I've delayed this too long, but Ingo found why the scheduler sometimes did
bad things, and this should all be good.
A lot of fairly small changes all over the map, see the full changelog for
details.
Linus
---
Summary of changes from v2.5.64 to v2.5.65
============================================
<[email protected]>:
o Re: hot scsi disk resize
<[email protected]>:
o io_edgeport.c diff to fix endianess bugs
<[email protected]>:
o [trivial] avoid a warning for each module on s390x
<[email protected]>:
o [ATM]: Get lec net_device names correct
o [ATM]: Obsolete some atm_vcc members
<[email protected]>:
o usb-midi.h: fixes for SC-8820/50
<[email protected]>:
o [VLAN]: Fix memory leak in procfs handling
o Memleak in KOBIL USB Smart Card Terminal Driver
o USB: more Edgeport USB Serial Converter driver stuff
o USB: memleak in Edgeport USB Serial Converter driver
<[email protected]>:
o ia64: CONFIG_NUMA build fix
<[email protected]>:
o [CRYPTO]: Move km_types out of header
<[email protected]>:
o [NETFILTER]: Fix ipv6 build
<[email protected]>:
o ia64: fsys-version of gettimeofday()
<[email protected]>:
o [NET]: Move SKB header init to allocation time for better cache
behavior
<[email protected]>:
o [IPSEC]: Add full ipv6 support
<[email protected]>:
o USB: Patch for DSBR-100 driver
<[email protected]>:
o Fix memory overwrite in readdir on large directories
Adam Belay <[email protected]>:
o Manual Resource Setting Update
o Interface Changes
o PnP Card Serivice Revisions
o ALS100 Updates
o OSS SB driver Updates
o Aditional Card Service Changes
Alan Cox <[email protected]>:
o [NETFILTER]: Remove 2.0 and 2.2 stuff from netfilter
o minor breakage fix from typo fixes
o minor typo fix that got missed
o remove 2.2 compatmac stuff from rio driver
o exterminate 2.2 gunk from video stuff
o look a typo 8)
o more 2.0 crap
o more 2.0/2.2 bits
o another typo that escaped
o fix aic7xxx aicasm build
o remove a pile of 2.0 and 2.2 support
o filter more 2.2 junk
o kill stdarg in intermezzo
o missed patch - static not extern inline in cia
o correct BUG doc
o kill the now dead ide_ioreg_t
o kill long unused macro
o add serial port table for PC9800
o correct BUG doc in parisc
o remove unused beep macro on sh
o Add ELF types for Hitach H8 series
o remove 2.2 junk from efs
o remove 2.2 bits fromw anpipe
o remove 2.0 and 2.2 stuff from netfilter
o add a new PCI quirk type for the ALi Magik series
o remove 2.0/2.2 stuff
o remove 2.0/2.2 stuff from wanrouter
o correct emu10k url
o correct file names in comments in mm
o clean up 2.2 stuff in wanrouter code
o ALi it turns out has a 31bit audio device
o update PCI quirks
o correct irq logic for x86
o correct building of the old ide/hd.c driver
o clean up all the console inits using an initcall variant
o bring ide-disk driver into line with 2.4.21pre
o ide-dma
o switch ide-floppy to ide_execute_command
o switch ide-io (core ioctls etc) to ide_execute_command
o add sensible names to the ide iops
o remove spare cast
o switch ide taskfile ioctls to ide_execute_command
o fix wrong type and statics in amd ide
o update via driver from 3.35-ac to 3.36
Alan Stern <[email protected]>:
o USB: Patch for auto-sense cmd_len
Alexey Kuznetsov <[email protected]>:
o [IPV4]: Fix deadlock in IGMP locking
Andi Kleen <[email protected]>:
o Fast path context switch - microoptimize FPU reload
o x86-64 updates for 2.5.64-bk3
Andrew Morton <[email protected]>:
o i386 IRQ balancing cleanup
o balance_irq lockup fix
o Fix nfs oops during mount
o rtc lock ranking fix
o sk98lin 64-bit divide fix
o cciss: fix initialization for PCI hotplug
o export pfn_to_nid to modules
o Reduced stack usage in random.c
o fix inode reclaim imbalance
o remove __pgd_offset
o remove __pmd_offset
o remove __pte_offset
o missed unlock_kernel() in ext3+htree
o reduce stack size: elf_core_dump()
o uninlining in fs/binfmt_elf.c
o ext3: fix error-path memory leak
o fix ioremap off by one bug
o register_blkdev cleanups
o fix possible latency in balance_dirty_pages()
o Implement sendfile() for NFS
o Allow VFS readahead to fall to zero
o Make diskstats per-cpu using kmalloc_percpu
o Fix vm_area_struct slab corruption
o slab use-after-free detector
o slab debug: track caller program counter
o slab debug: symbolic output in caller tracking
o Fix copy_page_range()'s handling of invalid pages
o move CONFIG_SWAP around
o fix div-by-zero in bonding.c
o usercopy checks in old_readdir()
o hugetlb unmap_vmas() SMP && PREEMPT fix
o ext2: fix error-path double-free
o fix memory leak in load_elf_binary()
o Extended attribute sharing and debug macro typo fixes
o protect 'action' in show_interrupts
o i386 show_interrupts() fix
o fix SMP lockup in eepro100 with ethtool on unused
o Larger buffer for /proc/interrupts display
o Disable the "Unknown IO_APIC" message
o raw.c: dev_t cleanup and oops fix
o missing spin_unlock() in sysfs_remove_dir()
o rpc_delete_timer race fix
o remove compile warning from serial console initcall
o revert the "remove kernel_flag" patch
o fix a warning in eepro100.c
o revert "noirqbalance still doesn't do anything"
o ACPI suspend/resume locking fix
o fix typo in init/Kconfig
o pnp warning fix
o fix console ordering default
o work around gcc-3.x inlining bugs
o NCPFS memleak fix
o Memleak in Windows Logical Disk Manager partition
o Fix memleak in ircomm_core
o Force cache alignment of task_structs
o ext3: error handling robustness
