Linux 2.5.71

From: Linus Torvalds (
Date: Sat Jun 14 2003 - 16:17:32 EST

I think I'll call this kernel the "sticky turtle", in honor of that
historic "greased weasel" kernel, and as a comment on how sadly dependent
I've become on the daily BK snapshots. It's been too long since 2.5.70.

I'll do better, I promise. While most developers happily use either the
daily snapshots or the BK tree itself, not everybody does, and a lot of
users want "real releases".

There's nothing hugely interesting here, but Al Viro ha sbeen cleaning up
the tty layer, and Stephen Hemminger has been fixing up some network
device alloc/free issues with the help of various people.

And obviously there are the normal usb/pcmcia/sound/architecture updates.
With driver models and networking thrown in as a bonus.



Summary of changes from v2.5.70 to v2.5.71 ============================================

Adrian Bunk: o kbuild: [PATCH] document modules_install in "make help" o [wan lmc] remove 2.0.x-era code o SECURITY_ROOTPLUG must depend on USB

Alan Cox: o [netdrvr tlan] fix 64-bit issues

Alan Iwi: o Missing magic number

Alan Stern: o USB: fix address assignment after device reset o USB: Don't allocate transfer buffers on the stack in hub.c o USB: Rename static functions in hub.c and increase timeouts o USB: Make hub.c DMA-aware

Alexander Viro: o callout removal: ircomm_tty o tty_driver refcounting o Fix sound lockup - missing chardev init o procfs bug exposed by cdev changes o Fix disk partitioning with multiple IDE disks o >minor_shift removal o [NET]: Eliminate {init,register,unregister}_fcdev o [NET]: Eliminate init_hippi_dev and {un,}register_hipdev o [NET]: Convert most tokenring drivers away from {init,register,unregister}_trdev, only ibmtr remains o [NET]: Eliminate init_fddidev o [NET]: Move sk98lin driver away from init_etherdev() o [NET]: Move 3c509 driver away from init_etherdev() o [NET]: Move sunqe.c driver away from init_etherdev() o [NET]: Move bmac.c away from init_etherdev() o [NET]: Convert USB drivers away from init_etherdev()

Alexey Kuznetsov: o tcp_output.c, tcp.c, tcp.h

Andi Kleen: o Work around gcc 3.3 bug on amd64 in binfmt_elf.c o Make spinlock debugging compile on x86-64 o x86-64 merge

Andreas Schultz: o I2C: fix unsafe usage of list_for_each in i2c-core

Andrew Morton: o truncate and timestamps o fix typo in coda o Fix suspend with pccardd running o fix oops on resume from apm bios initiated suspend o export mmu_cr4_features to modules o [VISWS] irqreturn_t conversion o Fix CONFIG_PROCFS=n o zoran user-pointer fix o irq balance logic fix o kill lock_kernel() in inode_setattr() o i2o memleak comment o write_one_page() fixlets o speed up the unlink speedup o Remove unneeded fcntl check o unregister_netdev cleanups o support 64 bit pci_alloc_consistent o svcsock use-after-free fix o Fix writev when a segment generates EFAULT o Fixes trivial error in o [NET]: Convert rtnl_lock/register_netdevice/rtnl_unlock to register_netdev o fix generic_file_write() o magazine layer for slab o Additional fields in slabinfo o initialise vma->vm_next in do_mmap_pgoff() o More irq balance fixes o dirty_writeback_centisecs fixes o Copy nanosecond stat values for i386 o Some subarch warning fixes o MTD build fix o improved core support for time-interpolation o ext3: fix for blocksize < PAGE_CACHE_SIZE o /proc/kcore fixes o remove 16-bit pid assumption from ipc/sem.c o fix broken networking o [NET]: Fix broken cleanups in net/core/iovec.c o aio: small cleanups o clean up timer interpolation code o radio-cadet.c: remove unnecessary copy_to_user() o cmpci: fix improper access to userspace o zr36120: fix improper access to userspace o remove unsafe BUG_ON() o pnpbios dereferencing user pointer o fix bw-qcam.c bad copy_to_user o Graceful failure in devfs_remove() o Fix generic_file_write() again o reiserfs option parser fix and ability to pass o reiserfs support for blocksizes other than 4096 bytes o hugetlbfs: mount options and permissions o DEADLINE: hash removal fix o ext3: fix deadlock in journal_create() o kmalloc_percpu: interface change o per-cpu support inside modules (minimal) o IRQs: handle bad return values from handlers o IRQs: fix up irq_desc initialisation for non-ia32 o force_successful_syscall_return() o fix wobbly /proc/stat:btime o Console blanking fix o Console privacy for braille users o Fix tty devfs mess o misc fixes o cs423x fixes o remove get_current_user() o remove triggerable BUG() from de_thread o Don't let processes be scheduled on CPU-less nodes (1-3) o DAC960 fix for fibre channel transfer rate o /proc/sys/vm/min_free_kbytes o loop: remove the balance_dirty_pages() call o Fix build for CONFIG_KALLSYMS=n o ppc64: fixup for family/sk_family rename o Fix the build with !CONFIG_PROC_FS o common 32-bit ioctl code o ioctl32 cleanup: sparc64 o x86_64: use common ioctl code o remove_proc_entry() fix o JFFS_PROC_FS must depend on JFFS_FS o fix apic handling for NUMA-Q o cleanup conditionals in summit subarch o provide bus to node mapping for Summit o rocket.c: devfs fix o add bootmem failure warning o eventpoll: fix possible use-after-free o Remove DRM ioctls for common compat ioctl code o fix possible busywait in rtc_read() o fix discontig with 0-sized nodes o fix TARGET_CPUS inconsistency o update MAINTAINERS for Compaq drivers o optimize fixed-sized kmalloc calls o fix scheduler bug not passing idle o fix numa meminfo o fix oops in driver/serial/core.c o fix hangs with nfs to localhost o devfs_mk_dir() fix o cpqarray.c: fix stack usage o dirty bit clearing on s390 o drivers/char/mem.c cleanup o irq_cpustat cleanup o MAINTAINERS: Compaq->HP o loop: file use highmem o loop: make bio_copy private to loop o loop: loop bio renaming o loop: copy bio not data o loop: remove an IV o loop: remove LO_FLAGS_BH_REMAP o loop: remove blk_queue_bounce o loop: copy_bio use highmem o loop: don't lose PF_MEMDIE o tmpfs: shmem_file_write EFAULT o tmpfs: swapoff-truncate race o tmpfs: misc fixes o cleanup seqfile usage in resource.c o x25 facilities parsing fix o eicon usercopy fix o intermezzo symlink fix o mdc800 usercopy fix o mpu401 usercopy fix o emu10k1 memleak fix o rio memleak fix o fix resource leak in i810 driver o tioclinux() numbers in <linux/tiocl.h> o fix writeback for dirty ramdisk blockdev inodes

