sfrench/cifs-2.6.git
18 years ago[MIPS] TX49XX has prefetch.
Atsushi Nemoto [Fri, 17 Mar 2006 03:59:22 +0000 (12:59 +0900)]
[MIPS] TX49XX has prefetch.

The TX49XX has the prefetch instruction.  It supports only Pref_Load
(hint 0).  Actually changes in this patch except for Kconfig are not
have any effects, I added these changes to prevent misuse of unsupported
hints.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Kill tlb-andes.c.
Thiemo Seufer [Tue, 14 Mar 2006 14:35:27 +0000 (14:35 +0000)]
[MIPS] Kill tlb-andes.c.

Basically identical to c-r4k.c, so maintaining one is really enough.

Signed-off-by: Thiemo Seufer <ths@networkno.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] War on whitespace: cleanup initial spaces followed by tabs.
Ralf Baechle [Sat, 11 Mar 2006 08:18:41 +0000 (08:18 +0000)]
[MIPS] War on whitespace: cleanup initial spaces followed by tabs.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Makefile crapectomy.
Ralf Baechle [Wed, 8 Mar 2006 11:35:00 +0000 (11:35 +0000)]
[MIPS] Makefile crapectomy.

Dump all the ridiculously complicated stuff that was needed support
compilers older and newer than 3.0.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Thiemo Seufer <ths@networkno.de>
18 years ago[MIPS] Reformat __xchg().
Ralf Baechle [Fri, 3 Mar 2006 09:42:05 +0000 (09:42 +0000)]
[MIPS] Reformat __xchg().

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Mention Broadcom part number for BigSur board
Martin Michlmayr [Wed, 1 Mar 2006 01:55:05 +0000 (01:55 +0000)]
[MIPS] Mention Broadcom part number for BigSur board

Mention the Broadcom part number for the BigSur board (BCM91480B)
in Kconfig, just like it's done for other Broadcom boards.

Signed-off-by: Martin Michlmayr <tbm@cyrius.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Remove CONFIG_BUILD_ELF64.
Ralf Baechle [Wed, 22 Feb 2006 23:06:55 +0000 (23:06 +0000)]
[MIPS] Remove CONFIG_BUILD_ELF64.

This option is no longer usable with supported compilers.  It will be
replaced by usage of -msym32 in a separate patch.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Further sparsification for 32-bit compat code.
Atsushi Nemoto [Tue, 21 Feb 2006 07:05:11 +0000 (16:05 +0900)]
[MIPS] Further sparsification for 32-bit compat code.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] fix wrong __user usage in _sysn32_rt_sigsuspend
Atsushi Nemoto [Mon, 20 Feb 2006 16:27:59 +0000 (01:27 +0900)]
[MIPS] fix wrong __user usage in _sysn32_rt_sigsuspend

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Signal cleanup
Atsushi Nemoto [Sun, 19 Feb 2006 14:46:44 +0000 (23:46 +0900)]
[MIPS] Signal cleanup

Move function prototypes to asm/signal.h to detect trivial errors and
add some __user tags to get rid of sparse warnings.  Generated code
should not be changed.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Reformat all of signal32.c with tabs instead of space for consistency
Martin Michlmayr [Sat, 18 Feb 2006 20:06:32 +0000 (20:06 +0000)]
[MIPS] Reformat all of signal32.c with tabs instead of space for consistency

Signed-off-by: Martin Michlmayr <tbm@cyrius.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Delete unused sys32_waitpid.
Ralf Baechle [Sun, 19 Feb 2006 03:20:07 +0000 (03:20 +0000)]
[MIPS] Delete unused sys32_waitpid.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Make I/O helpers more customizable
Atsushi Nemoto [Thu, 16 Feb 2006 16:36:24 +0000 (01:36 +0900)]
[MIPS] Make I/O helpers more customizable

1. Move ioswab*() and __mem_ioswab*() to mangle-port.h.  This gets rid
   of CONFIG_SGI_IP22 from include/asm-mips/io.h.

2. Pass a virtual address to *ioswab*().  Then we can provide
   mach-specific *ioswab*() and can do every evil thing based on its
   argument.  It could be useful on machines which have regions with
   different endian conversion scheme.

3. Call __swizzle_addr*() _after_ adding mips_io_port_base.  This
   unifies the meaning of the argument of __swizzle_addr*() (always
   virtual address).  Then mach-specific __swizzle_addr*() can do every
   evil thing based on the argument.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Symmetric Uniprocessor support for Qemu.
Ralf Baechle [Wed, 15 Feb 2006 15:17:19 +0000 (15:17 +0000)]
[MIPS] Symmetric Uniprocessor support for Qemu.

SMP bits needed to builds and run an SMP kernel.  While only a single
processor is supported ATM it's still useful for some SMP debugging using
Qemu.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] sc-rm7k.c cleanup
Atsushi Nemoto [Wed, 15 Feb 2006 09:25:48 +0000 (18:25 +0900)]
[MIPS] sc-rm7k.c cleanup

Use blast_scache_range, blast_inv_scache_range for rm7k scache routine.
Output code should be logically same.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] MIPS64 R2 optimizations for 64-bit endianess swapping.
Ralf Baechle [Wed, 15 Feb 2006 13:06:34 +0000 (13:06 +0000)]
[MIPS] MIPS64 R2 optimizations for 64-bit endianess swapping.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[MIPS] Add early console for Cobalt.
Peter Horton [Sun, 12 Feb 2006 17:10:25 +0000 (17:10 +0000)]
[MIPS] Add early console for Cobalt.