o ext2: fix directory handling bug
o unplugging fix
o remove scsi_eh_retry_cmd
o fix devfs oops
o Fix mem= options
o fix the fix for unmap_vmas & hugepages
o Early writeback initialisation
o Fix memleak in e100 driver
o Ext2/3 noatime and dirsync fixes
o ext2: block allocation fix
o kiocbClear should use clear_bit instead of set_bit
o initialise inode->i_rdev
o AFFS fixes
o fix raid0 oops
o miropcm20-rds.c compile fixes
Andries E. Brouwer <[email protected]>:
o scsi_error fix
o some more NAND flash IDs
o fix affs/super.c
Anton Blanchard <[email protected]>:
o ppc64: use fast version of mtmsrd when changing RI, from Milton
Miller
o ppc64: remove extra clear of RI, we do it later on, from Milton
Miller
o ppc64: Fix some warnings in the ppc64 build, from Andrew Morton
o ppc64: rework pci PHB probe code
o ppc64: add missing include
o ppc64: remove old irq balance code
o ppc64: move pSeries specific fixup_bus out of generic code
o ppc64: really move pSeries_pcibios_fixup_bus this time
o missing include
o ppc64: handle 8 byte loads and stores atomically in xmon
o ppc64: add/remove config.h where necessary
o ppc64: compat_sys_fcntl from Stephen Rothwell, remove socketcall
emulation
o ppc64: remove -finline-limit now that we force inlines
Art Haas <[email protected]>:
o [NETFILTER]: C99 initializers for ipv6 netfilter
o [NETFILTER]: C99 initializers for ipv4 netfilter
o [NETFILTER]: Really apply the ipv4 C99 patches this time. :-)
Bart De Schuymer <[email protected]>:
o [EBTABLES]: Trivial changes and cleanups
Benjamin LaHaise <[email protected]>:
o [NET]: Make sure nr_frags is accurate on paged SKB allocation
failure
o [NET]: Simplify scm handling and sendmsg/recvmsg invocation,
consolidate net compat syscalls
Bjorn Helgaas <[email protected]>:
o ia64: iosapic: make pcat_compat system property
o ia64: iosapic: remove find_iosapic duplication
o ia64: iosapic: simplify ISA IRQ init
o ia64: iosapic: self-documenting polarity/trigger arguments
o ia64: iosapic: rationalize __init/__devinit
o ia64: IA32 support without sysctl doesn't work
Christoph Hellwig <[email protected]>:
o update fdomain pcmcia support
o update qlogic pcmcia support
o remove some braindamage from drivers/scsi/pcmcia/Kconfig
o i2c-core.c procfs updates
o remove devfs_only()
o fix possible NULL pointer dereference in scsi_scan.c
o fix OOPS in i2c sysctl registration
o fix up the i2c locking changes
o switch over /proc/bus/i2c to seq_file interface
o fix kmem_cache_size() for new slab poisoning
o remaining bits of DEVFS_FL_AUTO_DEVNUM
o remove regular file support from devfs
o missing drivers/video/Makefile entry
o i2c ID updates
o update i2c algorithm drivers
o i2c-core locking updates
Dave Jones <[email protected]>:
o [CPUFREQ] fix cpufreq core breakage(s)
o [CPUFREQ] fix userspace governor
o [CPUFREQ] remove unneeded code
o [CPUFREQ] updated cpufreq ref-counting and locking scheme
o [CPUFREQ] add support for ICH4-M chipset in speedstep driver
o [CPUFREQ] allow cpufreq drivers to export sysfs files
o [CPUFREQ] update documentation
o [CPUFREQ] Move pci define to pci_ids.h
o [WATCHDOG] amd7xx_tco updates from Zwane, and nuke
EXPORT_NO_SYMBOLS
o [CPUFREQ] Fix documentation typos
o [CPUFREQ] powernow-k7 lazy voltage setting
o [CPUFREQ] More typos
o [CPUFREQ] Yet another typo From Steven Cole
o [CPUFREQ] Yet more typos
o [CPUFREQ] Fix formatting of 'nothing' output
o [CPUFREQ] powernow_decode_bios can be static
o [CPUFREQ] Fix signed comparison warning in powernow-k7
o [CPUFREQ] fix signed comparison warnings for longhaul
o [CPUFREQ] Drop FSB scaling from VIA longhaul driver
Dave Kleikamp <[email protected]>:
o JFS: Fix hang while flushing outstanding transactions under heavy
load
David Brownell <[email protected]>:
o USB: track usb ch9 device state
o USB ohci: "registers" sysfs file
David Gibson <[email protected]>:
o Squash warnings in usb-serial.c
o Squash warning in ohci-pci.c on PowerBooks
David Mosberger <[email protected]>:
o ia64: Minor whitespace & formatting fixups in asm-ia64/sal.h
o ia64: Minor cleanups
o ia64: Make signal deliver work when the current register frame is
incomplete (as a result of a faulting mandatory RSE load).
o ia64: Fix typo in #error message of page-fault handler
o ia64: Sync up with 2.5.60
o ia64: Fix do_gettimeoffset() to not update last_nsec_offset with
(potentially) invalid values.
o ia64: Fix fsys_gettimeofday() and tune it some more
o ia64: Add forgotten probe.w.fault checks in fsys_gettimeofday()
o ia64: Fix formatting inconsistencies introduced by my
fsys_gettimeofday() patch.
o ia64; Improve debug output from kernel unwinder. Based on patch by
Keith Owens.
o ia64: In kernel unwinder, replace dump_info_pt() with
get_scratch_regs() and reformat to make it fit in 100 columns.
o ia64: Correct region_start calculation in kernel unwinder
o ia64: Minor formatting fixes for the preemption patch
o ia64: Fix ia32 sysinfo() emulation
o ia64: Fix SAL processor-log info handling. Based on patch by Keith
Owens.
o ia64: Implement _raw_write_trylock(). Based on patch by Joel
Guillet
o ia64: Implement pcibios_prep_mwi() and define HAVE_ARCH_PCI_MWI to
ensure that PCI line-size gets programmed properly. Based
o ia64: Make ia64_fetch_and_add() simpler to optimize so lib/rwsem.c
can be optimized properly.
o ia64: Correct the value of siginfo.si_addr for SIGSEGV signals
triggered by NaT-page-consumption faults.