Andries E. Brouwer: o isa_writeb args interchanged o scsi: ten -> use_10_for_rw / use_10_for_ms

Andy Grover: o ACPI: fix extra semicolon (Pavel Machek) o ACPI: Trivial name init patch (Bjorn Helgaas) o ACPI: Disable ACPI sleep on SMP systems (Pavel Machek) o ACPI: Add access checking (Andi Kleen) o ACPI: Add ACPI PCI Subdriver. This enables acpiphp to work again. (Matthew Wilcox) o ACPI: Implement PCI Domain support (Matthew Wilcox)

Anton Blanchard: o ppc64: comment fix from Milton Miller o ppc64: Fix problem creating zImage/zImage.initrd multiples times from Bryan Logan o ppc64: clean up SLB reload code and remove some unnecessary isyncs o ppc64: fix compile warnings o ppc64: fix misreporting of unhandled IRQs in xics IPI o ppc64: cleanup some hardcoded constants o ppc64: replace MAX_PACAS with NR_CPUS o ppc64: Fix a bad shift against PCI_BASE_CLASS_BRIDGE from Will Schmidt o ppc64: fix for boot cpu > 31 o ppc64: Report cpus > 32 in proc/cpuinfo correctly o ppc64: merge conflicts o compat_wait4 fix o ppc64: Always pass non segment faults on the 0xc region up to do_page_fault o ppc64: Add some branch prediction o ppc64: remove ioperm o ppc64: threaded coredump support o ppc64: fix copy_from_user leak, from Milton Miller o ppc64: Add warning for unhandled irqs o ppc64: remove sys32.S o ppc64: ppc64 update o ppc64: copy_tofrom_user exception handling fix from Paul Mackerras o ppc64: kcore support o ppc64: Increase maximum allocation size to 16MB for largepage support o ppc64: FORCE_MAX_ZONEORDER is actually order + 1, from Dave Gibson o ppc64: fix compile error introduced in threaded coredump patch o fix matroxfb compile on ppc64] o ppc64: Call setup_kcore later in boot, we call kmalloc from it o ppc64: cputable support from Will Schmidt o ppc64: use an initcall to register ras irqs o update ppc64 MAINTAINERS entry o ppc64: forgot to add this in the cputable merge o ppc64: Fix overallocation of NUMA bootmem bitmap and fix for NUMA kernels on non NUMA boxes o ppc64: Fix for multiple zone 0 regions on many ppc64 NUMA boxes o ppc64: Fix ppc64 build o ppc64: Update BUG handling based on ppc32 o ppc64: use common 32bit ioctl code o ppc64: K&R to ANSI style conversions from Steven Cole o ppc64: copy_tofrom_user fix from Paul Mackerras o ppc64: various fixes to sys32_sysinfo, from Will Schmidt o ppc64: Rework inline syscall macros, fix clobbers & gcc3.3 from Franz Sirl o ppc64: rework user access functions

Arnaldo Carvalho de Melo: o wanrouter: fix bug introduced by latest namespace fix o net: abstract access to struct sock ->flags o USB drivers: initialize struct usb_driver ->owner field o drivers/usb/media/ibmcam: remove MOD_{INC,DEC}_USE_COUNT o drivers/usb/media/konicawc: remove MOD_{DEC,INC}_USE_COUNT o drivers/usb/media/ultracam: remove MOD_{INC,DEC}_USE_COUNT o n_hdlc: CodingStyle cleanups and removal of old stuff o net: reduce the data dependency of struct sock/struct tcp_tw_bucket o net: create struct sock_common and use in struct sock & tcp_tw_bucket o list.h: improve hlist o udpv6: use the right struct sock when testing if it is PF_INET6 family

Art Haas: o C99 patches for fs/ o C99 struct initializers for kernel files

Bart De Schuymer: o [BRIDGE]: Remove unnecessary code in br_input o [NETFILTER]: Fix ARPT_INV_MASK in arp_tables.h

Bartlomiej Zolnierkiewicz: o allow "hdX=scsi" for modular scsi/ide-scsi o kill "hdX=noremap" o fix two IDE list_head problems o create /proc/ide/hdX/capacity only once o kill recreate_proc_ide_device() o switch ide to taskfile IO

Ben Collins: o Update IEEE1394 (r939) o USB Multi-input quirk o Update IEEE1394 (r946) o [SPARC64]: Final image strip not to strip too much o USB: fix keyboard leds o Register scsi devices after naming them o Update IEEE1394 (r952)

Benjamin Herrenschmidt: o [SUNGEM]: Fix gcc3.3 warnings o Nuke check_highmem_ptes()