Signed-off-by: Peter Horton <pdh@colonel-panic.org>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
18 years ago[CRYPTO] aes: Fixed array boundary violation
David McCullough [Wed, 15 Mar 2006 10:08:51 +0000 (21:08 +1100)]
[CRYPTO] aes: Fixed array boundary violation

The AES setkey routine writes 64 bytes to the E_KEY area even though
there are only 60 bytes there.  It is in fact safe since E_KEY is
immediately follwed by D_KEY which is initialised afterwards.  However,
doing this may trigger undefined behaviour and makes Coverity unhappy.

So by combining E_KEY and D_KEY into one array we sidestep this issue
altogether.

This problem was reported by Adrian Bunk.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
18 years ago[CRYPTO] tcrypt: Fix key alignment
Atsushi Nemoto [Mon, 13 Mar 2006 10:39:23 +0000 (21:39 +1100)]
[CRYPTO] tcrypt: Fix key alignment

Force 32-bit alignment on keys in tcrypt test vectors.  Also rearrange the
structure to prevent unnecessary padding.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
18 years ago[CRYPTO] all: Add missing cra_alignmask
Atsushi Nemoto [Mon, 13 Mar 2006 10:30:29 +0000 (21:30 +1100)]
[CRYPTO] all: Add missing cra_alignmask

The "des3_ede" and "serpent" lack cra_alignmask.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
18 years ago[CRYPTO] all: Use kzalloc where possible
Eric Sesterhenn [Mon, 6 Mar 2006 10:42:07 +0000 (21:42 +1100)]
[CRYPTO] all: Use kzalloc where possible

this patch converts crypto/ to kzalloc usage.
Compile tested with allyesconfig.

Signed-off-by: Eric Sesterhenn <snakebyte@gmx.de>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
18 years ago[CRYPTO] api: Align tfm context as wide as possible
Herbert Xu [Wed, 25 Jan 2006 11:34:01 +0000 (22:34 +1100)]
[CRYPTO] api: Align tfm context as wide as possible

Since tfm contexts can contain arbitrary types we should provide at least
natural alignment (__attribute__ ((__aligned__))) for them.  In particular,
this is needed on the Xscale which is a 32-bit architecture with a u64 type
that requires 64-bit alignment.  This problem was reported by Ronen Shitrit.

The crypto_tfm structure's size was 44 bytes on 32-bit architectures and
80 bytes on 64-bit architectures.  So adding this requirement only means
that we have to add an extra 4 bytes on 32-bit architectures.

On i386 the natural alignment is 16 bytes which also benefits the VIA
Padlock as it no longer has to manually align its context structure to
128 bits.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
18 years ago[CRYPTO] twofish: Use rol32/ror32 where appropriate
Denis Vlasenko [Mon, 16 Jan 2006 06:42:28 +0000 (17:42 +1100)]
[CRYPTO] twofish: Use rol32/ror32 where appropriate

Convert open coded rotations to rol32/ror32.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
18 years ago[NETFILTER]: Add H.323 conntrack/NAT helper
Jing Min Zhao [Tue, 21 Mar 2006 07:41:17 +0000 (23:41 -0800)]
[NETFILTER]: Add H.323 conntrack/NAT helper

Signed-off-by: Jing Min Zhao <zhaojignmin@hotmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: Don't mark tg3_test_registers() as returning const.
David S. Miller [Tue, 21 Mar 2006 07:02:36 +0000 (23:02 -0800)]
[TG3]: Don't mark tg3_test_registers() as returning const.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV6]: Cleanups for net/ipv6/addrconf.c (kzalloc, early exit) v2
Ingo Oeser [Tue, 21 Mar 2006 07:01:47 +0000 (23:01 -0800)]
[IPV6]: Cleanups for net/ipv6/addrconf.c (kzalloc, early exit) v2

Here are some possible (and trivial) cleanups.
- use kzalloc() where possible
- invert allocation failure test like
  if (object) {
        /* Rest of function here */
  }
  to

  if (object == NULL)
        return NULL;

  /* Rest of function here */

Signed-off-by: Ingo Oeser <ioe-lkml@rameria.de>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV6]: Nearly complete kzalloc cleanup for net/ipv6
Ingo Oeser [Tue, 21 Mar 2006 07:01:32 +0000 (23:01 -0800)]
[IPV6]: Nearly complete kzalloc cleanup for net/ipv6

Stupidly use kzalloc() instead of kmalloc()/memset()
everywhere where this is possible in net/ipv6/*.c .

