2.6.31.4

[Tim Gardner]

Andy Whitcroft wrote:
> Stefan and I did a review of the 2.6.31.4 stable updates.  Most appear
> to either be fairly important fixes in core code, or small but very self
> contained updates for drivers and/or quirks and low risk.
> 
> There are three which could sensibly be ignored, the integrity ones are
> high end server style things, and a third which is actually a functional
> noop:
> 
>     ima: ecryptfs fix imbalance message
>     IMA: open new file for read
>     tun: Return -EINVAL if neither IFF_TUN nor IFF_TAP is set.
> 
> There is one which is definatly an ABI bumper, though the interface
> change is not in the core interfaces and likely could be ignored:
> 
>     tracing: correct module boundaries for ftrace_release
> 
> Overall it seems sensible to take all of these.  Some notes we took on
> each is below.
> 
> -apw
> 
> 1611e30 x86: fix csum_ipv6_magic asm memory clobber
> 	potential bad checksums with some gcc versions
> 4b5e77c tty: Avoid dropping ldisc_mutex over hangup tty re-initialization
> 	fixes a warn_on which is getting reported
> ee39c2f x86: Don't leak 64-bit kernel register values to 32-bit processes
> 	potiential security issue
> 23a27d5 ALSA: hda - Added quirk to enable sound on Toshiba NB200
> 	device specific quirk
> 6e4be6c tracing: correct module boundaries for ftrace_release
> 	potential leak/oops ABI BUMPER
> 1afc593 ftrace: check for failure for all conversions
> 	fails more seems reasonable
> 9cb696b futex: fix requeue_pi key imbalance
> 	locking balance issue
> 3e82b94 futex: Move exit_pi_state() call to release_mm()
> 	possible userspace futex hangs
> 1c4dacc futex: Nullify robust lists after cleanup
> 	corruption potiential
> 627376b futex: Fix locking imbalance
> 	locking fix
> 94c517d NOHZ: update idle state also when NOHZ is inactive
> 	looks like it could cause scheduler issues
> 99db179 ima: ecryptfs fix imbalance message
> 	integity stats, sounds like server thing
> 7bbd09e libata: fix incorrect link online check during probe
> 	yes please, may fix boot disk lossage
> 83a6b58 sound: via82xx: move DXS volume controls to PCM interface
> 	device specific
> 170fa2c ASoC: WM8350 capture PGA mutes are inverted
> 	probabally arm, not our arm
> 779632b KVM: Prevent overflow in KVM_GET_SUPPORTED_CPUID
> 	potential security
> ca9405b KVM: VMX: flush TLB with INVEPT on cpu migration
> 	nasty tbl flush change
> ddf2acb KVM: fix LAPIC timer period overflow
> 	simple cast
> 89cc564 KVM: SVM: Fix tsc offset adjustment when running nested
> 	server/eucalyptus thing -- clock handling
> 719a125 KVM: SVM: Handle tsc in svm_get_msr/svm_set_msr correctly
> 	server/eucalyptus thing -- clock handling
> 0833a1c net: Fix wrong sizeof
> 	iseries veth only
> 20330dc mm: add_to_swap_cache() must not sleep
> 	yes looks bad
> 5a69ea2 sis5513: fix PIO setup for ATAPI devices
> 	host adapter device specific
> 112a62d PIT fixes to unbreak suspend/resume (bug #14222)
> 	core suspend/resume fix, yes please
> 3ee41ba IMA: open new file for read
> 	integity stats, sounds like server thing
> 2f3102c ACPI: Clarify resource conflict message
> 	trivial message update, worth having
> 269beb9 ACPI: fix Compaq Evo N800c (Pentium 4m) boot hang regression
> 	boot hang, specific to later pentiums, upstream regression
> d649563 net: restore tx timestamping for accelerated vlans
> 	nice to have, but non-critical, may affect kvm, very minor
> 7d07d6b net: unix: fix sending fds in multiple buffers
> 	prevents multiple copies of fd's being passed, core component
> 0568d3b tun: Return -EINVAL if neither IFF_TUN nor IFF_TAP is set.
> 	BOGUS without 2b980dbd77d229eb60588802162c9659726b11f4 this patch is
> 	a noop
> e3d38b5 tcp: fix CONFIG_TCP_MD5SIG + CONFIG_PREEMPT timer BUG()
> 	preempt imbalance, obvious and simple fix, NULL for non-preempt
> 6574534 net: Fix sock_wfree() race
> 	use after free bug, in core sockets
> fe13bc4 smsc95xx: fix transmission where ZLP is expected
> 	transmit hangs for USB ethernet, device specific quirk only
> c839c5c sky2: Set SKY2_HW_RAM_BUFFER in sky2_init
> 	likely performance issue with sky2 ehternet
> fb0e870 appletalk: Fix skb leak when ipddp interface is not loaded
> 	memory leak in non-core protocol support
> 84b1767 ax25: Fix possible oops in ax25_make_new
> 	oops preveter in non-core component
> e928731 ax25: Fix SIOCAX25GETINFO ioctl
> 	trivial fix in non-core component
> 2d85289 sit: fix off-by-one in ipip6_tunnel_get_prl
> 	potential userspace corruption, trivial fix in non-core component
> 	ipv6 tunneling
> 5eee394 Linux 2.6.31.4

Thanks for the close study of these patches. I applied all of them.

rtg
-- 
Tim Gardner tim.gardner at canonical.com