Re: [PATCH] gcc 3.5 fixes

Re: [PATCH] gcc 3.5 fixes

[Mark Adler]

Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl> wrote in 
message news:<2ecq2-80i-1@gated-at.bofh.it>...
> Can't we just remove these variables?
...
>> -static const char inflate_copyright[] =
>> +static const char inflate_copyright[] __attribute_used__ =
>>     " inflate 1.1.3 Copyright 1995-1998 Mark Adler ";

It's there for a reason -- to have a copyright notice in the 
executable.  I would prefer it if it were not removed.  Thanks.

mark

Re: [PATCH] gcc 3.5 fixes

[Paul Jackson]

I'm not a lawyer, but I don't see why you'd want the copyright notice
in the executable:

 * It's probably the source text that you're intending to copyright, not
   the binary executable bits.

 * If you look, I believe you will find that there are almost no other
   copyright strings in the vmlinux executable.

 * By what right can I copy or distribute a kernel built with this string
   in it?  The comments in zlib.h let us use the source, but I don't see
   any authorization to use the resulting executable bits on which this
   static copyright string seems to assert copyright.

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373

Re: [PATCH] gcc 3.5 fixes

[Mark Adler]

On Jul 5, 2004, at 2:44 PM, Paul Jackson wrote:
> I'm not a lawyer, but I don't see why you'd want the copyright notice
> in the executable:

It's not about the copyright.  I like the string kept in the object 
code so that I and others can find out where the code is used and what 
version is in use.  This in fact turned out to be useful when a 
security vulnerability was found in versions 1.1.3 and earlier and a 
script could search for executables with the offending code.

mark

Re: [PATCH] gcc 3.5 fixes

[Paul Jackson]

Mark wrote:
> It's not about the copyright. 

So in theory, then, one could change the string:

  inflate 1.1.3 Copyright 1995-1998 Mark Adler

to have instead the form:

  inflate 1.1.3 1995-1998 Mark Adler

Or ... one could just ignore my considerations, since I
really am not a lawyer ;).

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373

Re: [PATCH] gcc 3.5 fixes

[Jesse Stockall]

On Mon, 2004-07-05 at 21:16, Mark Adler wrote:
> 
> It's not about the copyright.  I like the string kept in the object 
> code so that I and others can find out where the code is used and what 
> version is in use.  This in fact turned out to be useful when a 
> security vulnerability was found in versions 1.1.3 and earlier and a 
> script could search for executables with the offending code.
> 

<-- snip -->

> " inflate 1.1.3 Copyright 1995-1998 Mark Adler "

Did I miss something or are you saying that the version in the kernel
has a security vulnerability?

Jesse

-- 
Jesse Stockall <stockall@magma.ca>

Re: [PATCH] gcc 3.5 fixes

[Mark Adler]

On Jul 5, 2004, at 8:51 PM, Jesse Stockall wrote:
> Did I miss something or are you saying that the version in the kernel
> has a security vulnerability?

If the kernel has 1.1.3, then yes.  You can get 1.1.4 here, which 
remedies that vulnerability:

     http://www.gzip.org/zlib/zlib-1.1.4.tar.gz

mark

Re: [PATCH] gcc 3.5 fixes

[bert hubert]

On Mon, Jul 05, 2004 at 09:43:58PM -0700, Mark Adler wrote:
> On Jul 5, 2004, at 8:51 PM, Jesse Stockall wrote:
> >Did I miss something or are you saying that the version in the kernel
> >has a security vulnerability?
> 
> If the kernel has 1.1.3, then yes.  You can get 1.1.4 here, which 
> remedies that vulnerability:

I seem to recall that vulnerability was fixed in place, without upgrading to
1.1.4.

-- 
http://www.PowerDNS.com      Open source, database driven DNS Software 
http://lartc.org           Linux Advanced Routing & Traffic Control HOWTO

Re: [PATCH] gcc 3.5 fixes

[David Woodhouse]

On Tue, 2004-07-06 at 08:11 +0200, bert hubert wrote:
> On Mon, Jul 05, 2004 at 09:43:58PM -0700, Mark Adler wrote:
> > On Jul 5, 2004, at 8:51 PM, Jesse Stockall wrote:
> > >Did I miss something or are you saying that the version in the kernel
> > >has a security vulnerability?
> > 
> > If the kernel has 1.1.3, then yes.  You can get 1.1.4 here, which 
> > remedies that vulnerability:
> 
> I seem to recall that vulnerability was fixed in place, without upgrading to
> 1.1.4.

Indeed, but it's probably time we upgraded to 1.2.

-- 
dwmw2

Re: [PATCH] gcc 3.5 fixes

[Mark Adler]

On Jul 5, 2004, at 11:46 PM, David Woodhouse wrote:
> Indeed, but it's probably time we upgraded to 1.2.

