openwrt/staging/blogic.git
17 years ago[NETFILTER]: Remove useless comparisons before assignments
Jan Engelhardt [Wed, 7 Feb 2007 23:06:43 +0000 (15:06 -0800)]
[NETFILTER]: Remove useless comparisons before assignments

Remove unnecessary if() constructs before assignment.

Signed-off-by: Jan Engelhardt <jengelh@gmx.de>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: tcp conntrack: do liberal tracking for picked up connections
Patrick McHardy [Wed, 7 Feb 2007 23:05:33 +0000 (15:05 -0800)]
[NETFILTER]: tcp conntrack: do liberal tracking for picked up connections

Do liberal tracking (only RSTs need to be in-window) for connections picked
up without seeing a SYN to deal with window scaling. Also change logging
of invalid packets not to log packets accepted by liberal tracking to avoid
spamming the logs.

Based on suggestion from James Ralston <ralston@pobox.com>

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETFILTER]: Add SANE connection tracking helper
Michal Schmidt [Wed, 7 Feb 2007 23:05:12 +0000 (15:05 -0800)]
[NETFILTER]: Add SANE connection tracking helper

This is nf_conntrack_sane, a netfilter connection tracking helper module
for the SANE protocol used by the 'saned' daemon to make scanners available
via network. The SANE protocol uses separate control & data connections,
similar to passive FTP. The helper module is needed to recognize the data
connection as RELATED to the control one.

Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IRLAN]: handle out of memory errors
Akinobu Mita [Wed, 7 Feb 2007 08:12:13 +0000 (00:12 -0800)]
[IRLAN]: handle out of memory errors

This patch checks return values:

- irlmp_register_client()
- irlmp_register_service()
- irlan_open()

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Samuel Ortiz <samuel@sortiz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IRDA]: handle out of memory errors
Akinobu Mita [Wed, 7 Feb 2007 08:11:11 +0000 (00:11 -0800)]
[IRDA]: handle out of memory errors

This patch checks return value of memory allocation functions
for irda subsystem and fixes memory leaks in error cases.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Samuel Ortiz <samuel@sortiz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: unregister_netdevice as void
Stephen Hemminger [Wed, 7 Feb 2007 08:09:58 +0000 (00:09 -0800)]
[NET]: unregister_netdevice as void

There was no real useful information from the unregister_netdevice() return
code, the only error occurred in a situation that was a driver bug. So
change it to a void function.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] RAW: Add checksum default defines for MH.
Masahide NAKAMURA [Wed, 7 Feb 2007 08:07:39 +0000 (00:07 -0800)]
[IPV6] RAW: Add checksum default defines for MH.

Add checksum default defines for mobility header(MH) which
goes through raw socket. As the result kernel's behavior is
to handle MH checksum as default.

This patch also removes verifying inbound MH checksum at
mip6_mh_filter() since it did not consider user specified
checksum offset and was redundant check with raw socket code.

Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4/IPV6] multicast: Check add_grhead() return value
Alexey Dobriyan [Tue, 6 Feb 2007 22:35:25 +0000 (14:35 -0800)]
[IPV4/IPV6] multicast: Check add_grhead() return value

add_grhead() allocates memory with GFP_ATOMIC and in at least two places skb
from it passed to skb_put() without checking.

Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[XFRM]: Fix missed error setting in xfrm4_policy.c
David S. Miller [Tue, 6 Feb 2007 22:32:42 +0000 (14:32 -0800)]
[XFRM]: Fix missed error setting in xfrm4_policy.c

When we can't find the afinfo we should return EAFNOSUPPORT.
GCC warned about the uninitialized 'err' for this path as well.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPSEC]: IPv4 over IPv6 IPsec tunnel
Miika Komu [Tue, 6 Feb 2007 22:27:32 +0000 (14:27 -0800)]
[IPSEC]: IPv4 over IPv6 IPsec tunnel

This is the patch to support IPv4 over IPv6 IPsec.

Signed-off-by: Miika Komu <miika@iki.fi>
Signed-off-by: Diego Beltrami <Diego.Beltrami@hiit.fi>
Signed-off-by: Kazunori Miyazawa <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPSEC]: IPv6 over IPv4 IPsec tunnel
Miika Komu [Tue, 6 Feb 2007 22:27:02 +0000 (14:27 -0800)]
[IPSEC]: IPv6 over IPv4 IPsec tunnel

This is the patch to support IPv6 over IPv4 IPsec

Signed-off-by: Miika Komu <miika@iki.fi>
Signed-off-by: Diego Beltrami <Diego.Beltrami@hiit.fi>
Signed-off-by: Kazunori Miyazawa <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPSEC]: exporting xfrm_state_afinfo
Miika Komu [Tue, 6 Feb 2007 22:24:56 +0000 (14:24 -0800)]
[IPSEC]: exporting xfrm_state_afinfo

This patch exports xfrm_state_afinfo.

Signed-off-by: Miika Komu <miika@iki.fi>
Signed-off-by: Diego Beltrami <Diego.Beltrami@hiit.fi>
Signed-off-by: Kazunori Miyazawa <miyazawa@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[BONDING]: Replace kmalloc() + memset() pairs with the appropriate kzalloc() calls
Joe Jin [Tue, 6 Feb 2007 22:16:40 +0000 (14:16 -0800)]
[BONDING]: Replace kmalloc() + memset() pairs with the appropriate kzalloc() calls

Replace kmalloc() + memset() pairs with the appropriate kzalloc() calls in
the bonding driver.

Signed-off-by: Joe Jin <lkmaillist@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[TG3]: Avoid an expensive divide.
Eric Dumazet [Tue, 6 Feb 2007 21:29:21 +0000 (13:29 -0800)]
[TG3]: Avoid an expensive divide.

During an oprofile session of linux-2.6.20 on a dual opteron system, I noticed
an expensive divide was done in tg3_poll().

