2003-05-05 00:37:37

by Linus Torvalds

[permalink] [raw]
Subject: Linux 2.5.69


Ok,
I finally found the reason for why some of my machines had trouble with
restarting the X server, and it turns out that it's been around since very
early February. I bet others must have seen it too, with random crashes on
X server restart when the server used AGP (which means that it mainly hit
either hw-accelerated 3D setups or the intel integrated graphics which use
a UMA model with AGP as the backing store).

That's a big relief for me, as it was the major thing I personally worried
about for 2.6.x.

Anyway, that's fixed here, along with a lot of other updates. Much of
2.5.69 is small one-liners to drivers to handle the new IRQ semantics, but
there's a lot of other cleanups in there too (Christoph Hellwig continued
on his devfs rampage, for example).

NOTE! As of this release I think I'll want to have patches either be
_really_ obvious, or they should go through one of more people for
approval. In particular, I'm hoping that the paperwork stuff with Andrew
should be getting closer to finalized, and that we could start moving over
towards a 2.6.x release schedule..

Linus

---
Summary of changes from v2.5.68 to v2.5.69
============================================

<atulm:lsil.com>:
o Update megaraid to version 2.03

<ccheney:cheney.cx>:
o USB: vicam.c copyright patches

<dmo:osdl.org>:
o 2.5.68 scsi/gdth compile warnings and stack usage

<gj:pointblue.com.pl>:
o USB: fix usbkbd.c compilation error

<james:superbug.demon.co.uk>:
o USB: Add support for Pentax Still Camera to linux kernel

<l.s.r:web.de>:
o Remove unused function from fs/isofs/rock.c

<linux-usb:gemeinhardt.info>:
o USB: add support for Mello MP3 Player

<mikenc:us.ibm.com>:
o fixes compile errors in psi240i.c

<nstraz:sgi.com>:
o [XFS] Use "%p" to print out addresses from xfs_error_report().
This is so addresses don't get truncated on 64-bit archs.

<pixi:burble.org>:
o [quota] provide no-op sync_dquots_dev, one .config case wants it

<ralphs:org.rmk.(none)>:
o [NWFPE] Performance improvements [Parts 1-6]

Richard Henderson:
o Fix unwind info for sysenter entry point

<scole:zianet.com>:
o [NET]: Spelling fixes for net/

<valdis.kletnieks:vt.edu>:
o cpp cleanups for ia32/io_apic.c, sound/oss/trident.c
o cpp cleanups: use KERNEL_VERSION macro from linux/version.h
o [netdrvr typhoon] s/#if/#ifdef/ for a CONFIG_ var

<vinay-rc:naturesoft.net>:
o [NET]: Use mod_timer in dst.c
o [PKT_SCHED]: Use mod_timer in sch_cbq.c
o [PKT_SCHED]: Use mod_timer in sch_csz.c
o [PKT_SCHED]: Use mod_timer in sch_htb.c

Alan Stern:
o Trivial patch for scsi logging text string
o USB: Minor patch for uhci-hcd.c

Alex Williamson:
o 8250_pci include offset in iomap_base

Alexander Schulz:
o [ARM PATCH] 1517/1: Shark: new defconfig
o [ARM PATCH] 1518/1: Shark: cyberpro broken by machine_is_netwinder

Alexander Viro:
o tty cleanups (1-12)
o console cleanups (1-2)
o fbdev cleanup
o capifs cleanup
o invalidate_device()/check_disk_change() fixes
o ppc boot device selection cleanup
o simple_fill_super()
o pin_fs/release_fs
o open_by_devnum()
o blkmtd init cleanup
o bdget_disk()
o ataflop.c cleanup
o hd98 compile fixes

Andi Kleen:
o Runtime memory barrier patching
o Minor 32bit Opteron fixes
o Update alt_instr to handle SSE2 prefetch and better nops
o Fix prefetch patching in 2.5-bk
o x86-64 update
o discontigmem fix

Andrew Morton:
o 3c574_cs fixes
o Fix nc98 partition parser link error
o dmfe: don't free skb with local interrupts disabled
o dentry_stat accounting fix
o Fix and clean up DCACHE_REFERENCED usage
o Fix POSIX timers to give CLOCK_MONOTONIC full
o Fix jiffies_to_time[spec | val] and converse to use
o get_offset_pit and do_timer_overflow vs IRQ locking
o detect_lost_tick locking fixes
o Minor fix for driver/serial/core.c
o keyboard.c Fix SAK in raw mode
o Make PCI scanning order the same as 2.4
o Turn on NUMA rebalancing
o Move __set_page_dirty_buffers to fs/buffer.c
o Clean up various buffer-head dependencies
o follow_hugetlb_page fix
o hugetlb math overflow fix
o ATI Mach64 build fix
o quotactl(): sync all quotas
o AIO mmap fix
o shmdt() speedup
o implement __GFP_REPEAT, __GFP_NOFAIL, __GFP_NORETRY
o make alloc_buffer_head take gfp_flags
o use __GFP_REPEAT in pte_alloc_one()
o use __GFP_REPEAT in pmd_alloc_one()
o Disallow swapoff if there is insufficient memory
o Permit interruption of swapoff
o oom-kill: preferentially kill swapoff
o DAC960: add call to blk_queue_bounce_limit
o shm_get_stat-handle-hugetlb-pages.patch
o Allocate hd_structs dynamically
o fix CONFIG_NOMMU mismerges
o Extend map_vm_area()/get_vm_area()
o don't shrink slab for highmem allocations
o prepare device mapper for larger dev_t
o smbfs: larger dev_t preparation
o Fix nfsctl for larger dev_t
o Aggregated disk statistics
o fbdev build fix
o irqs: drivers/block
o irqs: sym2
o irqs: rtc
o irqs in sound/
o irqs: ipmi driver
o irqs: watchdog drivers
o irqs: various char drivers
o irqs: multimedia drivers
o irqs: video drivers
o irqs: 1394
o parport_serial fix
o ax25 build fix
o irqs: IRDA
o irqs: ISDN
o irqs: input drivers
o irqs: hotplug drivers
o More careful about VMA merging
o usb: minor usb stuff
o print IRQ handler addresses
o warning fixes
o fix typo in m68k mm code
o irqs: scsi
o Fix IRQ_NONE clash
o irqs: ATM
o irqs: drivers/block
o irqs: char drivers
o irqs: scsi
o sound driver fixes
o CPU flags fixes
o various irqreturn_t fixes
o parkbd.c jiffies fix
o watchdog driver compile fixes
o bttv warning fix
o jiffy type warning fixes
o net driver cleanup, volume 7
o [NETFILTER]: Put back missing list_head iterator local var
o irqs: i2c
o irqs: IRDA
o Fix slab-vs-gfp bitflag clash
o irqs: bttv
o APM locking fix
o Fix warnings in xd.c
o DAC960 patch to entry points with a new fix
o allow modular JBD
o generic HDLC module API update
o proc_file_read fix
o buffer.c unused vars
o simple mwave code cleanup
o fs/ext3/super.c fix for orphan recovery error path
o update nr_threads commentary
o lost_tick fixes
o zone accounting race fix
o aio support for block devices
o percpu counters cause UML compilation errors in with SMP
o config menu cleanups
o oom-killer locking fix
o cs46xx: fix incomplete search-and-replace

Andy Grover:
o ACPI: interpreter update to 20030418
o ACPI: Fix link devices on SMP systems (Dan Zink)
o ACPI: Add missing include
o ACPI: Indicate whether we handled the interrupt or not

Anton Altaparmakov:
o NTFS: Add handling for initialized_size != data_size in compressed
files
o NTFS: Update version
o NTFS: Minor updates
o NTFS: Remove compile warning for newer gcc
o unistr.c
o NTFS: Reduce function local stack usage from 0x3d4 bytes to just
noise in fs/ntfs/upcase.c. (Randy Dunlap <[email protected]>)
o NTFS: 2.1.2 release: Fix buggy free cluster and free inode
determination logic
o NTFS: Fix silly porting typo
o NTFS: Typo fix
o NTFS: Fix compiler warnings on big endian machines
o super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
clusters (Philipp Thomas)
o NTFS: load_attribute_list() bug fix from Szaka
o NTFS: Fix typo and release 2.1.3
o NTFS: 2.1.4 release - Reduce compiler requirements

Anton Blanchard:
o ppc64 needs setup-bus.c
o [netdrvr 8139cp] enable MWI via pci_set_mwi, rather than manually