o ia64: Hook up POSIX-timer syscalls. Take advantage of
ptrace_notify()
o ia64: Fix several small bugs/omissions from the 2.5.64 sync
o ia64: Minor formatting/whitespace fixes in ia64-version of acpi.c
o ia64: Don't output backspaces in palinfo output
David S. Miller <[email protected]>:
o [IPSEC]: Use dst_hold unless assigning result to something
o [FRAMEBUFFER]: No need for check_var/set_par ops in SBUS fb drivers
o [SBUSFB]: Implement enough ioctls to get X working
o [SBUSFB]: Fix up ioctl helper changes
o [SPARC64]: Fix cpufreq config deps
o [KERNEL]: Add typecheck macro for verifying types at compile time
o [JIFFIES]: Use typecheck in time_foo jiffies macros
o [NET]: Do not duplicate verify_compat_iovec in sparc64 solaris
module
o [IPSEC]: Fix build when ipsec is disabled
o [SPARC]: One too many chars in INIT_C_CC
o [FRAMEBUFFER]: Convert SBUS LEO driver to new APIs
o [SPARC64]: Use pci_remove_bus_device to delete, found by Ben
Collins
o [SPARC64]: Update defconfig
o [SPARC64]: Make sure update_process_times runs inside of
irq_{enter,exit} region
o [ATM]: Add missing $(obj) to driver makefile
o Fix time comparison typing bugs
Dominik Brodowski <[email protected]>:
o pcmcia: it works again!
Douglas Gilbert <[email protected]>:
o scsi_debug in 2.5.64
o sg version 3.5.28 for lk 2.5.64
o scsi_debug version 1.68 mark III
Duncan Sands <[email protected]>:
o USB speedtouch: send path optimization
Gerd Knorr <[email protected]>:
o v4l: video-buf update
o v4l: crunch MIN/MAX macros
o v4l: create include/media
Greg Kroah-Hartman <[email protected]>:
o LSM: fix merge where we lost a prototype in security.h
o USB: add support for Treo devices to the visor driver
o USB: added support for radio shack device to pl2303 driver
o USB: unfortunatly, we can't call usb_unlink_urb() right now all of
the time
o gen_init_cpio: Add the ability to add files to the cpio image
o kbuild: handle any failures of the gen_init_cpio or initramfs image
to stop the build
o LSM: restore d_instantiate function that got lost in the mege
o USB: added support for the palm M100
o USB: fix up a comment in usb_unlink()
o USB: Added support for the Sony Clie NZ90V device
o i2c: add bus driver for ALI15x3 devices
o i2c: get i2c-ali15x3 driver to actually bind to a PCI device
o i2c: add bus driver for Intel 801 devices
o i2c: get i2c-i801 driver to actually bind to a PCI device
o i2c: add bus driver for Intel PIIX4 devices
o i2c: get i2c-piix4 driver to actually bind to a PCI device
o i2c: i2c-piix4.c: Clean up the ibm dma scan logic
o i2c: add i2c sysfs bus support
o driver core: Export the legacy_bus structure for drivers to use
o i2c: add driver model support to i2c adapter drivers
o USB: fixup from previous io_ti.c patch
o USB: added support for Ericsson data cable to pl2303 driver
Greg Ungerer <[email protected]>:
o include unistd.h in m68knommu 68360 entry code
o include unistd.h in m68knommu 68328 entry code
o include errn0.h in m68knommu 68328 interrupt setup code
o m68knommu/ucdimm should use generic 68328 irq setup
o fix spelling in m68knommu Kconfig help
o m68knommu/de2 should use generic 68328 irq setup
o fix wrong argument prototype in m68knommu/68360 for interrupt
handler
o include errno.h in m68knommu 68360 interrupt setup code
o fix m68knommu/68VZ328 Makefile to traverse sub-dirs
o fix m68knommu COMEM-lite PCI bios code
o move common timer and vector code for m68knommu/ColdFire/5272
o fix m68knommu/68VZ328/de2 Makefile to compile local code files
o fix m68knommu/68VZ328/ucdimm Makefile to compile local code files
o call schedule_tail() in m68knommu return from fork code path
o move common timer and vector code for m68knommu/ColdFire/5307
o fix m68knommu/68360 Kconfig wrong define
o inline some mm functions for MMUless targets
o include stddef.h in include/linux/list.h
o fix m68knommu COMEM-lite PCI header code
o move common timer and vector code for m68knommu/ColdFire/5407
o fix m68knommu/68328 serial driver to use work_struct
o build m68knommu/ColdFire common vectors.c and timers.c
o create common timer code for m68knommu/ColdFire processors
o move common timer and vector code for m68knommu/ColdFire/5249
o move common timer and vector code for m68knommu/ColdFire/5206e
o move common timer and vector code for m68knommu/ColdFire/5206
o fix m68knommu/68360 serial driver to use work_struct
o fix spelling in m68knommu comem PCI support code
o fix spelling in m68knommu signal.c
o add support to m68knommu linker script for console init section
Henning Meier-Geinitz <[email protected]>:
o USB: Fix crash in read/write/ioctl in scanner driver
Ingo Molnar <[email protected]>:
o "interactivity changes", sched-2.5.64-A6
o NUMA scheduler breakage
o more "interactivity changes", sched-B2
o "interactivity changes", sched-2.5.64-B2
o sched-2.5.64-bk10-C4
o sched-2.5.64-bk10-D0
Ivan Kokshaysky <[email protected]>:
o nautilus poweroff
o alpha pcibios_claim_console_setup fix
James Bottomley <[email protected]>:
o Correct nested variable thinko in scsi_error.c
o Fix DMA to stack problem in scsi_error.c
o Fix loop problem in SCSI error handler sense collection
o revert fs/block_dev.c change for hot resize: breaks initrd
James Morris <[email protected]>:
o [IPSEC]: Add family argument to compile_policy
o [NET]: Convert volatile char socket flags to real bitops mask, work
by Pedro Hortas
o [NET]: dst_clone --> dst_hold where appropriate
o [IPV4]: Fix multicast route lookups
o [TCP]: Commonize duplicated code into a new function,
tcp_bucket_destroy
o [NET]: Nuke SO_BSDCOMPAT
o [CRYPTO]: Add encrypt_iv() and decrypt_iv() methods
o [CRYPTO]: Eliminate crypto_tfm.crt_ctx, from Adam Richter
o [CRYPTO]: Documentation updates
Jean Tourrilhes <[email protected]>:
o export platform_bus_type
Jeff Garzik <[email protected]>:
o [hw_random] shuffle files in preparation for hw_random driver
update
o [hw_random] update amd768_rng driver to be modular; add Intel
support
o [ia32] cpu capabilities cleanups and additions
o [hw_random] add support for VIA Nehemiah RNG ("xstore" instruction)
o [hw_random] fixes and cleanups
Jeff Wiedemeier <[email protected]>:
o enable setting of marvel/titan agp->type
Jens Axboe <[email protected]>:
o Fix x86-64 build
o remove redundant local_irq_disable in bio_kmap_irq()
o honor hard barrier in deadline
o extra tokens after endif
Jesse Barnes <[email protected]>:
o ia64: SN updates for mmzone.h
o ia64: SN update
o ia64: ACPI fix for no PCI
Johannes Erdfelt <[email protected]>:
o uhci-hcd.c 2.5 finish completions in correct order
John Levon <[email protected]>:
o Fix oprofile on UP, small additional fix
o fix oprofile on x86 > 1 counter
Joshua Uziel <[email protected]>:
o [SPARC64]: Fix warning during uniprocessor build of US3 cpufreq
o [SPARC64]: Need to export up_clock_tick on uniprocessor
Kai Germaschewski <[email protected]>:
o ISDN: Fix up missed return value
o kbuild: Make build stop on vmlinux link error
o kbuild: Add some missing FORCE
o kbuild: Make per-cpu-check ignore __crc_ symbols
o do_mounts: Fix boot from ramdisk
o kbuild: Fix asm/offset.h generation
o ISDN: Resolve name clash in hisax
o ISDN/HiSax: Fix some warnings
o ISDN: [PATCH] isdn_net_lib.c must include isdn_concap.h
o ISDN/AVMB1: Fix compile w/o CONFIG_ISDN_DRV_AVMB1_B1PCIV4
o ISDN: Fix compile error w/o CONFIG_ISDN_PPP_MP
o ISDN: Fix hisax_fcpcipnp PnP
o ISDN/HiSax: Introduce probe() functions
o ISDN/HiSax: config.c cleanup
o ISDN/HiSax: Remove unused ISDN_CHIP_* macros
o ISDN/HiSax: Get rid of CARD_* defines
o ISDN/HiSax: Remove amd7930.c
o kbuild: Fix output when linking vmlinux
Keith Owens <[email protected]>:
o ia64: fix scratch-regs handling in kernel unwinder
Kenneth W. Chen <[email protected]>:
o ia64: rwsem using atomic primitive
Linus Torvalds <[email protected]>:
o Fix up some timeouts to use the proper types
o Make the "interactive bonus" work both ways - both the process that
is actually showing interactive behaviour (sleeping) and the
process that wakes up the interative process get a bonus.
o Fix "con_init()" function type and stale (and incorrect)
declaration
o Fix a very theoretical race between the new RCU lookup and
concurrent renames in another directory.
o Avoid warning due to missing return value
o Ignore initramfs cpio file
o Fix up i387 task switching bug introduced by Andi Kleen's patch to
speed it up - use the proper bitmask for clearing "used-fpu" state.
o Fix nanosleep() problem noticed by Todd Mokros <[email protected]>
o Cache the MSR_IA32_SYSENTER_CS value in the per-CPU TSS (using the
otherwise unused cpl1 entry for SS), so that we can avoid
re-loading it on task switches if it doesn't change.
o Move "used FPU status" into new non-atomic thread_info->status
field
o Remove <asm-i386/xor.h>'s own home-made FPU begin/end macros, use
the real ones instead.
o Use a fixed per-cpu SYSENTER_MSR_ESP value by having the sysenter
entry routine load the real ESP0 off that per-cpu stack. Make this
even faster by putting the sysenter stack in the per-CPU TSS, so
that we can use the tss->esp0 value directly (which we have to
update on task switches anyway).
o Use cond_sched() instead of manual expansion of it
o Error out for the case of a gcc-2.96 compiler with
CONFIG_FRAME_POINTER set. A few versions of gcc-2.96 generate
seriously incorrect code.