I am using gcc-4.1.1, so the following comment from drivers/net/tg3.c seems
over-optimistic :

/* Do not place this n-ring entries value into the tp struct itself,
  * we really want to expose these constants to GCC so that modulo et
  * al.  operations are done with shifts and masks instead of with
  * hw multiply/modulo instructions.  Another solution would be to
  * replace things like '% foo' with '& (foo - 1)'.
  */
#define TG3_RX_RCB_RING_SIZE(tp)        \
         ((tp->tg3_flags2 & TG3_FLG2_5705_PLUS) ?  512 : 1024)

Assembly code before patch :
(oprofile results included)
   6434  0.0088 :ffffffff803684b9:       mov    0x6f0(%r15),%eax
    587 8.0e-04 :ffffffff803684c0:       and    $0x40000,%eax
   2170  0.0030 :ffffffff803684c5:       cmp    $0x1,%eax
                :ffffffff803684c8:       lea    0x1(%r13),%eax
                :ffffffff803684cc:       sbb    %ecx,%ecx
   2051  0.0028 :ffffffff803684ce:       xor    %edx,%edx
                :ffffffff803684d0:       and    $0x200,%ecx
     20 2.7e-05 :ffffffff803684d6:       add    $0x200,%ecx
   1986  0.0027 :ffffffff803684dc:       div    %ecx
103427  0.1410 :ffffffff803684de:       cmp    %edx,0xffffffffffffff7c(%rbp)

Assembly code after the suggested patch :

ffffffff803684b9:           mov    0x6f0(%r15),%eax
ffffffff803684c0:           and    $0x40000,%eax
ffffffff803684c5:           cmp    $0x1,%eax
ffffffff803684c8:           sbb    %eax,%eax
ffffffff803684ca:           inc    %r13d
ffffffff803684cd:           and    $0x200,%eax
ffffffff803684d2:           add    $0x1ff,%eax
ffffffff803684d7:           and    %eax,%r13d
ffffffff803684da:           cmp    %r13d,0xffffffffffffff7c(%rbp)

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Acked-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[DCCP]: Warning fixes.
Andrew Morton [Tue, 6 Feb 2007 02:18:21 +0000 (18:18 -0800)]
[DCCP]: Warning fixes.

net/dccp/ccids/ccid3.c: In function `ccid3_hc_rx_packet_recv':
net/dccp/ccids/ccid3.c:1007: warning: long int format, different type arg (arg 3)
net/dccp/ccids/ccid3.c:1007: warning: long int format, different type arg (arg 4)

opaque types must be suitably cast for printing.

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] slip: Replace kmalloc() + memset() pairs with the appropriate kzalloc() calls
Joe Jin [Tue, 6 Feb 2007 02:08:47 +0000 (18:08 -0800)]
[NET] slip: Replace kmalloc() + memset() pairs with the appropriate kzalloc() calls

This patch replace kmalloc() + memset() pairs with the appropriate
kzalloc().

Signed-off-by: Joe Jin <joe.jin@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] net/wanrouter/wanmain.c: cleanups
Adrian Bunk [Tue, 6 Feb 2007 02:07:27 +0000 (18:07 -0800)]
[NET] net/wanrouter/wanmain.c: cleanups

This patch contains the following cleanups:
- make the following needlessly global functions static:
  - lock_adapter_irq()
  - unlock_adapter_irq()
- #if 0 the following unused global functions:
  - wanrouter_encapsulate()
  - wanrouter_type_trans()

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[ATM]: Fix for crash in adummy_init()
Daniel Walker [Tue, 6 Feb 2007 02:04:06 +0000 (18:04 -0800)]
[ATM]: Fix for crash in adummy_init()

This was reported by Ingo Molnar here,

http://lkml.org/lkml/2006/12/18/119

The problem is that adummy_init() depends on atm_init() , but adummy_init()
is called first.

So I put atm_init() into subsys_initcall which seems appropriate, and it
will still get module_init() if it becomes a module.

Interesting to note that you could crash your system here if you just load
the modules in the wrong order.

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: user of the jiffies rounding code: Networking
Arjan van de Ven [Tue, 6 Feb 2007 01:59:51 +0000 (17:59 -0800)]
[NET]: user of the jiffies rounding code: Networking

This patch introduces users of the round_jiffies() function in the
networking code.

These timers all were of the "about once a second" or "about once
every X seconds" variety and several showed up in the "what wakes the
cpu up" profiles that the tickless patches provide.  Some timers are
highly dynamic based on network load; but even on low activity systems
they still show up so the rounding is done only in cases of low
activity, allowing higher frequency timers in the high activity case.

The various hardware watchdogs are an obvious case; they run every 2
seconds but aren't otherwise specific of exactly when they need to
run.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Don't apply FIN exception to full TSO segments.
John Heffner [Tue, 6 Feb 2007 01:53:11 +0000 (17:53 -0800)]
[TCP]: Don't apply FIN exception to full TSO segments.

Signed-off-by: John Heffner <jheffner@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Check num sacks in SACK fast path
Baruch Even [Mon, 5 Feb 2007 07:37:41 +0000 (23:37 -0800)]
[TCP]: Check num sacks in SACK fast path

We clear the unused parts of the SACK cache, This prevents us from mistakenly
taking the cache data if the old data in the SACK cache is the same as the data
in the SACK block. This assumes that we never receive an empty SACK block with
start and end both at zero.

Signed-off-by: Baruch Even <baruch@ev-en.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Seperate DSACK from SACK fast path
Baruch Even [Mon, 5 Feb 2007 07:36:42 +0000 (23:36 -0800)]
[TCP]: Seperate DSACK from SACK fast path

Move DSACK code outside the SACK fast-path checking code. If the DSACK
determined that the information was too old we stayed with a partial cache
copied. Most likely this matters very little since the next packet will not be
DSACK and we will find it in the cache. but it's still not good form and there
is little reason to couple the two checks.

Since the SACK receive cache doesn't need the data to be in host order we also
remove the ntohl in the checking loop.

Signed-off-by: Baruch Even <baruch@ev-en.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Advance fast path pointer for first block only
Baruch Even [Mon, 5 Feb 2007 07:35:57 +0000 (23:35 -0800)]
[TCP]: Advance fast path pointer for first block only

Only advance the SACK fast-path pointer for the first block, the
fast-path assumes that only the first block advances next time so we
should not move the cached skb for the next sack blocks.

