openwrt/staging/blogic.git
19 years ago[NETFILTER] {ip,nf}_conntrack TCP: Accept SYN+PUSH like SYN
Vlad Drukker [Sat, 12 Nov 2005 20:13:14 +0000 (12:13 -0800)]
[NETFILTER] {ip,nf}_conntrack TCP: Accept SYN+PUSH like SYN

Some devices (e.g. Qlogic iSCSI HBA hardware like QLA4010 up to firmware
3.0.0.4) initiates TCP with SYN and PUSH flags set.

The Linux TCP/IP stack deals fine with that, but the connection tracking
code doesn't.

This patch alters TCP connection tracking to accept SYN+PUSH as a valid
flag combination.

Signed-off-by: Vlad Drukker <vlad@storewiz.com>
Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[IPV6]: Fix rtnetlink dump infinite loop
Herbert Xu [Sat, 12 Nov 2005 20:12:05 +0000 (12:12 -0800)]
[IPV6]: Fix rtnetlink dump infinite loop

The recent change to netlink dump "done" callback handling broke IPv6
which played dirty tricks with the "done" callback.  This causes an
infinite loop during a dump.

The following patch fixes it.

This bug was reported by Jeff Garzik.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[PATCH] I8K: fix /proc reporting of blank service tags
Dmitry Torokhov [Sat, 12 Nov 2005 05:55:15 +0000 (00:55 -0500)]
[PATCH] I8K: fix /proc reporting of blank service tags

Make /proc/i8k display '?' when service tag is blank in BIOS.
This fixes segfault in i8k gkrellm plugin.

Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] fix b2c2 dvb undefined symbol
Prakash Punnoor [Sat, 12 Nov 2005 08:17:38 +0000 (09:17 +0100)]
[PATCH] fix b2c2 dvb undefined symbol

This fixes

  drivers/built-in.o: In function `flexcop_frontend_init':
  : undefined reference to `lgdt330x_attach'

[ Side note: I really dislike that dvb people want to include every
  possible frontend into the kernel - I only need the mt312 one for my
  Skystar2 card.  I'd highly appreciate it this would be made selectable
  again... ]

Signed-off-by: Prakash Punnoor <prakash@punnoor.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge branch 'upstream-fixes' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Sat, 12 Nov 2005 18:39:03 +0000 (10:39 -0800)]
Merge branch 'upstream-fixes' of /linux/kernel/git/jgarzik/libata-dev

19 years agoRevert "[PATCH] fbcon: Add rl (Roman Large) font"
Linus Torvalds [Sat, 12 Nov 2005 18:14:02 +0000 (10:14 -0800)]
Revert "[PATCH] fbcon: Add rl (Roman Large) font"

This reverts 998e6d51162707685336ff99c029c8911b270d32 commit.

19 years ago[BLOCK] elevator: elv_latter/former_request update
Tejun Heo [Thu, 10 Nov 2005 17:22:36 +0000 (18:22 +0100)]
[BLOCK] elevator: elv_latter/former_request update

With generic dispatch queue update, implicit former/latter request
handling using rq->queuelist.prev/next doesn't work as expected
anymore.  Also, the only iosched dependent on this feature was
noop-iosched and it has been reimplemented to have its own
latter/former methods.  This patch removes implicit former/latter
handling.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] noop-iosched: reimplementation of request dispatching
Tejun Heo [Thu, 10 Nov 2005 17:21:30 +0000 (18:21 +0100)]
[BLOCK] noop-iosched: reimplementation of request dispatching

The original implementation directly used dispatch queue.  As new
generic dispatch queue imposes stricter rules over ioscheds and
dispatch queue usage, this direct use becomes somewhat problematic.
This patch reimplements noop-iosched such that it complies to generic
iosched model better.  Request merging with q->last_merge and
rq->queuelist.prev/next work again now.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de
19 years ago[BLOCK] cfq-iosched: fix slice_left calculation
Tejun Heo [Thu, 10 Nov 2005 17:20:16 +0000 (18:20 +0100)]
[BLOCK] cfq-iosched: fix slice_left calculation

When cfq slice expires, remainder of slice is calculated and stored in
cfqq->slice_left.  Current code calculates the opposite of remainder -
how many jiffies the cfqq has used past slice end.  This patch fixes
the bug.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] fix string handling in elv_iosched_store
Tejun Heo [Thu, 10 Nov 2005 07:55:01 +0000 (08:55 +0100)]
[BLOCK] fix string handling in elv_iosched_store

elv_iosched_store doesn't terminate string passed from userspace if
it's too long.  Also, if the written length is zero (probably not
possible), it accesses elevator_name[-1].  This patch fixes both bugs.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] Implement elv_drain_elevator for improved switch error detection
Tejun Heo [Thu, 10 Nov 2005 07:52:05 +0000 (08:52 +0100)]
[BLOCK] Implement elv_drain_elevator for improved switch error detection

This patch adds request_queue->nr_sorted which keeps the number of
requests in the iosched and implement elv_drain_elevator which
performs forced dispatching.  elv_drain_elevator checks whether
iosched actually dispatches all requests it has and prints error
message if it doesn't.  As buggy forced dispatching can result in
wrong barrier operations, I think this extra check is worthwhile.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] cfq-iosched: cfq forced dispatching fix
Tejun Heo [Thu, 10 Nov 2005 07:49:19 +0000 (08:49 +0100)]
[BLOCK] cfq-iosched: cfq forced dispatching fix

cfq forced dispatching might not return all requests on the queue.
This bug can hang elevator switchinig and corrupt request ordering
during flush sequence.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] elevator: run queue in elevator_switch
Tejun Heo [Thu, 10 Nov 2005 07:48:21 +0000 (08:48 +0100)]
[BLOCK] elevator: run queue in elevator_switch

elevator_dispatch needs to run queue after forced dispatching;
otherwise, the queue might stall.

Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] Document the READ/WRITE splitup of the disk stats
Jens Axboe [Wed, 9 Nov 2005 12:38:47 +0000 (13:38 +0100)]
[BLOCK] Document the READ/WRITE splitup of the disk stats

Use the symbolic name where appropriate and add a comment to the
disk_stats structure.

Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] elevator init fixes #2
Zachary Amsden [Wed, 9 Nov 2005 12:24:20 +0000 (13:24 +0100)]
[BLOCK] elevator init fixes #2

In addition to the first patch, which is probably goodness, I found the
cause of my panic - applying this patch fixes it and now I am booting.
If the chosen_elevator[] is not found, fall back to noop.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[BLOCK] elevator init fixes
Zachary Amsden [Wed, 9 Nov 2005 12:23:01 +0000 (13:23 +0100)]
[BLOCK] elevator init fixes

I got a panic in the elevator code, backtrace :

Unable to handle kernel NULL pointer dereference at virtual address 00000060
..
EIP is at elevator_put+0x0/0x30 (null elevator_type passed)
..
elevator_init+0x38
blk_init_queu_node+0xc9
floppy_init+0xdb
do_initcalls+0x23
init+0x10a
init+0x0

Clearly if the kmalloc here fails, e->elevator_type is not yet set; this
appears to be the correct fix, but I think I probably hit the second case
due to a race condition.  Someone more familiar with the elevator code
should look at this more closely until I can determine if I can reproduce.

Signed-off-by: Zachary Amsden <zach@vmware.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jens Axboe <axboe@suse.de>
19 years ago[libata ahci] set port ATAPI bit correctly
Jeff Garzik [Sat, 12 Nov 2005 06:32:19 +0000 (01:32 -0500)]
[libata ahci] set port ATAPI bit correctly

Although according to the documentation this largely only affects
desktop LED control, let's make sure we set the ATAPI bit when we
have an ATAPI device attached to the port.

19 years ago[libata ahci, qstor] fix miscount of scatter/gather entries
Jeff Garzik [Sat, 12 Nov 2005 06:27:07 +0000 (01:27 -0500)]
[libata ahci, qstor] fix miscount of scatter/gather entries

Don't directly reference qc->n_elem, as that might cause an off-by-one
error for misaligned (padded) ATAPI transfers.

19 years agoLinux v2.6.15-rc1
Linus Torvalds [Sat, 12 Nov 2005 01:43:36 +0000 (17:43 -0800)]
Linux v2.6.15-rc1

As per the new release rules: two weeks of merging, and then an -rc1 and
calming down for the next release.

19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 12 Nov 2005 01:00:17 +0000 (17:00 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

19 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Sat, 12 Nov 2005 00:30:35 +0000 (16:30 -0800)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

19 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Sat, 12 Nov 2005 00:29:47 +0000 (16:29 -0800)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

19 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bunk/trivial
Linus Torvalds [Sat, 12 Nov 2005 00:29:22 +0000 (16:29 -0800)]
Merge git://git./linux/kernel/git/bunk/trivial

19 years ago[SCTP]: Include ulpevents in socket receive buffer accounting.
Neil Horman [Sat, 12 Nov 2005 00:08:24 +0000 (16:08 -0800)]
[SCTP]: Include ulpevents in socket receive buffer accounting.

Also introduces a sysctl option to configure the receive buffer
accounting policy to be either at socket or association level.
Default is all the associations on the same socket share the
receive buffer.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP]: Fix ia64 NaT consumption fault with sctp_sideffect commands.
Vladislav Yasevich [Sat, 12 Nov 2005 00:07:40 +0000 (16:07 -0800)]
[SCTP]: Fix ia64 NaT consumption fault with sctp_sideffect commands.

On ia64, it is possible to get NaT Consumption Fault and a kernel panic
when initializing sctp sideeffect commands arguments.  The union
sctp_arg_t contains different sized elements and when loading a smaller
sized element (32 or 16 bits), it is possible for a speculative load to
fail and result in a NaT bit set which causes a kernel crash.  The easy
way to get around it is to load the largerst member of the union.

Signed-off-by: Vladislav Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP]: Remove timeouts[] array from sctp_endpoint.
Vladislav Yasevich [Sat, 12 Nov 2005 00:06:16 +0000 (16:06 -0800)]
[SCTP]: Remove timeouts[] array from sctp_endpoint.

The socket level timeout values are maintained in sctp_sock and
association level timeouts are in sctp_association. So there is
no need for ep->timeouts.