Signed-off-by: Ingo Oeser <ioe-lkml@rameria.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV6]: Cleanup of net/ipv6/reassambly.c
Ingo Oeser [Tue, 21 Mar 2006 07:01:17 +0000 (23:01 -0800)]
[IPV6]: Cleanup of net/ipv6/reassambly.c

Two minor cleanups:

1. Using kzalloc() in fraq_alloc_queue()
   saves the memset() in ipv6_frag_create().

2. Invert sense of if-statements to streamline code.
   Inverts the comment, too.

Signed-off-by: Ingo Oeser <ioe-lkml@rameria.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: Remove duplicate const from is_link_local() argument type.
Andrew Morton [Tue, 21 Mar 2006 07:00:56 +0000 (23:00 -0800)]
[BRIDGE]: Remove duplicate const from is_link_local() argument type.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DECNET]: net/decnet/dn_route.c: fix inconsequent NULL checking
Adrian Bunk [Tue, 21 Mar 2006 07:00:29 +0000 (23:00 -0800)]
[DECNET]: net/decnet/dn_route.c: fix inconsequent NULL checking

The Coverity checker noted this inconsequent NULL checking in
dnrt_drop().

Since all callers ensure that NULL isn't passed, we can simply remove
the check.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: make drivers/net/tg3.c:tg3_request_irq() static
Adrian Bunk [Tue, 21 Mar 2006 07:00:14 +0000 (23:00 -0800)]
[TG3]: make drivers/net/tg3.c:tg3_request_irq() static

This patch makes the needlessly global function tg3_request_irq()
static.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: use LLC to send STP
Stephen Hemminger [Tue, 21 Mar 2006 06:59:49 +0000 (22:59 -0800)]
[BRIDGE]: use LLC to send STP

The bridge code can use existing LLC output code when building
spanning tree protocol packets.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[LLC]: llc_mac_hdr_init const arguments
Stephen Hemminger [Tue, 21 Mar 2006 06:59:36 +0000 (22:59 -0800)]
[LLC]: llc_mac_hdr_init const arguments

Cleanup of LLC.  llc_mac_hdr_init can take constant arguments,
and it is defined twice once in llc_output.h that is otherwise unused.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Acked-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: allow show/store of group multicast address
Stephen Hemminger [Tue, 21 Mar 2006 06:59:21 +0000 (22:59 -0800)]
[BRIDGE]: allow show/store of group multicast address

Bridge's communicate with each other using Spanning Tree Protocol
over a standard multicast address. There are times when testing or
layering bridges over existing topologies or tunnels, when it is
useful to use alternative multicast addresses for STP packets.

The 802.1d standard has some unused addresses, that can be used for this.
This patch is restrictive in that it only allows one of the possible
addresses in the standard.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: use llc for receiving STP packets
Stephen Hemminger [Tue, 21 Mar 2006 06:59:06 +0000 (22:59 -0800)]
[BRIDGE]: use llc for receiving STP packets

Use LLC for the receive path of Spanning Tree Protocol packets.
This allows link local multicast packets to be received by
other protocols (if they care), and uses the existing LLC
code to get STP packets back into bridge code.

The bridge multicast address is also checked, so bridges using
other link local multicast addresses are ignored. This allows
for use of different multicast addresses to define separate STP
domains.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: stp timer to jiffies cleanup
Stephen Hemminger [Tue, 21 Mar 2006 06:58:49 +0000 (22:58 -0800)]
[BRIDGE]: stp timer to jiffies cleanup

Cleanup the get/set of bridge timer value in the packets.
It is clearer not to bury the conversion in macro.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: forwarding remove unneeded preempt and bh diasables
Stephen Hemminger [Tue, 21 Mar 2006 06:58:36 +0000 (22:58 -0800)]
[BRIDGE]: forwarding remove unneeded preempt and bh diasables

Optimize the forwarding and transmit paths. Both places are
called with bottom half/no preempt so there is no need to use
spin_lock_bh or rcu_read_lock.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: netfilter inline cleanup
Stephen Hemminger [Tue, 21 Mar 2006 06:58:21 +0000 (22:58 -0800)]
[BRIDGE]: netfilter inline cleanup

Move nf_bridge_alloc from header file to the one place it is
used and optimize it.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: netfilter VLAN macro cleanup
Stephen Hemminger [Tue, 21 Mar 2006 06:58:05 +0000 (22:58 -0800)]
[BRIDGE]: netfilter VLAN macro cleanup

Fix the VLAN macros in bridge netfilter code. Macros should
not depend on magic variables.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: netfilter dont use __constant_htons
Stephen Hemminger [Tue, 21 Mar 2006 06:57:46 +0000 (22:57 -0800)]
[BRIDGE]: netfilter dont use __constant_htons

Only use__constant_htons() for initializers and switch cases.
For other uses, it is just as efficient and clearer to use htons

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: netfilter whitespace
Stephen Hemminger [Tue, 21 Mar 2006 06:57:32 +0000 (22:57 -0800)]
[BRIDGE]: netfilter whitespace

Run br_netfilter through Lindent to fix whitespace.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: optimize frame pass up
Stephen Hemminger [Tue, 21 Mar 2006 06:57:18 +0000 (22:57 -0800)]
[BRIDGE]: optimize frame pass up

The netfilter hook that is used to receive frames doesn't need to be a
stub.  It is only called in two ways, both of which ignore the return
value.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: use kzalloc
Stephen Hemminger [Tue, 21 Mar 2006 06:57:03 +0000 (22:57 -0800)]
[BRIDGE]: use kzalloc