Bruce D. Elliott: o [SPARC64]: Fix transmit handling in sunsab.c serial driver

Bruce Fields: o gss_marshal and gss_validate depend on gss_cred_get_ctx never returning NULL; but gss_refresh depends on gss_cred_get_ctx returning NULL whenever the cred is not up to date. So, I replaced the single gss_cred_get_ctx by a gss_cred_get_ctx and a gss_cred_get_uptodate_ctx. o add handling of the new CTXPROBLEM and CREDPROBLEM RPCSEC_GSS errors o Add a "protocol: udp/tcp" line so that gssd can use the same protocol for null calls that was specified in the mount options. o I believe we need to set a timeout before doing the sleep in gss-upcall o allow gssd to communicate failure to initialize contexts back to the kernel, so the kernel can return -EACCES when a user lacks credentials, instead of just hanging until they kinit. o This makes several changes to the gss upcalls o Trivial; I kept forgetting what each of the xdr_netobj's passed to the gss-api routine meant, so I thought I'd fool with the argument names in an effort to make them more helpful. o Trivial fix for a typo in fs/nfs/nfs4state.c

Chas Williams: o [ATM]: lane and mpoa module refcounting and locking cleanup o [ATM]: HE driver coding style conformance o [ATM]: HE driver misc irq handler cleanups o [ATM]: Move rategrid off stack in HE driver o [ATM]: more cleanup in HE driver

Chris Wright: o lsm: Early init for security modules (1/4) o lsm: Remove task_kmod_set_label hook (2/4) o lsm: Remove inode_permission_lite hook (3/4) o lsm: setfsuid/setgsuid bug fix (4/4)

Christoph Hellwig: o acpi serial stuff o kill register_pccard_driver o give ->proc_info a struct Scsi_Host * parameter o PPC32: Syscall cleanups o [NET]: Switch lanmedia driver to initcalls o [NET]: Remove sdla from setup.c o [SPARC64]: Kill sys_aplib o use second arg to scsi_add_host in usb storage o fix scsi_register_host abuse in usb scanner drivers o driver model for scsi upper drivers, take 2 o LDM-based scsi host lookup o switch /proc/scsi/scsi to seq_file and proper device model o kill scsi_host_get_next o kill remaining direct uses of scsi_register_host o fix signal.h compilation on PPC o [NET]: Move dmascc away from setup.c o [NET]: Fix non-modular sdla.c build o [NET]: Kill useless/wrong Version line from net/core/dv.c o [NET]: net/core/dst.c typo o [NET]: Fix coding style in net/core/filter.c o [NET]: Fix coding style in net/core/iovec.c o [NET]: Convert ppc early-init drivers to initcalls o [NET]: Fix accidental revert of init_etherdev killing in PPC net drivers o [NET]: Convert skfp over to initcalls, kill fddi cruft from Space.c

Daniele Bellucci: o USB: replaced BKL in rio500.c

Dave Engebretsen: o [netdrvr pcnet32] bug fixes

Dave Jones: o AM53C974 request region o [AGPGART] Compilation fix o [AGPGART] Remove useless early agp_init() from i810fb agp_init() just printk's a banner. This is unnecessary at this early stage. o [AGPGART] Yet another missed typedef compile fix o [AGPGART] Report fixing of errata, and add missing printk stuff (\n's, KERN_INFO, PFX) o [AGPGART] Kill useless printk in frontend o [AGPGART] Missing printk levels o [CPUFREQ] Make powernow-k7 leap big buildings^Wranges o [CPUFREQ] kill cpufreq_driver export o [CPUFREQ] Kill unused variables o [CPUFREQ] CodingStyle fixes o [CPUFREQ] Fix ACPI P-State driver o [CPUFREQ] missing export powernow-k7 needs dmi_broken o [AGPGART] Add webpage link o [CPUFREQ] Fix documentation filename o [CPUFREQ] Add Athlon to list of supported cpufreq drivers o [CPUFREQ] sysfs moved some files around. update documentation to reflect reality o Cset exclude:|ChangeSet|20030611121150|30244 o [CPUFREQ] Merge Jeremy's Centrino speedstep driver o [CPUFREQ] Move old speedstep driver to speedstep-ich o [AGPGART] Some Intel chipsets were using the wrong masks o [CPUFREQ] speedstep docu clarification

Dave Kleikamp: o Update JFS team members in jfs.txt o JFS: i_acl & i_default_acl are not being re-initialized o JFS: resize fixes o JFS: add back read_inode super_operation

David Brownell: o USB: ethernet "gadget", rx overflows happen o USB: kerneldoc for gadget API o PCI: pci pool, poison more like slab code o USB: usb/core/devio: identify process o USB: net2280 patch: control-out fix, minor cleanups o USB: ohci-hcd, remove FIXME

David Gibson: o Update orinoco driver to 0.13e

David Mosberger: o fix TCP roundtrip time update code o allow thread_info to be allocated as part of task_struct o [TG3]: Workaround 4g DMA bug more portably