Signed-off-by: Vladislav Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[SCTP]: Fix potential NULL pointer dereference in sctp_v4_get_saddr
Vladislav Yasevich [Sat, 12 Nov 2005 00:05:55 +0000 (16:05 -0800)]
[SCTP]: Fix potential NULL pointer dereference in sctp_v4_get_saddr

It is possible to get to sctp_v4_get_saddr() without a valid
association.  This happens when processing OOTB packets and
the cached route entry is no longer valid.
However, when responding to OOTB packets we already properly
set the source address based on the information in the OOTB
packet.  So, if we we get to sctp_v4_get_saddr() without an
association we can simply return.

Signed-off-by: Vladislav Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[IPV6]: Fix inet6_init missing unregister.
David S. Miller [Fri, 11 Nov 2005 23:05:47 +0000 (15:05 -0800)]
[IPV6]: Fix inet6_init missing unregister.

Based mostly upon a patch from Olaf Kirch <okir@suse.de>

When initialization fails in inet6_init(), we should
unregister the PF_INET6 socket ops.

Also, check sock_register()'s return value for errors.

Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[PATCH] kill libata scsi_wait_req usage (make libata compile with scsi-misc changes)
Mike Christie [Fri, 11 Nov 2005 22:38:53 +0000 (16:38 -0600)]
[PATCH] kill libata scsi_wait_req usage (make libata compile with scsi-misc changes)

scsi_wait_req does not exist any more in the SCSI layer.  This patch
makes it so libata can compile again.

Signed-off-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoppc64: default build as the merged 'powerpc' architecture
Linus Torvalds [Fri, 11 Nov 2005 22:27:32 +0000 (14:27 -0800)]
ppc64: default build as the merged 'powerpc' architecture

After the last merge of the new unified 'powerpc' architecture, ppc64 no
longer compiles cleanly as a standalone architecture.  Some bits and
pieces still exist as files under the old ppc64 hierarchy, but the old
"ARCH=ppc64" is dead.

So if "uname" says ppc64, that now implies that the default architecture
should be "powerpc".

Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-for-linus-2.6
Linus Torvalds [Fri, 11 Nov 2005 22:04:37 +0000 (14:04 -0800)]
Merge /linux/kernel/git/jejb/scsi-for-linus-2.6

19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Fri, 11 Nov 2005 22:03:49 +0000 (14:03 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

19 years ago[ARM] Update mach-types
Russell King [Fri, 11 Nov 2005 21:56:33 +0000 (21:56 +0000)]
[ARM] Update mach-types

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 3147/1: update ixp2000 defconfigs to 2.6.14-git13
Lennert Buytenhek [Fri, 11 Nov 2005 21:53:30 +0000 (21:53 +0000)]
[ARM] 3147/1: update ixp2000 defconfigs to 2.6.14-git13

Patch from Lennert Buytenhek

Update the ixp2000 defconfigs to 2.6.14-git13.

Signed-off-by: Lennert Buytenhek <buytenh@wantstofly.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 3152/1: make various assembly local labels actually local (the rest)
Nicolas Pitre [Fri, 11 Nov 2005 21:51:49 +0000 (21:51 +0000)]
[ARM] 3152/1: make various assembly local labels actually local (the rest)

Patch from Nicolas Pitre

For assembly labels to actually be local they must start with ".L" and
not only "." otherwise they still remain visible in the final link and
clutter kallsyms needlessly, and possibly make for unclear symbolic
backtrace. This patch simply inserts a"L" where appropriate. The code
itself is unchanged.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 3151/1: make various assembly local labels actually local (io-*.S)
Nicolas Pitre [Fri, 11 Nov 2005 21:51:48 +0000 (21:51 +0000)]
[ARM] 3151/1: make various assembly local labels actually local (io-*.S)

Patch from Nicolas Pitre

For assembly labels to actually be local they must start with ".L" and
not only "." otherwise they still remain visible in the final link and
clutter kallsyms needlessly, and possibly make for unclear symbolic
backtrace. This patch simply inserts a"L" where appropriate. The code
itself is unchanged.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[ARM] 3150/1: make various assembly local labels actually local (uaccess.S)
Nicolas Pitre [Fri, 11 Nov 2005 21:51:47 +0000 (21:51 +0000)]
[ARM] 3150/1: make various assembly local labels actually local (uaccess.S)

Patch from Nicolas Pitre

For assembly labels to actually be local they must start with ".L" and
not only "." otherwise they still remain visible in the final link and
clutter kallsyms needlessly, and possibly make for unclear symbolic
backtrace. This patch simply inserts a"L" where appropriate. The code
itself is unchanged.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
19 years ago[SPARC64]: Restore 2.4.x /proc/cpuinfo behavior for "ncpus probed" field.
David S. Miller [Fri, 11 Nov 2005 20:48:56 +0000 (12:48 -0800)]
[SPARC64]: Restore 2.4.x /proc/cpuinfo behavior for "ncpus probed" field.

Noticed by Tom 'spot' Callaway.

Even on uniprocessor we always reported the number of physical
cpus in the system via /proc/cpuinfo.  But when this got changed
to use num_possible_cpus() it always reads as "1" on uniprocessor.
This change was unintentional.

So scan the firmware device tree and count the number of cpu
nodes, and report that, as we always did.

Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[DECNET]: fix SIGPIPE
Patrick Caulfield [Fri, 11 Nov 2005 20:04:28 +0000 (12:04 -0800)]
[DECNET]: fix SIGPIPE

Currently recvmsg generates SIGPIPE whereas sendmsg does not; for the
other stacks it seems to be the other way round!