Use kzalloc versus kmalloc+memset. Also don't need to do
memset() of bridge address since it is in netdev private data
that is already zero'd in alloc_netdev.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: use kcalloc
Stephen Hemminger [Tue, 21 Mar 2006 06:56:50 +0000 (22:56 -0800)]
[BRIDGE]: use kcalloc

Use kcalloc rather than kmalloc + memset.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: use setup_timer
Stephen Hemminger [Tue, 21 Mar 2006 06:56:38 +0000 (22:56 -0800)]
[BRIDGE]: use setup_timer

Use the now standard setup_timer function.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE]: remove unneeded bh disables
Stephen Hemminger [Tue, 21 Mar 2006 06:56:25 +0000 (22:56 -0800)]
[BRIDGE]: remove unneeded bh disables

The STP timers run off softirq (kernel timers), so there is no need to
disable bottom half in the spin locks.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE] br_netfilter: Warning fixes.
Andrew Morton [Tue, 21 Mar 2006 06:55:24 +0000 (22:55 -0800)]
[BRIDGE] br_netfilter: Warning fixes.

net/bridge/br_netfilter.c: In function `br_nf_pre_routing':
net/bridge/br_netfilter.c:427: warning: unused variable `vhdr'
net/bridge/br_netfilter.c:445: warning: unused variable `vhdr'

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[BRIDGE] ebtables: Build fix.
Andrew Morton [Tue, 21 Mar 2006 06:55:02 +0000 (22:55 -0800)]
[BRIDGE] ebtables: Build fix.

net/bridge/netfilter/ebtables.c:1481: warning: initialization makes pointer from integer without a cast

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SPARC]: Fixup SO_PEERSEC value on 32-bit sparc.
David S. Miller [Tue, 21 Mar 2006 06:53:42 +0000 (22:53 -0800)]
[SPARC]: Fixup SO_PEERSEC value on 32-bit sparc.

Sparc64 and Sparc32 have to have identical socket call
numbering in order to handle compat layer stuff properly.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[INET]: Fix typo in Arnaldo's connection sock compat fixups.
David S. Miller [Tue, 21 Mar 2006 06:52:32 +0000 (22:52 -0800)]
[INET]: Fix typo in Arnaldo's connection sock compat fixups.

"struct inet_csk" --> "struct inet_connection_sock" :-)

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP] feat: Pass dccp_minisock ptr where only the minisock is used
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:51:53 +0000 (22:51 -0800)]
[DCCP] feat: Pass dccp_minisock ptr where only the minisock is used

This is in preparation for having a dccp_minisock embedded into
dccp_request_sock so that feature negotiation can be done prior to
creating the full blown dccp_sock.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP] minisock: Rename struct dccp_options to struct dccp_minisock
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:50:58 +0000 (22:50 -0800)]
[DCCP] minisock: Rename struct dccp_options to struct dccp_minisock

This will later be included in struct dccp_request_sock so that we can
have per connection feature negotiation state while in the 3way
handshake, when we clone the DCCP_ROLE_LISTEN socket (in
dccp_create_openreq_child) we'll just copy this state from
dreq_minisock to dccps_minisock.

Also the feature negotiation and option parsing code will mostly touch
dccps_minisock, which will simplify some stuff.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SELINUX]: selinux_socket_getpeer_{stream,dgram} fixup
Catherine Zhang [Tue, 21 Mar 2006 06:49:00 +0000 (22:49 -0800)]
[SELINUX]: selinux_socket_getpeer_{stream,dgram} fixup

Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
18 years ago[NET]: Identation & other cleanups related to compat_[gs]etsockopt cset
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:48:35 +0000 (22:48 -0800)]
[NET]: Identation & other cleanups related to compat_[gs]etsockopt cset

No code changes, just tidying up, in some cases moving EXPORT_SYMBOLs
to just after the function exported, etc.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SK_BUFF]: export skb_pull_rcsum
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:47:55 +0000 (22:47 -0800)]
[SK_BUFF]: export skb_pull_rcsum

*** Warning: "skb_pull_rcsum" [net/bridge/bridge.ko] undefined!
*** Warning: "skb_pull_rcsum" [net/8021q/8021q.ko] undefined!
*** Warning: "skb_pull_rcsum" [drivers/net/pppoe.ko] undefined!
*** Warning: "skb_pull_rcsum" [drivers/net/ppp_generic.ko] undefined!

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SECURITY] getpeersec: Fix build breakage
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:47:37 +0000 (22:47 -0800)]
[SECURITY] getpeersec: Fix build breakage

A recent changeset removes dummy_socket_getpeersec, replacing it with
two new functions, but still references the removed function in the
security_fixup_ops table, fix it by doing the replacement operation in
the fixup table too.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[INFINIBAND] ipoib: Remove leftover use of neigh_ops->destructor
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:46:40 +0000 (22:46 -0800)]
[INFINIBAND] ipoib: Remove leftover use of neigh_ops->destructor

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ICSK] compat: Introduce inet_csk_compat_[gs]etsockopt
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:46:16 +0000 (22:46 -0800)]
[ICSK] compat: Introduce inet_csk_compat_[gs]etsockopt

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SNAP]: Remove leftover unused hdr variable
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:45:37 +0000 (22:45 -0800)]
[SNAP]: Remove leftover unused hdr variable

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: {get|set}sockopt compatibility layer
Dmitry Mishin [Tue, 21 Mar 2006 06:45:21 +0000 (22:45 -0800)]
[NET]: {get|set}sockopt compatibility layer

This patch extends {get|set}sockopt compatibility layer in order to
move protocol specific parts to their place and avoid huge universal
net/compat.c file in the future.

