commit 424a8a4e49031103af677c476f4773964751539b Author: Stefan Bader acerhdf: limit modalias matching to supported Type: Correction - Prevents the module from autoloading on non-supported hw ! We already carry that patch commit 13a10b71269f1b91ecbceb540ac6bc00933ba589 Author: Eric Millbrandt ASoC: Do not write to invalid registers on the wm9712. Type: Register corruption - Adds constraint checks to one driver to prevent accessing the wrong registers. commit fb87ce81acf9ef43131f95b083433ef7a59cc837 Author: Jeff Layton cifs: NULL out tcon, pSesInfo, and srvTcp pointers when chasing DFS referrals Type: Oops - Fixes an Oops by moving NULL pointer assingment into the retry path. Also seems to fix a memory leak at the same time. commit d9f2a1fd4ac85f9da09a9f2c528c7fe516140511 Author: Thomas Gleixner clockevents: Prevent clockevent_devices list corruption on cpu hotplug Type: List corruption - Add a pass to find unused clockevent entries per cpu. commit 143690aec7cabdd51023afeded23578196dec85a Author: Nicolas Ferre dma: at_hdmac: correct incompatible type for argument 1 of 'spin_lock_bh' Type: Locking/Race - Passing the lock by reference instead by value. Effectively there was no locking before. commit ffa1b73815908157e483c740055eee5d2a88a10d Author: Julia Lawall drivers/net/usb: Correct code taking the size of a pointer Type: Incorrect size - Incorrect use of sizeof would result in the wrong size of data set to controller registers. commit 43f4c46841f108cb109432bcb665081b30523b31 Author: Zhu Yi iwmc3200wifi: fix array out-of-boundary access Type: Memory corruption - Increases the size of an array which would otherwise be accessed outside its boundaries. ! Was voted out of stable by the author commit c822a4618b4161b53276e56f7abe7106907fbd02 Author: Daniel Mack Libertas: fix buffer overflow in lbs_get_essid() Type: Memory corruption - Removes adding terminating zero to the SSID buffer which is zeroed and might be completely used by the SSID bytes. commit 00daf36d1c9c5b45ada8deab168f7d958376bb98 Author: NeilBrown md: Fix unfortunate interaction with evms Type: Userspace interaction - Prevent a md device from disappearing after only minor actions done via ioctl. commit c3a26f90cc7395590543870c341e506772aef1fa Author: Bartlomiej Zolnierkiewicz pata_cmd64x: fix overclocking of UDMA0-2 modes Type: Data corruption - Use a define (which is a transfer mode) instead of the number (which is a mode number). Otherwise a too high DMA mode can be selected. commit 314e117ec20dd54907b5fe61c2abcb5ea01ba3de Author: Sergei Shtylyov pata_hpt3x2n: fix clock turnaround Type: ??? - Several fixes around the DPLL flag. It seems to control some delays which could potentially cause transfer errors. - Only one pata driver affected. commit 1d889c2b105815406a5ed9ed6e44a550315f8909 Author: Mike Christie SCSI: fc class: fix fc_transport_init error handling Type: Memory leak - Adds cleanup code to unregister those structures already registered when an error occurs. commit c7b2f6157d18bfe7eaa60878fee67d78408ddf51 Author: Clemens Ladisch sound: sgio2audio/pdaudiocf/usb-audio: initialize PCM buffer Type: Memory handling - Replace vmalloc calls by vmalloc*_user variants to avoid applications to get access to the same (re-used) buffer. commit 5df5fc452ce04fb7751924bf1e9f309ac4432076 Author: Clemens Ladisch USB: emi62: fix crash when trying to load EMI 6|2 firmware Type: Oops - Replace end-of-loop condition to use the correctfinal record indication. commit c6d183251eb74af9d64b8d4b3c8b6d2b1ae31dd0 Author: pancho horrillo USB: Fix a bug on appledisplay.c regarding signedness Type: Enablement - Changes two variables from char to u8 to prevent failure on architectures where char is signed by default. commit 83166a0ed52759b1e044cfb93919201500d63e35 Author: Sergei Shtylyov USB: musb: gadget_ep0: avoid SetupEnd interrupt Type: Error handling - Sets the dataend bit unconditionally when a zero length setup request is detected. This avoids an error message. - Also fixes the missing newline on that error message while on it. - Only affects a single gadget driver. commit db44db2a7fd991eaccff8769e18e12af57d12d02 Author: Donny Kurnia USB: option: support hi speed for modem Haier CE100 Type: Enablement - Adds a new ID to the option driver. commit 0a43db04a3a71691421e0ba03e0dff2d750d1a2e Author: Suresh Siddha x86, cpuid: Add "volatile" to asm in native_cpuid() Type: Oops - Adds volatile to the cpuid assembly to indcate that the values returned can change by other means. Prevent gcc from optimizing away calls. commit 7e3b7a6899f2dc363069583ab797b8405d2f29ca Author: Roger Oksanen e100: Use pci pool to work around GFP_ATOMIC order 5 memory allocation failure Type: Error handling(???) - Work around the fact that pci_alloc_consistent() implicitely uses GFP_ATOMIC, so larger allocations are likely to fail. The patch moves to pci_pool_alloc, which takes a GFP argument. ! Requires the next patch as it misses the hidden fact that this change causes non-zeroed memory to be returned. commit c09fa0a066dad9311e652a824e6f1dc076cc5a94 Author: Roger Oksanen e100: Fix broken cbs accounting due to missing memset. Type: Fixup - Adds a memset to make sure memory allocated by pci_pool_alloc is correctly set to zero. commit 301c49072d4bc6949fbec2d55c5053aea5871e2d Author: Martin Decky hostap: Revert a toxic part of the conversion to net_device_ops