David S. Miller: o [NET]: One too many IRQ_HANDLED added to sunqe.c driver o [IPV4/IPV6]: Use Jenkins hash for fragment reassembly handling o [IPV6]: Input full addresses into TCP_SYNQ hash function o [IPV4]: Add sysctl to control ipfrag_secret_interval o [IPV6]: Fix typo in defragmentation changes o [TCP]: Do not access inet_sk() of a time-wait bucket o [SPARC64]: Fix probe error handling in envctrl.c driver o [SPARC64]: Fix probe error handling in bbc_{envctrl,i2c}.c driver o [SPARC64]: Do not export {un,}register_ioctl32_converstion twice o [NET}: Fix typo in sock_set_flag changes o [ATM]: Fix driver Makefile clean-files o [SPARC64]: Fix sys_shmat handling for 64-bit binaries o [XFRM]: Handle use_time expiration properly o [SOUND]: Revert buggy ALSA merge ioctl32 changes o [SOUND]: vx/vx_core.c needs linux/init.h o [SOUND]: Fix sparc cs4231 driver build due to pcm list changes o [SPARC64]: Update defconfig o [NET]: Kill PTI fddi driver, never completed and no plans to finish o [NET]: Fix CONFIG_HIPPI build o [NET]: Fix typos in init_trdev changes to lanstreamer.c o [NET]: Eliminate {init,register,unregister}_trdev() o [NET]: Eliminate init_etherdev usage from arch/um drivers o [IPV6]: In ipv6_add_dev dont call __ipv6_regen_rndid without initial reference held o [NET]: Use INIT_LIST_HEAD in arch/um/drivers/net_kern.c o [NET]: Typo in iph5527.c driver changes o [NET]: Use INIT_LIST_HEAD correctly in arch/um/drivers/net_kern.c o [IPSEC]: {esp,ah} --> {esp4,ah4} o [NET]: Move arch/cris drivers away from init_etherdev() o [NET]: Convert ia64 simeth.c away from init_etherdev() o [NET]: Convert PPC 8260_io/enet.c away from init_etherdev() o [NET]: Convert PPC 8260_io/fcc_enet.c away from init_etherdev() o [NET]: Convert PPC 8xx_io/enet.c away from init_etherdev() o [NET]: Convert PPC 8xx_io/fec.c away from init_etherdev() o [NET]: Actually apply Al's sunqe.c changes o [NET]: Fix thinkos in PPC 8260_io/fcc_enet.c changes o [PCI]: Move pci_remove_bus_device back to include/linux/pci.h, discussed with o [NET]: Kill drivers/net/setup.c, it no longer does anything o [IPSEC]: Implement xfrm type module autoloading o [NET]: Some stuff missed during acme's struct sock cleanup o [NET]: Missing __KERNEL__ ifdefs in linux/{tcp,udp}.h o [SCTP]: Kill unused local variable in init_sctp_mibs o [NET]: Process hotplug list in FIFO order o [NET]: Fix typo in neigh_sysctl_unregister changes o [LLC]: Fix typing error in procfs code

David Stevens: o [IGMP]: Make sock_alloc_send_skb calls non-blocking

David van Hoose: o Fix compilation errors in ppa and imm modules

David Woodhouse: o MTD and JFFS2 update o Final cleanups for MTD merge o Fix some accidental regressions which slipped in with the MTD merge

Davide Libenzi: o epoll race fix

Deepak Saxena: o [ARM PATCH] 1537/1: big-endian support for do_div64 o [ARM PATCH] 1538/1: arch/arm/Makefile and KConfig Big-Endian changes

Dominik Brodowski: o [PCMCIA] Move socket_info_t o [PCMCIA] Move get_socket_info_by_nr o [PCMCIA] Remove socket_table o [PCMCIA] True driver module locking o [PCMCIA] split_init o [PCMCIA] register o [PCMCIA] Callbacks use pcmcia_socket not integer o [PCMCIA] socket reference in client_t o [PCMCIA] Replace more socket numbers with pcmcia_socket o [PCMCIA] Make ds.c use pcmcia_socket->sock rather than local version o [PCMCIA] unify yenta.c and pci_socket.c o [PCMCIA] Remove socket_info_t o [PCMCIA] i82365 depends on ISA o [PCMCIA] rename i82365.c socket_info_t o [PCMCIA] Rename tcic.c socket_info_t

Douglas Gilbert: o scsi_mid_low_api.txt in lk 2.5.69 o sg driver for lk 2.5.70

Eddie Williams: o scsi: allow devices to restrict start on add

Edward Peng: o [netdrvr sundance] fix flow control bug o [netdrvr sundance] fix another flow control bug

Frank Cusack: o nfs_unlink() problem fix

Geert Uytterhoeven: o [NET]: asm/smp.h --> linux/smp.h in sch_ingress.c

Greg Kroah-Hartman: o USB: fix up unusual_devs.h merge mess o USB: remove some old references to /proc/bus/usb/drivers o Cset exclude:|ChangeSet|20030529215347|05329 o USB: add usb_find_device() function to USB core o Root plug: remove USB bus walking functions, now use usb_find_device() o USB: added .owner to kobil_sct driver o PCI: make pci_setup_device(), pci_alloc_primary_bus() and pci_alloc_primary_bus_parented() static o PCI: make pools_lock and pci_lock static o ACPI PCI Hotplug: remove hand made pci_find_bus function o IBM PCI hotplug: remove hand made pci_find_bus function o IBM PCI hotplug: remove direct access of pci_devices variable o PCI Hotplug: move drivers/hotplug/* to drivers/pci/hotplug/* o PCI: Remove a lot of PCI core only functions from include/linux/pci.h o PCI: remove CONFIG_PROC_FS checks in .c files o PCI: Move more functions out of include/linux/pci.h that don't need to be there o PCI: Grab reference count on pci_dev if the pci driver binds to the device o PCI: remove usage of pci_for_each_dev() in drivers o PCI: finally remove pci_for_each_dev() now that all users of it are gone o PCI: remove usage of pci_for_each_dev() in arch/ppc64/kernel/pci.c o PCI: remove direct access of pci_devices from arch/m68k/atari/hades-pci.c o PCI: remove direct access of pci_devices from drivers/macintosh/via-pmu68k.c o PCI: fix up previous fusion driver pci changes o PCI: add pci_find_device_reverse() for users of pci_find_each_dev_reverse() to use o PCI: remove usage of pci_for_each_dev_reverse() in o PCI: remove pci_for_each_dev_reverse() now that all users of it are gone o PCI: move pci_present() into drivers/pci/search.c o PCI: remove EXPORT_SYMBOL(pci_devices) o I2C: sync i2c-id.h with cvs version o TTY: add a release function for tty_class devices o TTY: release function should be set in the class, not the class_device o PCI: add pci_find_next_bus() function to prevent people from walking pci bus lists themselves o PCI: remove pci_for_each_bus() usage from arch/ia64/hp/common/sba_iommu.c o PCI: remove pci_present(), pci_for_each_bus() and pci_bus_b() usage o PCI: sparse fixups for drivers/pci/proc.c o USB: fix problem found by sparse in usb.h o USB: sparse fixups for drivers/usb/core/devices.c o USB: sparse fixups for drivers/usb/core/inode.c o USB: lots of sparse fixups for usbfs o I2C: coding style updates for i2c-iop3xx driver o I2C: fix some errors found by sparse in include/linux/i2c.h o I2C: fix up sparse warnings in drivers/i2c/i2c-core.c o I2C: fix up sparse warnings in the i2c-dev driver

