"I really don't want to do this, and I was actually hoping to release 2.6.25 last weekend (which is why -rc9 is a few days late - just me hoping to not do another -rc at all), but I've done an -rc9," Linus Torvalds said, announcing the 2.6.25-rc9 kernel. "The changes in -rc9 are pretty small (shortlog appended)," he continued, "and 60% of them are m68k updates - mostly defconfigs. And some doc updates. But there's some network driver updates (tg3 and wireless hostap stand out), some late XFS patches and a mvsas driver update (the mvsas driver is new in 2.6.25, so that's not going to regress anything ;). The rest is mostly one-liners, with a few reverts going on." Linus then explained why he was putting out another release candidate:
"The reason for not doing a 2.6.25 is that some people are making noises about slab/page-alloc setup issues, and I wanted something out this week, but didn't feel comfy doing a final release.
"That said, I think I'll have to do 2.6.25 early next week regardless, because we can't just keep holding things back forever. At some point it will have to turn into a 2.6.25.x issue, and the developers with stuff pending for the next version need to be able to start merging."
From: Linus Torvalds <torvalds@...> Subject: Linux 2.6.25-rc9 Date: Apr 11, 4:51 pm 2008 I really don't want to do this, and I was actually hoping to release 2.6.25 last weekend (which is why -rc9 is a few days late - just me hoping to not do another -rc at all), but I've done an -rc9. The changes in -rc9 are pretty small (shortlog appended), and 60% of them are m68k updates - mostly defconfigs. And some doc updates. But there's some network driver updates (tg3 and wireless hostap stand out), some late XFS patches and a mvsas driver update (the mvsas driver is new in 2.6.25, so that's not going to regress anything ;). The rest is mostly one-liners, with a few reverts going on. Dirstat: 4.2% Documentation/filesystems/ 3.2% Documentation/networking/ 11.4% Documentation/ 55.7% arch/m68k/configs/ 59.4% arch/m68k/ 2.0% arch/powerpc/ 65.7% arch/ 3.6% drivers/net/ 4.2% drivers/scsi/ 12.7% drivers/ 2.3% fs/ 2.2% include/ 3.9% net/ anyway, the reason for not doing a 2.6.25 is that some people are making noises about slab/page-alloc setup issues, and I wanted something out this week, but didn't feel comfy doing a final release. That said, I think I'll have to do 2.6.25 early next week regardless, because we can't just keep holding things back forever. At some point it will have to turn into a 2.6.25.x issue, and the developers with stuff pending for the next version need to be able to start merging. Linus PS. This has been a somewhat frustrating last week or so. So if I was even less polite to somebody than usual in public or private emails, my apologies. You know who you are. --- Abhijeet Kolekar (1): iwlwifi: fix n-band association problem Adrian Bunk (18): [ATM]: Make atm/he.c:read_prom_byte() static [ATM] atm/idt77252.c: Make 2 functions static [NET]: Remove Documentation/networking/sk98lin.txt V4L/DVB (7485): v4l2-int-device.c: add MODULE_LICENSE ide-pnp.c: add MODULE_LICENSE ide/pci/cmd640.c: add MODULE_LICENSE ide/ppc/mpc8xx.c: add MODULE_LICENSE ide/ppc/pmac.c: add MODULE_LICENSE ide/arm/ide_arm.c: add MODULE_LICENSE ide/cris/ide-cris.c: add MODULE_LICENSE ide/h8300/ide-h8300.c: add MODULE_LICENSE ide/legacy/gayle.c: add MODULE_LICENSE ide/legacy/buddha.c: add MODULE_LICENSE ide/legacy/falconide.c: add MODULE_LICENSE ide/legacy/macide: add MODULE_LICENSE ide/legacy/q40ide.c: add MODULE_LICENSE net/tokenring/olympic.c section fixes m68k: use KBUILD_DEFCONFIG Al Viro (1): fix endian lossage in forcedeth Alan Stern (1): USB: fix bug in sg initialization in usbtest Alexey Korolev (1): mtd: fix broken state in CFI driver caused by