It also fixes the bug where reading from a socket whose peer has shutdown
returned -EINVAL rather than 0.

Signed-off-by: Patrick Caulfield <patrick@tykepenguin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[IA64-SGI] set altix preferred console
Mark Maule [Fri, 11 Nov 2005 17:52:43 +0000 (11:52 -0600)]
[IA64-SGI] set altix preferred console

Fix default VGA console on SN platforms.  Since SN firmware does not pass
enough ACPI information to identify VGA cards and the associated legacy IO/MEM
addresses, we rely on the EFI PCDP table.  Since the linux pcdp driver is
optional (and overridden if console= directives are used) SN duplicates a
portion of the pcdp scan code to identify if there is a usable console VGA
adapter.  Additionally, dup necessary pcdp related structs to avoid dragging
drivers/pcdp.h into a more public location.

Signed-off-by: Mark Maule <maule@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] 4-level page tables
Robin Holt [Fri, 11 Nov 2005 15:35:43 +0000 (09:35 -0600)]
[IA64] 4-level page tables

This patch introduces 4-level page tables to ia64.  I have run
some benchmarks and found nothing interesting.  Performance has
consistently fallen within the noise range.

It also introduces a config option (setting the default to 3
levels).  The config option prevents having 4 level page
tables with 64k base page size.

Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Fri, 11 Nov 2005 17:26:39 +0000 (09:26 -0800)]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/netdev-2.6

19 years agoMerge branch 'upstream-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
Linus Torvalds [Fri, 11 Nov 2005 17:25:54 +0000 (09:25 -0800)]
Merge branch 'upstream-linus' of /linux/kernel/git/jgarzik/libata-dev

19 years agoMerge master.kernel.org:/home/rmk/linux-2.6-drvmodel
Linus Torvalds [Fri, 11 Nov 2005 17:24:26 +0000 (09:24 -0800)]
Merge kernel.org:/home/rmk/linux-2.6-drvmodel

19 years ago[PATCH] ipmi: fix inconsistent spinlock usage
Hironobu Ishii [Fri, 11 Nov 2005 14:12:21 +0000 (08:12 -0600)]
[PATCH] ipmi: fix inconsistent spinlock usage

Part of a patch was accidentally reverted, this corrects an
inconsistent spinlock use in the IPMI message handler.

Signed-off-by: Hironobu Ishii <hishii@soft.fujitsu.com>
Signed-off-by: Corey Minyard <minyard@acm.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc-merge
Linus Torvalds [Fri, 11 Nov 2005 17:22:27 +0000 (09:22 -0800)]
Merge git://git./linux/kernel/git/paulus/powerpc-merge

19 years agoMerge master.kernel.org:/home/rmk/linux-2.6-mmc
Linus Torvalds [Fri, 11 Nov 2005 17:22:08 +0000 (09:22 -0800)]
Merge master.kernel.org:/home/rmk/linux-2.6-mmc

19 years ago[PATCH] TCP: fix vegas build
Jeff Garzik [Fri, 11 Nov 2005 09:43:47 +0000 (04:43 -0500)]
[PATCH] TCP: fix vegas build

Recent TCP changes broke the build.

Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Input: convert hdaps to dynamic input_dev allocation.
Dmitry Torokhov [Fri, 11 Nov 2005 03:10:55 +0000 (22:10 -0500)]
[PATCH] Input: convert hdaps to dynamic input_dev allocation.

Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] gt96100eth.c: Don't concatenate __FUNCTION__ with strings.
Ralf Baechle [Tue, 8 Nov 2005 22:02:35 +0000 (22:02 +0000)]
[PATCH] gt96100eth.c: Don't concatenate __FUNCTION__ with strings.

As part of the ISO C9x conversion gcc deprecates concatenation with
__FUNCTION__ because __FUNCTION__ is not a preprocessor macro.

Signed-off-by: Clemens Buchacher <drizzd@aon.at>
Signed-off-by: Maximilian Attems <janitor@sternwelten.at>
Signed-off-by: Domen Puncer <domen@coderock.org>
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] smc91x: DB1200 support.
Pete Popov [Thu, 10 Nov 2005 03:46:05 +0000 (22:46 -0500)]
[PATCH] smc91x: DB1200 support.

The following patch support the SMC9111 present on DB1200 boards.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] SAA9730: Driver overhaul
Ralf Baechle [Wed, 9 Nov 2005 17:10:05 +0000 (17:10 +0000)]
[PATCH] SAA9730: Driver overhaul

 o Try to work around some of the undocumented "features" of the SAA9730
 o Use netdev_priv() instead of the previous broken mechanism to allocate
   the private data structure.
 o Try to make sure we don't leak resources on exit.
 o No more need to call SET_MODULE_OWNER in 2.6.
 o Use pci_free_consistent instead of homegrown architecture-specific
   allocation.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
 drivers/net/saa9730.c |  531 +++++++++++++++++++++++---------------------------
 1 files changed, 249 insertions(+), 282 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] SAA9730: Whitespace cleanup.
Ralf Baechle [Wed, 9 Nov 2005 16:44:02 +0000 (16:44 +0000)]
[PATCH] SAA9730: Whitespace cleanup.