o Revert duplicate <linux/stddef.h> addition
Luben Tuikov <[email protected]>:
o scsi_softirq queue is now list_head, eliminate bh_next
Marc Zyngier <[email protected]>:
o Fix fs/binfmt_elf.c build
o Fix arch/alpha/vmlinux.lds.S typos
o EISA/sysfs update
Marcel Holtmann <[email protected]>:
o [SPARC64]: Translate AUTOFS_IOC_EXPIRE_MULTI ioctl
Mark A. Greer <[email protected]>:
o PPC32: Fix our L2 / L3 cache updates for the bootloader
o PPC32: Fix a problem with 'next' and 'step' type KGDB commands
Mark Haverkamp <[email protected]>:
o aacraid driver for 2.5
Martin J. Bligh <[email protected]>:
o Fix NUMA scheduler problem after interactivity merge
o revert pfn_to_nid change
o 1/6 Share common physnode_map code between NUMA-Q and Summit
o 2/6 Make CONFIG_NUMA work on non-numa machines
o 3/6 Convert physnode_map to u8
o 4/6 Fix the type of get_zholes_size for NUMA-Q
o 5/6 Provide basic documentation for profiling
o 6/6 cacheline align files_lock
o NUMA scheduler fixup
Matthew Wilcox <[email protected]>:
o Remove naked GFP_DMA from drivers/scsi/sd.c
o Remove naked GFP_DMA from drivers/scsi/sr.c
o Some parisc updates for SCSI
o fs/locks.c: fix bugs
o neaten fs/locks.c a little
Mike Anderson <[email protected]>:
o Fix SCSI error handler abort case
Neil Brown <[email protected]>:
o md: Missing mddev_put in md resync code
o md: Convert /proc/mdstat to use seq_file
o md: C99 initiailzers for xor.h
o md: Opencode flush_curr_signals in md.c
o md: Tidy up recovery_running flags in md
o md: Include asm-i386/387.h in asm-i386/xor.h
o md: Remove md_recoveryd thread for md
o md: Fulltime delayed 'safe_mode' for md
o md: Fix bad interaction between sync checkpointing and recovery
o md: Allow components of MD raid array to have data start at offset
from start of device
o md: Allow md to select between superblock formats
o md: Add new superblock format for md
o kNFSd: Fix exit-without-free bug in nfsd
o kNFSd: Fix race in svcsock.c
o kNFSd: Fix deadlock problem in lockd
o kNFSd: Assort fixes to nfsd auth cache stuff
o kNFSd: Rename NFSEXP_CROSSMNT to NFSEXP_NOHIDE
o kNFSd: Introduce CROSSMNT flag for knfsd
Oleg Drokin <[email protected]>:
o reiserfs: Correctly free all the allocated memory if open of the
journal failed
o memleak in drivers/char/vt.c
Patrick Mochel <[email protected]>:
o sysfs: Make sure root inode is executable and readable by everyone
o driver model: fix platform_match()
o sysfs: don't complain when sysfs can't register
o cpufreq: fix compile error
o sysfs: fix up directory removal, once and for all
o sysfs: Fix binary file handling
o sysfs: fix BUG()s on directory creation and removal
o driver model: add bus_rescan_devices()
Paul Mackerras <[email protected]>:
o PPC32: Add a thread-pointer argument to the clone syscall, make a
prepare_to_copy()
o PPC32: Better check for when we should expand the stack
o PPC32: Set max_pfn correctly
o PPC32: Make sure interrupts are disabled in IPI handlers
o PPC32: Don't reregister existing /proc/irq entries
Peter Anvin <[email protected]>:
o bootsect removal
o Fix $(src) versus $(obj)
Peter Chubb <[email protected]>:
o ia64: Preemption patch against ~2.5.60
Randy Dunlap <[email protected]>:
o reduce stack in qlogicfc.c
o update filesystems config. menu
o typos only
o fsmenu update
o [IPV4/IPV6]: ICMP cleanups
o [SNMP]: Fix SNMP_STAT_{USR,BH}PTR
Robert Love <[email protected]>:
o no need for kernel_flag on UP
Rohit Seth <[email protected]>:
o ia64: HugeTLB Page patch for IA-64 2.5.60 kernel
o ia64: 2nd update for HugeTLB Page patch for IA-64 2.5.60 kernel
Roland McGrath <[email protected]>:
o signal fix for wedge on multithreaded core dump
Roman Zippel <[email protected]>:
o restore old config behaviour for dependencies on 'm'
o menu structure fix
o add menuconfig support
o add menuconfig support to the front ends
o gtk front end
Russell King <[email protected]>:
o Separate out pci bus resource allocator
o Eliminate stack allocation of struct pci_dev
o Only add devices to bus->devices while scanning
o Convert setup-bus resource allocation to use bus->devices
o Fix up pci_scan_bridge and friends
o [ARM] Remove set_mac_address from acorn drivers
o [ARM] Convert ecard.c shutdown/reboot to use driver model
o [ARM] Always call schedule_tail() after a fork()
o [ARM] Always claim the timer IRQ using SA_INTERRUPT
o [ARM] Add new kernel typedefs for __kernel_{timer,clockid}_t
o [ARM] Add dummy set_pgd() implementation
o [ARM] Make ARM's pci_controller_num dev argument appear to be used
o [ARM] Kill old mkdep dependency hack
o [ARM] Fix ARM ide.h
o [ARM] Make TLB instruction selection more fine-grained
o [ARM] Update mach-types to latest vesion
o [ARM] Ensure we preserve other CPSR bits when switching to SVC mode
o [ARM] Power management updates
o [ARM] Add better PM support to SA1111 and SA11x0
o [ARM] Clean up ARM PCI support (bios32.c)
o [CPUFREQ] Update ARM CPUFREQ drivers
o [ARM] CPUFREQ - allow ARM to work with userspace governor
o [CPUFREQ] Make sa11x0_ppcr_to_freq return in units of kHz, not
100kHz
o [ARM] Add generic SSP "PIO" mode driver
o [SERIAL] Overhaul 8250_pci.c
o [SERIAL] Update 8250_acorn.c
o [SERIAL] Add ttydriver->owner initialisation
o [SERIAL] Make tty->driver_data point at the uart_state structure
o [SERIAL] Make uart_tasklet_action take uart_state
o [SERIAL] Eliminate some more passing of struct uart_info
o [SERIAL] Add per-port semaphore
o [SERIAL] Remove remaining notifier-based PM support
o [SERIAL] Four bug fixes
o [SERIAL] Prevent multiple calls to tty_{un,}register_device()
o [SERIAL] Add new device model based power management infrastructure
o [SERIAL] Add sa1100 serial PM support using device model
o [SERIAL] Add uart_console(port) macro
o [SERIAL] Add PCI serial power management support
o [SERIAL] Add Xircom RBM56G PCI ID
o [SERIAL] Only update the console termios cflag once
o [TTY] Register tty devclass before use
o [PCI] pci-6 - Fix scanning of non-zero functions
o [PCI] pci-7: Remove second argument to pcibios_update_resource()
o [PCI] pci-8: pci_resource_to_bus()
o [PCI] pci-9: Kill per-architecture pcibios_update_resource()
o [PCI] pci-10: Miscellaneous cleanups to probe.c
o [PCI] pci-11: use u32 for bus numbers/latency not unsigned long