Arnaldo Carvalho de Melo:
o net: new module infrastructure for net_proto_family
o rtnetlink: use C99 struct init style
o atm/lec.c: use C99 struct init style
o rtnetlink_dev: use C99 struct init style
o net: module refcounting for sk_alloc/sk_free
o net/socket: make sys_accept bump the net proto family module
usage count
o ipx: remove MOD_{INC,DEC}_USE_COUNT
o llc: remove MOD_{INC,DEC}_USE_COUNT
o af_llc: initialize ->owner in llc_ui_family_ops
o appletalk: remove MOD_{INC,DEC}_USE_COUNT
o af_llc: add missing include module.h
o af_unix: remove MOD_{INC,DEC}_USE_COUNT
o wireless: make the ioctl tables more resilient to errors using
C99 style init
o af_ax25: remove MOD_{INC,DEC}_USE_COUNT
o af_econet: remove MOD_{INC,DEC}_USE_COUNT
o af_irda: remove MOD_{INC,DEC}_USE_COUNT
o af_key: remove MOD_{INC,DEC}_USE_COUNT
o af_netrom: remove MOD_{INC,DEC}_USE_COUNT
o af_packet: remove MOD_{INC,DEC}_USE_COUNT
o af_rose: remove MOD_{INC,DEC}_USE_COUNT
o af_wanpipe: remove MOD_{INC,DEC}_USE_COUNT
o af_x25: remove MOD_{INC,DEC}_USE_COUNT
o netrom/nr_dev: use SET_MODULE_OWNER, removing calls to
MOD_{INC,DEC}_USE_COUNT
o rose/rose_dev: use SET_MODULE_OWNER, removing calls to
MOD_{INC,DEC}_USE_COUNT
o net: several C99 struct init style conversions and cleanups
o net: save some more bytes in the kernel image moving global zero
inits to .bss
o af_decnet: remove MOD_{INC,DEC}_USE_COUNT
o ipx: several simple cleanups
o pppox: simple code cleanups
o af_pppox: create module infrastructure for protocol modules
o af_pppox: return -EPROTONOSUPPORT if try_module_get fails at
pppox_create
o net/socket: return -EAFNOSUPPORT if net_family_get fails at
sock_create and sys_accept
o net/llc: simple cleanups
o net/sched: some trivial code cleanups, making some code smaller
o net/core/dev: fix obvious bug in dev_get_idx
o net/core/dev: add missing ++*pos in dev_seq_next
o net/core/dev: another fix for the seq_file handling of
/proc/net/dev
o net: improve the current module infrastructure
o pppoe: use revised net module infrastructure
o bluetooth: use revised net module infrastructure
o appletalk: use revised net module infrastructure
o ax25: use revised net module infrastructure
o decnet: use revised net module infrastructure
o econet: use revised net module infrastructure
o ipx: use revised net module infrastructure
o irda: use revised net module infrastructure
o af_key: use revised net module infrastructure
o llc: use revised net module infrastructure
o netlink: use revised net module infrastructure
o netrom: use revised net module infrastructure
o packet: use revised net module infrastructure
o rose: use revised net module infrastructure
o unix: use revised net module infrastructure
o wanpipe: use revised net module infrastructure
o x25: use revised net module infrastructure
o sctp: use revised net module infrastructure
o ipv6: use revised net module infrastructure
o ipv4: use revised net module infrastructure
o vlan: fix comment about understanding shared skbs

Art Haas:
o C99 initializers for drivers/scsi
o C99 initializers for drivers/block/genhd.c
o Fix C99 initializers in fs/nfs/nfs4proc.c
o C99 initializers for fs/proc/proc_misc.c

Bart De Schuymer:
o [NETFILTER]: Add ipt_physdev extension
o [BRIDGE]: Always set BRNF_BRIDGED mask when bridging
o [NETFILTER]: Possible use of freed skbuff in netfilter.c
o [EBTABLES]: Add ebtables match for the pkt_type member of an skbuff
o [EBTABLES]: Add ARP MAC address filtering

Bartlomiej Zolnierkiewicz:
o fix init_irq
o fix mismatched access_ok() checks in sg_io()
o fix DMA for taskfile IO
o fix compilation of taskfile IO
o remove duplicated defines from ide.h
o Remove duplication of generic ide funcs from ide-taskfile.c
o Kill dups of read_24(), rename it to ide_read_24()
o make floppy driver useable for 2.5

Ben Collins:
o 1394 updates
o IEEE1394/Firewire updates
o Merge to current SVN repo (r915)
o [SPARC64]: Fix ioctl32.c in latest BK
o [VIDEO]: Revert cfbimgblt.c back to a working state on 64-bit
o [VIDEO]: Revert atyfb back to known working clean base
o Fix compat_ioctl
o add ieee1394 module dev table
o ieee1394 update (r925)

Benjamin Herrenschmidt:
o PPC32: Handle CPUs that have extra BAT (block address translation)
registers
o PPC32: flush the cache more thoroughly on sleep
o PPC32: Updates for newer PowerMac/PowerBook machines
o PPC32: Fix for older SMP powermacs

Christoph Hellwig:
o devfs: remove devfs_unregister
o devfs: switch over ubd to ->devfs_name
o devfs: remove dead devfs code in dasd
o devfs: superflous devfs_remove in scsi
o devfs: introduce devfs_mk_bdev
o devfs: gendisk.devfs_name updates
o devfs: kill devfs_register_partition
o devfs: warn on block modes in devfs_register
o devfs: convert input, dvb, usb, sound, videodev, miscdev, s390,
ipmi, swim3, uml, loop, nbd, rd, swim3, aztcd, gscd, optcd, sjcd,
sonycd, mtdblock, xpram, floppy, device-mapper, md
o initrd.h
o rename end_request in floppy() and raid1
o replace __blk_run_queue with blk_run_queue
o remove dasd_get_kdev
o remove some junk from hd98.c's ioctl implementation
o remove a tiny bit of kdev_t abuse from the floppy driver
o scsi_scan.c coding style fixes
o unexport scsi_host_get_next
o kill ASSERT_LOCK
o i2c: remove dead junk from i2c-sensors.h
o i2c: remove dead code from adm1021
o i2c: remove dead init code from i2c-sensors.c
o i2c: bring i2c-viapro uptodate with the style guide
o [PCMCIA] remove unused files
o split initrd from ramdisk driver
o kill LOCAL_END_REQUEST
o don't use mem_map_reserve/mem_map_unreserve
o don't include devfs_fs_kernel.h in global headers
o fix devfs_mk_dir prototype
o update s390 tape_block for 2.5 APIs
o remove a wrong invalidate_bdev from ide-disk.c
o fix dasd open/release
o remove proc_print_scsidevice abuse from drivers
o Fix devfs botch in IDE naming
o kill <linux/wrapper.h>
o use file->private_data in ide-tape
o [PCMCIA] consolidate cs_error()
o de-uglify scsi.c
o [netdrvr pcmcia] switch drivers to using pcmcia_register_driver
o fix devfs_register_tape stub
o update dcache documentation
o improved bdevname
o use .devfs_name in struct miscdevice
o remove devfs hack from misc_register
o add an missing prototype to initrd.h
o switch drivers/input/serio/serport.c to new-style module handling
o remove useless MOD_{INC,DEC}_USE_COUNT from sunrpc
o make __bdevname output more similar to bdevname
o make <linux/blk.h> obsolete

Dave Kleikamp:
o JFS: Avoid rare deadlock

David Brownell:
o USB: fix for deadlock in v2.5.67
o USB: hcd-pci.c catch up to dev_printk changes
o usb: fix (rare?) disconnect
o USB: usbnet, config changes for CDC Ether

David S. Miller:
o [PKT_SCHED]: Proper module refcounting for packet classifiers
o [PKT_SCHED]: Proper module refcounting for packet schedulers
o [SPARC64]: A few missing pgtable __GFP_REPEAT
o [SPARC]: Rename signal macros SV_foo --> _SV_foo
o [NET]: In sock_alloc_send_pskb, add __GFP_REPEAT when __GFP_WAIT
o [SPARC]: setup.c needs linux/initrd.h
o [SOUND SPARC]: Update for irqreturn_t
o [SPARC]: Fix dumb typo in sun4c mm code
o [SPARC]: Platform code changes for irqreturn_t
o [SERIAL SPARC]: Update for irqreturn_t
o [SOUND]: mpu401.h needs linux/interrupt.h
o [CHAR SPARC]: Update for irqreturn_t
o [RTC]: Update for irqreturn_t
o [FC4 SPARC]: Update for irqreturn_t
o [MESSAGE FUSION]: Update for irqreturn_t
o [SCSI ESP]: Update for irqreturn_t
o [SCSI QLOGICFC]: Update for irqreturn_t
o [SCSI QLOGICISP]: Update for irqreturn_t
o [SCSI QLOGICPTI]: Update for irqreturn_t
o [SCSI AIC7XXX_OLD]: Update for irqreturn_t
o [SCSI SYM53C8XX_2]: Update for irqreturn_t
o [TG3]: Update to irqreturn_t
o [NET SUN]: Update for irqreturn_t
o [MYRI_SBUS]: Update for irqreturn_t
o [SOUND PCI]: Update several drivers for irqreturn_t
o [SOUND TRIDENT]: Update for irqreturn_t
o [SPARC64]: Update defconfig
o [NETLINK]: Fix minor numbers in netlink_dev.c
o [SPARC]: CLOCK_MONOTONIC fixes, from x86
o [SPARC64]: Add LOOP_{GET,SET}_STATUS64 to ioctl32
o [NET]: Do not let GCC reload pointers after NULL checks
o [SPARC64]: Kill unnecessary MOD_{INC,DEC}_USE_COUNT in cpwatchdog
and envctrl drivers
o [NET]: SG without checksum support is illegal
o [USB INPUT]: hiddev.c needs dev_fs_kernel.h
o [SCTP]: ICMP6 per-device changes for sctp
o [IPV6]: Export in6_dev_finish_destroy
o [BRIDGE]: br_if.c needs linux/init.h
o [EBTABLES]: Make ebt_vlan.c use correct printf format for size_t
o [DECNET]: Kill warning with gcc-3.x in dn_route.c
o [NETFILTER]: Make ip_conntrack_core.c use correct printf format for
size_t
o [SPARC64]: Update defconfig
o [NETFILTER]: Kill unused var in nf_reinject
o [NETFILTER]: Use proper size_t printf format in ip6t_LOG.c
o [IPV6]: Kill unused vars in mcast procfs code
o [IPV4]: Use dst_pmtu not dev->mtu to determine if fragmentation is
needed
o [IPV4]: Fix typos in ipip.c commented out code
o [PKT SCHED]; Missing semicolon in acme cleanups
o [SPARC64]: Update defconfig
o [NET]: Fix hashing exploits in ipv4 routing, IP conntrack, and TCP
synq
o [IPV4]: Fix ip_rt_acct reading
o [IPV4]: Fix typo in hashing changes
o [IPV4]: Add missing init_timer for rt_secret_timer
o [OPROFILE]: timer_int.c needs profile.h and init.h
o [SPARC64]: Update defconfig
o [SPARC64]: oprofile/init.c needs errno.h