Signed-off-by: Dmitry Mishin <dim@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV6]: remove useless test in ip6_append_data
Dave Jones [Tue, 21 Mar 2006 06:44:52 +0000 (22:44 -0800)]
[IPV6]: remove useless test in ip6_append_data

We've already dereferenced 'np' a dozen
times at this point, so it's safe to say it's not null.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[PKT_SCHED]: Let NET_CLS_ACT no longer depend on EXPERIMENTAL
Adrian Bunk [Tue, 21 Mar 2006 06:44:24 +0000 (22:44 -0800)]
[PKT_SCHED]: Let NET_CLS_ACT no longer depend on EXPERIMENTAL

This option should IMHO no longer depend on EXPERIMENTAL.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
ACKed-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: Replace skb_pull/skb_postpull_rcsum with skb_pull_rcsum
Herbert Xu [Tue, 21 Mar 2006 06:43:56 +0000 (22:43 -0800)]
[NET]: Replace skb_pull/skb_postpull_rcsum with skb_pull_rcsum

We're now starting to have quite a number of places that do skb_pull
followed immediately by an skb_postpull_rcsum.  We can merge these two
operations into one function with skb_pull_rcsum.  This makes sense
since most pull operations on receive skb's need to update the
checksum.

I've decided to make this out-of-line since it is fairly big and the
fast path where hardware checksums are enabled need to call
csum_partial anyway.

Since this is a brand new function we get to add an extra check on the
len argument.  As it is most callers of skb_pull ignore its return
value which essentially means that there is no check on the len
argument.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DECnet]: Use RCU locking in dn_rules.c
Steven Whitehouse [Tue, 21 Mar 2006 06:43:28 +0000 (22:43 -0800)]
[DECnet]: Use RCU locking in dn_rules.c

As per Robert Olsson's patch for ipv4, this is the DECnet
version to keep the code "in step". It changes the list
of rules to use RCU rather than an rwlock.

Inspired-by: Robert Olsson <robert.olsson@its.uu.se>
Signed-off-by: Steven Whitehouse <steve@chygwyn.com>
Signed-off-by: Patrick Caulfield <patrick@tykepenguin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DECnet]: Patch to fix recvmsg() flag check
Patrick Caulfield [Tue, 21 Mar 2006 06:43:05 +0000 (22:43 -0800)]
[DECnet]: Patch to fix recvmsg() flag check

This patch means that 64bit kernel/32bit userland platforms will now
work correctly with DECnet.

Signed-off-by: Patrick Caulfield <patrick@tykepenguin.com>
Signed-off-by: Steven Whitehouse <steve@chygwyn.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DECnet]: Endian annotation and fixes for DECnet.
Steven Whitehouse [Tue, 21 Mar 2006 06:42:39 +0000 (22:42 -0800)]
[DECnet]: Endian annotation and fixes for DECnet.

The typedef for dn_address has been removed in favour of using __le16
or __u16 directly as appropriate. All the DECnet header files are
updated accordingly.

The byte ordering of dn_eth2dn() and dn_dn2eth() are both changed
since just about all their callers wanted network order rather than
host order, so the conversion is now done in the functions themselves.

Several missed endianess conversions have been picked up during the
conversion process. The nh_gw field in struct dn_fib_info has been
changed from a 32 bit field to 16 bits as it ought to be.

One or two cases of using htons rather than dn_htons in the routing
code have been found and fixed.

There are still a few warnings to fix, but this patch deals with the
important cases.

Signed-off-by: Steven Whitehouse <steve@chygwyn.com>
Signed-off-by: Patrick Caulfield <patrick@tykepenguin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SECURITY]: TCP/UDP getpeersec
Catherine Zhang [Tue, 21 Mar 2006 06:41:23 +0000 (22:41 -0800)]
[SECURITY]: TCP/UDP getpeersec

This patch implements an application of the LSM-IPSec networking
controls whereby an application can determine the label of the
security association its TCP or UDP sockets are currently connected to
via getsockopt and the auxiliary data mechanism of recvmsg.

Patch purpose:

This patch enables a security-aware application to retrieve the
security context of an IPSec security association a particular TCP or
UDP socket is using.  The application can then use this security
context to determine the security context for processing on behalf of
the peer at the other end of this connection.  In the case of UDP, the
security context is for each individual packet.  An example
application is the inetd daemon, which could be modified to start
daemons running at security contexts dependent on the remote client.

Patch design approach:

- Design for TCP
The patch enables the SELinux LSM to set the peer security context for
a socket based on the security context of the IPSec security
association.  The application may retrieve this context using
getsockopt.  When called, the kernel determines if the socket is a
connected (TCP_ESTABLISHED) TCP socket and, if so, uses the dst_entry
cache on the socket to retrieve the security associations.  If a
security association has a security context, the context string is
returned, as for UNIX domain sockets.

