kernel weekly news – 03.12.2011

Posted: December 3, 2011 in kernel

Hello people and welcome to the 1st edition of December!

-Greg Kroah-Hartman announces the release of kernels 3.1.3, 3.0.11 and, Takashi Iwai has sound updates for -rc4, Tejun Heo has
-rc3 fixes for percpu and cgroup and Greg KH announces kernels
3.1.4 and 3.0.12.

-Arnaldo Carvalho de Melo has a 25-piece perf/core fixes and
improvements pull request, Pekka Enberg has slab fixes for
-rc4, Roland Dreier updates infiniband and John W. Linville has
wireless updates.

-David Brown announces msm updates for 3.3, Chris Mason has
btrfs updates, Joel Becker has ocfs updates for -rc3 and
Linus Torvalds announces Linux 3.2-rc4:

This doesn’t look tons smaller than -rc2 or -rc3, but it really is.
Yes, there are some ARM updates and fixups to the new Exonys DRI code,
and a questionably late ocfs2 update, but if you ignore those three
areas (and most people can happily ignore them), things really are
calming down pretty nicely.

There are some small sound updates, and btrfs is still getting fixups
(but nowhere near -rc2 levels), but other than that it was almost
eerily quiet.

I’m waiting for the other shoe to drop. Maybe Davem and GregKH are
holding back – they’ve been suspiciously quiet, and I think I can hear
some evil chuckling going on there. But maybe it’s just time for my

-David Miller has misc fixes for the networking tree:

Lots of tiny fixes, pretty much everywhere. Nothing really stands out, but
each bug is certainly a bummer if you happen to run into it.

I hope all the PMTU and redirect regressions I added to the ipv4 code
are really fixed now :-) A huge thanks to Steffen Klassert, Eric Dumazet,
and others for helping to resolve these problems.

The netfilter ADVANCED depency adjustments are, of course, in here too.

1) ARCNET and PHYLIB were inadvertantly changed to bool, revert back to
tristate. From Ben Hutchings.

2) Two netfilter ADVANCED adjustments.

3) Cached route lookups must validate referenced inetpeer.

4) Revert old udp_recvmsg() ‘redundant variable’ change, because in fact
the ‘copied’ variable was not redundant and this broke things when
encountering UDP packets with bad checksums.

5) TCP ipv6 input packet processing could crash under various
circumstances in error paths, fixes from Eric Dumazet.

6) Fix bad decnet socket refcounting, also from Eric Dumazet.

7) L2TP needs to validate attached socket route, fix from Florian Westphal.

8) Proxy ARP entries listed improperly in network namespace, fix from
Jorge Boncompte.

9) IGMP’s ip_mc_add_src() incorrectly decrements the entry refcount,
fix from Jun Zhao.

10) bnx2x LED handling fix from Yaniv Rosner.

11) TX descriptor list was mis-sized in qlge driver, fix from Thadeu
Lima de Souza Cascardo.

12) ehea per-TX-queue memory consumption was enormous, scale it
down to be more reasonable, from Anton Blanchard.

13) We have to adjust the IP header ->daddr in ip_forward_options()
not ip_options_rcv_srr() as the latter is too early. Fix from
Li Wei.

14) Netlabel adds to ipv4 map list in ipv6 code :-) Fix from
Dan Carpenter.

15) Several MTU handling regression fixes in ipv4 from Steffen Klassert.

16) DM9000 driver tests bit using && instead of &. Fix from Mark Brown.

17) Concurrent stream socket reads may cause poll() to incorrectly block
for AF_UNIX sockets, report and fix from Alexey Moiseytsev.

18) Invalidate cached ipv4 redirects properly, fix from Eric Dumazet.

19) Some 802.11 netlink attributes were not being validated properly,
fix from Eliad Peller.

20) Wireless TX aggregation accidental double-stop fix from Johannes Berg.

21) rtlwifi can deadlock on lps_lock, fix from Stanislaw Gruszka.

22) /proc/*/net/dev_mcast output corruption fix from Anton Blanchard.

23) “-1” means default for multicast hops in ipv6, fix from Li Wei.

24) Integer overflow fix in SCTP from Xi Wang, although it took me two
tries to get it committed properly, my bad.

25) ISDN string NULL termination fixes from Dan Carpenter.

26) TEQL driver needs to perform dst_get_neighbour() under proper
RCU protection, fix from Eric Dumazet.

27) Likewise, fix a similar lockdep splat in rt_cache_seq_show, also
from Eric Dumazet.

28) When the PSCHED_SHIFT was adjusted from 10 to 6 the RED packet scheduler did
not have it’s idle period handling adjusted properly. Fix from Eric Dumazet.

29) When bridge user-stp mode is used, we can get the device stuck in no-carrier
state forever. Fix from Vitalii Demianets.

30) When the RED packet scheduler was made classful, we didn’t update the test
for queue empty. It must check q->qdisc->q.len not whether sch->q is empty.
Fix from Eric Dumazet.

31) Missing spin_lock_init() and workqueue deadlock fix in p54spi driver from
Michael Buesch.

32) ip_route_me_harder() in netfilter needs to use HH_DATA_ALIGN when expanding
the SKB header area, otherwise packet headers become misaligned. Fix
from Paul Guo.

33) iseries_veth uses wrong length argument to memset() call, from Thomas Jarosch.

34) When DCCP ipv4 connect fails to lookup a route, we don’t record the error
from the ‘rt’ error pointer properly, fix from RongQing Li.

35) Fix namespace based OOPS in nf_conntrack by making event callback
registration per-ns. Fix from Pablo Neira Ayuso.

36) In netlink socket dumps, report the TOS attribute of inet sockets,
regardless of whether they are ipv4 or ipv6, because this value
has meaning for ipv4 mapped ipv6 connections. Fix from Maciej

37) ndisc_send_redirect() has reversed rate limiting check, oops. Fix from
Li Wei.

38) B44 needs to use dev_kfree_skb_irq() when releasing TX buffers because
the free runs with an IRQ safe spinlock held, fix from Xander Hover.

39) If we reuse a TCP time-wait socket in ipv6, we erroneously drop the SYN
because treq->iif is not set early enough in tcp_v6_conn_request(). Fix
from Eric Dumazet. This was largely not noticed because 1 second later the
client would resend the SYN and that would work since the first SYN
killed off the time-wait entry.

40) Fix PHY initialization in jme driver, from Aries Lee.

-Takashi Iwai has sound fixes, Dave Airlie has drm fixes,
mainly nouveau and radeon, and this concludes this week’s news.
Enjoy your weekend!


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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s