Signed-off-by: Baruch Even <baruch@ev-en.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PACKET]: Fix skb->cb clobbering between aux and sockaddr
Herbert Xu [Mon, 5 Feb 2007 07:33:10 +0000 (23:33 -0800)]
[PACKET]: Fix skb->cb clobbering between aux and sockaddr

Both aux data and sockaddr tries to use the same buffer which
obviously doesn't work.  We just happen to have 4 bytes free in
the skb->cb if you take away the maximum length of sockaddr_ll.
That's just enough to store the one piece of info from aux data
that we can't generate at recvmsg(2) time.

This is what the following patch does.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PACKET]: Add optional checksum computation for recvmsg
Herbert Xu [Mon, 5 Feb 2007 07:31:32 +0000 (23:31 -0800)]
[PACKET]: Add optional checksum computation for recvmsg

This patch is needed to make ISC's DHCP server (and probably other
DHCP servers/clients using AF_PACKET) to be able to serve another
client on the same Xen host.

The problem is that packets between different domains on the same
Xen host only have partial checksums.  Unfortunately this piece of
information is not passed along in AF_PACKET unless you're using
the mmap interface.  Since dhcpd doesn't support packet-mmap, UDP
packets from the same host come out with apparently bogus checksums.

This patch adds a mechanism for AF_PACKET recvmsg(2) to return the
status along with the packet.  It does so by adding a new cmsg that
contains this information along with some other relevant data such
as the original packet length.

I didn't include the time stamp information since there is already
a cmsg for that.

This patch also changes the mmap code to set the CSUMNOTREADY flag
on all packets instead of just outoing packets on cooked sockets.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4/IPV6]: Always wait for IPSEC SA resolution in socket contexts.
David S. Miller [Thu, 8 Feb 2007 10:09:21 +0000 (02:09 -0800)]
[IPV4/IPV6]: Always wait for IPSEC SA resolution in socket contexts.

Do this even for non-blocking sockets.  This avoids the silly -EAGAIN
that applications can see now, even for non-blocking sockets in some
cases (f.e. connect()).

With help from Venkat Tekkirala.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: remove tcp header from tcp_v4_check (take #2)
Frederik Deweerdt [Mon, 5 Feb 2007 04:15:27 +0000 (20:15 -0800)]
[TCP]: remove tcp header from tcp_v4_check (take #2)

The tcphdr struct passed to tcp_v4_check is not used, the following
patch removes it from the parameter list.

This adds the netfilter modifications missing in the patch I sent
for rc3-mm1.

Signed-off-by: Frederik Deweerdt <frederik.deweerdt@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] ROUTE: Do not route packets to link-local address on other device.
YOSHIFUJI Hideaki [Mon, 5 Feb 2007 04:15:04 +0000 (20:15 -0800)]
[IPV6] ROUTE: Do not route packets to link-local address on other device.

With help from Wei Dong <weid@np.css.fujitsu.com>.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETLINK]: Don't BUG on undersized allocations
Patrick McHardy [Thu, 1 Feb 2007 07:16:40 +0000 (23:16 -0800)]
[NETLINK]: Don't BUG on undersized allocations

Currently netlink users BUG when the allocated skb for an event
notification is undersized. While this is certainly a kernel bug,
its not critical and crashing the kernel is too drastic, especially
when considering that these errors have appeared multiple times in
the past and it BUGs even if no listeners are present.

This patch replaces BUG by WARN_ON and changes the notification
functions to inform potential listeners of undersized allocations
using a unique error code (EMSGSIZE).

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET_SCHED] sch_prio: class statistics printing enabled
Jarek Poplawski [Wed, 31 Jan 2007 20:21:24 +0000 (12:21 -0800)]
[NET_SCHED] sch_prio: class statistics printing enabled

This patch adds a dump_stats callback to enable
printing of basic statistics of prio classes.
(With help of Patrick McHardy).

Signed-off-by: Jarek Poplawski <jarkao2@o2.pl>
Acked-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/mfasheh...
Linus Torvalds [Thu, 8 Feb 2007 18:37:22 +0000 (10:37 -0800)]
Merge branch 'upstream-linus' of /linux/kernel/git/mfasheh/ocfs2

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/mfasheh/ocfs2: (22 commits)
  configfs: Zero terminate data in configfs attribute writes.
  [PATCH] ocfs2 heartbeat: clean up bio submission code
  ocfs2: introduce sc->sc_send_lock to protect outbound outbound messages
  [PATCH] ocfs2: drop INET from Kconfig, not needed
  ocfs2_dlm: Add timeout to dlm join domain
  ocfs2_dlm: Silence some messages during join domain
  ocfs2_dlm: disallow a domain join if node maps mismatch
  ocfs2_dlm: Ensure correct ordering of set/clear refmap bit on lockres
  ocfs2: Binds listener to the configured ip address
  ocfs2_dlm: Calling post handler function in assert master handler
  ocfs2: Added post handler callable function in o2net message handler
  ocfs2_dlm: Cookies in locks not being printed correctly in error messages
  ocfs2_dlm: Silence a failed convert
  ocfs2_dlm: wake up sleepers on the lockres waitqueue
  ocfs2_dlm: Dlm dispatch was stopping too early
  ocfs2_dlm: Drop inflight refmap even if no locks found on the lockres
  ocfs2_dlm: Flush dlm workqueue before starting to migrate
  ocfs2_dlm: Fix migrate lockres handler queue scanning
  ocfs2_dlm: Make dlmunlock() wait for migration to complete
  ocfs2_dlm: Fixes race between migrate and dirty
  ...

