kernel weekly news – 05.11.2011

Posted: November 5, 2011 in kernel

Hello and welcome to the first November edition of KWN!

-Alex Elder has XFS updates for 3.2:

The repository is If you would prefer I use something
you feel is more trustworthy let me know and I will arrange that. I
have tagged the top commit “xfs_for_3.2-rc1_v1”, signing it with a
key I established last year for signing XFS user space releases.

-James Bottomley announces the first round of SCSI updates
for the merge window:

This represents the usual round of driver updates (fusion, mvsas, isci,
qla2 qla4, iscsi, lpfc, be2iscsi, bnx2fc, megaraid, fcoe and ipr).
There is also a new Marvell UMI driver and this also contains the gendev
preferred_name update, which will set how the user wants their disk
device to be reported by the kernel printk system. One of the
highlights of the release is the increasing use of the block generic
device to transport configuration information to HBAs.

-Chris Ball also has some updates, this time the tree is MMC and
the destination is -rc1:

There are currently no merge conflicts,
and the patches have been tested in linux-next. Thanks.

Does the below method of verifying a signed pull request work well for you?
Would you prefer signed git tags? To summarize, if you:

% git checkout 41ebe9c # rewind to the mainline commit I branched at
% git pull git:// for-linus
% git rev-parse HEAD

and the top-most commit hash you see matches the one above, then git has
guaranteed that the commits I’m sending you are the commits I intended
to send you, and this mail carrying a good gpg signature from my key
guarantees that the hash above is unaltered.

-Takashi Iwai also has updates for -rc1, but for the sound tree:

One of the scattered changes over the tree are one-liner IRQF_DISALBED
removals. Other than that, the changes are mostly concentrated on
the usual stuff, HD-audio and ASoC. Pretty many commits for ASoC at
this time.

In the sound core side, the ENUM-type user-space control was newly
supported. It’s needed for the better firewire sound driver support
in user-space.

In HD-audio, we improved the auto-parser for Realtek codecs, and
reduced lots of static configurations. The size halved, finally.
Also there are a few regression fixes in Conexant and other codecs.

Some highlights for ASoC changes are;
– Substantial optimization of the DAPM algorithm, substantially reducing
the CPU usage when power states change. This is especially beneficial
with larger modern devices.
– Support for CODEC drivers using the regmap API.
– Some smaller API updates – support for larger register maps, support
for specifying a source when setting a sysclk.
– New CPU for Alchemy and Freescale MXS.
– New CODEC drivers for Analog ADAU1373, Realtek RT5631 and Wolfson
Microelectronics WM1811 and WM5100.

Other than that, small fixes and imporvements here and there, e.g.
– More flexible buffer management in hdspm driver
– Support for non-SRC mode in some old drivers
– USB-audio driver refacotring, and the addition of a few new quirks

-Grant Likely has GPIO —

The majority of gpio changes this
cycle are actually going through rmk’s tree since they are arch/arm
cleanups. Nothing terribly exciting here except for one patch that
touches kernel/irq to export a handful of symbols which tglx acked.
That’s here since an earlier commit in this tree caused a failure when
building modules.

and spi driver changes for 3.2, while Arnaldo Carvalho de Melo has
perf/core fixes and Konrad Rzeszutek Wilk has xen updates, also
for 3.2:

a). make the backend driver use the API provided by xenbus module instead of its
own handling (which was duplicated across backends),
b). removes dead code (ia64),
c). lays out the groundwork for removing the ‘vmalloc_sync_all’ patch
that was added in v3.1 as a bug-fix.

Note: The patches touch other maintainers trees and I’ve gotten
Acks from them to do this as one patchset instead of cutting it in
small pieces and parceling out.

Note2: The git history is a bit fresh (4 days old) in the branch b/c
I’ve edited the git commits to add axboe and davem’s Ack-by.

-Florian Tobias Schandinat announces fbdev updates for 3.2:

It contains

– a series from Laurent which touches many different drivers to sanitize their
panning behavior

– two patches to fix the suspend/resume locking issue

– a little fix to our EDID parsing

– a new driver for SMSC UFX6000/7000 USB display devices

– a new driver for Aeroflex Gaisler GRVGA IP core

– a lot of driver updates for OMAP, sh_mobile, fsl-diu-fb, au1200fb, udlfb and
viafb, a few patches to others

It’s the same tree (only a different branch) you used to pull viafb updates from
so, I think, you can trust it. You will probably experience some (trivial) merge
conflicts related to xgifb and OMAP, please let me know if you want me to handle

-Avi Kivity has KVM updates (all these are for 3.2, unless otherwise
stated), Joerg Roedel has IOMMU updates and Ohad Ben-Cohen has
hwspinlock changes:

The only noticeable change is a new hwspinlock driver for
ST-Ericsson’s u8500 platform by Mathieu; the rest of the changes are
either cleanups or trivial fixes (one of the fixes also trivially
changes a mach-omap2 file, and is Acked by Tony).

