Re: + uml-make-execvp-safe-for-our-usage.patch added to -mm tree

Re: + uml-make-execvp-safe-for-our-usage.patch added to -mm tree

[Jeff Dike]

On Tue, Oct 17, 2006 at 05:14:35PM -0700, akpm@osdl.org wrote:
> The patch titled
> 
>      uml: make execvp safe for our usage
> 
> has been added to the -mm tree.  Its filename is
> 
>      uml-make-execvp-safe-for-our-usage.patch
> 

I had previously objected to this patch on grounds of taste.  Since
this fixes a number of problems, and I don't have any solutions which
are huge improvements, I withdraw those objections.

Feel free to send to Linus.

				Jeff

-- 
Work email - jdike at linux dot intel dot com

Re: + uml-make-execvp-safe-for-our-usage.patch added to -mm tree

[Andrew Morton]

On Fri, 24 Nov 2006 14:50:11 -0500
Jeff Dike <jdike@addtoit.com> wrote:

> On Tue, Oct 17, 2006 at 05:14:35PM -0700, akpm@osdl.org wrote:
> > The patch titled
> > 
> >      uml: make execvp safe for our usage
> > 
> > has been added to the -mm tree.  Its filename is
> > 
> >      uml-make-execvp-safe-for-our-usage.patch
> > 
> 
> I had previously objected to this patch on grounds of taste.  Since
> this fixes a number of problems, and I don't have any solutions which
> are huge improvements, I withdraw those objections.
> 
> Feel free to send to Linus.

OK.  Is it needed for 2.6.19?

Also, I'm still sitting on the below.  I have a note that you nacked it,
but I forget why.


From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>

Fix prototypes in user.h - was needed when including user.h in kernelspace, as
we did in previous patch.

Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
Cc: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
---

 arch/um/include/user.h |    9 ++++++---
 1 files changed, 6 insertions(+), 3 deletions(-)

diff -puN arch/um/include/user.h~uml-fix-prototypes arch/um/include/user.h
--- a/arch/um/include/user.h~uml-fix-prototypes
+++ a/arch/um/include/user.h
@@ -6,6 +6,10 @@
 #ifndef __USER_H__
 #define __USER_H__
 
+/* Both on kernelspace and userspace this will provide the size_t definition. It should, at
+ * least. But on userspace it won't hurt surely. */
+#include <linux/types.h>
+
 extern void panic(const char *fmt, ...)
 	__attribute__ ((format (printf, 1, 2)));
 extern int printk(const char *fmt, ...)
@@ -13,9 +17,8 @@ extern int printk(const char *fmt, ...)
 extern void schedule(void);
 extern int in_aton(char *str);
 extern int open_gdb_chan(void);
-/* These use size_t, however unsigned long is correct on both i386 and x86_64. */
-extern unsigned long strlcpy(char *, const char *, unsigned long);
-extern unsigned long strlcat(char *, const char *, unsigned long);
+extern size_t strlcpy(char *, const char *, size_t);
+extern size_t strlcat(char *, const char *, size_t);
 
 #endif
 
_

Re: + uml-make-execvp-safe-for-our-usage.patch added to -mm tree

[Jeff Dike]

On Fri, Nov 24, 2006 at 01:46:21PM -0800, Andrew Morton wrote:
> OK.  Is it needed for 2.6.19?

It fixes bugs, so I would say yes.

> Also, I'm still sitting on the below.  I have a note that you nacked it,
> but I forget why.

I nacked it mostly based on the changelog, which turned out to be
incorrect.  The substance of the change I think is OK, except that it
looks like stddef (according to the C standard - 4.1.5) , not
linux/types is the official way to get size_t.   I'll send a follow-up
patch to fix that.

I would hold this patch until after 2.6.19.

				Jeff
-- 
Work email - jdike at linux dot intel dot com