David Stevens:
o [IGMPv3/MPDv2]: Bug fixes and ipv4 multiprotocol API
o [IGMP]: Fix bug in broadcast handling

Douglas Gilbert:
o scsi_mid_low_api.txt update for 2.5.67

Duncan Sands:
o USB speedtouch: bump the version number
o USB speedtouch: crc optimization
o USB speedtouch: compile fix

Edward Peng:
o [netdrvr via-rhine] fix promisc mode
o [netdrvr sundance] bug fixes, VLAN support

Eli Carter:
o [ARM PATCH] 1508/1: use #define's for iq80321
o [ARM PATCH] 1511/1: iop321 #define cleanup
o [ARM PATCH] 1510/1: use a #define for asm jump address
o [ARM PATCH] 1513/1: iq80310 fix missing header
o [ARM PATCH] 1514/1: iq80321 MTD C99 fix

Eric Brower:
o [SPARC]: Refactor AUXIO support

Eric Sandeen:
o [XFS] Make MODULE_AUTHOR consistent with other SGI modules

Florin Iucha:
o i2c: added it87 driver

Fran?ois Romieu:
o [wan dscc4] irqreturn_t update
o [DECNET]: Fix build with CONFIG_DECNET_ROUTE_FWMARK enabled

Ganesh Venkatesan:
o [netdrvr ixgb] add new driver for Intel's 10 gig ethernet

Greg Kroah-Hartman:
o USB: add support for new tty tiocmget and tiocmset functions
o i2c: fix up it87.c check_region mess
o i2c: removed unused flags paramater in found_proc callback
o i2c: fix up the media drivers due to removing flags paramater of
callback function
o i2c: removed unneeded typedef from i2c-sensor.h
o i2c: remove a lot of dupliated macros from i2c-sensor.h and use the
current values in i2c.h
o USB: add error reporting functionality to the pl2303 driver
o tty: let tiocmset pass TIOCM_LOOP changes to the tty drivers
o kobject: kobj_lock needs to be grabed using spinlock_irq
o driver core: rework driver class structures and logic
o driver core: fix up cpu.c, memblk.c, and node.c due to the class
changes
o driver core: fix up the input_class logic due to the class changes
o driver core: fix up cpufreq code to work with new class changes
o driver core: fix up tty code to work with the new class changes
o driver core: fix up scsi code to compile due to the class changes
o driver core: fix up the pcmcia code to work with the new class
changes
o driver core: removed drivers/base/fs/*, drivers/base/intf.c and
drivers/base/hotplug.c
o USB: fix CHECKER found bug in the empeg.c driver
o USB: fix CHECKER found bug in the io_edgeport.c driver
o USB: fix CHECKER found bug in the ipaq.c driver
o USB: fix CHECKER found bug in the keyspan.c driver
o USB: create usb_init_urb() for those people who like to live
dangerously (like the bluetooth stack.)
o USB: added support for Sony DSC-P8
o USB: add comment to storage/unusual_devs.h that specifies how to
add new entries

Harald Welte:
o [NETFILTER]: Makefile and build fixes
o [NETFILTER]: Trivial but important state fix for ipt_conntrack

Hideaki Yoshifuji:
o [IPSEC]: nexthdr in xfrm6_input needs to be int
o [IPV6]: dst_alloc() clean-up
o [IPV6]: SNMP6 clean-up
o [IPV6]: Per-interface statistics infrastructure

James Bottomley:
o Fix megaraid compile warnings
o Fix megaraid module ownership
o scsi_scan.c: cope with second inquiry failure
o irqreturn_t fixup for 53c700
o Add irqreturn_t to scsi/psi240i
o Fix ncr53c8xx for PA-RISC Zalon SCSI driver
o Compile fix for 53c700 on PA-RISC
o convert Megaraid to irqreturn_t
o Fix mismerge in megaraid.c

James Morris:
o [IPSEC]: allow only tunnel mode in xfrm4_tunnels
o [IPSEC]: pmtu discovery support at local tunnel gateway
o [NET]: Cosmetic cleanups of jhash code
o [IPV4]: Choose new rt_hash_rnd every rt_run_flush
o [IPSEC]: Consolidate some output code into xfrm_check_output

Jeff Garzik:
o Modernize rcpci45 I2O LAN driver (#204)
o [rcpci45] typo fix: s/virual/virtual/
o net driver cleanup, volumes 1-6
o fix printk when an irq doesn't get responded to
o [netdrvr tg3] detect shared (and screaming) interrupts
o [netdrvr tg3] fix omission in board shutdown sequence
o s/#if/#ifdef/ for a few CONFIG_SMP tests in public headers
o [hw_random] fix bug, bump version
o [netdrvr ixgb] Lindent, then fix up obvious indent uglies by hand
o [netdrvr ixgb] use standard kernel u8/u16/u32 types
o [netdrvr ixgb] more cleanups

Jeff Smith:
o [NETFILTER IPV4]: Fix typo in Kconfig

Jens Axboe:
o request structure stack corruption
o cleanup bio_map_user and helper

Joe Perches:
o USB: fix up usbnet's macros for older compilers
o USB: fix up usb_serial.h's dbg macro to take up less space
o USB: fix up usb.h's dbg macro to take up less space

John Levon:
o OProfile updates

Keith M. Wesolowski:
o [SPARC]: Replace "magic" values

Linus Torvalds:
o Fix-ups for i830 from Arjan
o Interrupt handlers should return whether the interrupt was for them
or not, so that the irq subsystem can properly handle screaming
shared interrupts.
o Update ensoniq driver to return whether the interrupt was for it
o Fix IO-APIC vector allocation boundary case - we never want to
allocate FIRST_SYSTEM_VECTOR as an external interrupt. It's
unlikely, but could happen if we have a _ton_ of interrupt sources.
o Return IRQ_NONE for ieee1394 driver when the interrupt was for
somebody else.
o Allow gcc to generate better code for irq handling
o Add the Xeon variations (Pentium-III and P4-based) to the list
o Fix irq event debug print-out, and add stack dump which can give a
clue about what the context was that might have caused the spurious
interrupt.
o Never merge vma's that have mapping-private data
o Since "apply_alternatives()" also runs at module load time it must
not be marked __init.
o Merge with DRI CVS tree: remove stale old context switching code
and DMA histogramming. Be more careful about DMA page-list
allocations, and remove old and broken (not SMP-safe, and unused)
DRM read(), write() and poll() support.
o 'hw_status_page' looks like a pointer, quacks like a pointer and
walks like a pointer. It _is_ a pointer. So make it one, and remove
a lot of silly casts.
o Fix up some mixing of ramdisk/initrd. They have nothing in common,
but the build was confused by the fact that they did share some
files.
o Avoid warning: print out hw_status_page as the pointer it now is
o scsi.c needs <linux/interrupt.h>. Somebody was a bit over-eager at
cleanups.
o DRI CVS merge: move more gamma-only functions away from generic dri
files and into gamma driver files.
o Previous DRI CVS merge improperly removed some sparc-only support.
Add it back in now that DRI is synched up again.
o Remove old (disabled) debugging code
o DRI CVS merge: make sure to clean up irq and DMA on final close
o DRI CVS merge: only free pages when we _have_ pages to free
o DRI CVS merge: make sure the device is properly initialized before
opening it.
o DRI CVS merge: memory barrier updates
o DRI texmem branch merge cleanups. Texture ages are unsigned, and
radeon should use generic texture structure now.
o PCI ID's for Quadrics from Daniel Blueman
o Revert pmd/pgd slabification. wli will fix it properly

Maksim Krasnyanskiy:
o [Bluetooth] Use very short disconnect timeout for SCO connections.
They cannot be reused and therefor there is no need to keep them
around.
o [Bluetooth] Kill incoming SCO connection when SCO socket is closed
o [Bluetooth] HCI USB driver update. Support for SCO over HCI USB
o [Bluetooth] Don't forget to set HCI device owner in USB driver
o [Bluetooth] Update BT PCMCIA drivers to use
pcmcia_register_driver()
o [Bluetooth] Improved RFCOMM TTY TX buffer management
o [Bluetooth] Fix race condition in RFCOMM session and dcl scheduler
o [Bluetooth] USB drivers cannot call usb_unlink_urb() under spin
lock
o [Bluetooth] Initialize net_proto_family->owner field. This covers
only HCI sockets
o [Bluetooth] Initialize ->owner field of the RFCOMM tty driver

Marc Zyngier:
o Convert Alpha to the new 2.5 IRQ API
o EISA/sysfs update

Marcel Holtmann:
o [Bluetooth] Add support for the Ultraport Module from IBM
o [Bluetooth] Use R1 for default value of pscan_rep_mode
o [Bluetooth] Respond correctly to RLS packets
o [Bluetooth] Fix L2CAP binding to local address
o [Bluetooth] Correction of the HCI USB driver description

