[patch 1/1] fix syscallN() macro errno value checking for i386

[patch 1/1] fix syscallN() macro errno value checking for i386

[blaisorblade]

From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Cc: David Howells <dhowells@redhat.com>

The errno values which are visible for userspace are actually in the range 
-1 - -129, not until -128 (): this value was added:

#define	EKEYREJECTED	129	/* Key was rejected by service */

And this would break ucLibc (for what I heard).

This is just a quick-fix, because putting a macro inside errno.h instead of
having it copied in two places would be probably nicer.

However, I've heard by D. Howells it wasn't accepted, so this is the solution for now.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
---

 linux-2.6.11-paolo/include/asm-i386/unistd.h |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff -puN include/asm-i386/unistd.h~fix-syscall-macro include/asm-i386/unistd.h
--- linux-2.6.11/include/asm-i386/unistd.h~fix-syscall-macro	2005-01-29 00:42:48.000000000 +0100
+++ linux-2.6.11-paolo/include/asm-i386/unistd.h	2005-01-29 00:44:51.000000000 +0100
@@ -298,12 +298,12 @@
 #define NR_syscalls 289
 
 /*
- * user-visible error numbers are in the range -1 - -128: see
- * <asm-i386/errno.h>
+ * user-visible error numbers are in the range -1 - -129: see
+ * <asm-i386/errno.h> (currently it includes <asm-generic/errno.h>)
  */
 #define __syscall_return(type, res) \
 do { \
-	if ((unsigned long)(res) >= (unsigned long)(-(128 + 1))) { \
+	if ((unsigned long)(res) >= (unsigned long)(-(129 + 1))) { \
 		errno = -(res); \
 		res = -1; \
 	} \
_

Re: [patch 1/1] fix syscallN() macro errno value checking for i386

[Arnd Bergmann]

[-- Attachment #1: Type: text/plain, Size: 852 bytes --]

On Sünnavend 29 Januar 2005 02:01, blaisorblade@yahoo.it wrote:
> 
> From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
> Cc: David Howells <dhowells@redhat.com>
> 
> The errno values which are visible for userspace are actually in the range 
> -1 - -129, not until -128 (): this value was added:
> 
> #define	EKEYREJECTED	129	/* Key was rejected by service */
> 
> And this would break ucLibc (for what I heard).
> 
> This is just a quick-fix, because putting a macro inside errno.h instead of
> having it copied in two places would be probably nicer.

Yes. Note that your patch only fixes the bug on i386. The code has been
copied to many other architectures, and some of them have been updated
less recently and are checking for values lower than 128. There should
really be a way to keep them all in sync.

	Arnd <><

[-- Attachment #2: signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [patch 1/1] fix syscallN() macro errno value checking for i386

[jerome lacoste]

On Sun, 30 Jan 2005 18:00:22 +0100, Arnd Bergmann <arnd@arndb.de> wrote:
> On Sünnavend 29 Januar 2005 02:01, blaisorblade@yahoo.it wrote:
> >
> > From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
> > Cc: David Howells <dhowells@redhat.com>
> >
> > The errno values which are visible for userspace are actually in the range
> > -1 - -129, not until -128 (): this value was added:
> >
> > #define       EKEYREJECTED    129     /* Key was rejected by service */
> >
> > And this would break ucLibc (for what I heard).
> >
> > This is just a quick-fix, because putting a macro inside errno.h instead of
> > having it copied in two places would be probably nicer.
> 
> Yes. Note that your patch only fixes the bug on i386. The code has been
> copied to many other architectures, and some of them have been updated
> less recently and are checking for values lower than 128. There should
> really be a way to keep them all in sync.


what about something along?

#define EKEYNEXT    130     /* key counter */

and 

 if ((unsigned long)(res) >= (unsigned long)(-EKEYNEXT)) {

JL

Re: [patch 1/1] fix syscallN() macro errno value checking for i386

[H. Peter Anvin]

Followup to:  <5a2cf1f605013010305f8270de@mail.gmail.com>
By author:    jerome lacoste <jerome.lacoste@gmail.com>
In newsgroup: linux.dev.kernel
> 
> what about something along?
> 
> #define EKEYNEXT    130     /* key counter */
> 
> and 
> 
>  if ((unsigned long)(res) >= (unsigned long)(-EKEYNEXT)) {
> 

What you really need is EMAX.

	-hpa

Re: [patch 1/1] fix syscallN() macro errno value checking for i386

[Blaisorblade]

On Sunday 30 January 2005 19:30, jerome lacoste wrote:
> On Sun, 30 Jan 2005 18:00:22 +0100, Arnd Bergmann <arnd@arndb.de> wrote:
> > On Sünnavend 29 Januar 2005 02:01, blaisorblade@yahoo.it wrote:
> > > From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
> > > Cc: David Howells <dhowells@redhat.com>
> > >
> > > The errno values which are visible for userspace are actually in the
> > > range -1 - -129, not until -128 (): this value was added:
> > >
> > > #define       EKEYREJECTED    129     /* Key was rejected by service */
> > >
> > > And this would break ucLibc (for what I heard).
> > >
> > > This is just a quick-fix, because putting a macro inside errno.h
> > > instead of having it copied in two places would be probably nicer.
> >
> > Yes. Note that your patch only fixes the bug on i386. The code has been
> > copied to many other architectures, and some of them have been updated
> > less recently and are checking for values lower than 128. There should
> > really be a way to keep them all in sync.
>
> what about something along?
>
> #define EKEYNEXT    130     /* key counter */
>
> and
>
>  if ((unsigned long)(res) >= (unsigned long)(-EKEYNEXT)) {
Yes, I agree with you... I didn't do it that way because of this mail:

"Subject: Re: Fw: Re: [patch 02/11] uml: fix compilation for missing headers
From: David Howells <dhowells@redhat.com>
To: Andrew Morton <akpm@osdl.org>
CC: blaisorblade_spam@yahoo.it
Date: Fri Jan 14 12:19:33 2005
[....]
> I think that the max errno value should become a macro defined in errno.h.

I agree with him on this, but I seem to remember that this didn't go down very
well.

David"

Now, I don't know why it was not done that way, but I wanted to do a quick-fix 
to merge it. Somebody please fix it definitively... in 2.6.11 official this 
is still unfixed.

-- 
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
http://www.user-mode-linux.org/~blaisorblade