Kill trailing whitespace, replace leading whitespace with tabs.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
 drivers/net/saa9730.c |   98 +++++++++++++++++++++++++-------------------------
 1 files changed, 49 insertions(+), 49 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[netdrvr forcedeth] phy address scan range
Ayaz Abdulla [Fri, 11 Nov 2005 13:31:11 +0000 (08:31 -0500)]
[netdrvr forcedeth] phy address scan range

Added phy address 0 to the phy scan.

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
19 years ago[netdrvr forcedeth] support for irq mitigation
Ayaz Abdulla [Fri, 11 Nov 2005 13:30:38 +0000 (08:30 -0500)]
[netdrvr forcedeth] support for irq mitigation

This patch contains support for different modes of interrupt mitigation
of forcedeth. It includes changes based on Jeff's comments. Currently,
the modes are changed through module parameters since ethtool does not
support something similar.

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
19 years ago[netdrvr forcedeth] remove superfluous rx engine stop/start
Ayaz Abdulla [Fri, 11 Nov 2005 13:29:59 +0000 (08:29 -0500)]
[netdrvr forcedeth] remove superfluous rx engine stop/start

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
19 years ago[PATCH] gianfar mii needs to zero out the mii_bus structure
Kumar Gala [Wed, 9 Nov 2005 18:13:11 +0000 (12:13 -0600)]
[PATCH] gianfar mii needs to zero out the mii_bus structure

To ensure that phy_mask and any future elements of the mii_bus
structure are initialized use kzalloc() instead of kmalloc().

This fixes an issue in which phy_mask was not being initialized
and we would skip random phy addresses when scanning.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] s390: mail address changed
Frank Pavlic [Thu, 10 Nov 2005 12:51:42 +0000 (13:51 +0100)]
[PATCH] s390: mail address changed

[patch 7/7] s390: mail address changed

From: Frank Pavlic <fpavlic@de.ibm.com>
- mail address changed to fpavlic@de.ibm.com

Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
diffstat:
 lcs.c       |    4 ++--
 qeth_main.c |    4 ++--
 qeth_mpc.c  |    2 +-
 qeth_mpc.h  |    2 +-
 qeth_sys.c  |    2 +-
 qeth_tso.h  |    2 +-
 6 files changed, 8 insertions(+), 8 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] s390: introduce guestLan sniffer support in qeth
Frank Pavlic [Thu, 10 Nov 2005 12:51:25 +0000 (13:51 +0100)]
[PATCH] s390: introduce guestLan sniffer support in qeth

[patch 6/7] s390: introduce guestLan sniffer support in qeth

From: Peter Tiedemann  <ptiedem@de.ibm.com>
- introduce guestLan sniffer support in qeth
  feature allows a linux in a virtual machine
  guest to become a network LAN sniffer,
  monitoring and recording the networking traffic
  within an entire guestLan.

Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
diffstat:
 qeth.h      |    2 +
 qeth_main.c |   93 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 qeth_mpc.h  |   11 ++++---
 3 files changed, 102 insertions(+), 4 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] s390: fix recovery failure of non-guestLAN devices
Frank Pavlic [Thu, 10 Nov 2005 12:51:17 +0000 (13:51 +0100)]
[PATCH] s390: fix recovery failure of non-guestLAN devices

[patch 5/7] s390: fix recovery failure of non-guestLAN devices

From: Frank Pavlic <fpavlic@de.ibm.com>
- Recovery of non-guestLAN Layer 2 device failed due to
  trying to register the real MAC address we got from
  the READ_MAC adapter parameters command.
  We have to keep the "old" MAC address when we process
  the reply of a READ_MAC.

Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
diffstat:
 qeth.h      |   12 ++++++------
 qeth_main.c |   27 ++++++++++++++++-----------
 2 files changed, 22 insertions(+), 17 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] s390: some more qeth fixes
Frank Pavlic [Thu, 10 Nov 2005 12:50:58 +0000 (13:50 +0100)]
[PATCH] s390: some more qeth fixes

[patch 4/7] s390: some more qeth fixes

From: Frank Pavlic <fpavlic@de.ibm.com>
From: Peter Tiedemann <ptiedem@de.ibm.com>
- possible race on list fixed by reset
  list processing after every operation
- traffic hang fixed

Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
diffstat:
 qeth_main.c |   11 +++++++----
 1 files changed, 7 insertions(+), 4 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] s390: qeth multicast address registration fixed
Frank Pavlic [Thu, 10 Nov 2005 12:49:28 +0000 (13:49 +0100)]
[PATCH] s390: qeth multicast address registration fixed

[patch 3/7] s390: qeth multicast address registration fixed

From: Klaus Dieter Wacker <kdwacker@de.ibm.com>
- when running in Layer2 mode we don't have to register
  the multicast IP address but only group mac address.
  Therefore for Layer 2 devices it is enough to go
  through dev->mc_list list and register these entries.

Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
diffstat:
 qeth_main.c |  106 +++++++++++++++++++++++++++++++++++++++++++++---------------
 1 files changed, 80 insertions(+), 26 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] s390: minor modification in qeth layer2 code
Frank Pavlic [Thu, 10 Nov 2005 12:49:15 +0000 (13:49 +0100)]
[PATCH] s390: minor modification in qeth layer2 code

[patch 2/7] s390: minor modification in qeth layer2 code

From: Frank Pavlic <fpavlic@de.ibm.com>
- use qeth_layer2_send_setdelvlan_cb to check
  return code of a SET/DELVLAN IP Assist command.
  It fits better in qeth's design and mechanism of IP Assist
  command handling.

Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
diffstat:
 qeth_main.c |   40 ++++++++++++++++++++++++++--------------
 1 files changed, 26 insertions(+), 14 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] s390: synthax checking for VIPA addresses fixed
Frank Pavlic [Thu, 10 Nov 2005 12:49:02 +0000 (13:49 +0100)]
[PATCH] s390: synthax checking for VIPA addresses fixed

[patch 1/7] s390: synthax checking for VIPA addresses fixed

From: Peter Tiedemann <ptiedem@de.ibm.com>
- synthax checking for VIPA addresses fixed

Signed-off-by: Frank Pavlic <fpavlic@de.ibm.com>
diffstat:
 qeth.h     |   65 ++++++++++++++++++++++++++++++++++++++++++++++++-------------
 qeth_sys.c |    6 ++---
 2 files changed, 55 insertions(+), 16 deletions(-)
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] libata: propogate host private data from probe function
Alan Cox [Tue, 8 Nov 2005 14:09:44 +0000 (14:09 +0000)]
[PATCH] libata: propogate host private data from probe function

This will let me chop the code size of several drivers right down. In
many cases the actual private data is very useful and constant for a
given host controller so being able to just pass it at probe time would
be very useful indeed (eg with the via driver would could pass the udma
clocking and reduce the code size, or with the AMD one the UDMA
multiplier and the offset)

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years agoMerge branch 'atmel'
Jeff Garzik [Fri, 11 Nov 2005 13:14:56 +0000 (08:14 -0500)]
Merge branch 'atmel'

19 years ago[PATCH] Atmel wireless update
simon@thekelleys.org.uk [Sun, 30 Oct 2005 15:50:15 +0000 (15:50 +0000)]
[PATCH] Atmel wireless update

* Merge PCMCIA card table with new Brodowski PCMCIA id table.
* Add missing entries to PCMCIA id table.
* Other tweaks to conform with Documentation/driver-changes.txt
  (types, call request_region, etc)
* Fix size of requested IO region.
* Reduce printk verbosity.
* Remove EXPERIMENTAL
* tweak to association code - don't force shared key authentication
  when wep in use.

Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[libata sata_mv] add Adaptec 1420SA PCI ID
Jeff Garzik [Fri, 11 Nov 2005 13:08:03 +0000 (08:08 -0500)]
[libata sata_mv] add Adaptec 1420SA PCI ID

Contributed by Jeroen <dekien@pandora.be>

19 years agopowerpc: Fix some compile problems with the VDSO stuff
Paul Mackerras [Fri, 11 Nov 2005 12:04:40 +0000 (23:04 +1100)]
powerpc: Fix some compile problems with the VDSO stuff

We needed the VDSO symbols in the arch/ppc asm-offsets.c, and there
were a few usages of _systemcfg still left lying around.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agopowerpc: Fix reading and writing SPRs from xmon on 32-bit
Paul Mackerras [Fri, 11 Nov 2005 11:36:34 +0000 (22:36 +1100)]
powerpc: Fix reading and writing SPRs from xmon on 32-bit

When we created the instructions to read/write SPRs in xmon, we were
setting up a ppc64-style procedure descriptor and calling that, which
doesn't work in 32-bit.  For 32-bit a function pointer just points
to the instructions of the function.  This fixes it to do the right
thing for both 32-bit and 64-bit.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agopowerpc: Initialize secondary CPU setup for 32-bit SMP
Paul Mackerras [Fri, 11 Nov 2005 11:34:43 +0000 (22:34 +1100)]
powerpc: Initialize secondary CPU setup for 32-bit SMP

32-bit SMP powermacs weren't booting with ARCH=powerpc because the
boot cpu wasn't saving away the state of various control registers,
but the secondary CPUs were loading them from the uninitialized
state.  This adds the necessary save-state call.

Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] powerpc: Merge vdso's and add vdso support to 32 bits kernel
Benjamin Herrenschmidt [Fri, 11 Nov 2005 10:15:21 +0000 (21:15 +1100)]
[PATCH] powerpc: Merge vdso's and add vdso support to 32 bits kernel

This patch moves the vdso's to arch/powerpc, adds support for the 32
bits vdso to the 32 bits kernel, rename systemcfg (finally !), and adds
some new (still untested) routines to both vdso's: clock_gettime() with
support for CLOCK_REALTIME and CLOCK_MONOTONIC, clock_getres() (same
clocks) and get_tbfreq() for glibc to retreive the timebase frequency.

Tom,Steve: The implementation of get_tbfreq() I've done for 32 bits
returns a long long (r3, r4) not a long. This is such that if we ever
add support for >4Ghz timebases on ppc32, the userland interface won't
have to change.

I have tested gettimeofday() using some glibc patches in both ppc32 and
ppc64 kernels using 32 bits userland (I haven't had a chance to test a
64 bits userland yet, but the implementation didn't change and was
tested earlier). I haven't tested yet the new functions.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] powerpc: remove initrd debug printk
David Woodhouse [Fri, 11 Nov 2005 08:07:11 +0000 (08:07 +0000)]
[PATCH] powerpc: remove initrd debug printk

This removes a stray debugging printk which offended Anton.