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6
Linus Torvalds [Thu, 8 Feb 2007 18:04:48 +0000 (10:04 -0800)]
Merge /pub/scm/linux/kernel/git/gregkh/driver-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6:
  Revert "Driver core: convert SPI code to use struct device"

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Thu, 8 Feb 2007 18:04:20 +0000 (10:04 -0800)]
Merge /pub/scm/linux/kernel/git/paulus/powerpc

* master.kernel.org:/pub/scm/linux/kernel/git/paulus/powerpc: (116 commits)
  [POWERPC] Add export of vgacon_remap_base
  [POWERPC] Remove bogus comment about page_is_ram
  [POWERPC] windfarm: don't die on suspend thread signal
  [POWERPC] Fix comment in kernel/irq.c
  [POWERPC] ppc: Fix booke watchdog initialization
  [POWERPC] PPC: Use ARRAY_SIZE macro when appropriate
  [POWERPC] Use ARRAY_SIZE macro when appropriate
  [POWERPC] Fix ppc64's writing to struct file_operations
  [POWERPC] ppc: use syslog macro for the printk log level
  [POWERPC] ppc: cs4218_tdm remove extra brace
  [POWERPC] Add mpc52xx/lite5200 PCI support
  [POWERPC] Only use H_BULK_REMOVE if the firmware supports it
  [POWERPC] Fixup error handling when emulating a floating point instruction
  [POWERPC] Enable interrupts if we are doing fp math emulation
  [POWERPC] Added kprobes support to ppc32
  [POWERPC] Make pSeries use the H_BULK_REMOVE hypervisor call
  [POWERPC] Clear RI bit in MSR before restoring r13 when returning to userspace
  [POWERPC] Fix performance monitor exception
  [POWERPC] Compile fixes for arch/powerpc dcr code
  [POWERPC] Maple: use mmio nvram
  ...

17 years agoMerge branch 'drm-patches' of master.kernel.org:/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Thu, 8 Feb 2007 18:03:28 +0000 (10:03 -0800)]
Merge branch 'drm-patches' of /linux/kernel/git/airlied/drm-2.6

* 'drm-patches' of master.kernel.org:/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm: Allow for 44 bit user-tokens (or drm_file offsets)
  drm/via: Disable AGP DMA for chips with the new 3D engine.
  drm: update core memory manager from git drm tree
  drm: remove drm_ioremap and drm_ioremapfree
  i810/i830: use drm_core_ioremap instead of drm_ioremap
  drm: use vmalloc_user instead of vmalloc_32 for DRM_SHM
  via: allow for npot texture pitch alignment
  via: add some new chipsets
  via: some PCI posting flushes

17 years agokbuild: fix space for good (take 103)
Linus Torvalds [Thu, 8 Feb 2007 16:16:44 +0000 (08:16 -0800)]
kbuild: fix space for good (take 103)

Michal Ostrowski points out what the real problem was: the spaces at the
start of the definition of the 'checker-shell' make function.

Cc: Michal Ostrowski <mostrows@watson.ibm.com>
Acked-by: David Miller <davem@davemloft.net>
Acked-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Acked-by: Oleg Verych <olecom@flower.upol.cz>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoRevert "Driver core: convert SPI code to use struct device"
Greg Kroah-Hartman [Thu, 8 Feb 2007 05:34:08 +0000 (21:34 -0800)]
Revert "Driver core: convert SPI code to use struct device"

This reverts commit 2943ecf2ed32632473c06f1975db47a7aa98c10f.

This should go through the SPI maintainer, it was my fault that it did
not.  Especially as it conflicts with other patches he has pending.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agodrm: Allow for 44 bit user-tokens (or drm_file offsets)
Thomas Hellstrom [Thu, 8 Feb 2007 05:14:05 +0000 (16:14 +1100)]
drm: Allow for 44 bit user-tokens (or drm_file offsets)

17 years ago[POWERPC] Add export of vgacon_remap_base
Mathieu Desnoyers [Thu, 8 Feb 2007 00:04:44 +0000 (19:04 -0500)]
[POWERPC] Add export of vgacon_remap_base

The following macro :
include/asm-powerpc/vga.h:#define VGA_MAP_MEM(x,s) (x + vgacon_remap_base)
is used by drivers/video/console/vgacon.c which can be compiled as a module
(drivers/video/vga16fb.ko).

Therefore, vgacon_remap_base should be exported.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Remove bogus comment about page_is_ram
Johannes Berg [Wed, 7 Feb 2007 12:45:45 +0000 (13:45 +0100)]
[POWERPC] Remove bogus comment about page_is_ram

arch/powerpc/mm/mem.c states that page_is_ram is called by the code that
implements /dev/mem, which isn't true.  Remove the comment.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] windfarm: don't die on suspend thread signal
Johannes Berg [Mon, 5 Feb 2007 18:30:29 +0000 (19:30 +0100)]
[POWERPC] windfarm: don't die on suspend thread signal

When the windfarm thread gets a suspend signal it will die instead of
freezing. This fixes it.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix comment in kernel/irq.c
Johannes Berg [Wed, 13 Dec 2006 12:38:22 +0000 (13:38 +0100)]
[POWERPC] Fix comment in kernel/irq.c

kernel/irq.c contains a comment that speaks of -1 and -2 as interrupt
numbers, but this is actually dependent on configuration options now.
Replace by NO_IRQ and NO_IRQ_ENABLED.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] ppc: Fix booke watchdog initialization
Stefan Roese [Wed, 7 Feb 2007 08:45:55 +0000 (09:45 +0100)]
[POWERPC] ppc: Fix booke watchdog initialization

Fix two problems in the book-e watchdog driver.

a) The 4xx default period was defined wrong
b) Clear status before enabling the watchdog exception

Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] PPC: Use ARRAY_SIZE macro when appropriate
Ahmed S. Darwish [Tue, 6 Feb 2007 00:14:10 +0000 (16:14 -0800)]
[POWERPC] PPC: Use ARRAY_SIZE macro when appropriate

Use ARRAY_SIZE macro already defined in linux/kernel.h

Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Kumar Gala <galak@gate.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Use ARRAY_SIZE macro when appropriate
Ahmed S. Darwish [Tue, 6 Feb 2007 00:14:09 +0000 (16:14 -0800)]
[POWERPC] Use ARRAY_SIZE macro when appropriate

Use ARRAY_SIZE macro already defined in linux/kernel.h

Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Fix ppc64's writing to struct file_operations
Stephen Rothwell [Tue, 6 Feb 2007 00:14:05 +0000 (16:14 -0800)]
[POWERPC] Fix ppc64's writing to struct file_operations

In preparation for marking file_operations as const.

Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] ppc: use syslog macro for the printk log level
Robert P. J. Day [Tue, 6 Feb 2007 00:14:04 +0000 (16:14 -0800)]
[POWERPC] ppc: use syslog macro for the printk log level

Use the appropriate logging macro for the priority level for that
printk call.

Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] ppc: cs4218_tdm remove extra brace
Mariusz Kozlowski [Tue, 6 Feb 2007 00:14:03 +0000 (16:14 -0800)]
[POWERPC] ppc: cs4218_tdm remove extra brace

Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years ago[POWERPC] Add mpc52xx/lite5200 PCI support
Grant Likely [Tue, 12 Dec 2006 22:13:19 +0000 (15:13 -0700)]
[POWERPC] Add mpc52xx/lite5200 PCI support

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Acked-by: Sylvain Munaut <tnt@246tNt.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years agokbuild: make $(checker-shell ) strip spaces around the result
Linus Torvalds [Thu, 8 Feb 2007 04:24:25 +0000 (20:24 -0800)]
kbuild: make $(checker-shell ) strip spaces around the result

It looks like GNU make version 3.80 (but apparently not 3.81) adds
leading whitespace to the result of the checker-shell execution.  This
strips them off explicitly.

Also, don't bother symlinking the output file to /dev/null.  It's likely
as expensive as just writing the temp-file, and we need to remove it
anyway afterwards.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoMerge branch 'for_paulus' of master.kernel.org:/pub/scm/linux/kernel/git/galak/powerp...
Paul Mackerras [Thu, 8 Feb 2007 04:03:11 +0000 (15:03 +1100)]
Merge branch 'for_paulus' of /linux/kernel/git/galak/powerpc into for-2.6.21

17 years ago[POWERPC] Only use H_BULK_REMOVE if the firmware supports it
Paul Mackerras [Thu, 8 Feb 2007 04:02:35 +0000 (15:02 +1100)]
[POWERPC] Only use H_BULK_REMOVE if the firmware supports it

The previous patch changing pSeries to use H_BULK_REMOVE broke the
JS20 blade, where the firmware doesn't support H_BULK_REMOVE.  This
adds a firmware check so that on machines that don't have H_BULK_REMOVE,
we just use the H_REMOVE call as before.

Signed-off-by: Paul Mackerras <paulus@samba.org>
17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Thu, 8 Feb 2007 03:32:36 +0000 (19:32 -0800)]
Merge /pub/scm/linux/kernel/git/bart/ide-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/bart/ide-2.6: (23 commits)
  ide-acpi support warning fix
  ACPI support for IDE devices
  IDE Driver for Delkin/Lexar/etc.. cardbus CF adapter
  ide: it8213 IDE driver update (version 2)
  ide: add it8213 IDE driver
  tc86c001: add missing __init tag for tc86c001_ide_init()
  tc86c001: mark init_chipset_tc86c001() with __devinit tag
  tc86c001: init_hwif_tc86c001() can be static
  ide: add Toshiba TC86C001 IDE driver (take 2)
  pdc202xx_new: remove check_in_drive_lists abomination
  pdc202xx_new: remove useless code
  slc90e66: carry over fixes from piix driver
  piix: tuneproc() fixes/cleanups
  piix: fix 82371MX enablebits
  hpt366: HPT36x PCI clock detection fix
  hpt366: init code rewrite
  hpt366: clean up DMA timeout handling for HPT370
  hpt366: merge HPT37x speedproc handlers
  hpt366: cache channel's MCR address
  hpt366: switch to using pci_get_slot
  ...

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/pci-2.6
Linus Torvalds [Thu, 8 Feb 2007 03:23:44 +0000 (19:23 -0800)]
Merge /pub/scm/linux/kernel/git/gregkh/pci-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/pci-2.6: (41 commits)
  Revert "PCI: remove duplicate device id from ata_piix"
  msi: Make MSI useable more architectures
  msi: Kill the msi_desc array.
  msi: Remove attach_msi_entry.
  msi: Fix msi_remove_pci_irq_vectors.
  msi: Remove msi_lock.
  msi: Kill msi_lookup_irq
  MSI: Combine pci_(save|restore)_msi/msix_state
  MSI: Remove pci_scan_msi_device()
  MSI: Replace pci_msi_quirk with calls to pci_no_msi()
  PCI: remove duplicate device id from ipr
  PCI: remove duplicate device id from ata_piix
  PCI: power management: remove noise on non-manageable hw
  PCI: cleanup MSI code
  PCI: make isa_bridge Alpha-only
  PCI: remove quirk_sis_96x_compatible()
  PCI: Speed up the Intel SMBus unhiding quirk
  PCI Quirk: 1k I/O space IOBL_ADR fix on P64H2
  shpchp: delete trailing whitespace
  shpchp: remove DBG_XXX_ROUTINE
  ...

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Thu, 8 Feb 2007 03:23:21 +0000 (19:23 -0800)]
Merge /pub/scm/linux/kernel/git/gregkh/usb-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/usb-2.6: (70 commits)
  USB: remove duplicate device id from zc0301
  USB: remove duplicate device id from usb_storage
  USB: remove duplicate device id from keyspan
  USB: remove duplicate device id from ftdi_sio
  USB: remove duplicate device id from visor
  USB: a bit more coding style cleanup
  usbcore: trivial whitespace fixes
  usb-storage: use first bulk endpoints, not last
  EHCI: fix interrupt-driven remote wakeup
  USB: switch ehci-hcd to new polling scheme
  USB: autosuspend for usb printer driver
  USB Input: Added kernel module to support all GTCO CalComp USB InterWrite School products
  USB: Sierra Wireless auto set D0
  USB: usb ethernet gadget recognizes HUSB2DEV
  USB: list atmel husb2_udc gadget controller
  USB: gadgetfs AIO tweaks
  USB: gadgetfs behaves better on userspace init bug
  USB: gadgetfs race fix
  USB: gadgetfs simplifications
  USB: gadgetfs cleanups
  ...

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6
Linus Torvalds [Thu, 8 Feb 2007 03:22:26 +0000 (19:22 -0800)]
Merge /pub/scm/linux/kernel/git/gregkh/driver-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6: (28 commits)
  sysfs: Shadow directory support
  Driver Core: Increase the default timeout value of the firmware subsystem
  Driver core: allow to delay the uevent at device creation time
  Driver core: add device_type to struct device
  Driver core: add uevent vars for devices of a class
  SYSFS: Fix missing include of list.h in sysfs.h
  HOWTO: Add a reference to Harbison and Steele
  sysfs: error handling in sysfs, fill_read_buffer()
  kobject: kobject_put cleanup
  sysfs: kobject_put cleanup
  sysfs: suppress lockdep warnings
  Driver core: fix race in sysfs between sysfs_remove_file() and read()/write()
  driver core: Change function call order in device_bind_driver().
  driver core: Don't stop probing on ->probe errors.
  driver core fixes: device_register() retval check in platform.c
  driver core fixes: make_class_name() retval checks
  /sys/modules/*/holders
  USB: add the sysfs driver name to all modules
  SERIO: add the sysfs driver name to all modules
  PCI: add the sysfs driver name to all modules
  ...