If you are using 1.1.4 or a properly patched 1.1.3, then that's good 
enough for now.  I recommend waiting a little while for 1.2.x to 
mature.

mark

[PATCH] gcc 3.5 fixes

[Anton Blanchard]

gcc 3.5 is warning about static vs non static function declarations. The
following patch removes function prototypes in .h files where possible
and changes prototypes to be static elsewhere.

Anton

Signed-off-by: Anton Blanchard <anton@samba.org>

diff -ur /root/toolchain/linux-2.5-bk/drivers/net/pppoe.c linux-2.5-bk/drivers/net/pppoe.c
--- /root/toolchain/linux-2.5-bk/drivers/net/pppoe.c	2004-06-25 09:13:14.000000000 +0000
+++ linux-2.5-bk/drivers/net/pppoe.c	2004-07-03 11:20:06.780997728 +0000
@@ -86,6 +86,7 @@
 static struct proto_ops pppoe_ops;
 static rwlock_t pppoe_hash_lock = RW_LOCK_UNLOCKED;
 
+static struct ppp_channel_ops pppoe_chan_ops;
 
 static inline int cmp_2_addr(struct pppoe_addr *a, struct pppoe_addr *b)
 {
diff -ur /root/toolchain/linux-2.5-bk/drivers/serial/8250.c linux-2.5-bk/drivers/serial/8250.c
--- /root/toolchain/linux-2.5-bk/drivers/serial/8250.c	2004-04-20 02:56:09.000000000 +0000
+++ linux-2.5-bk/drivers/serial/8250.c	2004-07-03 11:23:07.247987240 +0000
@@ -2013,7 +2013,7 @@
 	return uart_set_options(port, co, baud, parity, bits, flow);
 }
 
-extern struct uart_driver serial8250_reg;
+static struct uart_driver serial8250_reg;
 static struct console serial8250_console = {
 	.name		= "ttyS",
 	.write		= serial8250_console_write,
diff -ur /root/toolchain/linux-2.5-bk/drivers/usb/input/hiddev.c linux-2.5-bk/drivers/usb/input/hiddev.c
--- /root/toolchain/linux-2.5-bk/drivers/usb/input/hiddev.c	2004-06-19 05:48:19.000000000 +0000
+++ linux-2.5-bk/drivers/usb/input/hiddev.c	2004-07-03 11:22:28.699984216 +0000
@@ -66,9 +66,6 @@
 
 static struct hiddev *hiddev_table[HIDDEV_MINORS];
 
-/* forward reference to make our lives easier */
-extern struct usb_driver hiddev_driver;
-
 /*
  * Find a report, given the report's type and ID.  The ID can be specified
  * indirectly by REPORT_ID_FIRST (which returns the first report of the given
diff -ur /root/toolchain/linux-2.5-bk/fs/cifs/cifsfs.c linux-2.5-bk/fs/cifs/cifsfs.c
--- /root/toolchain/linux-2.5-bk/fs/cifs/cifsfs.c	2004-06-16 06:27:55.000000000 +0000
+++ linux-2.5-bk/fs/cifs/cifsfs.c	2004-07-03 11:04:41.790902048 +0000
@@ -48,8 +48,6 @@
 static struct quotactl_ops cifs_quotactl_ops;
 #endif
 
-extern struct file_system_type cifs_fs_type;
-
 int cifsFYI = 0;
 int cifsERROR = 1;
 int traceSMB = 0;
diff -ur /root/toolchain/linux-2.5-bk/include/linux/idr.h linux-2.5-bk/include/linux/idr.h
--- /root/toolchain/linux-2.5-bk/include/linux/idr.h	2004-06-21 13:50:10.000000000 +0000
+++ linux-2.5-bk/include/linux/idr.h	2004-07-03 11:16:54.561931696 +0000
@@ -76,5 +76,3 @@
 int idr_get_new_above(struct idr *idp, void *ptr, int starting_id, int *id);
 void idr_remove(struct idr *idp, int id);
 void idr_init(struct idr *idp);
-
-extern kmem_cache_t *idr_layer_cache;
diff -ur /root/toolchain/linux-2.5-bk/include/linux/if_pppox.h linux-2.5-bk/include/linux/if_pppox.h
--- /root/toolchain/linux-2.5-bk/include/linux/if_pppox.h	2004-02-22 13:48:05.000000000 +0000
+++ linux-2.5-bk/include/linux/if_pppox.h	2004-07-03 11:20:12.454966584 +0000
@@ -159,8 +159,6 @@
     PPPOX_DEAD		= 16  /* dead, useless, please clean me up!*/
 };
 