- Design for UDP
Unlike TCP, UDP is connectionless.  This requires a somewhat different
API to retrieve the peer security context.  With TCP, the peer
security context stays the same throughout the connection, thus it can
be retrieved at any time between when the connection is established
and when it is torn down.  With UDP, each read/write can have
different peer and thus the security context might change every time.
As a result the security context retrieval must be done TOGETHER with
the packet retrieval.

The solution is to build upon the existing Unix domain socket API for
retrieving user credentials.  Linux offers the API for obtaining user
credentials via ancillary messages (i.e., out of band/control messages
that are bundled together with a normal message).

Patch implementation details:

- Implementation for TCP
The security context can be retrieved by applications using getsockopt
with the existing SO_PEERSEC flag.  As an example (ignoring error
checking):

getsockopt(sockfd, SOL_SOCKET, SO_PEERSEC, optbuf, &optlen);
printf("Socket peer context is: %s\n", optbuf);

The SELinux function, selinux_socket_getpeersec, is extended to check
for labeled security associations for connected (TCP_ESTABLISHED ==
sk->sk_state) TCP sockets only.  If so, the socket has a dst_cache of
struct dst_entry values that may refer to security associations.  If
these have security associations with security contexts, the security
context is returned.

getsockopt returns a buffer that contains a security context string or
the buffer is unmodified.

- Implementation for UDP
To retrieve the security context, the application first indicates to
the kernel such desire by setting the IP_PASSSEC option via
getsockopt.  Then the application retrieves the security context using
the auxiliary data mechanism.

An example server application for UDP should look like this:

toggle = 1;
toggle_len = sizeof(toggle);

setsockopt(sockfd, SOL_IP, IP_PASSSEC, &toggle, &toggle_len);
recvmsg(sockfd, &msg_hdr, 0);
if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) {
    cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr);
    if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) &&
        cmsg_hdr->cmsg_level == SOL_IP &&
        cmsg_hdr->cmsg_type == SCM_SECURITY) {
        memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext));
    }
}

ip_setsockopt is enhanced with a new socket option IP_PASSSEC to allow
a server socket to receive security context of the peer.  A new
ancillary message type SCM_SECURITY.

When the packet is received we get the security context from the
sec_path pointer which is contained in the sk_buff, and copy it to the
ancillary message space.  An additional LSM hook,
selinux_socket_getpeersec_udp, is defined to retrieve the security
context from the SELinux space.  The existing function,
selinux_socket_getpeersec does not suit our purpose, because the
security context is copied directly to user space, rather than to
kernel space.

Testing:

We have tested the patch by setting up TCP and UDP connections between
applications on two machines using the IPSec policies that result in
labeled security associations being built.  For TCP, we can then
extract the peer security context using getsockopt on either end.  For
UDP, the receiving end can retrieve the security context using the
auxiliary data mechanism of recvmsg.

Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com>
Acked-by: James Morris <jmorris@namei.org>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[XFRM]: Fix aevent related crash
Patrick McHardy [Tue, 21 Mar 2006 06:40:54 +0000 (22:40 -0800)]
[XFRM]: Fix aevent related crash

When xfrm_user isn't loaded xfrm_nl is NULL, which makes IPsec crash because
xfrm_aevent_is_on passes the NULL pointer to netlink_has_listeners as socket.
A second problem is that the xfrm_nl pointer is not cleared when the socket
is releases at module unload time.

Protect references of xfrm_nl from outside of xfrm_user by RCU, check
that the socket is present in xfrm_aevent_is_on and set it to NULL
when unloading xfrm_user.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TCP]: sysctl to allow TCP window > 32767 sans wscale
Rick Jones [Tue, 21 Mar 2006 06:40:29 +0000 (22:40 -0800)]
[TCP]: sysctl to allow TCP window > 32767 sans wscale

Back in the dark ages, we had to be conservative and only allow 15-bit
window fields if the window scale option was not negotiated.  Some
ancient stacks used a signed 16-bit quantity for the window field of
the TCP header and would get confused.

Those days are long gone, so we can use the full 16-bits by default
now.

There is a sysctl added so that we can still interact with such old
stacks

Signed-off-by: Rick Jones <rick.jones2@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV4] ARP: Documentation for new arp_accept sysctl variable.
Neil Horman [Tue, 21 Mar 2006 06:40:03 +0000 (22:40 -0800)]
[IPV4] ARP: Documentation for new arp_accept sysctl variable.

As John pointed out, I had not added documentation to describe the
arp_accpet sysctl that I posted in my last patch.  This patch adds
that documentation.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV4] ARP: Alloc acceptance of unsolicited ARP via netdevice sysctl.
Neil Horman [Tue, 21 Mar 2006 06:39:47 +0000 (22:39 -0800)]
[IPV4] ARP: Alloc acceptance of unsolicited ARP via netdevice sysctl.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: netif_carrier_off runs too early; could still be queued when init fails
Jeff Mahoney [Tue, 21 Mar 2006 06:39:21 +0000 (22:39 -0800)]
[TG3]: netif_carrier_off runs too early; could still be queued when init fails

Move the netif_carrier_off() call from tg3_init_one()->
tg3_init_link_config() to tg3_open() as is the convention for most other
network drivers.