17 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Thu, 8 Feb 2007 03:21:56 +0000 (19:21 -0800)]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

* 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6: (116 commits)
  sk98lin: planned removal
  AT91: MACB support
  sky2: version 1.12
  sky2: add new chip ids
  sky2: Yukon Extreme support
  sky2: safer transmit timeout
  sky2: TSO support for EC_U
  sky2: use dev_err for error reports
  sky2: add Wake On Lan support
  fix unaligned exception in /drivers/net/wireless/orinoco.c
  Remove unused kernel config option DLCI_COUNT
  z85230: spinlock logic
  mips: declance: Driver model for the PMAD-A
  Spidernet: Rework RX linked list
  NET: turn local_save_flags() + local_irq_disable() into local_irq_save()
  NET-3c59x: turn local_save_flags() + local_irq_disable() into local_irq_save()
  hp100: convert pci_module_init() to pci_register_driver()
  NetXen: Added ethtool support for user level tools.
  NetXen: Firmware crb init changes.
  maintainers: add atl1 maintainers
  ...

17 years agodrm/via: Disable AGP DMA for chips with the new 3D engine.
Thomas Hellstrom [Thu, 8 Feb 2007 01:57:40 +0000 (12:57 +1100)]
drm/via: Disable AGP DMA for chips with the new 3D engine.

17 years agodrm: update core memory manager from git drm tree
Thomas Hellstrom [Mon, 8 Jan 2007 11:25:47 +0000 (22:25 +1100)]
drm: update core memory manager from git drm tree

Remove the memory manager parameter from the put_block function, as this
makes the client code a lot cleaner. Prepare buffer manager for lock and
unlock calls.
Fix buggy aligned allocations.
Remove the stupid root_node field from the core memory manager.
Support multi-page buffer offset alignments
Add improved alignment functionality to the core memory manager.
This makes an allocated block actually align itself and returns any
wasted space to the manager.

Signed-off-by: Dave Airlie <airlied@linux.ie>
17 years agodrm: remove drm_ioremap and drm_ioremapfree
Christoph Hellwig [Mon, 8 Jan 2007 10:56:59 +0000 (21:56 +1100)]
drm: remove drm_ioremap and drm_ioremapfree

hch originally submitted this for paravirt ops work, airlied took it
and cleaned up a lot of unused code caused by using this.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Airlie <airlied@linux.ie>
17 years agoi810/i830: use drm_core_ioremap instead of drm_ioremap
Dave Airlie [Mon, 8 Jan 2007 10:31:13 +0000 (21:31 +1100)]
i810/i830: use drm_core_ioremap instead of drm_ioremap

This makes the i810/i830 use the drm_core_ioremap functions.

Signed-off-by: Dave Airlie <airlied@linux.ie>
17 years agodrm: use vmalloc_user instead of vmalloc_32 for DRM_SHM
Thomas Hellstrom [Mon, 8 Jan 2007 10:22:50 +0000 (21:22 +1100)]
drm: use vmalloc_user instead of vmalloc_32 for DRM_SHM

Signed-off-by: Dave Airlie <airlied@linux.ie>
17 years agovia: allow for npot texture pitch alignment
Thomas Hellstrom [Mon, 8 Jan 2007 10:21:41 +0000 (21:21 +1100)]
via: allow for npot texture pitch alignment

Signed-off-by: Dave Airlie <airlied@linux.ie>
17 years agovia: add some new chipsets
Thomas Hellstrom [Mon, 8 Jan 2007 10:19:57 +0000 (21:19 +1100)]
via: add some new chipsets

Disable 3D functionality and AGP DMA for chipsets with the DX9 3D engine.

Signed-off-by: Dave Airlie <airlied@linux.ie>
17 years agovia: some PCI posting flushes
Thomas Hellstrom [Mon, 8 Jan 2007 10:03:23 +0000 (21:03 +1100)]
via: some PCI posting flushes

Signed-off-by: Dave Airlie <airlied@linux.ie>
17 years agoRevert "PCI: remove duplicate device id from ata_piix"
Greg Kroah-Hartman [Thu, 8 Feb 2007 00:17:27 +0000 (16:17 -0800)]
Revert "PCI: remove duplicate device id from ata_piix"

This reverts commit b11056355ea149c37edf0ef54976a49f5258cd54.

It was incorrect, the proper fix is coming through the SATA tree, sorry
about that.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agosk98lin: planned removal
Stephen Hemminger [Wed, 7 Feb 2007 17:18:30 +0000 (09:18 -0800)]
sk98lin: planned removal

Document planned removal of sk98lin driver.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agoAT91: MACB support
Andrew Victor [Wed, 7 Feb 2007 15:40:44 +0000 (16:40 +0100)]
AT91: MACB support

The Atmel MACB Ethernet peripheral is also integrated in the AT91SAM9260
and AT91SAM9263 processors.  The differences from the AVR32 version are:
      * Single peripheral clock.
      * MII/RMII selection bit is inverted.
      * Clock enable bit.

Original patch from Patrice Vilchez.

Signed-off-by: Andrew Victor <andrew@sanpeople.com>
Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosky2: version 1.12
Stephen Hemminger [Tue, 6 Feb 2007 18:45:45 +0000 (10:45 -0800)]
sky2: version 1.12

Updated version for WOL and new id's

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosky2: add new chip ids
Stephen Hemminger [Tue, 6 Feb 2007 18:45:44 +0000 (10:45 -0800)]
sky2: add new chip ids

More new chip id's from vendor driver version 10.0.4.3

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosky2: Yukon Extreme support
Stephen Hemminger [Tue, 6 Feb 2007 18:45:43 +0000 (10:45 -0800)]
sky2: Yukon Extreme support

This is basic support for the new Yukon Extreme
chip, extracted from the new vendor driver 10.0.4.3.