Matthew Wilcox:
o fix iomem_resource

Michael Hunold:
o Fix mxb.c stack usage

Mike Anderson:
o 2.5.67+ scsi_release_request call queue next

Muli Ben-Yehuda:
o [NETFILTER]: ip_queue memory leaks

Nathan Scott:
o [XFS] UUID cleanup - remove unused functions, create a decent table
abstraction and make the mount code simpler in the process.
o [XFS] Fix build for big endian platforms; make
xfs_xlate_dinode_core consistent
o [XFS] Add a validity check for unwritten extents, trying to trap a
problem
o [XFS] Fix compile for Alpha architecture

Neil Brown:
o Update umem to new request_irq interface
o Update umem driver for newer cards

Nivedita Singhvi:
o [AF_UNIX]: Fix max_dgram_qlen procfs permissions

Oliver Neukum:
o add DC395 SCSI driver

Patrick Mansfield:
o fix ppa locking and oops
o scsi-misc-2.5 fix repeat_inquiry bflags setting
o scsi-misc-2.5 remove scsi_scan.c EVPD code

Patrick McHardy:
o [NETFILTER]: Multiple ipt_REJECT fixes

Paul Fulghum:
o synclink update
o synclinkmp update
o synclink_cs update
o n_hdlc update
o Added new PCI ID
o pci.ids update

Paul Mackerras:
o PPC32: Change interrupt handlers to return irqreturn_t
o PPC32: Reduce __MAX_NDELAY a little to avoid compiler warnings
o drivers/macintosh irq handler type
o PPC32: Move xmon declarations to their own header file
o [PPP]: Module owners for ppp compressors

Pavel Machek:
o Fix SWSUSP & !SWAP
o ioctl32 cleanups
o ioctl32: leftovers

Pavel Roskin:
o [PCMCIA] Fix compilation of cardmgr
o [PCMCIA] Fix oops in validate_mem when CONFIG_PCMCIA_PROBE=n

Pete Zaitcev:
o [SPARC]: Colin Gibbs gcc-3.x support
o [SPARC]: Openprom drivers needs linux/fs.h
o [SPARC]: The iommu rewrite

Randy Dunlap:
o replace URLs in Kconfig
o [IPV6]: Per-interfave icmpv6 statistics support
o sidewinder: reduce stack usage
o uinput.c: reduce stack usage

Richard Henderson:
o Fix unwind info for sysenter entry point

Rik van Riel:
o [wireless airo] make end-of-array test more portable

Rob Radez:
o [SPARC]: Kill initialize_secondary, unused

Robert Love:
o trivial task_prio() fix

Robert Olsson:
o [NET]: Remove skb_head_pool

Roland McGrath:
o i386 vsyscall DSO implementation
o Fix the DSO patch
o allow ptrace and /proc/PID/mem to read fixmap pages

Russell Cattelan:
o [XFS] Whitespace cleanup Merge whitespace cleanup to 2.5 tree
o [XFS] Rework the way xfs includes xfs_<blah>.h headers

Russell King:
o [ARM] NWFPE 1: Convert instruction decoding from switch() to table
o [ARM] NWFPE 2: Take advantage of the CPDO functions behaviour
o [ARM] NWFPE 3: Eliminate setting of fType in CPDO worker functions
o [ARM] NWFPE 4: Eliminate getFd from CPDO worker functions
o [ARM] NWFPE 5: Eliminate use of Fd
o [NWFPE] Clean up indentation in assembly files
o [PCMCIA] Don't cache CIS bytes found to be invalid
o [PCMCIA] Make cb_release_cis_mem() local to cardbus.c
o [ARM] Fix two makefile problems
o [ARM] Bypass cache cleaning if cache/mmu was disabled
o [ARM] Fix another case of looking at task_struct instead of
thread_info
o [ARM] Provide more early command line parsing
o [ARM] lock up() functions should be memory barriers
o [ARM] Ensure gcc does not assume asm() is conditional
o [ARM] Fix integrator cpufreq build errors
o [ARM] Fix includes
o [ARM] Make tlb_start_vma() flush the cache
o [ARM] Inline PMD entry cache handling
o [ARM] Clean up ARM cache handling interfaces (part 1)
o [ARM] Part 2 in the cache API changes
o [ARM] Remove check_bugs()
o [ARM] set_pgd is confusing; rename it switch_mm
o [ARM] Clean up nwfpe makefile
o [ARM] Don't allow FPE modules to be built as a module
o [ARM] Remove unused msleep() function in h3600.c
o [ARM] Switch to SVC mode using read/modify/write
o [ARM] Fix a collection of missed changes from cache API changes
o [ARM] Fix elf_fpregset_t
o [ARM] Update mach-types to latest version

Rusty Russell:
o [NETFILTER]: Add owner field to nf_hook_ops
o complete modinfo section
o __module_get

Scott Feldman:
o [netdrvr e1000] mark e1000 NAPI feature not-experimental
o [netdrvr e1000] add a bit of source cross-version compat

Shachar Shemesh:
o Fix IRDA irq handler prototype

Stephen Hemminger:
o [BRIDGE]: New maintainership
o [BRIDGE]: Missing unlocks in ioctl error paths
o [BRIDGE]: Bridge confuses kernel user HZ
o [BRIDGE]: Get write lock in config PDU processing
o [BRIDGE]: Possible race with timer on shutdown
o [BRIDGE]: Use list macros for ports
o [BRIDGE]: Use RCU for port table
o [BRIDGE]: Use C99 initializers for netfilter bridge
o [NETFILTER]: Use Read Copy Update
o [BRIDGE]: Inline and _rcu change
o [BRIDGE}: More user hz conversions
o Replace br_lock() in snap with Read Copy Update
o [BRIDGE}: Change bridge forwarding table to use hlist

Stephen Lord:
o [XFS] Rework the remount path to better seperate the linux vfs
portion and the xfs portion of it. Move the code to more
appropriate places in the tree.
o [XFS] Fix a use after free in the unwritten extent code. Also
rework the interface to the allocator to have its own flag set, and
always go through the same interface in all cases rather than
having unwritten extent requests take a different path from all
others.
o [XFS] report extended attribute existence in the xattr flags field

Steve French:
o Add resume key support for readdir to workaround Windows 2000 and
XP server problem. Update oplock handling code. Reduce excessive
stack usage in link.c
o fix readdir on empty directories to only issue one network search
o Unload nls if mount fails
o fix hang in truncate setting file size
o Fix delete of files with readonly attribute. Reflect setting of
readonly dos attribute in mode when server does not support CIFS
Unix extensions. Fix abbreviated readdir to servers that do
support CIFS Unix extensions.

Steven Cole:
o [ARM] spelling fixes for arm
o [SPARC64]: Spelling fixes
o Avast there ye swabs, prepare to fire a broadside!

Steven Whitehouse:
o [IP_GRE]: Kill duplicate update_pmtu call

Tom Rini:
o PPC32: Correct BASE_BAUD on IBM Redwood platforms



2003-05-05 04:06:19

by Voluspa

[permalink] [raw]
Subject: Re: Linux 2.5.69


Quick compile crash on a PII400, gcc version 2.95.3 20010315 (release). Have compiled previous 2.5.xx since about 65 cleanly.

[...]

AME=sysenter -c -o arch/i386/kernel/sysenter.o arch/i386/kernel/sysenter.c
gcc -Wp,-MD,arch/i386/kernel/.vsyscall-int80.o.d -D__ASSEMBLY__ -D__KERNEL__ -Iinclude -Iinclude/asm-i386/mach-default -nostdinc -iwithprefix include -traditional -c -o arch/i386/kernel/vsyscall-int80.o arch/i386/kernel/vsyscall-int80.S
gcc -nostdlib -shared -s -Wl,-soname=linux-vsyscall.so.1 \
-o arch/i386/kernel/vsyscall-int80.so -Wl,-T,arch/i386/kernel/vsyscall.lds arch/i386/kernel/vsyscall-int80.o
gcc -Wp,-MD,arch/i386/kernel/.vsyscall-sysenter.o.d -D__ASSEMBLY__ -D__KERNEL__ -Iinclude -Iinclude/asm-i386/mach-default -nostdinc -iwithprefix include -traditional -c -o arch/i386/kernel/vsyscall-sysenter.o arch/i386/kernel/vsyscall-sysenter.S
gcc -nostdlib -shared -s -Wl,-soname=linux-vsyscall.so.1 \
-o arch/i386/kernel/vsyscall-sysenter.so -Wl,-T,arch/i386/kernel/vsyscall.lds arch/i386/kernel/vsyscall-sysenter.o
gcc -Wp,-MD,arch/i386/kernel/.vsyscall.o.d -D__ASSEMBLY__ -D__KERNEL__ -Iinclude -Iinclude/asm-i386/mach-default -nostdinc -iwithprefix include -traditional -c -o arch/i386/kernel/vsyscall.o arch/i386/kernel/vsyscall.S
/tmp/ccKug4Ma.s: Assembler messages:
/tmp/ccKug4Ma.s:1102: Error: Unknown pseudo-op: `.incbin'
/tmp/ccKug4Ma.s:1107: Error: Unknown pseudo-op: `.incbin'
make[1]: *** [arch/i386/kernel/vsyscall.o] Error 1
make: *** [arch/i386/kernel] Error 2

Regards,
Mats Johannesson

2003-05-05 04:18:31

by Florin Iucha

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Sun, May 04, 2003 at 05:48:53PM -0700, Linus Torvalds wrote:
> I finally found the reason for why some of my machines had trouble with
> restarting the X server, and it turns out that it's been around since very
> early February. I bet others must have seen it too, with random crashes on
> X server restart when the server used AGP (which means that it mainly hit
> either hw-accelerated 3D setups or the intel integrated graphics which use
> a UMA model with AGP as the backing store).
>
> That's a big relief for me, as it was the major thing I personally worried
> about for 2.6.x.