FL_SHUTDOWN Alok Kataria (1): acpi: fix "buggy BIOS check" when CPUs are hot removed Andi Kleen (1): [NET]: srandom32 fixes for networking v2 Andrea Arcangeli (1): Fix bounce setting for 64-bit Andrew Morton (1): alpha: get_current(): don't add zero to current_thread_info()->task Andrew Paprocki (1): [WATCHDOG] it8712f_wdt Zero MSB timeout byte when disabling watchdog Anthony Liguori (1): virtio_net: remove overzealous printk Anton Vorontsov (1): [POWERPC] 83xx: Fix wrong USB phy type in mpc837xrdb dts Arjan van de Ven (1): es1968: fix sleep-while-holding-lock bug Balbir Singh (1): memory controller: make memory resource control aware of boot options Bartlomiej Sieka (1): [POWERPC] mpc5200: Amalgamated DTS fixes and updates Bartlomiej Zolnierkiewicz (3): ide-h8300: 32-bit I/O is unsupported ide: use ->ata_input_data in ide_driveid_update() libata: fix IDENTIFY order in ata_bus_probe() Benjamin Herrenschmidt (3): [POWERPC] Fix iSeries hard irq enabling regression [POWERPC] Fix deadlock with mmu_hash_lock in hash_page_sync pata_sil680: Fix build on arch/ppc Benoit Boissinot (2): IPv6: only update the lifetime of the relevant temporary address IPv6: do not create temporary adresses with too short preferred lifetime Bjorn Helgaas (2): V4L/DVB (7486): radio-cadet: wrap PNP probe code in #ifdef CONFIG_PNP pnp: increase number of devices supported per protocol Brad Sawatzky (1): USB: serial: fix regression in Visor/Palm OS module for kernels >= 2.6.24 Bryan Wu (1): fix bug - executing FDPIC ELF on NFS mount triggers BUG() at mm/nommu.c:862:/do_mmap_private() Carol Hebert (1): ipmi: change device node ordering to reflect probe order Christian Borntraeger (1): kvm: provide kvm.h for all architecture: fixes headers_install Chuck Lever (1): SUNRPC: Fix a memory leak in rpc_create() Clark Rawlins (1): USB: Allow initialization of broken keyspan serial adapters. Cyrill Gorcunov (2): V4L/DVB (7461): bttv: fix missed index check V4L/DVB (7460): bttv: Bt832 - fix possible NULL pointer deref Dan Williams (1): md: close a livelock window in handle_parity_checks5 Daniel Drake (1): ipw2200: set MAC address on radiotap interface Daniel Lezcano (1): [NETNS][IPV6] tcp - assign the netns for timewait sockets Daniel Wagner (1): rt61pci: rt61pci_beacon_update do not free skb twice Dave Young (2): bluetooth : use lockdep sub-classes for diffrent bluetooth protocol bluetooth : __rfcomm_dlc_close lock fix David Brownell (3): drivers/net/usb/pegasus.c: remove in_atomic() check USB: another ehci_iaa_watchdog fix spi: documentation tweaks David Chinner (1): [XFS] Fix superblock features2 field alignment problem David Howells (5): FRV: Handle update_mmu_cache() being called when current->mm is NULL [try #2] FRV: Move STACK_TOP_MAX up [try #2] FRV: Add support for emulation of userspace atomic ops [try #2] FRV: Make NOMMU-mode work with base addresses other than 0xC0000000 [try #2] FRV: Don't make smp_{r, w, }mb() interpolate MEMBAR when CONFIG_SMP=n [try #2] David S. Miller (11): [NET]: Use local_irq_{save,restore}() in napi_complete(). [SPARC64]: Don't open-code {get,put}_cpu_var() in flush_tlb_pending(). [BOND]: Fix warning in bond_sysfs.c [ATM] firestream: Fix uninitialized var warning. [INET]: inet_frag_evictor() must run with BH disabled [LLC]: Kill llc_station_mac_sa symbol export. [SPARC64]: Fix FPU saving in 64-bit signal handling. [SPARC64]: Fix user accesses in regset code. [NET]: Undo code bloat in hot paths due to print_mac(). [SPARC]: Fix several regset and ptrace bugs. [IPV4]: Fix byte value boundary check in do_ip_getsockopt(). Davide Libenzi (2): eventfd/kaio integration fix signalfd: fix for incorrect SI_QUEUE user data reporting Denis V. Lunev (4): [NETFILTER]: Replate direct proc_fops assignment with proc_create call. [IPV6]: Event type