Since this is untested hardware, it has a big fat warning for now.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosky2: safer transmit timeout
Stephen Hemminger [Tue, 6 Feb 2007 18:45:42 +0000 (10:45 -0800)]
sky2: safer transmit timeout

Rather than trying to be "smart" about possible transmit timeout
causes. Just clear all pending frames and reset the PHY.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosky2: TSO support for EC_U
Stephen Hemminger [Tue, 6 Feb 2007 18:45:41 +0000 (10:45 -0800)]
sky2: TSO support for EC_U

The Yukon EC_U chipset apparently supports TSO but only for non-Jumbo
frame sizes because it lacks a Ram buffer.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agosky2: use dev_err for error reports
Stephen Hemminger [Tue, 6 Feb 2007 18:45:40 +0000 (10:45 -0800)]
sky2: use dev_err for error reports

Use the standard dev_xxx functions instead of printk directly for
error reports. Fix a bug where the initialization would return 0
if allocation of network device failed.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>
17 years agomsi: Make MSI useable more architectures
Eric W. Biederman [Sun, 28 Jan 2007 19:56:37 +0000 (12:56 -0700)]
msi: Make MSI useable more architectures

The arch hooks arch_setup_msi_irq and arch_teardown_msi_irq are now
responsible for allocating and freeing the linux irq in addition to
setting up the the linux irq to work with the interrupt.

arch_setup_msi_irq now takes a pci_device and a msi_desc and returns
an irq.

With this change in place this code should be useable by all platforms
except those that won't let the OS touch the hardware like ppc RTAS.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agomsi: Kill the msi_desc array.
Eric W. Biederman [Sun, 28 Jan 2007 19:52:03 +0000 (12:52 -0700)]
msi: Kill the msi_desc array.

We need to be able to get from an irq number to a struct msi_desc.
The msi_desc array in msi.c had several short comings the big one was
that it could not be used outside of msi.c.  Using irq_data in struct
irq_desc almost worked except on some architectures irq_data needs to
be used for something else.

So this patch adds a msi_desc pointer to irq_desc, adds the appropriate
wrappers and changes all of the msi code to use them.

The dynamic_irq_init/cleanup code was tweaked to ensure the new
field is left in a well defined state.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agomsi: Remove attach_msi_entry.
Eric W. Biederman [Sun, 28 Jan 2007 19:47:52 +0000 (12:47 -0700)]
msi: Remove attach_msi_entry.

The attach_msi_entry has been reduced to a single simple assignment,
so for simplicity remove the abstraction and directory perform the
assignment.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agomsi: Fix msi_remove_pci_irq_vectors.
Eric W. Biederman [Sun, 28 Jan 2007 19:45:54 +0000 (12:45 -0700)]
msi: Fix msi_remove_pci_irq_vectors.

Since msi_remove_pci_irq_vectors is designed to be called during
hotplug remove it is actively wrong to query the hardware and expect
meaningful results back.

To that end remove the pci_find_capability calls.  Testing
dev->msi_enabled and dev->msix_enabled gives us all of the information
we need.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agomsi: Remove msi_lock.
Eric W. Biederman [Sun, 28 Jan 2007 19:44:21 +0000 (12:44 -0700)]
msi: Remove msi_lock.

With the removal of msi_lookup_irq all of the functions using msi_lock
operated on a single device and none of them could reasonably be
called on that device at the same time.

Since what little synchronization that needs to happen needs to happen
outside of the msi functions, msi_lock could never be contended and as
such is useless and just complicates the code.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agomsi: Kill msi_lookup_irq
Eric W. Biederman [Sun, 28 Jan 2007 19:42:52 +0000 (12:42 -0700)]
msi: Kill msi_lookup_irq

The function msi_lookup_irq was horrible.  As a side effect of running
it changed dev->irq, and then the callers would need to change it
back.  In addition it does a global scan through all of the irqs,
which seems to be the sole justification of the msi_lock.

To remove the neede for msi_lookup_irq I added first_msi_irq to struct
pci_dev.  Then depending on the context I replaced msi_lookup_irq with
dev->first_msi_irq, dev->msi_enabled, or dev->msix_enabled.

msi_enabled and msix_enabled were already present in pci_dev for other
reasons.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoMSI: Combine pci_(save|restore)_msi/msix_state
Michael Ellerman [Thu, 25 Jan 2007 08:34:08 +0000 (19:34 +1100)]
MSI: Combine pci_(save|restore)_msi/msix_state

The PCI save/restore code doesn't need to care about MSI vs MSI-X, all
it really wants is to say "save/restore all MSI(-X) info for this device".

This is borne out in the code, we call the MSI and MSI-X save routines
side by side, and similarly with the restore routines.

So combine the MSI/MSI-X routines into pci_save_msi_state() and
pci_restore_msi_state(). It is up to those routines to decide what state
needs to be saved.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoMSI: Remove pci_scan_msi_device()
Michael Ellerman [Thu, 25 Jan 2007 08:34:08 +0000 (19:34 +1100)]
MSI: Remove pci_scan_msi_device()

pci_scan_msi_device() doesn't do anything anymore, so remove it.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoMSI: Replace pci_msi_quirk with calls to pci_no_msi()
Michael Ellerman [Thu, 25 Jan 2007 08:34:07 +0000 (19:34 +1100)]
MSI: Replace pci_msi_quirk with calls to pci_no_msi()

I don't see any reason why we need pci_msi_quirk, quirk code can just
call pci_no_msi() instead.

Remove the check of pci_msi_quirk in msi_init(). This is safe as all
calls to msi_init() are protected by calls to pci_msi_supported(),
which checks pci_msi_enable, which is disabled by pci_no_msi().

The pci_disable_msi routines didn't check pci_msi_quirk, only
pci_msi_enable, but as far as I can see that was a bug not a feature.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI: remove duplicate device id from ipr
Greg Kroah-Hartman [Fri, 26 Jan 2007 00:15:24 +0000 (16:15 -0800)]
PCI: remove duplicate device id from ipr

As pointed out by Kay Sievers <kay.sievers@suse.de>

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI: remove duplicate device id from ata_piix
Greg Kroah-Hartman [Fri, 26 Jan 2007 00:15:24 +0000 (16:15 -0800)]
PCI: remove duplicate device id from ata_piix

As pointed out by Kay Sievers <kay.sievers@suse.de>

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI: power management: remove noise on non-manageable hw
Pavel Machek [Tue, 16 Jan 2007 11:17:13 +0000 (12:17 +0100)]
PCI: power management: remove noise on non-manageable hw

Return early from pci_set_power_state() if hardware does not support
power management. This way, we do not generate noise in the logs.