Unfortunately it is not the same reason that locks up my machine ;(

On SIS 735 motherboard, with agpgart, sis-agp and radeon loaded, I get
this on the serial console before the machine freezes:
agpgart: Found an AGP 2.0 compliant device.
agpgart: Putting AGP V2 device at 00:00.0 into 4x mode
agpgart: Putting AGP V2 device at 01:00.0 into 4x mode
Without these modules loaded, the machine is stable.

I am using XFree86 4.3.0 with a Radeon 8500.

Please let me know if you need more details or you have a patch for
testing.

Cheers,
florin

--

"NT is to UNIX what a doughnut is to a particle accelerator."


Attachments:
(No filename) (1.18 kB)
(No filename) (189.00 B)
Download all attachments

2003-05-05 04:28:46

by Linus Torvalds

[permalink] [raw]
Subject: Re: Linux 2.5.69


On Sun, 4 May 2003, Florin Iucha wrote:
>
> On SIS 735 motherboard, with agpgart, sis-agp and radeon loaded, I get
> this on the serial console before the machine freezes:
> agpgart: Found an AGP 2.0 compliant device.
> agpgart: Putting AGP V2 device at 00:00.0 into 4x mode
> agpgart: Putting AGP V2 device at 01:00.0 into 4x mode
> Without these modules loaded, the machine is stable.

Make sure to also test with regular 1x AGP (and no fast write stuff etc).
A lot of motherboards really aren't going to like 4x and some other
settings (in particular, enabling fast writes seems to be a very iffy
proposition indeed).

Also, check if the same setup is stable under 2.4.x and possibly using the
DRI CVS tree. Radeon in particular seems to be a lot stabler in DRI these
days than it has historically been.

Indeed, one of the reasons it took me so long to figure out the stability
issues I saw was simply that there have been real bugs in direct
rendering, and I was blaming them instead and I spent a lot of time trying
to chase down the bug as an AGP or DRI issue.

Linus

2003-05-05 06:19:24

by Anders Karlsson

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, 2003-05-05 at 05:41, Linus Torvalds wrote:

> Make sure to also test with regular 1x AGP (and no fast write stuff etc).
> A lot of motherboards really aren't going to like 4x and some other
> settings (in particular, enabling fast writes seems to be a very iffy
> proposition indeed).

Will try that in case that fixes the problems I see.

> Also, check if the same setup is stable under 2.4.x and possibly using the
> DRI CVS tree. Radeon in particular seems to be a lot stabler in DRI these
> days than it has historically been.
>
> Indeed, one of the reasons it took me so long to figure out the stability
> issues I saw was simply that there have been real bugs in direct
> rendering, and I was blaming them instead and I spent a lot of time trying
> to chase down the bug as an AGP or DRI issue.

Would you think that the problems I have been seeing with a Radeon
Mobility LY on kernel 2.4.2[01] could be down to exactly these issues as
well?

The problems are that the X server can be started only once and that
there seems to be a chance of it locking the machine hard at start after
that, especially after unloading the radeon.o module.

I will do more testing on that to find out exactly what does what, but
would 2.5.69 be a good test to see what has been fixed?

Regards,

/Anders


Attachments:
signature.asc (198.00 B)
This is a digitally signed message part

2003-05-05 06:55:53

by Anders Karlsson

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, 2003-05-05 at 07:31, Anders Karlsson wrote:
> On Mon, 2003-05-05 at 05:41, Linus Torvalds wrote:
>
> > Make sure to also test with regular 1x AGP (and no fast write stuff etc).
> > A lot of motherboards really aren't going to like 4x and some other
> > settings (in particular, enabling fast writes seems to be a very iffy
> > proposition indeed).
>
> Will try that in case that fixes the problems I see.

Hi there again,

This did indeed fix the problem seen on the IBM X31
with Radeon Mobility LY. Setting AGPMode to 1 cured it
of the "black screen on 2nd and later starts". This is
on kernel 2.4.21-rc1, acpi and apm both switched off.

Many many thanks for that tip Linus. :-)))

Regards,

/Anders


Attachments:
signature.asc (198.00 B)
This is a digitally signed message part

2003-05-05 08:02:21

by Benjamin Herrenschmidt

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, 2003-05-05 at 02:48, Linus Torvalds wrote:

> Anyway, that's fixed here, along with a lot of other updates. Much of
> 2.5.69 is small one-liners to drivers to handle the new IRQ semantics, but
> there's a lot of other cleanups in there too (Christoph Hellwig continued
> on his devfs rampage, for example).
>
> NOTE! As of this release I think I'll want to have patches either be
> _really_ obvious, or they should go through one of more people for
> approval. In particular, I'm hoping that the paperwork stuff with Andrew
> should be getting closer to finalized, and that we could start moving over
> towards a 2.6.x release schedule..

You still plan to get Patrick's Power Management updates in there ?

Ben.

2003-05-05 12:49:45

by walt

[permalink] [raw]
Subject: Re: Linux 2.5.69

Linus Torvalds wrote:

<much snippage>

> Jeff Garzik:
> o [netdrvr tg3] detect shared (and screaming) interrupts
> o [netdrvr tg3] fix omission in board shutdown sequence

I had high hopes for these patches but unfortunately they didn't
fix the problem I've had with the built-in Broadcom chip on the
ASUS A7V8X motherboard: after bootup I still need to do an
'ifconfig eth0 down' followed by 'ifconfig eth0 up' before the
chip will actually start passing packets. From then 'til the
next reboot the chip works fine.

> Patrick Mansfield:
> o fix ppa locking and oops

My parallel-port ZIP drive finally works, thanks!

2003-05-05 12:55:29

by Florin Iucha

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Sun, May 04, 2003 at 09:41:10PM -0700, Linus Torvalds wrote:
>
> On Sun, 4 May 2003, Florin Iucha wrote:
> >
> > On SIS 735 motherboard, with agpgart, sis-agp and radeon loaded, I get
> > this on the serial console before the machine freezes:
> > agpgart: Found an AGP 2.0 compliant device.
> > agpgart: Putting AGP V2 device at 00:00.0 into 4x mode
> > agpgart: Putting AGP V2 device at 01:00.0 into 4x mode
> > Without these modules loaded, the machine is stable.
>
> Make sure to also test with regular 1x AGP (and no fast write stuff etc).
> A lot of motherboards really aren't going to like 4x and some other
> settings (in particular, enabling fast writes seems to be a very iffy
> proposition indeed).

On your suggestion I did use AGPMode 1 and 2. No difference.

> Also, check if the same setup is stable under 2.4.x and possibly using the
> DRI CVS tree. Radeon in particular seems to be a lot stabler in DRI these
> days than it has historically been.

The machine was stable in 2.5.30 - 2.5.40 timeframe, using DRI modules
from DRI nightly builds on top of XFree 4.2 . Direct rendering was
working as well.

I will try with the DRI modules again.

Thank you,
florin

--

"NT is to UNIX what a doughnut is to a particle accelerator."


Attachments:
(No filename) (1.24 kB)
(No filename) (189.00 B)
Download all attachments

2003-05-05 13:08:01

by Florin Iucha

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, May 05, 2003 at 08:07:56AM -0500, Florin Iucha wrote:
> The machine was stable in 2.5.30 - 2.5.40 timeframe, using DRI modules
> from DRI nightly builds on top of XFree 4.2 . Direct rendering was
> working as well.
>
> I will try with the DRI modules again.

DRI snapshot works! Yipeeeeee....

Thank you,
florin

--

"NT is to UNIX what a doughnut is to a particle accelerator."


Attachments:
(No filename) (389.00 B)
(No filename) (189.00 B)
Download all attachments

2003-05-05 15:44:20

by Linus Torvalds

[permalink] [raw]
Subject: Re: Linux 2.5.69


[ Linux-kernel added to the cc, since I got several queries about what the
crashes were.. ]

On Mon, 5 May 2003, David van Hoose wrote:
>
> Can I get some details regarding the AGP problem? I had some really bad
> random crashes, panics, and hardlocks up through 2.5.68, and I'm
> wondering if this is the same issue. I first noticed them around 2.5.63.

They actually started in 2.5.60 if it's the same bug.

And yes, you'd get random crashes, panics, lockups and even reboots. The
problem was that the pmd/pgd's were put in the slab cache in between
2.5.59 and 2.5.60, and that was simply wrong because the AGP code changes
the cacheability of the kernel pages when it maps stuff into the AGP
aperture. That in turn will change the page tables but it won't update the
cached entries in the pmd slab caches.

So what happens is that once you exit X, and the page tables are put back
together without the cacheability changes, and you start a new program,
that program may get a page table with partly bogus kernel page table
entries.

That, in turn, when it happens will cause _major_ memory corruption, and
your machine is toast, often in very interesting ways because the internal
kernel data structures got corrupted. It can also cause random SIGSEGV's
etc.

But it only happens with AGP, and a lot of people either don't use it or
run only one X session.