-extern struct ppp_channel_ops pppoe_chan_ops;
-
 #endif /* __KERNEL__ */
 
 #endif /* !(__LINUX_IF_PPPOX_H) */
diff -ur /root/toolchain/linux-2.5-bk/include/linux/netfilter_ipv4/ip_tables.h linux-2.5-bk/include/linux/netfilter_ipv4/ip_tables.h
--- /root/toolchain/linux-2.5-bk/include/linux/netfilter_ipv4/ip_tables.h	2004-06-21 13:50:10.000000000 +0000
+++ linux-2.5-bk/include/linux/netfilter_ipv4/ip_tables.h	2004-07-03 11:18:47.644908672 +0000
@@ -284,8 +284,6 @@
 	struct ipt_entry entrytable[0];
 };
 
-extern struct semaphore ipt_mutex;
-
 /* Standard return verdict, or do jump. */
 #define IPT_STANDARD_TARGET ""
 /* Error verdict. */
diff -ur /root/toolchain/linux-2.5-bk/include/linux/sunrpc/svcauth.h linux-2.5-bk/include/linux/sunrpc/svcauth.h
--- /root/toolchain/linux-2.5-bk/include/linux/sunrpc/svcauth.h	2004-06-13 14:37:00.000000000 +0000
+++ linux-2.5-bk/include/linux/sunrpc/svcauth.h	2004-07-03 11:27:44.772992792 +0000
@@ -93,7 +93,6 @@
 	int	(*release)(struct svc_rqst *rq);
 	void	(*domain_release)(struct auth_domain *);
 };
-extern struct auth_ops	*authtab[RPC_AUTH_MAXFLAVOR];
 
 #define	SVC_GARBAGE	1
 #define	SVC_SYSERR	2
diff -ur /root/toolchain/linux-2.5-bk/net/ipv4/xfrm4_state.c linux-2.5-bk/net/ipv4/xfrm4_state.c
--- /root/toolchain/linux-2.5-bk/net/ipv4/xfrm4_state.c	2004-05-26 04:15:19.000000000 +0000
+++ linux-2.5-bk/net/ipv4/xfrm4_state.c	2004-07-03 11:21:27.492969016 +0000
@@ -11,7 +11,7 @@
 #include <linux/pfkeyv2.h>
 #include <linux/ipsec.h>
 
-extern struct xfrm_state_afinfo xfrm4_state_afinfo;
+static struct xfrm_state_afinfo xfrm4_state_afinfo;
 
 static void
 __xfrm4_init_tempsel(struct xfrm_state *x, struct flowi *fl,
diff -ur /root/toolchain/linux-2.5-bk/net/netlink/af_netlink.c linux-2.5-bk/net/netlink/af_netlink.c
--- /root/toolchain/linux-2.5-bk/net/netlink/af_netlink.c	2004-06-19 05:48:21.000000000 +0000
+++ linux-2.5-bk/net/netlink/af_netlink.c	2004-07-03 11:24:21.701977136 +0000
@@ -176,7 +176,7 @@
 	return sk;
 }
 
-extern struct proto_ops netlink_ops;
+static struct proto_ops netlink_ops;
 
 static int netlink_insert(struct sock *sk, u32 pid)
 {

Re: [PATCH] gcc 3.5 fixes

[Anton Blanchard]

gcc 3.5 is warning about unused static variables, add __attribute_unused__
to the 2 places to silence it.

Signed-off-by: Anton Blanchard <anton@samba.org>

diff -ur /root/toolchain/linux-2.5-bk/kernel/configs.c linux-2.5-bk/kernel/configs.c
--- /root/toolchain/linux-2.5-bk/kernel/configs.c	2004-06-25 09:13:15.000000000 +0000
+++ linux-2.5-bk/kernel/configs.c	2004-07-03 11:03:45.019878184 +0000
@@ -58,7 +58,7 @@
 /**************************************************/
 /* globals and useful constants                   */
 
-static const char IKCONFIG_VERSION[] __initdata = "0.7";
+static const char IKCONFIG_VERSION[] __attribute_used__ __initdata = "0.7";
 
 static ssize_t
 ikconfig_read_current(struct file *file, char __user *buf,
diff -ur /root/toolchain/linux-2.5-bk/lib/zlib_inflate/inftrees.c linux-2.5-bk/lib/zlib_inflate/inftrees.c
--- /root/toolchain/linux-2.5-bk/lib/zlib_inflate/inftrees.c	2004-02-22 13:48:07.000000000 +0000
+++ linux-2.5-bk/lib/zlib_inflate/inftrees.c	2004-07-03 11:27:06.633933752 +0000
@@ -7,7 +7,7 @@
 #include "inftrees.h"
 #include "infutil.h"
 
-static const char inflate_copyright[] =
+static const char inflate_copyright[] __attribute_used__ =
    " inflate 1.1.3 Copyright 1995-1998 Mark Adler ";
 /*
   If you use the zlib library in a product, an acknowledgment is welcome

Re: [PATCH] gcc 3.5 fixes

[Anton Blanchard]

gcc 3.5 ICEd on the ppc64 version of __ptep_set_access_flags, but it does
look like we want to use *ptep here.