o [PCI] pci-12: Add #defines for cardbus specifics
o [PCI] pci-13: unuse pci_do_scan_bus()
o [PCI] pci-14: Add the Mobility Electronics EV1000 PCI device
numbers
o [PCI] pci-15: Fix setup-bus.c resource sizing
Sam Ravnborg <[email protected]>:
o kbuild: Smart notation for non-verbose output
o kbuild: touch-module after successfull creation only
o kbuild: Do not clutter output with make -jN
o kbuild/all arch: Use filechk rule for offsets generation
o kbuild: Use targets := to tell kbuild about additional targets
o kbuild: Introduced extra-y, as replacement for EXTRA_TARGETS
o kbuild: build-targets replaced with always
o kbuild: Updated Documentation/kbuild/makefiles.txt
Stephen D. Smalley <[email protected]>:
o Add LSM hook to do_kern_mount
o Replace inode_post_lookup hook with d_instantiate hook
o allocate and free security structures for private files
o Restore LSM hook calls to setpriority and setpgid
o LSM: Add LSM sysctl hook to 2.5.59
o LSM: Add LSM syslog hook to 2.5.59
o LSM: coding style fixups in sb_kern_mount
Stephen Hemminger <[email protected]>:
o Turn off aio printk meant for debugging (2.5.64)
Stephen Rothwell <[email protected]>:
o ia64: compat_sys_futex() support
o compat_sys_fcntl{,64} Generic part
o compat_sys_fcntl{,64} x86_64 part
o compat_sys_fcntl{,64} s390x part
o compat_sys_fcntl{,64} parisc part
o [COMPAT]: Sparc64 part of fcntl changes
Steve French <[email protected]>:
o Remember to free mapping in all writepage paths
o Fix oops in getdfs when null path passed in on mount. Fix oops
when changed readsize caused readpages problem. Add support for
altering rsize so can reduce pages read across net below default of 4
Steven Cole <[email protected]>:
o Documentation spelling cleanup
o USB: spelling fixes for drivers/usb
St?phane Eranian <[email protected]>:
o ia64: perfmon patch for 2.5.59
Suresh Siddha <[email protected]>:
o ia64: 1/2 fix for generic kernel
o ia64: 2/2 fix in machvec.h
Tom Rini <[email protected]>:
o PPC32: Fix a typo in modular IDE support on Motorola LoPEC
o PPC32: Fix building of the Motorola LoPEC and CONFIG_DUMMY_CONSOLE
o PPC32: Replace 2 inline functions with their normal macro
equivalents
Willem Riede <[email protected]>:
o fix jiffies compare warning in osst
Zwane Mwaikambo <[email protected]>:
o noirqbalance still doesn't do anything
Compile statistics: 2.5.65
2.5.64 2.5.65
-------------------- -----------------
bzImage (defconfig) 14 warnings 14 warnings
0 errors 0 errors
bzImage (allmodconfig) 30 warnings 30 warnings
9 errors 12 errors
modules (allmodconfig) 2356 warnings 2421 warnings
99 errors 100 errors
Compile statistics have been for kernel releases from 2.5.46 to 2.5.65
at: http://www.osdl.org/archive/cherry/stability (will be updated by 6PM PST).
Other stability-related links:
OSDL Stability page:
http://osdl.org/projects/26lnxstblztn/results/
Nightly linux-2.5 bk build:
http://www.osdl.org/archive/cherry/stability/linus-tree/running.txt
2.5 porting items:
http://www.osdl.org/archive/cherry/stability/linus-tree/port_items.txt
2.5 porting items history:
http://www.osdl.org/archive/cherry/stability/linus-tree/port_history.txt
John
On Mon, Mar 17, 2003 at 02:31:01PM -0800, Linus Torvalds wrote:
> I've delayed this too long, but Ingo found why the scheduler sometimes did
Yaay! I was getting withdrawal symptoms. :)
> bad things, and this should all be good.
>
> A lot of fairly small changes all over the map, see the full changelog for
> details.
Aye. Most .config options got lost in the upgrade (I've had to reset
most of them so far).
One question. Should PCMCIA_AHA152X only be compilable as a module? I
found this in Kconfig:
config PCMCIA_AHA152X
tristate "Adaptec AHA152X PCMCIA support"
depends on m
help
Say Y here if you intend to attach this type of PCMCIA SCSI host
adapter to your computer.
...
The help and the tristate seems to indicate that I should be able to
compile it into the kernel, but menuconfig wont let me. This is
presumably due to the dependancy but is it right?
--
"Other countries of course, bear the same risk. But there's no doubt his
hatred is mainly directed at us. After all this is the guy who tried to
kill my dad."
- George W. Bush Jr, 'President' of Regime of the United States
September 26, 2002 (from a political fundraiser in Huston, Texas)
On Tue, Mar 18, 2003 at 04:22:57PM +1100, CaT wrote:
> One question. Should PCMCIA_AHA152X only be compilable as a module? I
> found this in Kconfig:
>
> config PCMCIA_AHA152X
> tristate "Adaptec AHA152X PCMCIA support"
> depends on m
> help
> Say Y here if you intend to attach this type of PCMCIA SCSI host
> adapter to your computer.
> ...
>
> The help and the tristate seems to indicate that I should be able to
> compile it into the kernel, but menuconfig wont let me. This is
> presumably due to the dependancy but is it right?
I think all pcmcia drivers currently are compilable only as module.
This was because historically they need cardmgr to work properly, but
someone is working on fixing that IIRC.
On Tue, Mar 18, 2003 at 09:15:16AM +0100, Christoph Hellwig wrote:
> On Tue, Mar 18, 2003 at 04:22:57PM +1100, CaT wrote:
> > The help and the tristate seems to indicate that I should be able to
> > compile it into the kernel, but menuconfig wont let me. This is
> > presumably due to the dependancy but is it right?
>
> I think all pcmcia drivers currently are compilable only as module.
All the others I've used can be compiled in.
> This was because historically they need cardmgr to work properly, but
> someone is working on fixing that IIRC.
Ok. That might explain it then. :)
--
"Other countries of course, bear the same risk. But there's no doubt his
hatred is mainly directed at us. After all this is the guy who tried to
kill my dad."
- George W. Bush Jr, 'President' of Regime of the United States
September 26, 2002 (from a political fundraiser in Huston, Texas)
Hi,
On Tue, 18 Mar 2003, CaT wrote:
> One question. Should PCMCIA_AHA152X only be compilable as a module? I
> found this in Kconfig:
>
> config PCMCIA_AHA152X
> tristate "Adaptec AHA152X PCMCIA support"
> depends on m
> help
> Say Y here if you intend to attach this type of PCMCIA SCSI host
> adapter to your computer.
> ...