I was getting a panic after a tg3 device failed to initialize due to DMA
failure.  The oops pointed to the link watch queue with spinlock debugging
enabled.  Without spinlock debugging, the Oops didn't occur.

I suspect that the link event was getting queued but not executed until
after the DMA test had failed and the device was freed.  The link event was
then operating on freed memory, which could contain anything.  With this
patch applied, the Oops no longer occurs.

[ Based upon feedback from Michael Chan, we move netif_carrier_off()
  to the end of tg3_init_one() instead of moving it to tg3_open() -DaveM ]

Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TIPC]: Avoid compiler warning
Per Liden [Tue, 21 Mar 2006 06:38:33 +0000 (22:38 -0800)]
[TIPC]: Avoid compiler warning

Signed-off-by: Per Liden <per.liden@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TIPC]: Reduce stack usage
Per Liden [Tue, 21 Mar 2006 06:38:14 +0000 (22:38 -0800)]
[TIPC]: Reduce stack usage

The node_map struct can be quite large (516 bytes) and allocating two of
them on the stack is not a good idea since we might only have a 4K stack
to start with.

Signed-off-by: Per Liden <per.liden@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TIPC]: Cleanups
Adrian Bunk [Tue, 21 Mar 2006 06:37:52 +0000 (22:37 -0800)]
[TIPC]: Cleanups

This patch contains the following possible cleanups:
- make needlessly global code static
- #if 0 the following unused global functions:
  - name_table.c: tipc_nametbl_print()
  - name_table.c: tipc_nametbl_dump()
  - net.c: tipc_net_next_node()

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Per Liden <per.liden@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TIPC]: Remove unused functions
Per Liden [Tue, 21 Mar 2006 06:37:27 +0000 (22:37 -0800)]
[TIPC]: Remove unused functions

Signed-off-by: Per Liden <per.liden@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TIPC]: Remove inlines from *.c
Sam Ravnborg [Tue, 21 Mar 2006 06:37:04 +0000 (22:37 -0800)]
[TIPC]: Remove inlines from *.c

With reference to latest discussions on linux-kernel with respect to
inline here is a patch for tipc to remove all inlines as used in
the .c files. See also chapter 14 in Documentation/CodingStyle.

Before:
   text        data     bss     dec     hex filename
 102990        5292    1752  110034   1add2 tipc.o

Now:
   text        data     bss     dec     hex filename
 101190        5292    1752  108234   1a6ca tipc.o

This is a nice text size reduction which will improve icache usage.
In some cases bigger (> 4 lines) functions where declared inline
and used in many places, they are most probarly no longer inlined by gcc
resulting in the size reduction.
There are several one liners that no longer are declared inline, but gcc
should inline these just fine without the inline hint.

With this patch applied one warning is added about an unused static
function - that was hidded by utilising inline before.
The function in question were kept so this patch is solely a
inline removal patch.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Per Liden <per.liden@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TIPC]: Fix simple sparse warnings
Sam Ravnborg [Tue, 21 Mar 2006 06:36:47 +0000 (22:36 -0800)]
[TIPC]: Fix simple sparse warnings

Tried to run the new tipc stack through sparse.
Following patch fixes all cases where 0 was used
as replacement of NULL.
Use NULL to document this is a pointer and to silence sparse.

This brough sparse warning count down with 127 to 24 warnings.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Per Liden <per.liden@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NETFILTER]: Fix warnings in ip_nat_snmp_basic.c
David S. Miller [Tue, 21 Mar 2006 06:36:21 +0000 (22:36 -0800)]
[NETFILTER]: Fix warnings in ip_nat_snmp_basic.c

net/ipv4/netfilter/ip_nat_snmp_basic.c: In function 'asn1_header_decode':
net/ipv4/netfilter/ip_nat_snmp_basic.c:248: warning: 'len' may be used uninitialized in this function
net/ipv4/netfilter/ip_nat_snmp_basic.c:248: warning: 'def' may be used uninitialized in this function
net/ipv4/netfilter/ip_nat_snmp_basic.c: In function 'snmp_translate':
net/ipv4/netfilter/ip_nat_snmp_basic.c:672: warning: 'l' may be used uninitialized in this function
net/ipv4/netfilter/ip_nat_snmp_basic.c:668: warning: 'type' may be used uninitialized in this function

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP]: Fix uninitialized var warnings in dccp_parse_options().
David S. Miller [Tue, 21 Mar 2006 06:36:01 +0000 (22:36 -0800)]
[DCCP]: Fix uninitialized var warnings in dccp_parse_options().

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: sem2mutex part 2
Ingo Molnar [Tue, 21 Mar 2006 06:35:41 +0000 (22:35 -0800)]
[NET]: sem2mutex part 2

Semaphore to mutex conversion.

The conversion was generated via scripts, and the result was validated
automatically via a script as well.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM] suni: cast arg properly in SONET_SETFRAMING
Alexey Dobriyan [Tue, 21 Mar 2006 06:35:16 +0000 (22:35 -0800)]
[ATM] suni: cast arg properly in SONET_SETFRAMING

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[WAN]: fix section mismatch warning in sbni
Sam Ravnborg [Tue, 21 Mar 2006 06:34:52 +0000 (22:34 -0800)]
[WAN]: fix section mismatch warning in sbni

In latest -mm sbni gives following warning: WARNING:
drivers/net/wan/sbni.o - Section mismatch: reference to \ .init.data:
from .text between 'init_module' (at offset 0x14ef) and \
'cleanup_module'

The warning is caused by init_module() calling a function declared
__init.  Declare init_module() __init too to fix warning.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SUNGEM]: sem2mutex
Ingo Molnar [Tue, 21 Mar 2006 06:34:25 +0000 (22:34 -0800)]
[SUNGEM]: sem2mutex

Semaphore to mutexes conversion.