in addrconf_ifdown is mis-used. [IPV6]: inet6_dev on loopback should be kept until namespace stop. [IPV6]: Fix refcounting for anycast dst entries. Dmitri Vorobiev (1): Fix typos in Documentation/unaligned-memory-access.txt Dmitry Adamushko (1): mtd/chips: add missing set_current_state() to cfi_{amdstd,staa}_sync() Eliezer Tamir (1): BNX2X: Correct bringing chip out of reset Eric Paris (1): SELinux: don't BUG if fs reuses a superblock Eric Sandeen (2): [XFS] remove shouting-indirection macros from xfs_sb.h [XFS] Ensure "both" features2 slots are consistent FUJITA Tomonori (2): sparc64: add the segment boundary checking to IOMMUs while merging SG entries sparc64: remove unused calc_npages() in iommu_common.h Fabio Checconi (2): cfq-iosched: fix rcu freeing of cfq io contexts cfq-iosched: do not leak ioc_data across iosched switches Florian Fainelli (1): USB: cp2101: Add identifiers for the Telegesys ETRX2USB Geert Uytterhoeven (1): m68k: update defconfigs for 2.6.25 Grant Grundler (1): [netdrvr] tulip_read_eeprom fixes for BUG 4420 Harvey Harrison (2): net: marvell.c fix sparse shadowed variable warning lzo: fix typo in decompressor Heiko Carstens (1): Fix "$(AS) -traditional" compile breakage caused by asmlinkage_protect Herbert Xu (3): [NET]: Add preemption point in qdisc_run [IPV6]: Fix ICMP relookup error path dst leak [ICMP]: Ensure that ICMP relookup maintains status quo Holger Schurig (2): libertas: fix spinlock recursion bug libertas: fix mode initialization problem Hugh Dickins (2): splice: use mapping_gfp_mask scsi: fix sense_slab/bio swapping livelock Ilpo Järvinen (4): [TCP]: Restore 2.6.24 mark_head_lost behavior for newreno/fack [TCP]: Fix NewReno's fast rexmit/recovery problems with GSOed skb [TCP]: tcp_simple_retransmit can cause S+L [TCP]: Don't allow FRTO to take place while MTU is being probed Ingo Molnar (4): forcedeth: fix locking bug with netconsole x86: fix nmi_watchdog=2 on Pentium-D CPUs x86: print message if nmi_watchdog=2 cannot be enabled revert "x86: tsc prevent time going backwards" Ivan Kokshaysky (1): alpha: fix ALSA DMA mmap crash Ivo van Doorn (1): rt2x00: Ignore set_state(STATE_SLEEP) failure J. Bruce Fields (4): Spell out behavior of atomic_dec_and_lock() in kerneldoc Documentation: move nfsroot.txt to filesystems/ Documentation: move rpc-cache.txt to filesystems/ Move sched-rt-group.txt to scheduler/ James Bottomley (2): Revert "[SCSI] fix bsg queue oops with iscsi logout" [SCSI] libsas: Warn if ATA device detected but CONFIG_SCSI_SAS_ATA not set James Smart (1): [SCSI] hosts.c: fixes for "no error" reported after error scenarios Jan Engelhardt (2): Fixes to the seq_file document [NETFILTER]: nf_nat: autoload IPv4 connection tracking Jan Niehusmann (1): mac80211: trigger ieee80211_sta_work after opening interface Jarek Poplawski (1): [ROSE/AX25] af_rose: rose_release() fix Jeff Layton (1): NFS: initialize flags field in nfs_open_context Jens Axboe (1): splice: fix infinite loop in generic_file_splice_read() Jesse Barnes (1): fix IS_I9XX macro in i915 DRM driver Jim Meyering (1): doc: fix typo in SubmittingPatches Jiri Slaby (2): Char: ip2, fix sparse warnings Char: rio, fix sparse warnings Johann Felix Soden (1): ata/sata_fsl: Remove unused variable in sata_fsl_probe Johannes Berg (1): nl80211: fix STA AID bug Jonathan Corbet (2): Add the seq_file documentation Fill out information on patch tags in SubmittingPatches Joonwoo Park (3): [LLC]: bogus llc packet length [LLC]: station source mac address [LLC]: skb allocation size for responses Josef Bacik (1): selinux: prevent rentry into the FS Joy Latten (1): [CRYPTO] xcbc: Fix crash when ipsec uses xcbc-mac with big data chunk Julia Lawall (1): drivers/net/wireless/iwlwifi/iwl-4965.c: correct use of ! and & KAMEZAWA Hiroyuki (1): memcg: fix node_state handling Karsten Wiese (1): x86: fix call to set_cyc2ns_scale() from time_cpufreq_notifier() Kay Sievers (5): spi: fix platform driver hotplug/coldplug usb gadget: fix platform driver hotplug/coldplug usb host: fix platform driver hotplug/coldplug watchdog: fix platform driver hotplug/coldplug rtc: fix platform driver hotplug/coldplug Ke Wei (8): [SCSI] mvsas: fill in error info record and phy mode6 bits. [SCSI] mvsas: a tag handler implementation [SCSI] mvsas : interrupt handling [SCSI] mvsas: check hd whether unplugged [SCSI] mvsas: retry aborting task. [SCSI] mvsas: fix the buffer of rx DMA overflow bug [SCSI] mvsas: get phy info. [SCSI] mvsas: check subsystem id Ken'ichi Ohmichi (1): vmcoreinfo: add the symbol "phys_base" Kim Phillips (2): [POWERPC] 83xx: enable usb in 837x rdb and 83xx defconfigs [POWERPC] sata_fsl: reduce compatibility to fsl,pq-sata Krzysztof Helt (1): acpi thermal: fix result check Kumar Gala (2): [POWERPC] fsldma: Use compatiable binding as spec [POWERPC] Fix defconfigs so we dont set both GENRTC and RTCLIB Larry Finger (1): ssb-pcicore: Fix IRQ TPS flag handling Laurent Pinchart (1): [POWERPC] Fix CPM2 SCC1 clock initialization. Leonardo Potenza (1): [ATM] drivers/atm/iphase.c: compilation warning fix Linus Torvalds (6): parport_pc: make sure to release IO ports after probing for IT87XX Be more careful about marking buffers dirty Revert "ACPI: Ignore _BQC object when registering backlight device" Revert "smc91x: fix build breakage from the SMC_GET_MAC_ADDR API upgrade" Add commentary about the new "asmlinkage_protect()" macro Linux 2.6.25-rc9 Marc Pignat (1): atmel_serial: fix uart/console concurrent access Mark Fasheh (1): Add additional examples in Documentation/spinlocks.txt Mark McLoughlin (3): xen: refactor xen_{alloc,release}_{pt,pd}() xen: Do not pin/unpin PMD pages xen: Clear PG_pinned in release_{pt,pd}() Mathieu Desnoyers (1): markers: use synchronize_sched() Matt Carlson (1): [TG3]: Add PHY workaround for 5784 Matti Linnanvuori (1): net: Comment dev_kfree_skb_irq and dev_kfree_skb_any better Mauro Carvalho Chehab (1): V4L/DVB (7499): v4l/dvb Kconfig: Fix bugzilla #10067 Maxim Shchetynin (1): [POWERPC] Fix rtas_flash procfs interface Michael Abd-El-Malek (1): xen: fix grant table bug Michael Buesch (3): b43: Add DMA mapping failure messages b43: Fix PCMCIA IRQ routing ssb-mipscore: Fix interrupt vectors Michael Ellerman (1): Make pasemi_mac.c depend on PPC_PASEMI to prevent link errors Michael Hennerich (1): spi: spi_bfin5xx: remove unused label Michael Krufky (4): V4L/DVB (7495): s5h1409: fix blown-away bit in function s5h1409_set_gpio V4L/DVB (7496): pvrusb2: add new usb pid for 75xxx models V4L/DVB (7497): pvrusb2: add new usb pid for 73xxx models pvrusb2: fix broken build due to patch order dependency Michael Trimarchi (1): atmel_serial: avoid stopping pdc during transmission Mike Frysinger (3): Blackfin EMAC Driver: delete unused variables to fixup gcc warnings usb net: asix does not really need 10/100mbit spi: spi_bfin5xx build fix Mike Pagano (1): cciss: error: implicit declaration of function 'sg_init_table' Mike Snitzer (1): nbd: prevent sock_xmit from attempting to use a NULL socket Patrick McHardy (3): [LLC]: Restrict LLC sockets to root [XFRM]: xfrm_user: fix selector family initialization [NETFILTER]: xt_hashlimit: fix mask calculation Paul Menage (2): cgroups: add cgroup support for enabling controllers at boot time cgroups: include hierarchy ids in /proc/<pid>/cgroup Paul Mundt (1): i2c-ibm_iic: Fast mode parm desc fixup Pavel Emelyanov (2): [VLAN]: Proc entry is not renamed when vlan device name changes. [VLAN]: Fix egress priority mappings leak. Pavel Machek (1): x86, agpgart: scary messages are fortunately obsolete Pete Zaitcev (1): ub: remove BUG() after __blk_end_request and fix the condition causing it Peter Korsgaard (1): dm9601: Fix multicast hash table handling Ralf Baechle (1): [MIPS] Handle aliases in vmalloc correctly. Randy Dunlap (1): rtc: rtc-s35390a.c needs the bitreverse library Ravikiran G Thirumalai (1): x86: fix breakage of vSMP irq operations René Bürgel (1): [POWERPC] Fix MPC5200 (not B!) device tree so FEC ethernet works Richard Kennedy (1): USB: ohci: fix 2 timers to fire at jiffies + 1s Robert Fitzsimons (3): V4L/DVB (7277): bttv: Re-enabling radio support requires the use of struct bttv_fh V4L/DVB (7278): bttv: Re-enable radio tuner support for VIDIOCGFREQ/VIDIOCSFREQ ioctls V4L/DVB (7400): bttv: Add a radio compat_ioctl file operation Robert P. J. Day (3): lapb: use the shorter LIST_HEAD form for brevity bluetooth: replace deprecated RW_LOCK_UNLOCKED macros efs: update error msg to not refer to deleted read_inode() Robert Spanton (1): USB: serial: ti_usb_3410_5052: Correct TUSB3410 endpoint requirements. Roland McGrath (3): x86 ptrace: avoid unnecessary wrmsr asmlinkage_protect replaces prevent_tail_call asmlinkage_protect sys_io_getevents Roman Zippel (1): HFS+: fix unlink of links Rusty Russell (3): [NET]: Don't send ICMP_FRAG_NEEDED for GSO packets Fix booting pentium+ with dodgy TSC virtio: remove overzealous BUG_ON. Sergei Shtylyov (3): [MIPS] Pb1200: Fix header breakage [MIPS] Make KGDB compile on UP siimage: fix kernel oops on PPC 44x Sreenivasa Honnur (1): S2io: Version update for Tx completion patch Stefano Brivio (1): b43legacy: fix bcm4303 crash Stephen Hemminger (1): sky2: fix suspend/resume races Stephen Smalley (2): selinux: handle files opened with flags 3 by checking ioctl permission SELinux: more GFP_NOFS fixups to prevent selinux from re-entering the fs code Steven Rostedt (1): pop previous section in alternative.c Steven Toth (3): V4L/DVB (7464): Convert driver to use a single SRAM memory map V4L/DVB (7465): Fix eeprom parsing and errors on the HVR1800 products V4L/DVB (7466): Avoid minor model number warning when an OEM HVR1250 board is detected Suresh Siddha (1): x86: fix 64-bit asm NOPS for CONFIG_GENERIC_CPU Sven Schnelle (3): afs: add missing up_write() on return afs: remove smp_prcessor_id() from debug macro rxrpc: remove smp_processor_id() from debug macro Tejun Heo (3): libata: uninline atapi_cmd_type() libata: ATA_12/16 doesn't fall into ATAPI_MISC pata_ali: disable ATAPI DMA Thomas Gleixner (2): x86: tsc prevent time going backwards x86: revert assign IRQs to hpet timer Thomas Graf (1): [ESP]: Ensure IV is in linear part of the skb to avoid BUG() due to OOB access Till Harbaum (1): i2c-tiny-usb: New VID/PID pair Tobias Mueller (1): Input: appletouch - add product IDs for the 4th generation MacBooks Trond Myklebust (1): SUNRPC: don't call flush_dcache_page() with an invalid pointer Troy Kisky (1): i2c-davinci: Fix lost interrupt Valentine Barshak (1): ibm_newemac: emac_tx_csum typo fix Venki Pallipadi (1): x86: Simplify cpu_idle_wait Vitja Makarov (1): spi: spi_bfin5xx: fix probe() sequencing Vladimir Koutny (2): mac80211: correct use_short_preamble handling mac80211: use short_preamble mode from capability if ERP IE not present Will Newton (1): Fix a typo in highres.txt YOSHIFUJI Hideaki (2): [IPV6] MCAST: Ensure to check multicast listener(s). [IP] UDP: Use SEQ_START_TOKEN. --
Glad
I am glad that Linus slightly delays the release when he feels that it is whats needed. Rather than to release it, when its not ready.
Dur
"I am glad that Linus slightly delays the release when he feels that it is whats needed. Rather than to release it, when its not ready."
I think you missunderstand. Linus _WANTED_ to release it. He'd just take more flak than he'd care for if he did make a release with memory management regressions.