>
> The help and the tristate seems to indicate that I should be able to
> compile it into the kernel, but menuconfig wont let me. This is
> presumably due to the dependancy but is it right?
Yes, this was the behaviour of the old config tools, which was restored
with 2.5.65. This means 'm' is a marker that this thing works only as a
module.
If you want the other behaviour, that it can only be built as a module in
a modular kernel, but compile it into a nonmodular kernel, you can use "m
|| !MODULES" instead.
bye, Roman
On Tue, Mar 18, 2003 at 11:08:08AM +0100, Roman Zippel wrote:
> > The help and the tristate seems to indicate that I should be able to
> > compile it into the kernel, but menuconfig wont let me. This is
> > presumably due to the dependancy but is it right?
>
> Yes, this was the behaviour of the old config tools, which was restored
> with 2.5.65. This means 'm' is a marker that this thing works only as a
> module.
> If you want the other behaviour, that it can only be built as a module in
> a modular kernel, but compile it into a nonmodular kernel, you can use "m
> || !MODULES" instead.
Ahhh. So if I want module support but not use it as a module then I'm
SOL?
--
"Other countries of course, bear the same risk. But there's no doubt his
hatred is mainly directed at us. After all this is the guy who tried to
kill my dad."
- George W. Bush Jr, 'President' of Regime of the United States
September 26, 2002 (from a political fundraiser in Huston, Texas)
Hi,
On Tue, 18 Mar 2003, CaT wrote:
> Ahhh. So if I want module support but not use it as a module then I'm
> SOL?
Yes, or you have to complain to the module author.
bye, Roman
> > > The help and the tristate seems to indicate that I should be able to
> > > compile it into the kernel, but menuconfig wont let me. This is
> > > presumably due to the dependancy but is it right?
> >
> > Yes, this was the behaviour of the old config tools, which was restored
> > with 2.5.65. This means 'm' is a marker that this thing works only as a
> > module.
> > If you want the other behaviour, that it can only be built as a module in
> > a modular kernel, but compile it into a nonmodular kernel, you can use "m
> > || !MODULES" instead.
>
> Ahhh. So if I want module support but not use it as a module then I'm
> SOL?
vi .config
John.
On Mon, Mar 17, 2003 at 02:31:01PM -0800, Linus Torvalds wrote:
>...
> Summary of changes from v2.5.64 to v2.5.65
> ============================================
>...
> Christoph Hellwig <[email protected]>:
>...
> o remaining bits of DEVFS_FL_AUTO_DEVNUM
>...
The following files in 2.5.65 still contain DEVFS_FL_AUTO_DEVNUM:
arch/ia64/sn/io/xbow.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/sn2/pcibr/pcibr_dvr.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/sn2/xbow.c: DEVFS_FL_AUTO_DEVNUM, 0, 0,
arch/ia64/sn/io/sn2/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/sn1/pcibr.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/ifconfig_net.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/ioconfig_bus.c: 0, DEVFS_FL_AUTO_DEVNUM,
arch/ia64/sn/io/hcl.c: 0, DEVFS_FL_AUTO_DEVNUM,
drivers/media/dvb/dvb-core/dvbdev.c: #define DVB_DEVFS_FLAGS (DEVFS_FL_DEFAULT|DEVFS_FL_AUTO_DEVNUM)
The last one causes a compile error on i386 with CONFIG_DVB_DEVFS_ONLY
enabled.
cu
Adrian
--
"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed
On Tue, 18 Mar 2003, CaT wrote:
>
> Ahhh. So if I want module support but not use it as a module then I'm
> SOL?
Well, does it _work_ as a built-in? If it does, just send me a patch for
the Kconfig file.
A lot of the PCMCIA stuff (16-bit) historically _only_ works as modules,
because the old PCMCIA code depended on module unload to do a lot of the
cleanups that the regular internal eject handling didn't do. But if that
driver works for you built-in, then the Kconfig file itself is simply just
wrong.
Linus
On Tue, Mar 18, 2003 at 07:16:06AM -0800, Linus Torvalds wrote:
> On Tue, 18 Mar 2003, CaT wrote:
> > Ahhh. So if I want module support but not use it as a module then I'm
> > SOL?
>
> Well, does it _work_ as a built-in? If it does, just send me a patch for
> the Kconfig file.
Well, I would if I could. I don't actually have the card, I was just
going to compile it in incase I had to use one one day. Saves time not
having to recompile the kernel and I don't wanna waste disk space on
modules (and having to clean them up from old kernel versions etc). My
laptop HD space is precious. :)
> A lot of the PCMCIA stuff (16-bit) historically _only_ works as modules,
> because the old PCMCIA code depended on module unload to do a lot of the
> cleanups that the regular internal eject handling didn't do. But if that
> driver works for you built-in, then the Kconfig file itself is simply just
> wrong.
It might be an idea to modify the help and maybe the error message right
now as at the moment the help indicates that it can be compiled in
without stating any conditions for it, and the error message talks about
some other mysterious component that this one depends on that was selected
as a module and so this one will be, which would send the user off on a
goose chace (like it did me for a wee bit :). Is there a way to
customize that? If so I could make up a patch that clarifies things
more.
--
"Other countries of course, bear the same risk. But there's no doubt his
hatred is mainly directed at us. After all this is the guy who tried to
kill my dad."
- George W. Bush Jr, 'President' of Regime of the United States
September 26, 2002 (from a political fundraiser in Houston, Texas)
On Tue, Mar 18, 2003 at 01:45:16PM +0100, Adrian Bunk wrote:
> On Mon, Mar 17, 2003 at 02:31:01PM -0800, Linus Torvalds wrote:
> >...
> > Summary of changes from v2.5.64 to v2.5.65
> > ============================================
> >...
> > Christoph Hellwig <[email protected]>:
> >...
> > o remaining bits of DEVFS_FL_AUTO_DEVNUM
> >...
>
> The following files in 2.5.65 still contain DEVFS_FL_AUTO_DEVNUM:
>
> arch/ia64/sn/io/xbow.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/sn2/pcibr/pcibr_dvr.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/sn2/xbow.c: DEVFS_FL_AUTO_DEVNUM, 0, 0,
> arch/ia64/sn/io/sn2/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/klgraph.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/sn1/pcibr.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/ifconfig_net.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/ioconfig_bus.c: 0, DEVFS_FL_AUTO_DEVNUM,
> arch/ia64/sn/io/hcl.c: 0, DEVFS_FL_AUTO_DEVNUM,
This code is currently not working at all on 2.5 and will get it's
own filesystem instead.