Greg Ungerer: o create m68knommu/coldfire specific ints.c o remove common m68knommu ints.c o don't compile m68knommu/kernel ints.c o compile m68knommu/ColdFire ints.c o fix calls to do_fork() o remove obsolete BLKMEM driver reference o cleanup is_in_rom() checker o fix broken trace flag check in 68328 system call entry o security init call support in linker script o conditional ROMfs copy for ARNEWSH/5206 setup o support BOOTPARAM's on m68knommu/5206 targets o support BOOTPARAM's on m68knommu/5206e targets o m68knommu/pilot startup copy init segment to RAM

Hartmut Wahl: o USB: Patch for Samsung Digimax 410

Henning Meier-Geinitz: o USB: new vendor/product ids for scanner driver

Herbert Xu: o [IPSEC]: Order SPD using priority o [NET]: Missing refcount bump in flow cache o [XFRM]: u64 --> __u64 in linux/xfrm.h o [XFRM_USER]: Fix xfrm_state_lookup args in xfrm_add_sa o [XFRM_USER]: Rename confusing member of struct xfrm_usersa_id o [XFRM]: Too many reference drops of delpol in xfrm_policy_insert o Fix ide-mod unload crash o [IPSEC]: Include linux/slab.h where necessary o [XFRM_USER]: Allow del policy by id and get policy by selector o [IPSEC]: Zap killed policies from the flow cache properly o [IPSEC]: Fix preempt race in flow_flush_cache o [IPSEC]: Kill object argument from flow_cache_flush o [IPSEC]: Proper percpu handling in flow cache o [IPSEC]: Initialize flow key properly in decode_session

Hideaki Yoshifuji: o [IPV6]: Convert /proc/net/if_inet6 to seq_file o [IPV6]: Fix order of destruction of procfs o [IPV6]: Make procfs destructors return void o [CRYPTO]: Fix compiler warnings in sha512.c o [IPV6]: Fix possible idev leakage in icmp.c o [IPV6]: Fix possible oops in ndisc_send_na o [IPV6]: Clean up ip6_dst_alloc() calls o [IPV6]: Always remove fragment header o [IPV6]: Fix possible dst leakage in ndisc_send_redirect o [IPV6]: Fix default router selection in some cases o [IPV6]: Add ip6frag sysctls o [NET]: Add ip-sysctl.txt entries for missing ip{,6}frag_* sysctls o [IPV6]: Set dead flag on idev if snmp6_register_dev() fails o [IPV6]: Fix several errors in udpv6_connect() o [IPV6]: typo, unrequired #undef and bad operator precedence o [netdrvr] C99 initializers for arcnet o [IPV6]: dev_get_by_name("lo") --> dev_hold(&loopback_dev) o [IPV6]: ipv6_addr_prefix() cleanup, eliminate duplication o [NET]: Make neigh_parms setup/teardown handling symmetric o [IPV6]: Fix payload length of reassembled packet o [IPV6]: Use sizeof(struct frag_hdr) instead of magic value

Hirofumi Ogawa: o Adds the large partition (> 128GB) support to FAT (1/5) o Fix VFAT_IOCTL_READDIR_BOTH/_SHORT ioctl (2/5) o Remove Documentation/filesystems/fat_cvf.txt (3/5) o FAT cluster chain cache per superblock (4/5) o FAT cluster chain cache per inode (5/5)

Hollis Blanchard: o awe_wave.c user pointer dereference

Ian Molton: o ARM26 architecture

Ivan Kokshaysky: o alpha: compile warning fix o alpha: fix panic on smp boot (fork_by_hand) o alpha: typo in EISA bridge detection o alpha: single-step breakpoints - updated fix o PCI domains warning

James Bottomley: o Fix up proc_info conversion in 53c700 o Fix while in spinup loop of sd o SCSI: Make sysfs attributes mutable o Use of the new attribute modifiers on the 53c700 o Fix up 53c700 compile o scsi sysfs add attribute release function o qla1280: convert the driver to be endian neutral o qla1280: set the data direction correctly o qla1280: convert the driver to the new SCSI error handler o Fix __exit routine of NCR_D700 o fix character subsystem initialisation

James Morris: o [CRYPTO]: Use "select" kconfig facility instead of fragile defaults

Jaroslav Kysela: o Several ALSA updates

Jeff Garzik: o [netdrvr eepro] update MODULE_AUTHOR per old-author request o [netdrvr tlan] cleanup o [netdrvr] s/init_etherdev/alloc_etherdev/ in code comments, in 8139too and pci-skeleton drivers. o [netdrvr 8139too] respond to "isn't this racy?" comment o [netdrvr r8169] use alloc_etherdev, pci_disable_device o [ROSE]: Kill kfree of net_device->name o Cset exclude:|ChangeSet|20030529205634|46794 o [netdrvr] gcc 3.3 cleanups o [netdrvr skge] add ULL modifier to 64-bit constant o [netdrvr] add MAINTAINERS entry for atmel wireless driver

Jeff Wiedemeier: o compile fix for agp_memory struct definitition change

Jens Axboe: o remove buggy BUG_ON in ide-cd o blk layer tag resize o copy the tag_map o ide-cd buglets o scsi_ioctl HZ fixes o ide-cd/scsi/block fixups for SG_IO o kill old stuff

Jim Houston: o preallocate signal queue resource - Posix timers

Joe Burks: o USB: vicam.c patch

Joe Thornber: o dm: Replace __HIGH() and __LOW() macros o dm: signed/unsigned audit o dm: new suspend/resume target methods o dm: Lift dm_div_up() o dm: Fix memory leak in dm_register_target() o dm: Remove some debug messages o dm: Remove an old FIXME