Linus

2003-05-05 15:49:58

by William Lee Irwin III

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, May 05, 2003 at 08:56:41AM -0700, Linus Torvalds wrote:
> They actually started in 2.5.60 if it's the same bug.
> And yes, you'd get random crashes, panics, lockups and even reboots. The
> problem was that the pmd/pgd's were put in the slab cache in between
> 2.5.59 and 2.5.60, and that was simply wrong because the AGP code changes
> the cacheability of the kernel pages when it maps stuff into the AGP
> aperture. That in turn will change the page tables but it won't update the
> cached entries in the pmd slab caches.
> So what happens is that once you exit X, and the page tables are put back
> together without the cacheability changes, and you start a new program,
> that program may get a page table with partly bogus kernel page table
> entries.
> That, in turn, when it happens will cause _major_ memory corruption, and
> your machine is toast, often in very interesting ways because the internal
> kernel data structures got corrupted. It can also cause random SIGSEGV's
> etc.
> But it only happens with AGP, and a lot of people either don't use it or
> run only one X session.

Any chance one of you could try out the fixed slabification patches?
Message-ID: <[email protected]>

-- wli

2003-05-05 15:54:28

by David van Hoose

[permalink] [raw]
Subject: Re: Linux 2.5.69

Linus Torvalds wrote:
> [ Linux-kernel added to the cc, since I got several queries about what the
> crashes were.. ]
>
> On Mon, 5 May 2003, David van Hoose wrote:
>
>>Can I get some details regarding the AGP problem? I had some really bad
>>random crashes, panics, and hardlocks up through 2.5.68, and I'm
>>wondering if this is the same issue. I first noticed them around 2.5.63.
>
>
> They actually started in 2.5.60 if it's the same bug.
>
> And yes, you'd get random crashes, panics, lockups and even reboots. The
> problem was that the pmd/pgd's were put in the slab cache in between
> 2.5.59 and 2.5.60, and that was simply wrong because the AGP code changes
> the cacheability of the kernel pages when it maps stuff into the AGP
> aperture. That in turn will change the page tables but it won't update the
> cached entries in the pmd slab caches.
>
> So what happens is that once you exit X, and the page tables are put back
> together without the cacheability changes, and you start a new program,
> that program may get a page table with partly bogus kernel page table
> entries.
>
> That, in turn, when it happens will cause _major_ memory corruption, and
> your machine is toast, often in very interesting ways because the internal
> kernel data structures got corrupted. It can also cause random SIGSEGV's
> etc.
>
> But it only happens with AGP, and a lot of people either don't use it or
> run only one X session.

Okay.. I kept having all of those problems. I guess the memory
corruption could also explain the buffer overflow in the panic
information I posted around the time of 2.5.65 then. I'll stick the beta
kernel back on my system and play with it again. If I find anything
else, I'll be sure to mention it. Don't want any nasty bugs like that
hiding in 2.6.x.

Thanks!
David

2003-05-05 16:05:16

by Steven Cole

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, 2003-05-05 at 09:56, Linus Torvalds wrote:
> [ Linux-kernel added to the cc, since I got several queries about what the
> crashes were.. ]
>
> On Mon, 5 May 2003, David van Hoose wrote:
> >
> > Can I get some details regarding the AGP problem? I had some really bad
> > random crashes, panics, and hardlocks up through 2.5.68, and I'm
> > wondering if this is the same issue. I first noticed them around 2.5.63.
>
> They actually started in 2.5.60 if it's the same bug.
>
> And yes, you'd get random crashes, panics, lockups and even reboots. The
> problem was that the pmd/pgd's were put in the slab cache in between
> 2.5.59 and 2.5.60, and that was simply wrong because the AGP code changes
> the cacheability of the kernel pages when it maps stuff into the AGP
> aperture. That in turn will change the page tables but it won't update the
> cached entries in the pmd slab caches.
>
> So what happens is that once you exit X, and the page tables are put back
> together without the cacheability changes, and you start a new program,
> that program may get a page table with partly bogus kernel page table
> entries.
>
> That, in turn, when it happens will cause _major_ memory corruption, and
> your machine is toast, often in very interesting ways because the internal
> kernel data structures got corrupted. It can also cause random SIGSEGV's
> etc.
>
> But it only happens with AGP, and a lot of people either don't use it or
> run only one X session.
>
> Linus

This fixes a problem I reported here:
http://marc.theaimsgroup.com/?l=linux-kernel&m=104990441203933&w=2

Thanks,
Steven

2003-05-05 18:41:27

by Linus Torvalds

[permalink] [raw]
Subject: Re: Linux 2.5.69


On Mon, 5 May 2003, Dave Jones wrote:
>
> The integrated graphics seemed to be a key too.

No, there were reports for radeon users etc, so it wasn't a requirement. I
think the real requirement is that it uses a fair _amount_ of AGP.

Linus

2003-05-05 18:37:29

by Dave Jones

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, May 05, 2003 at 08:56:41AM -0700, Linus Torvalds wrote:

> But it only happens with AGP, and a lot of people either don't use it or
> run only one X session.

The integrated graphics seemed to be a key too. I test the agpgart
changes I make on a half dozen boxes before asking you to pull them,
and still couldn't reproduce this bug. The only difference was that
I only have boxes without onboard graphics. All of them worked just
fine, and all of them used agp & dri. Maybe I was hitting the bug,
but the integrated chipsets thrash agpgart a little harder..

Dave

2003-05-05 20:53:33

by John Cherry

[permalink] [raw]
Subject: Re: Linux 2.5.69

Compile statistics: 2.5.69
Compiler: gcc 3.2.2
Script: http://www.osdl.org/archive/cherry/stability/compregress.sh


2.5.68 2.5.69
-------------------- -----------------
bzImage (defconfig) 7 warnings 7 warnings
0 errors 0 errors

bzImage (allmodconfig) 11 warnings 11 warnings
6 errors 0 errors

modules (allmodconfig) 1975 warnings 1567 warnings
60 errors 57 errors

Compile statistics have been for kernel releases from 2.5.46 to 2.5.69
at: http://www.osdl.org/archive/cherry/stability

Failure summary:

drivers/block: 6 warnings, 1 errors
drivers/bluetooth: 0 warnings, 1 errors
drivers/char: 319 warnings, 6 errors
drivers/hotplug: 16 warnings, 2 errors
drivers/isdn: 246 warnings, 6 errors
drivers/media: 128 warnings, 6 errors
drivers/mtd: 31 warnings, 1 errors
drivers/net: 363 warnings, 6 errors
drivers/video: 82 warnings, 13 errors
drivers/video/matrox: 3 warnings, 10 errors
sound: 5 warnings, 3 errors
sound/oss: 49 warnings, 3 errors



Warning summary:

drivers/atm: 39 warnings, 0 errors
drivers/cdrom: 25 warnings, 0 errors
drivers/i2c: 3 warnings, 0 errors
drivers/ide: 33 warnings, 0 errors
drivers/md: 3 warnings, 0 errors
drivers/message: 1 warnings, 0 errors
drivers/pcmcia: 3 warnings, 0 errors
drivers/serial: 1 warnings, 0 errors
drivers/telephony: 10 warnings, 0 errors
drivers/usb: 17 warnings, 0 errors
drivers/video/aty: 4 warnings, 0 errors
drivers/video/riva: 3 warnings, 0 errors
drivers/video/sis: 3 warnings, 0 errors
fs/intermezzo: 1 warnings, 0 errors
fs/lockd: 4 warnings, 0 errors
fs/nfs: 1 warnings, 0 errors
fs/nfsd: 4 warnings, 0 errors
fs/reiserfs: 1 warnings, 0 errors
fs/smbfs: 2 warnings, 0 errors
net: 56 warnings, 0 errors
security: 2 warnings, 0 errors
sound/isa: 3 warnings, 0 errors
sound/pci: 1 warnings, 0 errors


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




2003-05-05 21:32:48

by Arnaldo Carvalho de Melo

[permalink] [raw]
Subject: Re: Linux 2.5.69

Em Mon, May 05, 2003 at 11:39:10PM +0200, Sam Ravnborg escreveu:
> On Mon, May 05, 2003 at 02:05:52PM -0700, John Cherry wrote:
> >
> > modules (allmodconfig) 1975 warnings 1567 warnings
> > 60 errors 57 errors
>
> Is it possible to see a diff of .68 and .69 to see where the
> improvements came from?
> I did not find it on the web-page. If not possible take this as a
> feature request.

Removing MOD_{INC,USE}_COUNT with proper module accounting, using
SET_MODULE_OWNER in net_devices, I removed all of the related to struct sock
and struct socket accounting, etc.

- Arnaldo

2003-05-05 21:26:47

by Sam Ravnborg

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Mon, May 05, 2003 at 02:05:52PM -0700, John Cherry wrote:
>
> modules (allmodconfig) 1975 warnings 1567 warnings
> 60 errors 57 errors

Is it possible to see a diff of .68 and .69 to see where the
improvements came from?
I did not find it on the web-page. If not possible take this as a
feature request.

Sam

2003-05-05 23:09:50

by John Cherry

[permalink] [raw]
Subject: Re: Linux 2.5.69

Yes, the diff would be useful in many cases. I'll shoot for this by the
next kernel drop.

In the mean time, you can do your own diffs of changes you are
interested in. The last link in the table (the Files tab) will give you
a list of files that are logs of the modules that have warnings or
errors. For instance, warnings/errors for the md module would be in
2.5.69.log.md.txt.