Type: Regression - Fixes a change that was accidentally made in 2.6.30 which caused the queue length to be set the opposite way than before, which results in dropping all packets in AP mode. commit e3689fd2c6dc80054c334a5cd015b3f286b81bba Author: Roel Kluin hwmon: (fschmd) Fix check on unsigned in watchdog_write() Type: Error handling - Changes variable from size_t (which is unsigned) to int. Maybe a change to ssize_t (which is signed) would have been better as this is the type of the return value. But it should work. commit 140605c9fff2ab8c5ffa18496924559162af771e Author: Jonathan Cameron hwmon: (sht15) Off-by-one error in array index + incorrect constants Type: ??? - Fixes some calculations and constants and also preventing to exit a loop too soon for one hwmon driver. commit a3642c520693b17cf92d09d04f740c058e5d9291 Author: Michele Jr De Candia i2c/tsl2550: Fix lux value in extended mode Type: Correctness - Multiplies a value by 5 if in a certain mode. - Only affects this one driver when in that one mode. commit eae7b4a9bc0242596ca8c62f53f59e6970dfa9fc Author: Patrick McHardy ipv6: reassembly: use seperate reassembly queues for conntrack and local delivery Type: ??? (what is the result of packet jumps?) - ABI bumper - Adds a user field and seems to use this in the fragment matching code. - Said to be following ipv4 code. commit 4478b262db6e3f96e0e1b5ae8c521d9d4357e60a Author: Stefan Weinhuber S390: dasd: support DIAG access for read-only devices Type: Enablement - Only affects s/390 - Allow disks to be accessed when read-only instead of failing on them. commit 7d7f1223bb4fe5c1373d035f30b7bda15c54d3eb Author: Jan Kara udf: Try harder when looking for VAT inode Type: Enablement - Adds a work-around for bad udf filesystems looking for a valid inode a little before the expected position. Invalid inodes (the error case) look like causing udf_iget to do the unlocking correctly. commit 486618463cd58da0dc664c0fab2c699c38492a0e Author: Dan Carpenter V4L/DVB (13596): ov511.c typo: lock => unlock Type: Locking - Looks like an obvious copy paste error. Replaces lock on the way out with unlock. commit d1c146e569000d129841de948e90c3f03de79fb3 Author: Linus Torvalds x86/ptrace: make genregs[32]_get/set more robust Type: Enhancement - Is not addressing a real bug but replaces the loop exit condition in several places to ensure it cannot be missed. Changes should not change anything as long as the callers were correct. commit 4a20908b40ce95b002e2edd8201f1d74d218a553 Author: Jan Rekorajski XFS bug in log recover with quota (bugzilla id 855) Type: Data corruption - Sounds like causing a revocery to fail in some cases. Replaces the structure used to derive the expected buffer size by the correct one (which is smaller). commit 2c33e079948848749cb9ef94f003811ec88cca54 Author: Serge E. Hallyn generic_permission: MAY_OPEN is not write access Type: Regression - Sounds like masking away MAY_OPEN restores a previous behavior. commit 6fdf43071b721ae9cca0eb4386d25ad84d24a380 Author: Gertjan van Wingerde rt2x00: Disable powersaving for rt61pci and rt2800pci. Type: Regression - Opposed to description only changes rt61pci. - Disables powersaving by default as this turned out to cause problems. ! Was quickly reverted by a 2.6.31.11 as it seemed to have caused build failures in some cases. commit df91c1b9d6e126ea8ccb62741416e735fdd014d3 Author: Daisuke Nishimura memcg: avoid oom-killing innocent task in case of use_hierarchy Type: Error handling - The checking for which process to kill in OOM situations seemd to be wrong. - Patch makes the changes as described. commit b97fe86bfa12b70a3efab1af0b2cd82ac089f52f Author: Moiseev Vladimir Input: atkbd - add force relese key quirk for Samsung R59P/R60P/R61P Type: Enablement - BugLink: http://bugs.launchpad.net/bugs/253874 - Adds key release quirk. commit a92a695e8052bb9b97f87c4f11c0e2f99775a1f0 Author: Dmitry Monakhov Add unlocked version of inode_add_bytes() function Type: Enhancement - In effect adds an unlocked variant of the inode_add_bytes function. - No operational change at this point. commit 2aa6532583f77a16696a9df2147cad7d74679c76 Author: Dmitry Monakhov quota: decouple fs reserved space from quota reservation Type: Enablement - A rather big hunk in generic code. However seems to be only preparing a new interface. The unifying of functions should be without immediate effect. ! Was bisected to introduce a regression with ext3 -> quota: fix reserved space management for ordinary fs commit a91b0b1bc83c1b2a9c457bec38cfc5246c9e0ef1 Author: Dmitry Monakhov ext4: Convert to generic reserved quota's space management. Type: Race - By using the new quota interface added by the previous two patches this is supposed to fix some write vs. chown race. commit c788e1f8506cf828d6479cdcf8c28a4cc3043d83 Author: Dmitry Monakhov ext4: Fix potential quota deadlock Type: Locking ! Looks like this will get dropped for now as it requires additional patches and more testing. ! Prereq: 0637c6f4135f592f094207c7c21e7c0fc5557834 ee5f4d9cdf32fd99172d11665c592a288c2b1ff4 9d0be50230b333005635967f7ecd4897dbfd181b commit 3a9ee46a6c996eae9c50b7bf450b6c6aad52c129 Author: Dmitry Monakhov ext4: fix sleep inside spinlock issue with quota and dealloc (#14739) Type: Locking - Drop a lock during one section of code and re-aquire it later as the code might sleep in between.