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] powerpc: Move udbg code to arch/powerpc
David Gibson [Fri, 11 Nov 2005 05:42:12 +0000 (16:42 +1100)]
[PATCH] powerpc: Move udbg code to arch/powerpc

Since the udbg code in ppc64 has no ppc32 equivalent, move it straight
over into arch/powerpc (and include/asm-powerpc for udbg.h).  In time,
we probably want to meld the various bits and pieces of 32-bit early
debugging code into udbg, but for now only include it on
CONFIG_PPC64=y builds.  The only change during the move is to
standardise the protecting #ifdef/#define in udbg.h, and move its
banner comment above the initial #ifdef (which seems to be normal
practice).

Built and booted on POWER5 LPAR (ARCH=powerpc and ARCH=ppc64).  Built
for 32bit multiplatform (ARCH=powerpc).

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc64: Increase sparsemem defaults
Anton Blanchard [Fri, 11 Nov 2005 04:02:03 +0000 (15:02 +1100)]
[PATCH] ppc64: Increase sparsemem defaults

The definitions in sparsemem.h arent sufficient. We currently sell
machines with 2TB of RAM, and in order to give us room for a few years
growth lets set it to 16TB.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc64: Convert NUMA to sparsemem (3)
Anton Blanchard [Fri, 11 Nov 2005 03:22:35 +0000 (14:22 +1100)]
[PATCH] ppc64: Convert NUMA to sparsemem (3)

Convert to sparsemem and remove all the discontigmem code in the
process. This has a few advantages:

- The old numa_memory_lookup_table can go away
- All the arch specific discontigmem magic can go away

We also remove the triple pass of memory properties and instead create a
list of per node extents that we iterate through. A final cleanup would
be to change our lmb code to store extents per node, then we can reuse
that information in the numa code.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc64: prep for NUMA sparsemem rework 2
Anton Blanchard [Fri, 11 Nov 2005 03:13:20 +0000 (14:13 +1100)]
[PATCH] ppc64: prep for NUMA sparsemem rework 2

Remove ppc64 specific version of nr_cpus_node and use the generic one
provided.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc64: prep for NUMA sparsemem rework
Anton Blanchard [Fri, 11 Nov 2005 03:12:26 +0000 (14:12 +1100)]
[PATCH] ppc64: prep for NUMA sparsemem rework

Remove an unused numa define and move a discontigmem specific define
inside the relevant ifdef.

I will submit a separate patch to remove them from other architectures,
but the ppc64 patches to follow depend on this.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc64: Cleanup kprobe assembly
Anton Blanchard [Fri, 11 Nov 2005 03:07:20 +0000 (14:07 +1100)]
[PATCH] ppc64: Cleanup kprobe assembly

The kprobes code is doing ".previous .text". While the assembler doesnt
warn at the moment (and it seems to work), it might in the future.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc64: Remove debug boot message
Anton Blanchard [Fri, 11 Nov 2005 02:56:16 +0000 (13:56 +1100)]
[PATCH] ppc64: Remove debug boot message

We have been printing the raw ppc64_firmware_features during boot. Since
we can work it out from the device tree, lets remove it.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc64: Quieten lparcfg
Anton Blanchard [Fri, 11 Nov 2005 02:53:11 +0000 (13:53 +1100)]
[PATCH] ppc64: Quieten lparcfg

If we dont have permission to read some information from the hypervisor,
lparcfg outputs a warning on the console. Now that lparcfg is world
readable this is a problem.

Dont warn in the case of H_Authority, remove some unnecessary function
prototypes and fix whitespace damage in a structure as well.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years ago[PATCH] ppc32: fix PQ2 PCI DMA interrupt handling
Kumar Gala [Thu, 10 Nov 2005 16:34:33 +0000 (10:34 -0600)]
[PATCH] ppc32: fix PQ2 PCI DMA interrupt handling

The bit position in the status register corresponding to the
PCI DMA interrupt was incorrect.  Additionally, we did not
have a define for the PCI DMA interrupt.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
19 years agoMerge branch 'master'
Jeff Garzik [Fri, 11 Nov 2005 10:50:22 +0000 (05:50 -0500)]
Merge branch 'master'

19 years ago[PATCH] libata.h needs dma-mapping.h
Andrew Morton [Fri, 11 Nov 2005 00:21:38 +0000 (16:21 -0800)]
[PATCH] libata.h needs dma-mapping.h

On Alpha:

include/linux/libata.h: In function `ata_pad_alloc':
include/linux/libata.h:785: warning: implicit declaration of function `dma_alloc_coherent'
include/linux/libata.h:786: warning: assignment makes pointer from integer without a cast
include/linux/libata.h: In function `ata_pad_free':
include/linux/libata.h:792: warning: implicit declaration of function `dma_free_coherent'

(I have a decouple-some-header-files cleanup in -mm, so it's causing some
fallout of this nature)

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years ago[PATCH] disable DEBUG in ibmveth
Anton Blanchard [Fri, 11 Nov 2005 03:02:04 +0000 (14:02 +1100)]
[PATCH] disable DEBUG in ibmveth

At the moment ibmveth has DEBUG enabled which is rather verbose. Disable
it.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
19 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Fri, 11 Nov 2005 05:37:50 +0000 (21:37 -0800)]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

19 years ago[PATCH] lpfc build fix
Jeff Garzik [Fri, 11 Nov 2005 01:34:12 +0000 (20:34 -0500)]
[PATCH] lpfc build fix

Current upstream 'allmodconfig' build is broken.  This is the obvious
patch...

Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Fri, 11 Nov 2005 05:24:21 +0000 (21:24 -0800)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

19 years ago[SPARC]: display7seg build fix
Andrew Morton [Fri, 11 Nov 2005 05:14:16 +0000 (21:14 -0800)]
[SPARC]: display7seg build fix

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[CIFS] Fix spaces in cifs kconfig entry
Steve French [Fri, 11 Nov 2005 03:36:39 +0000 (19:36 -0800)]
[CIFS] Fix spaces in cifs kconfig entry

Signed-off-by: Steve French <sfrench@us.ibm.com>
19 years ago[CIFS] Reduce sparse endian warnings
Steve French [Fri, 11 Nov 2005 03:28:44 +0000 (19:28 -0800)]
[CIFS] Reduce sparse endian warnings

Signed-off-by: Steve French <sfrench@us.ibm.com>
19 years agoMerge with /pub/scm/linux/kernel/git/torvalds/linux-2.6.git
Steve French [Fri, 11 Nov 2005 01:31:49 +0000 (17:31 -0800)]
Merge ... /linux/kernel/git/torvalds/linux-2.6.git

19 years ago[CIFS] Update CIFS change log
Steve French [Fri, 11 Nov 2005 01:31:33 +0000 (17:31 -0800)]
[CIFS] Update CIFS change log

Signed-off-by: Steve French <sfrench@us.ibm.com>
19 years ago[TCP]: speed up SACK processing
Stephen Hemminger [Fri, 11 Nov 2005 01:14:59 +0000 (17:14 -0800)]
[TCP]: speed up SACK processing

Use "hints" to speed up the SACK processing. Various forms
of this have been used by TCP developers (Web100, STCP, BIC)
to avoid the 2x linear search of outstanding segments.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TCP]: spelling fixes
Stephen Hemminger [Fri, 11 Nov 2005 01:13:47 +0000 (17:13 -0800)]
[TCP]: spelling fixes

Minor spelling fixes for TCP code.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TCP]: receive buffer growth limiting with mixed MTU
John Heffner [Fri, 11 Nov 2005 01:11:48 +0000 (17:11 -0800)]
[TCP]: receive buffer growth limiting with mixed MTU

This is a patch for discussion addressing some receive buffer growing issues.
This is partially related to the thread "Possible BUG in IPv4 TCP window
handling..." last week.

Specifically it addresses the problem of an interaction between rcvbuf
moderation (receiver autotuning) and rcv_ssthresh.  The problem occurs when
sending small packets to a receiver with a larger MTU.  (A very common case I
have is a host with a 1500 byte MTU sending to a host with a 9k MTU.)  In
such a case, the rcv_ssthresh code is targeting a window size corresponding
to filling up the current rcvbuf, not taking into account that the new rcvbuf
moderation may increase the rcvbuf size.

One hunk makes rcv_ssthresh use tcp_rmem[2] as the size target rather than
rcvbuf.  The other changes the behavior when it overflows its memory bounds
with in-order data so that it tries to grow rcvbuf (the same as with
out-of-order data).

These changes should help my problem of mixed MTUs, and should also help the
case from last week's thread I think.  (In both cases though you still need
tcp_rmem[2] to be set much larger than the TCP window.)  One question is if
this is too aggressive at trying to increase rcvbuf if it's under memory
stress.

Orignally-from: John Heffner <jheffner@psc.edu>
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TCP]: Appropriate Byte Count support
Stephen Hemminger [Fri, 11 Nov 2005 01:09:53 +0000 (17:09 -0800)]
[TCP]: Appropriate Byte Count support

This is an updated version of the RFC3465 ABC patch originally
for Linux 2.6.11-rc4 by Yee-Ting Li. ABC is a way of counting
bytes ack'd rather than packets when updating congestion control.

The orignal ABC described in the RFC applied to a Reno style
algorithm. For advanced congestion control there is little
change after leaving slow start.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TCP]: add tcp_slow_start helper
Stephen Hemminger [Fri, 11 Nov 2005 01:07:24 +0000 (17:07 -0800)]
[TCP]: add tcp_slow_start helper

Move all the code that does linear TCP slowstart to one
inline function to ease later patch to add ABC support.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TCP]: simplify microsecond rtt sampling
Stephen Hemminger [Fri, 11 Nov 2005 00:56:12 +0000 (16:56 -0800)]
[TCP]: simplify microsecond rtt sampling

Simplify the code that comuputes microsecond rtt estimate used
by TCP Vegas. Move the callback out of the RTT sampler and into
the end of the ack cleanup.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TCP]: fix congestion window update when using TSO deferal
Stephen Hemminger [Fri, 11 Nov 2005 00:53:30 +0000 (16:53 -0800)]
[TCP]: fix congestion window update when using TSO deferal

TCP peformance with TSO over networks with delay is awful.
On a 100Mbit link with 150ms delay, we get 4Mbits/sec with TSO and
50Mbits/sec without TSO.

The problem is with TSO, we intentionally do not keep the maximum
number of packets in flight to fill the window, we hold out to until
we can send a MSS chunk. But, we also don't update the congestion window
unless we have filled, as per RFC2861.

This patch replaces the check for the congestion window being full
with something smarter that accounts for TSO.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>