John Hawkes: o 2.5.70 remove smp_send_reschedule() cruft

John Levon: o OProfile: Export task->tgid in the buffer o OProfile: update Changes o OProfile: remove useless code o OProfile: fix init / exit routine

Jon Grimm: o [SCTP] Verify contents of SACK against underrun o [SCTP] Use slab cache for sctp_chunk & sctp_bind_bucket o [SCTP] Bug fix for bind_bucket leak & heartbeat error count o [SCTP] Use non-prefetch list walker for short list o [SCTP] CANT_STR_ASSOC could/should return error up to user. ( o [SCTP] Multiple causes may be embedded in an ERROR chunk ( o [SCTP] Don't use path thresholds determine the overall error thresh o [SCTP] Let T5 timer affect SHUTDOWN-PENDING associations o [SCTP] sctp_jitter() needs to protect against div by 0 o [SCTP] Various bakeoff fixes o [SCTP] Fix double free of chunk along unexepected INIT path o [SCTP] Incorrect WORD_ROUND on a network endian value o [SCTP] Special case the handling of HOST_NAME_ADDRESS parm o [SCTP] Hand merge bk pull conflict. sk->state now sk->sk_state

Justin T. Gibbs: o Aic6xxx and Aic79xx Driver Update o Bump aic79xx driver version to 1.3.9

Jörn Engel: o zlib cleanup: remove FAR macro o zlib cleanup: __32BIT__ and STDC o zlib cleanup: ZEXTERN o zlib cleanup: ZEXPORT o zlib cleanup: z_off_t o zlib cleanup: OF o zlib cleanup: C++ workarounds o zlib merge: turboc o zlib merge: inffast.c o Mark Compaq MAINTAINERS entries stale o zlib cleanup: local o zlib cleanup: Z_NULL removal o zlib cleanup: unnecessary cast removal o zlib merge: return code o zlib merge: avoid 8-bit window errors o zlib changes: memlevel o Fix typo in comment o adjust ppp to zlib change

Krishna Kumar: o [NET]: Initialize sysctl_table to NULL in neigh_parms_alloc

Kurt Robideau: o Rocketport driver against 2.5.70-bk7 o Rocket patch against 2.5.70-bk18

Linus Torvalds: o Remove a few zero-sized files, as noted by David Gibson o Make zlib_inflate look more like ANSI C code o zlib cleanup: final fixups o Remove half-deleted zero-sized sound file o Quick response to de-listing the Compaq FC/RAID controllers from the MAINTAINERS list. How they're HP, and maintained by Stephen Cameron. o Re-introduce debugging code in list handling, poisoning stale list pointers to give us a nice oops if somebody is doing something bad. o Fix __d_drop() to properly initialize the d_hash fields, so that __d_drop() can safely be done multiple times on a dentry without corrupting other hash entries. o Don't make the source checker default path be quite so hackish. o Avoid warning by using an inline function rather than a macro for the default "pci_domain_nr()" definition. The inline function will evaluate the argument. o Make chr_dev_init() happen after PCI init, but before low-level driver initializations o Fix rcu list poisoning - since another CPU may be traversing the list as we delete the entry, we can only poison the back pointer, not the traversal pointer (rcu traversal only ever walks forward). o Make d_move() be able to gracefully handle the case of the dentry already being unhashed on entry. o Remove strange and broken ACPI rule from serial Makefile o Linux 2.5.71

Manfred Spraul: o Prepare for page unmapper

Margit Schubert-While: o I2C: add LM85 driver

Mark Haverkamp: o megaraid driver fix for 2.5.70

Mark M. Hoffman: o I2C: fix oops w83781d during rmmod o I2C: more w83781d fixes

Martin Schlemmer: o ethertap: fix struct sock cleanup leftover

Matt Domsch: o dynids: use list_add_tail o dynids: free dynids on driver unload

Matt Porter: o PPC32: Fix a compile error on 4xx embedded PowerPC o PPC32: Resolve colliding includes in boot wrappers

Matthew Dharm: o USB: storage: abort and disconnect handling o USB: storage: collapse one-use functions o USB: usb-storage: fix typo o USB: usb-storage: timeouts and aborts o USB: usb-storage: usb_stor_control_msg() and stuff o USB: usb-storage: change result codes o USB: usb-storage: handle babble o USB: usb-storage: re-organize probe/disconnect o USB: usb-storage: remove dead code

Matthew Wilcox: o PCI: domain support for sysfs

Mike Anderson: o Call release on scsi legacy LLDD

Miles Bader: o Remove some unneeded register saving on the v850 o Include <linux/fs.h> in arch/v850/kernel/rte_cb_leds.c o Miscellaneous v850 whitespace and comment cleanups o Handle new do_fork return value on v850 o Add __KERNEL__ guard to nb85e_cache.h on v850 o Add leading underline to new linker-script symbols on the v850 o Whitespace and comment cleanups for v850 entry.S o Add v850 support for hardware single-step (via ptrace) o Update irq.c on v850 to use irqreturn_t o const-qualify memory arg in v850's __test_bit

Mitsuru Kanda: o [IPV6]: Fix esp6 extension headers handling

Moritz Mühlenhoff: o [CRYPTO]: Default CRYPTO and MD5 to y if IPV6_PRIVACY is enabled

Neil Brown: o md: Export bio_split_pool for md to use o md: Use new single page bio splitting for raid0 and linear o md: Handle concurrent failure of two drives in raid5 o md: Improve test for which raid1 device doesn't need to be written to o md: Fix simple off-by-one error in md driver o md: Get rid of vmalloc/vfree from raid0 o md: Always allow a half-built md array to be stopped o md: Improve raid0 mapping code to simplify and reduce mem usage o md: Remove dependancy on MD_SB_DISKS from multipath o md: Remove dependancy on MD_SB_DISKS from raid5 o md: Remove dependancy on MD_SB_DISKS from raid0 o md: Remove MD_SB_DISKS limits from raid1 o md: Remove dependance on MD_SB_DISKS in linear personality o md: Replace bdev_partition_name with calls to bdevname o Fix raid5 bug where wrong 'dev' is used o Fix raid1 handling of writing to multiple devices o Fix up freeing of kmalloc structures o Fix bug in /proc/mdstat o md - Zero out some kmalloced space in md driver