Signed-off-by: Pavel Machek <pavel@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI: cleanup MSI code
Satoru Takeuchi [Thu, 18 Jan 2007 04:50:05 +0000 (13:50 +0900)]
PCI: cleanup MSI code

Cleanup MSI code as follows:

 - fix some types
 - fix strange local variable definition
 - delete unnecessary blank line
 - add comment to #endif which is far from corresponding #ifdef

Signed-off-by: Satoru Takeuchi <takeuchi_satoru@jp.fujitsu.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI: make isa_bridge Alpha-only
Adrian Bunk [Sat, 6 Jan 2007 20:48:41 +0000 (21:48 +0100)]
PCI: make isa_bridge Alpha-only

Since isa_bridge is neither assigned any value !NULL nor used on !Alpha,
there's no reason for providing it.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI: remove quirk_sis_96x_compatible()
Adrian Bunk [Sun, 14 Jan 2007 13:46:32 +0000 (14:46 +0100)]
PCI: remove quirk_sis_96x_compatible()

Since 2.6.0-test10, all quirk_sis_96x_compatible() had any effect on
was a printk().

This patch therefore removes it.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Mark M. Hoffman <mhoffman@lightlink.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI: Speed up the Intel SMBus unhiding quirk
Jean Delvare [Fri, 5 Jan 2007 10:23:15 +0000 (11:23 +0100)]
PCI: Speed up the Intel SMBus unhiding quirk

Speed up the Intel SMBus PCI quirk by avoiding tests which can only
fail. This also makes the compiled code significantly smaller when
using gcc 3.2/3.4. gcc 4.x appears to optimize the code by itself so
this change doesn't make a difference there.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCI Quirk: 1k I/O space IOBL_ADR fix on P64H2
Daniel Yeisley [Thu, 21 Dec 2006 19:34:57 +0000 (14:34 -0500)]
PCI Quirk: 1k I/O space IOBL_ADR fix on P64H2

There's an existing quirk for the kernel to use 1k IO space granularity
on the Intel P64H2.  It turns out however that pci_setup_bridge() in
drivers/pci/setup-bus.c reads in the IO base and limit address register
masks it off to the nearest 4k, and writes it back.  This causes the
kernel to be on 1k boundaries and the hardware to be 4k aligned.  The
patch below fixes the problem.

Signed-off-by: Dan Yeisley <dan.yeisley@unisys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoshpchp: delete trailing whitespace
Kenji Kaneshige [Tue, 9 Jan 2007 21:03:10 +0000 (13:03 -0800)]
shpchp: delete trailing whitespace

This patch deletes trailing white space in SHPCHP driver. This has no
functional change.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoshpchp: remove DBG_XXX_ROUTINE
Kenji Kaneshige [Tue, 9 Jan 2007 21:03:01 +0000 (13:03 -0800)]
shpchp: remove DBG_XXX_ROUTINE

This patch removes DBG_ENTER_ROUTINE, DBG_LEAVE_ROUTINE and related
code.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoshpchp: remove CONFIG_HOTPLUG_PCI_SHPC_POLL_EVENT_MODE
Kenji Kaneshige [Tue, 9 Jan 2007 21:02:48 +0000 (13:02 -0800)]
shpchp: remove CONFIG_HOTPLUG_PCI_SHPC_POLL_EVENT_MODE

The CONFIG_HOTPLUG_PCI_SHPC_POLL_EVENT_MODE config option is not
needed because polling mechanism for shpc hotplug events can be
enabled through module option 'shpchp_poll_mode'. This patch removes
CONFIG_HOTPLUG_PCI_SHPC_POLL_EVENT_MODE.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agoPCIEHP: Add Electro Mechanical Interlock (EMI) support to the PCIE hotplug driver.
Kristen Carlson Accardi [Tue, 9 Jan 2007 21:02:36 +0000 (13:02 -0800)]
PCIEHP: Add Electro Mechanical Interlock (EMI) support to the PCIE hotplug driver.

Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agopciehp: fix wait command completion
Kenji Kaneshige [Fri, 22 Dec 2006 01:01:10 +0000 (17:01 -0800)]
pciehp: fix wait command completion

This patch fixes this problem that pciehp driver will sleep
unnecessarily long when waiting for command completion. With this
patch, modprobe pciehp driver becomes very faster as follows for
instance.

  o Without this patch
    # time /sbin/modprobe pciehp

    real    0m4.976s
    user    0m0.000s
    sys     0m0.004s

  o With this patch
    # time /sbin/modprobe pciehp

    real    0m0.640s
    user    0m0.000s
    sys     0m0.004s

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agopciehp: cleanup wait command completion
Kenji Kaneshige [Fri, 22 Dec 2006 01:01:09 +0000 (17:01 -0800)]
pciehp: cleanup wait command completion

This patch cleans up the code to wait for command completion.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agopciehp: remove unused pcie_cap_base
Kenji Kaneshige [Fri, 22 Dec 2006 01:01:08 +0000 (17:01 -0800)]
pciehp: remove unused pcie_cap_base

This patch removes unused pcie_cap_base variable.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agopciehp: cleanup pciehp.h
Kenji Kaneshige [Fri, 22 Dec 2006 01:01:07 +0000 (17:01 -0800)]
pciehp: cleanup pciehp.h

This patch cleans up pciehp.h. This has no functional change.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agopciehp: cleanup register access
Kenji Kaneshige [Fri, 22 Dec 2006 01:01:06 +0000 (17:01 -0800)]
pciehp: cleanup register access

This patch cleans up register access functions. This has no functional
change.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
17 years agopciehp: remove unused pci_bus from struct controller
Kenji Kaneshige [Fri, 22 Dec 2006 01:01:05 +0000 (17:01 -0800)]
pciehp: remove unused pci_bus from struct controller

This patch removes unused pci_bus member from struct controller.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>