Cheers,
John

On Mon, 2003-05-05 at 14:39, Sam Ravnborg wrote:
> On Mon, May 05, 2003 at 02:05:52PM -0700, John Cherry wrote:
> >
> > modules (allmodconfig) 1975 warnings 1567 warnings
> > 60 errors 57 errors
>
> Is it possible to see a diff of .68 and .69 to see where the
> improvements came from?
> I did not find it on the web-page. If not possible take this as a
> feature request.
>
> Sam

2003-05-06 06:46:28

by Voluspa

[permalink] [raw]
Subject: Re: Linux 2.5.69


Mon, 5 May 2003 06:19:51 +0200 I wrote:

>/tmp/ccKug4Ma.s:1102: Error: Unknown pseudo-op: `.incbin'
>/tmp/ccKug4Ma.s:1107: Error: Unknown pseudo-op: `.incbin'
>make[1]: *** [arch/i386/kernel/vsyscall.o] Error 1
>make: *** [arch/i386/kernel] Error 2

Turns out I got bitten by too old binutils. Could you do a "feet/metre" convertion in
Documentation/Changes under _Current Minimal Requirements_, something like:

o binutils 2.9.5.0.25 # ld -v
+or
+o GNU/binutils 2.13.(wherever 'as' got to know about incbin)

I had the GNU 2.11.2 which is higher than 2.9.x... right ;-)

Regards,
Mats Johannesson

2003-05-06 09:55:57

by Paweł Gołaszewski

[permalink] [raw]
Subject: Re: Linux 2.5.69

That kernel fails for me when building...
[cut]

gcc -Wp,-MD,fs/lockd/.clntproc.o.d -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686 -malign-functions=4 -Iinclude/asm-i386/mach-default -nostdinc -iwithprefix include -DMODULE -DKBUILD_BASENAME=clntproc -DKBUILD_MODNAME=lockd -c -o fs/lockd/.tmp_clntproc.o fs/lockd/clntproc.c
In file included from fs/lockd/clntproc.c:17:
include/linux/sunrpc/svc.h: In function `svc_take_page':
include/linux/sunrpc/svc.h:180: invalid lvalue in assignment
make[3]: *** [fs/lockd/clntproc.o] Error 1
make[2]: *** [fs/lockd] Error 2
make[1]: *** [fs] Error 2
make: *** [modules] Error 2

My kernel config:
http://piorun.ds.pg.gda.pl/~blues/linux-2.5.69.txt

--
pozdr. Pawe? Go?aszewski
---------------------------------
worth to see: http://www.againsttcpa.com/
CPU not found - software emulation...

Subject: [PATCH] include/linux/sunrpc/svc.h compilation error

On Tue, 2003-05-06 at 11:08, Pawe? Go?aszewski wrote:
> That kernel fails for me when building...
> [cut]
>
> gcc -Wp,-MD,fs/lockd/.clntproc.o.d -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686 -malign-functions=4 -Iinclude/asm-i386/mach-default -nostdinc -iwithprefix include -DMODULE -DKBUILD_BASENAME=clntproc -DKBUILD_MODNAME=lockd -c -o fs/lockd/.tmp_clntproc.o fs/lockd/clntproc.c
> In file included from fs/lockd/clntproc.c:17:
> include/linux/sunrpc/svc.h: In function `svc_take_page':
> include/linux/sunrpc/svc.h:180: invalid lvalue in assignment
> make[3]: *** [fs/lockd/clntproc.o] Error 1
> make[2]: *** [fs/lockd] Error 2
> make[1]: *** [fs] Error 2
> make: *** [modules] Error 2

Looks like gcc fault, can You Pawel give as gcc version ?

this patch as reported by Pawel helps:


--- ./include/linux/sunrpc/svc.h.buildfix Mon May 6 12:45:11 2003
+++ ./include/linux/sunrpc/svc.h Tue May 6 12:42:13 2003
@@ -176,8 +176,14 @@
{
if (rqstp->rq_arghi <= rqstp->rq_argused)
return -ENOMEM;
+
+ rqstp->rq_arghi--;
+
rqstp->rq_respages[rqstp->rq_resused++] =
rqstp->rq_argpages[--rqstp->rq_arghi];
+
+ rqstp->rq_resused++;
+
return 0;
}


but still, it looks strange - i am sure it is just an gcc issue

--
Grzegorz Jaskiewicz <[email protected]>
K4 labs

Subject: Re: [PATCH] include/linux/sunrpc/svc.h compilation error

Sorry for that Linus, patch below is a right one.


diff -r -u 2/include/linux/sunrpc/svc.h 1/include/linux/sunrpc/svc.h
--- 2/include/linux/sunrpc/svc.h 2003-05-05 00:53:31.000000000
+0100
+++ 1/include/linux/sunrpc/svc.h 2003-05-06 13:06:27.000000000
+0100
@@ -176,8 +176,14 @@
{
if (rqstp->rq_arghi <= rqstp->rq_argused)
return -ENOMEM;
- rqstp->rq_respages[rqstp->rq_resused++] =
- rqstp->rq_argpages[--rqstp->rq_arghi];
+
+ rqstp->rq_arghi--;
+
+ rqstp->rq_respages[rqstp->rq_resused] =
+ rqstp->rq_argpages[rqstp->rq_arghi];
+
+ rqstp->rq_resused++;
+
return 0;
}


On Tue, 2003-05-06 at 13:02, Grzegorz Jaskiewicz wrote:
> On Tue, 2003-05-06 at 11:08, Pawe? Go?aszewski wrote:
> > That kernel fails for me when building...
> > [cut]
> >
> > gcc -Wp,-MD,fs/lockd/.clntproc.o.d -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686 -malign-functions=4 -Iinclude/asm-i386/mach-default -nostdinc -iwithprefix include -DMODULE -DKBUILD_BASENAME=clntproc -DKBUILD_MODNAME=lockd -c -o fs/lockd/.tmp_clntproc.o fs/lockd/clntproc.c
> > In file included from fs/lockd/clntproc.c:17:
> > include/linux/sunrpc/svc.h: In function `svc_take_page':
> > include/linux/sunrpc/svc.h:180: invalid lvalue in assignment
> > make[3]: *** [fs/lockd/clntproc.o] Error 1
> > make[2]: *** [fs/lockd] Error 2
> > make[1]: *** [fs] Error 2
> > make: *** [modules] Error 2
>
> Looks like gcc fault, can You Pawel give as gcc version ?
>
> this patch as reported by Pawel helps:
> but still, it looks strange - i am sure it is just an gcc issue
--
Grzegorz Jaskiewicz <[email protected]>
K4 labs

2003-05-06 12:45:57

by Paweł Gołaszewski

[permalink] [raw]
Subject: Re: [PATCH] include/linux/sunrpc/svc.h compilation error

On Tue, 6 May 2003, Grzegorz Jaskiewicz wrote:
> > That kernel fails for me when building...
> > [cut]
> >
> > gcc -Wp,-MD,fs/lockd/.clntproc.o.d -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686 -malign-functions=4 -Iinclude/asm-i386/mach-default -nostdinc -iwithprefix include -DMODULE -DKBUILD_BASENAME=clntproc -DKBUILD_MODNAME=lockd -c -o fs/lockd/.tmp_clntproc.o fs/lockd/clntproc.c
> > In file included from fs/lockd/clntproc.c:17:
> > include/linux/sunrpc/svc.h: In function `svc_take_page':
> > include/linux/sunrpc/svc.h:180: invalid lvalue in assignment
> > make[3]: *** [fs/lockd/clntproc.o] Error 1
> > make[2]: *** [fs/lockd] Error 2
> > make[1]: *** [fs] Error 2
> > make: *** [modules] Error 2
>
> Looks like gcc fault, can You Pawel give as gcc version ?

gcc-2.95.4-20010823

At least now it builds - I'll check later if it works :)

Thanks for your cooperation.

--
pozdr. Pawe? Go?aszewski
---------------------------------
worth to see: http://www.againsttcpa.com/
CPU not found - software emulation...

2003-05-06 13:27:22

by Roger Luethi

[permalink] [raw]
Subject: Re: Linux 2.5.69

I'm seeing "kernel BUG at include/linux/module.h:284!" with 2.5.69.

I first suspected the early summer in Europe made my hardware flaky, but I
can't reproduce with 2.5.68.