Nicolas Pitre: o [ARM PATCH] 1545/1: correct compiler flags for ARMv5TE/XScale o [ARM PATCH] 1540/2: fixes for gcc-3.3

Olaf Hering: o USB: incorrect ethtool -i driver name

Oliver Neukum: o fix irq handling for DC395 o improve Documentation for DC395 o 01-debug-cleanup.patch o major cleanup of the module code o USB: allocate memory for reset earlier o USB: return errors when disabling a port o USB: cut usb_set_config from hpusbscsi o USB: usb_set_configuration in empeg.c o USB: kill a compiler warning in hpusbscsi

Patrick McHardy: o [PPP] fix memory leak in ioctl error path

Patrick Mochel: o drver model: Add release method for class devices o sysfs: Fix binary file handling o [driver model] Clean up CPU unregistration o Driver Class: don't call put_device() when we never called get_device() o [driver model] Clean up class release handling o [kobject] Update Documentation and licenses o [driver model] Update copyrights and license statements o [driver model] fix comment in device.h o [fs] Remove kobject support for filesystems o [kobject] Remove kobj_lock and use lockless refcounting o [driver model] Add device_for_each_child iterator o [kobject] Add warning + back trace if kobject_get() is called with 0 refcount o [driver model] Remove extraneous class device release method o [driver model] Rewrite system device API o [list.h] Add list_for_each_entry_reverse o [kobject] Add set_kset_name o [driver model] Make sure that system devices are handled specially power-wise o [driver model] Convert to new system device API o [lapic] Convert to new system device API o [i8259] Convert to use new system device API o [nmi] Convert to use new system device API o [timer] Convert to use new system device API o [oprofile] Convert to use new system device API o [x86-64 i8259] Convert to use new system device API o [s390 xpram] Convert to use new system device API o [driver model] Create include/linux/sysdev.h and define sysdev_attribute o [driver model] Make sure right header is used for cpu.c o [memblk] Convert to use new system device API o [numa nodes] Convert to use new system device API o [driver model] Remove system device definitions from device.h o [apic] Use sysdev.h instead of device.h o [i8259] Use sysdev.h instead of device.h o [nmi] Use sysdev.h instead of device.h o [timer] Use sysdev.h instead of device.h o [oprofile] Use sysdev.h instead of device.h o [x86-64 i8259] Use sysdev.h instead of device.h o [s390 xpram] Use sysdev.h instead of device.h o [driver model] Add save() and restore() methods for system device drivers o [driver model] Don't Oops when registering global sysdev drivers o [cpu] Use sysdev.h instead of device.h and export cpu_sysdev_class o [mtrr] Add save()/restore() methods o [cpufreq] Convert to use new system device API o [sysfs] Get zeroed page for file read/write buffers o hand merge o [driver model] Actually implement sysdev_{create,remove}_file() o [driver model] Compile fixes for NUMA o [sa1100 irq] Convert to new system device API o [driver model] Turn off debugging by defualt for device iterations o [kobject] Don't specially order objects in lists based on parent o [driver model] Make sure we walk lists on shutdown in right order o [sysfs] Add __user tag to appropriate parameters o [driver model] Remove extraneous get_device() from class_device_add() o [driver model] Make sure system device drivers are added if registered late

Paul Fulghum: o tty_register_driver

Paul Mackerras: o PPC32: Fix preempt bugs identified by Milton Miller o PPC32: Discard the __ksymtab* sections when we are linking the boot wrapper o PPC32: Re-open I/O windows on PCI-PCI bridges, needed for some powermacs o PPC32: Better handling of program check exceptions on 4xx, patch from Kumar Gala o PPC32: Simplify the BUG() implementation for now (a better one is coming) o get rid of CONFIG_ALL_PPC o PPC32: Compile fix for array initialization from Christoph Hellwig o PPC32: Cleanups from Christoph Hellwig and Tom Rini o PPC32: Start adding __user to mark pointers from userspace o PPC32: Fix irq_desc initialization o PPC32: Fix various minor problems pointed out by Linus' check program o PPC32: Convert some K&R-style functions to ANSI-style. From Steven Cole o fix check warnings in drivers/macintosh o Fix check warnings in PPP code o Move BUG/BUG_ON/WARN_ON to asm headers o [PPP]: Fix PPP Deflate sequence number checking

Paul Mundt: o toplevel SH update o Move SH board-specific code around

Pavel Roskin: o Fix crash when unloading yenta_socket in Linux 2.5.69 o USB: name uninitialized in scanner.c

Peter Milne: o [ARM PATCH] 1546/1: iop321 additional Hardware PMMR defines o I2C: add New bus driver: XSCALE iop3xx

Petko Manolov: o USB: pegasus patch

Petr Vandrovec: o matroxfb update to new API

Randy Dunlap: o [NET]: add RFC references for Linux SNMP MIBs o [NET]: Typo corrections only o [IPV6]: Add IPv6 routing table statistic: fib_discarded_routes o [IPV6]: Fix spelling/typos

Reeja John: o [netdrvr amd8111e] interrupt coalescing, libmii, bug fixes o [netdrvr amd8111e] link against mii lib

René Scharfe: o hugetlbfs: fix error reporting in case of invalid mount o Some more stuff missed during the struct sock cleanup

Richard Henderson: o [ALPHA] Add posix timer and clock syscalls o [ALPHA] Add semtimedop syscall o [ALPHA] Implement bcopy o [ALPHA] Avoid warning in asm/unaligned.h o [ALPHA] Fix missed __ex_table to conversion to pc-relative relocs o [ALPHA] Streamline calls to __copy_user and __do_clear_user o [ALPHA] Fixup fallout from force_successful_syscall_return change