> drivers/media/dvb/dvb-core/dvbdev.c: #define DVB_DEVFS_FLAGS (DEVFS_FL_DEFAULT|DEVFS_FL_AUTO_DEVNUM)
>
>
> The last one causes a compile error on i386 with CONFIG_DVB_DEVFS_ONLY
> enabled.
I thought I removed that config option, need to check whether that hunk
got left.
On Tue, Mar 18, 2003 at 04:40:37PM +0100, Christoph Hellwig wrote:
> > drivers/media/dvb/dvb-core/dvbdev.c: #define DVB_DEVFS_FLAGS (DEVFS_FL_DEFAULT|DEVFS_FL_AUTO_DEVNUM)
> >
> >
> > The last one causes a compile error on i386 with CONFIG_DVB_DEVFS_ONLY
> > enabled.
>
> I thought I removed that config option, need to check whether that hunk
> got left.
Ok, here's the lost hunk:
--- 1.1/drivers/media/dvb/dvb-core/Kconfig Wed Oct 30 02:16:55 2002
+++ edited/drivers/media/dvb/dvb-core/Kconfig Tue Mar 18 16:56:51 2003
@@ -5,13 +5,3 @@
DVB core utility functions for device handling, software fallbacks etc.
Say Y when you have a DVB card and want to use it. If unsure say N.
-
-config DVB_DEVFS_ONLY
- bool "devfs only"
- depends on DVB_CORE=y && DEVFS_FS
- help
- Drop support for old major/minor device scheme and support only devfs
- systems. This saves some code.
-
- If unsure say N.
-
--- 1.3/drivers/media/dvb/dvb-core/dvbdev.c Mon Nov 25 16:57:37 2002
+++ edited/drivers/media/dvb/dvb-core/dvbdev.c Tue Mar 18 16:57:45 2003
@@ -21,8 +21,6 @@
*
*/
-/*#define CONFIG_DVB_DEVFS_ONLY 1*/
-
#include <linux/config.h>
#include <linux/version.h>
#include <linux/module.h>
@@ -56,17 +54,8 @@
};
-#ifdef CONFIG_DVB_DEVFS_ONLY
-
- #define DVB_MAX_IDS ~0
- #define nums2minor(num,type,id) 0
- #define DVB_DEVFS_FLAGS (DEVFS_FL_DEFAULT|DEVFS_FL_AUTO_DEVNUM)
-
-#else
-
- #define DVB_MAX_IDS 4
- #define nums2minor(num,type,id) ((num << 6) | (id << 4) | type)
- #define DVB_DEVFS_FLAGS (DEVFS_FL_DEFAULT)
+#define DVB_MAX_IDS 4
+#define nums2minor(num,type,id) ((num << 6) | (id << 4) | type)
static
@@ -234,8 +223,7 @@
sprintf(name, "%s%d", dnames[type], id);
dvbdev->devfs_handle = devfs_register(adap->devfs_handle, name,
- DVB_DEVFS_FLAGS,
- DVB_MAJOR,
+ 0, DVB_MAJOR,
nums2minor(adap->num, type, id),
S_IFCHR | S_IRUSR | S_IWUSR,
dvbdev->fops, dvbdev);
32 functions using >=1k of kernel stack on i386
The list is _much_ shorter than the 2.5.64 one. But I don't know how
to interpret that yet.
oldconfig created about the same amount of work as allyesconfig and
hand-pruning of breaking drivers. The resulting .config might have
changed a lot, enough to account for 19 lost functions.
0xc01b6994 presto_get_fileid: sub $0x1164,%esp
0xc01b5634 presto_copy_kml_tail: sub $0x1018,%esp
0xc06b2704 i2o_proc_read_ddm_table: sub $0xb48,%esp
0xc056daf9 v4l_compat_translate_ioctl: sub $0x924,%esp
0xc05beed6 ide_unregister: sub $0x8a0,%esp
0xc05403d4 writerids: sub $0x820,%esp
0xc05855f4 w9966_v4l_read: sub $0x820,%esp
0xc06b53f4 i2o_proc_read_lan_alt_addr: sub $0x81c,%esp
0xc06b4c74 i2o_proc_read_lan_mcast_addr: sub $0x81c,%esp
0xc0540224 readrids: sub $0x810,%esp
0xc06b2bb4 i2o_proc_read_groups: sub $0x810,%esp
0xc0105409 huft_build: sub $0x598,%esp
0xc0106e59 huft_build: sub $0x598,%esp
0xc023e794 dohash: sub $0x580,%esp
0xc0438ad4 ida_ioctl: sub $0x53c,%esp
0xc0107bd4 inflate_dynamic: sub $0x52c,%esp
0xc0106104 inflate_dynamic: sub $0x51c,%esp
0xc08eb604 device_new_if: sub $0x518,%esp
0xc01ad654 presto_ioctl: sub $0x4fc,%esp
0xc05ced04 ide_config: sub $0x494,%esp
0xc0107a67 inflate_fixed: sub $0x490,%esp
0xc0105f97 inflate_fixed: sub $0x490,%esp
0xc08e7bf9 br_ioctl_device: sub $0x490,%esp
0xc0425034 parport_config: sub $0x47c,%esp
0xc07c6d04 ixj_config: sub $0x478,%esp
0xc0333b66 sha512_transform: sub $0x440,%esp
0xc09326e4 gss_pipe_downcall: sub $0x43c,%esp
0xc02a8c74 ciGetLeafPrefixKey: sub $0x41c,%esp
0xc064a804 aic7xxx_detect: sub $0x404,%esp
0xc050a998 hex_dump: sub $0x400,%esp
0xc0323942 befs_warning: sub $0x400,%esp
0xc03238f2 befs_error: sub $0x400,%esp
J?rn
--
Courage is not the absence of fear, but rather the judgement that
something else is more important than fear.
-- Ambrose Redmoon