kernel_long_t only if __BITS_PER_LONG == 64.

That seems wrong in multiple ways... not only does it add more clutter,

It is true for more clutter.

Quoted 3 lines

but it would seem to give the wrong types when __BITS_PER_LONG in userspace is 32.

For x32, __BITS_PER_LONG is 64, not 32. If __BITS_PER_LONG
is 32, my patch doesn't change anything. If it works before,
it still works. If it is broken before, it remains broken.

I prefer my first patch, which is less clutter. But I can't guarantee it is correct for all x3-like ABIs. My second patch has more clutter, but it has no impact on other ABIs.

Signature · 1 line
H.J.
#4 lore raw
Quoted 5 lines · H.J. Lu
Quoted 4 lines

but it would seem to give the wrong types when __BITS_PER_LONG in userspace is 32.

Quoted 8 lines

For x32, __BITS_PER_LONG is 64, not 32. If __BITS_PER_LONG
is 32, my patch doesn't change anything. If it works before,
it still works. If it is broken before, it remains broken.

I prefer my first patch, which is less clutter. But I can't guarantee it is correct for all x3-like ABIs. My second patch has more clutter, but it has no impact on other ABIs.

It's rather simple to prove, which is to consider the generic definition of __kernel_[u]long_t.

-hpa

author
#5 lore raw
Quoted 10 lines · H. Peter Anvin
Quoted 9 lines · H.J. Lu
Quoted 4 lines

but it would seem to give the wrong types when __BITS_PER_LONG in userspace is 32.

For x32, __BITS_PER_LONG is 64, not 32. If __BITS_PER_LONG
is 32, my patch doesn't change anything. If it works before,
it still works. If it is broken before, it remains broken.

Quoted 8 lines
Quoted 4 lines

I prefer my first patch, which is less clutter. But I can't guarantee it is correct for all x3-like ABIs. My second patch has more clutter, but it has no impact on other ABIs.

It's rather simple to prove, which is to consider the generic definition of __kernel_[u]long_t.

Then. I withdrew my second alternative with __BITS_PER_LONG == 64 check.. Please use my first alternative for all my kernel_long_t and kernel_ulong_t patches.

Thanks.

Signature · 1 line
H.J.