The conversion was generated via scripts, and the result was validated
automatically via a script as well.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[CASSINI]: sem2mutex
Ingo Molnar [Tue, 21 Mar 2006 06:34:09 +0000 (22:34 -0800)]
[CASSINI]: sem2mutex

Semaphore to mutexes conversion.

The conversion was generated via scripts, and the result was validated
automatically via a script as well.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IRDA]: remove MODULE_PARM()
Andrew Morton [Tue, 21 Mar 2006 06:33:41 +0000 (22:33 -0800)]
[IRDA]: remove MODULE_PARM()

MODULE_PARM() is deprecated and is about to go away altogether.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET] sem2mutex: net/
Arjan van de Ven [Tue, 21 Mar 2006 06:33:17 +0000 (22:33 -0800)]
[NET] sem2mutex: net/

Semaphore to mutex conversion.

The conversion was generated via scripts, and the result was validated
automatically via a script as well.

Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IRDA] sem2mutex: drivers/net/irda
Arjan van de Ven [Tue, 21 Mar 2006 06:32:53 +0000 (22:32 -0800)]
[IRDA] sem2mutex: drivers/net/irda

Semaphore to mutex conversion.

The conversion was generated via scripts, and the result was validated
automatically via a script as well.

Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[NET]: dev_put/dev_hold cleanup
Stephen Hemminger [Tue, 21 Mar 2006 06:32:28 +0000 (22:32 -0800)]
[NET]: dev_put/dev_hold cleanup

Get rid of the old __dev_put macro that is just a hold over from pre 2.6
kernel.  And turn dev_hold into an inline instead of a macro.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP] options: Make dccp_insert_options & friends yell on error
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:32:06 +0000 (22:32 -0800)]
[DCCP] options: Make dccp_insert_options & friends yell on error

And not the silly LIMIT_NETDEBUG and silently return without inserting
the option requested.

Also drop some old debugging messages associated to option insertion.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP]: Remove leftover dccp_send_response prototype
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:31:46 +0000 (22:31 -0800)]
[DCCP]: Remove leftover dccp_send_response prototype

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP]: ditch dccp_v[46]_ctl_send_ack
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:31:26 +0000 (22:31 -0800)]
[DCCP]: ditch dccp_v[46]_ctl_send_ack

Merging it with its only user: dccp_v[46]_reqsk_send_ack.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP]: Use sk->sk_prot->max_header consistently for non-data packets
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:31:09 +0000 (22:31 -0800)]
[DCCP]: Use sk->sk_prot->max_header consistently for non-data packets

Using this also provides opportunities for introducing
inet_csk_alloc_skb that would call alloc_skb, account it to the sock
and skb_reserve(max_header), but I'll leave this for later, for now
using sk_prot->max_header consistently is enough.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[DCCP] options: Fix handling of ackvecs in DATA packets
Arnaldo Carvalho de Melo [Tue, 21 Mar 2006 06:30:51 +0000 (22:30 -0800)]
[DCCP] options: Fix handling of ackvecs in DATA packets

I.e. they should be just ignored, but we have to use 'break', not 'continue',
as we have to possibly reset the mandatory flag.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM]: Fix build after neigh->parms->neigh_destructor change.
David S. Miller [Tue, 21 Mar 2006 06:30:23 +0000 (22:30 -0800)]
[ATM]: Fix build after neigh->parms->neigh_destructor change.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: update version and reldate
Michael Chan [Tue, 21 Mar 2006 06:29:52 +0000 (22:29 -0800)]
[TG3]: update version and reldate

Update version to 3.52.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: Add firmware version info
Michael Chan [Tue, 21 Mar 2006 06:29:32 +0000 (22:29 -0800)]
[TG3]: Add firmware version info

Add fw_version information to ethtool -i.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: nvram cleanup
Michael Chan [Tue, 21 Mar 2006 06:29:15 +0000 (22:29 -0800)]
[TG3]: nvram cleanup

Some nvram related cleanup:

1. Add a tg3_nvram_read_swab() since swabing the data is frequently
done.

2. Add a function to convert nvram address to physical address
instead of doing it in 2 separate places.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: Fixup memory test for 5787
Michael Chan [Tue, 21 Mar 2006 06:28:57 +0000 (22:28 -0800)]
[TG3]: Fixup memory test for 5787

Ethtool memory test on 5787 requires a new memory table.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[TG3]: Add new one-shot MSI handler
Michael Chan [Tue, 21 Mar 2006 06:28:41 +0000 (22:28 -0800)]
[TG3]: Add new one-shot MSI handler

Support one-shot MSI on 5787.

This one-shot MSI idea is credited to David Miller. In this mode, MSI
disables itself automatically after it is generated, saving the driver
a register access to disable it for NAPI.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>