Signed-off-by: Anton Blanchard <anton@samba.org>

diff -ur /root/toolchain/linux-2.5-bk/include/asm-ppc64/pgtable.h linux-2.5-bk/include/asm-ppc64/pgtable.h
--- /root/toolchain/linux-2.5-bk/include/asm-ppc64/pgtable.h	2004-06-13 14:37:00.000000000 +0000
+++ linux-2.5-bk/include/asm-ppc64/pgtable.h	2004-07-03 12:01:10.522949800 +0000
@@ -424,7 +424,7 @@
 		stdcx.	%0,0,%4\n\
 		bne-	1b"
 	:"=&r" (old), "=&r" (tmp), "=m" (*ptep)
-	:"r" (bits), "r" (ptep), "m" (ptep), "i" (_PAGE_BUSY)
+	:"r" (bits), "r" (ptep), "m" (*ptep), "i" (_PAGE_BUSY)
 	:"cc");
 }
 #define  ptep_set_access_flags(__vma, __address, __ptep, __entry, __dirty) \

Re: [PATCH] gcc 3.5 fixes

[Bartlomiej Zolnierkiewicz]

Hi,

On Sunday 04 of July 2004 09:01, Anton Blanchard wrote:
> gcc 3.5 is warning about unused static variables, add __attribute_unused__
> to the 2 places to silence it.

Can't we just remove these variables?

> Signed-off-by: Anton Blanchard <anton@samba.org>
>
> diff -ur /root/toolchain/linux-2.5-bk/kernel/configs.c
> linux-2.5-bk/kernel/configs.c ---
> /root/toolchain/linux-2.5-bk/kernel/configs.c	2004-06-25 09:13:15.000000000
> +0000 +++ linux-2.5-bk/kernel/configs.c	2004-07-03 11:03:45.019878184 +0000
> @@ -58,7 +58,7 @@
>  /**************************************************/
>  /* globals and useful constants                   */
>
> -static const char IKCONFIG_VERSION[] __initdata = "0.7";
> +static const char IKCONFIG_VERSION[] __attribute_used__ __initdata =
> "0.7";
>
>  static ssize_t
>  ikconfig_read_current(struct file *file, char __user *buf,
> diff -ur /root/toolchain/linux-2.5-bk/lib/zlib_inflate/inftrees.c
> linux-2.5-bk/lib/zlib_inflate/inftrees.c ---
> /root/toolchain/linux-2.5-bk/lib/zlib_inflate/inftrees.c	2004-02-22
> 13:48:07.000000000 +0000 +++
> linux-2.5-bk/lib/zlib_inflate/inftrees.c	2004-07-03 11:27:06.633933752
> +0000 @@ -7,7 +7,7 @@
>  #include "inftrees.h"
>  #include "infutil.h"
>
> -static const char inflate_copyright[] =
> +static const char inflate_copyright[] __attribute_used__ =
>     " inflate 1.1.3 Copyright 1995-1998 Mark Adler ";
>  /*
>    If you use the zlib library in a product, an acknowledgment is welcome

Re: [PATCH] gcc 3.5 fixes

[Anton Blanchard]

 
> Can't we just remove these variables?

Certainly possible. Randy, does IKCONFIG_VERSION serve a purpose?

Anton

Re: [PATCH] gcc 3.5 fixes

[Randy.Dunlap]

On Sun, 4 Jul 2004 21:04:43 +1000 Anton Blanchard wrote:

|  
| > Can't we just remove these variables?
| 
| Certainly possible. Randy, does IKCONFIG_VERSION serve a purpose?

Just kill it.

Thanks,
--
~Randy

Re: [PATCH] gcc 3.5 fixes

[Anton Blanchard]

 
> Just kill it.

OK here we go.

--

Remove unused IKCONFIG_VERSION.

Signed-off-by: Anton Blanchard <anton@samba.org>

===== kernel/configs.c 1.9 vs edited =====
--- 1.9/kernel/configs.c	Mon Jul  5 20:34:16 2004
+++ edited/kernel/configs.c	Sat Jul 10 09:51:56 2004
@@ -55,11 +55,6 @@
 
 #ifdef CONFIG_IKCONFIG_PROC
 
-/**************************************************/
-/* globals and useful constants                   */
-
-static const char IKCONFIG_VERSION[] __attribute_used__ __initdata = "0.7";
-
 static ssize_t
 ikconfig_read_current(struct file *file, char __user *buf,
 		      size_t len, loff_t * offset)