All of the patches have been tested in linux-next and currently there
are no merge conflicts, though I expect one to happen with the arm-sco
tree, as reported (and fixed) by Stephen earlier this month (see

I’ve flagged one of the fixes as stable material (the one from Juan).
Normally I would only send you stable-worthy patches during the rc
cycle and not wait until the merge window, but since the hwspinlock
tree and its upstream path have only materialized quite late in the
cycle (see, I’ve decided this
time to wait for the merge window.

-Vinod Koul updates slave-dma:
Here is the slave-dmaengine updates which include consolidation of
samsung dma drivers from arm drivers to use dmaengine, enhancements and
fixes to amba-pl08x driver along with other usual fixes.

Since I have not been able to get my key signed with enough people yet,
so not sending this with key. Hopefully next time it would be a signed
message which you can verify. Also my tree is on infradead servers (same
one you pulled last time), let me know if you are okay with this server
or i should consider moving it.

-David Miller announces networking updates:

The majority of the bits here are driver bug fixes, but more notably:

1) icmp6_dst_alloc needs to set the destination address in the route
before trying to binding the route to an inetpeer entry, since the
inetpeer is found by destination. Fix from Gao Feng.

2) Traffic class not set properly for TIME_WAIT sockets, from Eric

3) Fix vlan over bonding ARP regression, also from Eric Dumazet.

4) ip6_ufo_append_data() does not propagate errors properly, resulting
in signal interrupts and hangups looking like memory allocation
errors. Fix from Zheng Yan.

5) Refcounting and hash lookup fixes in batman-adv from Simon Wunderlich.

7) Fix races in bond_close() and workqueue deadlocks. From Jay

8) IPV6 addrconf prefix handling needs to explicitly lookup routes
in the RT6_TABLE_PREFIX routing table, otherwise it might find
unrelated routes. Fix from Andreas Hofmeister.

-Martin Schwidefsky updates the s390 tree, Stefan Richter has
firewire updates:

We have got
– some janitorial work,
– support for the TI TSB41BA3D S400beta phy as local phy,
– bandwidth improvement of IPv4-over-1394 between 2%…30%
(depending on particular hardware; the gains are apparently at the
lower end of this range in case of typical PC hardware),
– a fix for certain topology scanning border cases,
– DMA mapping fixes.

The latter fix introduces false runtime warnings from the DMA debug
facility. Clemens posted a fix for that in and I will keep any eye on it
making its way into your tree eventually.

Everything has been in linux-next for a while, also during the downtime. Since most or all of the patches were not Cc’d
to LKML, I will post the full log and diff in a follow-up.

and Mark Brown updates regulator:

Please pull from the tree below to get updates to the regulator API for
version 3.2. Since Liam has been slammed with work until very recently
and is still catching up on things like restoring hosting for his git
trees after the breakin I’m sending the request instead of
him; hopefully he’ll get things sorted out soon but we weren’t sure
that’d happen during the merge window.

This has been a quiet release cycle for the regulator API – nothing much
going on in the core and not many driver updates. The biggest things
are that we’ve got a new gpio-regulator driver (for devices that select
voltages with GPIOs) and the tps65023 driver gained support for TPS65020
as well. The two commits from Marcus Folkesson with the same subject
are actually separate commits with similar changelogs.

-Roland Dreier updates infiniband, Gustavo Padovan updates the bluetooth tree,
Jan Kara has ext{2,3} and UDF fixes, James Morris updates linux-security with
some tomoyo fixes, Arnd Bergmann has lots of arm-soc fixes/cleanups and
Wu Fengguang has writeback patches.

-Richard Weinberger has UML changes, Mauro Carvalho Chehab updates EDAC,
Theodore Ts’o has ext4 updates, Paul Mundt has sh updates and Tejun Heo
updates pm-freezer.

-David Miller has updates for networking:

1) NULL crash in x25_recvmsg(), fix from Dave Jones.

2) __alloc_skb() alignments need adjusting for SLOB, fix from Tony Lindgren.

3) l2tp_recv_dequeue() can leave queue in inconsistent state, resulting in
a subsequent crash, fix from Eric Dumazet.

4) Missing bh_unlock_sock() in inet_csk_route_child_sock(), from Eric

5) Build fix for 3c505 driver from Andrew Morton.

6) Fix docbook comment in skbuff.h, from Marcos Paulo de Souza.

7) IPVS netfilter fixes from Krzysztof Wilczynski and Simon Horman.

8) Fix refcount leak in ipv6 netfilter core, from Florian Westphal.

9) Fix multicast/promisc state refcounting bug due to vlan propagating
flag state changes on downed devices. Fix from Matthijs Kooijman.

10) Fix build error due to typo in ll_temac driver, from Stephen Rothwell.

11) Fix build of AF_PACKET due to bogus inline prototypes, from Olof

12) be2net driver bug fixes from Padmanabh Ratnakar.

13) Intel ethernet fixes from Bruce Allan, Jesse Brandeburg, Tomasz Kantecki,
Emil Tantilov, John Fastabend, and Greg Rose.

-That’s it for this week, see ya!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s