Roland McGrath: o User FIXMAP area simplification

Roman Zippel: o Remove old code and macros o Change P_ROOTMENU into a MENU_ROOT flag o add new keywords to parser o expression support o reverse dependency support o support for 'range' o add more warnings o front end updates o create configuration in the destination directory o update kconfig documentation o choice handling fixes o boolean symbol state fix o ignore attempts to change unchangable symbols

Russell King: o [ARM] Fix GCC3.3 build error o [ARM] Remove old 26-bit ARM keyboard drivers o [ARM] Declare mmu_gathers using DEFINE_PER_CPU o [ARM] Fix more missing irqreturn_t and remove a static no_action func o [ARM] Move dma_alloc_coherent() to consistent.c o [ARM] Convert platform devices to use platform_device o [ARM] Tidy up Integrator core support o [PCMCIA] Fix sa11xx_core.c oops when changing cpu frequency o [PCMCIA] Convert sa11xx platforms to use new class code o [PCMCIA] Update SA11xx PCMCIA support for recent changes o [PCMCIA] sa11xx driver now takes pcmcia_socket instead of int socket

Rusty Russell: o kallsyms in proc o Move cpu notifiers et al to cpu.h o Fix module load failure case o [NETFILTER]: Delete unnecessary skb_linearize() calls in iptables_{filter,mangle}.c o sched.c neatening and fixes

Sam Ravnborg: o kbuild: [PATCH] Remove duplicate definitions in o kbuild: Updated make help o kbuild: Silence kbuild with make V=0 o kbuild: CROSS_COMPILE and ARCH definitions o kbuild: Silence output with make 3.80 o kbuild: Utilise kbuild infrastructure for vsyscall o kbuild: Nice output when generating PCI device list o kbuild: Nice output when generating crc32 table o docbook/kernel-api: include files updated o docbook: Recognize sis900 functions o docbook: Warn about missing parameter definitions o docbook: Move definition of MODULENAME_SIZE o kbuild: Kill use of do_cmd in drivers/char/Makefile o kbuild/i386: Add missing dependency in kernel/Makefile o kbuild: Enable modules to be build using the "make dir/" syntax o be more flexible about creating library archives o all archs: Replace O_TARGET with lib-y o kbuild: Document newly added lib-y o kbuild: kill do_cmd

Samuel Thibault: o cpufreq: correct initialization on Intel Coppermines o speedstep_detect_speed might not reenable interrupts

Scott Feldman: o [netdrvr e100] move register_netdev below netdev struct init o 10GbE ethtool support o remove ethtool privileged references

Shmulik Hen: o [netdrvr bonding] fix long failover in 802.3ad mode o [netdrvr bonding] fix ABI version control problem

Simon Kelley: o [netdrvr] add atmel[_cs], new wireless driver

Sridhar Samudrala: o [SCTP] Support for socket options that pass both addr and associd o [SCTP] Support for SCTP_GET_PEER_ADDR_INFO socket option o [SCTP] Rename struct sctp_protocol as struct sctp_globals and define macros for all the global fields in the structure. o [SCTP] SCTP_SHUTDOWN_EVENT notification support o [SCTP] /proc interface to display associations/endpoints o [IPV6]: Allow ipv6 fragmentation via ip6_xmit() when ipfragok is set

Stephen Hemminger: o [BRIDGE]: Make delete bridge work with current unregister semantics o [NET]: Sysfs netdev cleanup and bugfix o [NET]: Kill deprecated if_port_text and users o [NET]: Fix device unregister in TUN driver o typo in new class_device_release o sb1000 driver bugs o [DECNET]: Fix build warnings o [NET]: Fix sysfs kobj parent refcount handling o [NET]: Cleanup net-sysfs show and change functions o [NET]: Expose alloc_netdev() for use by drivers o [BRIDGE]: Bridge using alloc_netdev o [VLAN]: vlan network device using alloc_netdev o [TUN]: tun using alloc_netdev o [ACENIC]: Convert to alloc_etherdev o [NET]: Dynamic allocation for dummy net device o [IPV6]: Dynamic allocation for SIT net device o [IPV4]: Dynamic allocation for IPIP net device o [IPV4]: Dynamic allocation for IPGRE net device o [NET]: More reasonable error handling in SLIP driver unload

Steve French: o Make return code on failed cifs mounts more specific and fix incorrect smb to posix return code conversions o adjust for change of devname to const char (new mount format)

Steven Cole: o Use '#ifdef' to test for CONFIG options o K&R to ANSI C conversions for zlib o Yet more K&R to ANSI C conversions o More ANSI C cleanup of zlib o [SPARC]: Fix non-ansi parameter lists o Another final K&R to ANSI C cleanup of zlib o Two more sources of "non-ANSI parameter list" warnings o K&R to ANSI conversions for fs/jfs/jfs_dmap.c and jfs_xtree.c

Steven Whitehouse: o [AX25]: Sanitize ax25 netdevice private handling

Thomas Osterried: o [AX25]: AX.25 bug fixes

Thomas Schlichter: o [NET]: One missed non-netdev SET_MODULE_OWNER case o [NET]: IPSEC protocol module owner cleanup

Thomas Wahrenbruch: o USB: kobil_sct.c added support for KAAN SIM Reader

Trond Myklebust: o Fix udp_data_ready() to use the correct skbuff interface for extracting the XID. Following the introduction of zero-copy under UDP, the data may be entirely located in pages under skb_shinfo(skb)->frags[].

Ville Nuorvala: o [IPV6]: Add ip6ip6 tunnel driver

Zwane Mwaikambo: o cli/sti cleanup for fmvj18x

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to More majordomo info at Please read the FAQ at

This archive was generated by hypermail 2b29 : Sun Jun 15 2003 - 22:00:40 EST