Roger

May 6 14:22:52 k3 kernel: ------------[ cut here ]------------
May 6 14:22:52 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:22:52 k3 kernel: invalid operand: 0000 [#1]
May 6 14:22:52 k3 kernel: CPU: 0
May 6 14:22:52 k3 kernel: EIP: 0060:[sys_accept+113/336] Not tainted
May 6 14:22:52 k3 kernel: EFLAGS: 00010246
May 6 14:22:52 k3 kernel: EIP is at sys_accept+0x71/0x150
May 6 14:22:52 k3 kernel: eax: 00000000 ebx: f88bee40 ecx: 00000001 edx: f88bee80
May 6 14:22:52 k3 kernel: esi: f5d9e044 edi: f66a1434 ebp: f6459f7c esp: f6459ee8
May 6 14:22:52 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:22:52 k3 kernel: Process medusa-idled (pid: 1205, threadinfo=f6458000 task=f6457980)
May 6 14:22:52 k3 kernel: Stack: bffff704 f6459fa4 00000004 f6459f94 00000000 f6459f60 f6459f60 bffff5a8
May 6 14:22:52 k3 kernel: f6458000 f6459fb4 f7fff080 f7210808 f6458000 c03ad300 f645801c 00000000
May 6 14:22:52 k3 kernel: 00000002 00000000 fffeff9a 00000000 00000000 f6457980 c011bfd0 00000000
May 6 14:22:52 k3 kernel: Call Trace:
May 6 14:22:52 k3 kernel: [default_wake_function+0/32] default_wake_function+0x0/0x20
May 6 14:22:52 k3 kernel: [default_wake_function+0/32] default_wake_function+0x0/0x20
May 6 14:22:52 k3 kernel: [sys_socketcall+187/448] sys_socketcall+0xbb/0x1c0
May 6 14:22:52 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:22:52 k3 kernel:
May 6 14:22:52 k3 kernel: Code: 0f 0b 1c 01 4d 36 2e c0 8d b4 26 00 00 00 00 ff 83 c0 00 00
May 6 14:22:52 k3 kernel: ------------[ cut here ]------------
May 6 14:22:52 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:22:52 k3 kernel: invalid operand: 0000 [#2]
May 6 14:22:52 k3 kernel: CPU: 0
May 6 14:22:52 k3 kernel: EIP: 0060:[_end+944664612/1069725248] Not tainted
May 6 14:22:52 k3 kernel: EFLAGS: 00010246
May 6 14:22:52 k3 kernel: EIP is at unix_create1+0x84/0x160 [unix]
May 6 14:22:52 k3 kernel: eax: 00000000 ebx: f5cf8700 ecx: 00000001 edx: f88bee80
May 6 14:22:52 k3 kernel: esi: 00000000 edi: f5cf84c0 ebp: f6065ea4 esp: f6065e94
May 6 14:22:52 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:22:52 k3 kernel: Process gkrellm (pid: 1240, threadinfo=f6064000 task=f6456000)
May 6 14:22:52 k3 kernel: Stack: 00000000 f6065efc f5cf84c0 f88ba340 f6065ed4 f88ba398 00000000 f66a1a1c
May 6 14:22:52 k3 kernel: f6065efc 0000006e 00000011 7fffffff 00000000 00000000 fffffff4 f66a1a1c
May 6 14:22:52 k3 kernel: f6065f7c c0230d1f f66a1a1c f6065efc 0000006e 00000002 bffff0c4 f6065fa4
May 6 14:22:52 k3 kernel: Call Trace:
May 6 14:22:52 k3 kernel: [_end+944667008/1069725248] unix_stream_connect+0x30/0x470 [unix]
May 6 14:22:52 k3 kernel: [_end+944667096/1069725248] unix_stream_connect+0x88/0x470 [unix]
May 6 14:22:52 k3 kernel: [sys_connect+95/144] sys_connect+0x5f/0x90
May 6 14:22:52 k3 kernel: [sys_socket+46/80] sys_socket+0x2e/0x50
May 6 14:22:52 k3 kernel: [sys_socketcall+157/448] sys_socketcall+0x9d/0x1c0
May 6 14:22:52 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:22:52 k3 kernel:
May 6 14:22:52 k3 kernel: Code: 0f 0b 1c 01 f2 cd 8b f8 8d 74 26 00 ff 05 00 ef 8b f8 0f b7
May 6 14:23:09 k3 kernel: ------------[ cut here ]------------
May 6 14:23:09 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:23:09 k3 kernel: invalid operand: 0000 [#3]
May 6 14:23:09 k3 kernel: CPU: 0
May 6 14:23:09 k3 kernel: EIP: 0060:[sys_accept+113/336] Not tainted
May 6 14:23:09 k3 kernel: EFLAGS: 00010246
May 6 14:23:09 k3 kernel: EIP is at sys_accept+0x71/0x150
May 6 14:23:09 k3 kernel: eax: 00000000 ebx: f88bee40 ecx: 00000001 edx: f88bee80
May 6 14:23:09 k3 kernel: esi: f6f8323c edi: f6f83044 ebp: f6f1bf7c esp: f6f1bee8
May 6 14:23:09 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:23:09 k3 kernel: Process trivial-rewrite (pid: 900, threadinfo=f6f1a000 task=f7444000)
May 6 14:23:09 k3 kernel: Stack: bffffb94 f6f1bfa4 00000004 f590a008 00000000 f6f1bf04 c014307a f6f1bf0c
May 6 14:23:09 k3 kernel: c01435e5 f6f1bf14 f7fff080 f72107dc f7210000 00000020 f6f1bf5c c0147d3d
May 6 14:23:09 k3 kernel: f7fff080 f72107dc 00000282 00000001 00000004 f72107e0 0000002c 00000028
May 6 14:23:09 k3 kernel: Call Trace:
May 6 14:23:09 k3 kernel: [free_hot_page+10/16] free_hot_page+0xa/0x10
May 6 14:23:09 k3 kernel: [__free_pages+53/80] __free_pages+0x35/0x50
May 6 14:23:09 k3 kernel: [kfree+525/608] kfree+0x20d/0x260
May 6 14:23:09 k3 kernel: [select_bits_free+12/16] select_bits_free+0xc/0x10
May 6 14:23:09 k3 kernel: [select_bits_free+12/16] select_bits_free+0xc/0x10
May 6 14:23:09 k3 kernel: [sys_select+1086/1104] sys_select+0x43e/0x450
May 6 14:23:09 k3 kernel: [sys_socketcall+187/448] sys_socketcall+0xbb/0x1c0
May 6 14:23:09 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:23:09 k3 kernel:
May 6 14:23:09 k3 kernel: Code: 0f 0b 1c 01 4d 36 2e c0 8d b4 26 00 00 00 00 ff 83 c0 00 00
May 6 14:23:09 k3 kernel: ------------[ cut here ]------------
May 6 14:23:09 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:23:09 k3 kernel: invalid operand: 0000 [#4]
May 6 14:23:09 k3 kernel: CPU: 0
May 6 14:23:09 k3 kernel: EIP: 0060:[_end+944664612/1069725248] Not tainted
May 6 14:23:09 k3 kernel: EFLAGS: 00210246
May 6 14:23:09 k3 kernel: EIP is at unix_create1+0x84/0x160 [unix]
May 6 14:23:09 k3 kernel: eax: 00000000 ebx: f6f814c0 ecx: 00000001 edx: f88bee80
May 6 14:23:09 k3 kernel: esi: 00000000 edi: f6f51280 ebp: f5c85ea4 esp: f5c85e94
May 6 14:23:09 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:23:09 k3 kernel: Process gkrellm (pid: 1713, threadinfo=f5c84000 task=f6456cc0)
May 6 14:23:09 k3 kernel: Stack: 00000000 f5c85efc f6f51280 f88ba340 f5c85ed4 f88ba398 00000000 f6f3be0c
May 6 14:23:09 k3 kernel: f5c85efc 0000006e 0000001a 7fffffff 00000000 00000000 fffffff4 f6f3be0c
May 6 14:23:09 k3 kernel: f5c85f7c c0230d1f f6f3be0c f5c85efc 0000006e 00000002 bfffed04 f5c85fa4
May 6 14:23:09 k3 kernel: Call Trace:
May 6 14:23:09 k3 kernel: [_end+944667008/1069725248] unix_stream_connect+0x30/0x470 [unix]
May 6 14:23:09 k3 kernel: [_end+944667096/1069725248] unix_stream_connect+0x88/0x470 [unix]
May 6 14:23:09 k3 kernel: [sys_connect+95/144] sys_connect+0x5f/0x90
May 6 14:23:09 k3 kernel: [sys_socket+46/80] sys_socket+0x2e/0x50
May 6 14:23:09 k3 kernel: [sys_socketcall+157/448] sys_socketcall+0x9d/0x1c0
May 6 14:23:09 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:23:09 k3 kernel:
May 6 14:23:09 k3 kernel: Code: 0f 0b 1c 01 f2 cd 8b f8 8d 74 26 00 ff 05 00 ef 8b f8 0f b7
May 6 14:23:30 k3 gconfd (rl-1740): starting (version 1.0.9), pid 1740 user 'rl'
May 6 14:23:30 k3 kernel: ------------[ cut here ]------------
May 6 14:23:30 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:23:30 k3 kernel: invalid operand: 0000 [#5]
May 6 14:23:30 k3 kernel: CPU: 0
May 6 14:23:30 k3 kernel: EIP: 0060:[_end+944664612/1069725248] Not tainted
May 6 14:23:30 k3 kernel: EFLAGS: 00210246
May 6 14:23:30 k3 kernel: EIP is at unix_create1+0x84/0x160 [unix]
May 6 14:23:30 k3 kernel: eax: 00000000 ebx: f5cf8940 ecx: 00000001 edx: f88bee80
May 6 14:23:30 k3 kernel: esi: f6f3b62c edi: c03cc7e0 ebp: f0c33f10 esp: f0c33f00
May 6 14:23:30 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:23:30 k3 kernel: Process gconfd-1 (pid: 1740, threadinfo=f0c32000 task=f73b6660)
May 6 14:23:30 k3 kernel: Stack: f6f3b62c 00000004 c03cc7e0 00000001 f0c33f1c f88b9b1d f6f3b62c f0c33f5c
May 6 14:23:30 k3 kernel: c0230847 f6f3b62c 00000000 00000001 f0c33fa4 00000000 ffffff9f 00000048
May 6 14:23:30 k3 kernel: c0152741 f7d837b0 f1358824 00200292 f0c33f64 c0152741 f7ffd8f0 f0c33f7c
May 6 14:23:30 k3 kernel: Call Trace:
May 6 14:23:30 k3 kernel: [_end+944664925/1069725248] unix_create+0x5d/0x80 [unix]
May 6 14:23:30 k3 kernel: [sock_create+279/448] sock_create+0x117/0x1c0
May 6 14:23:30 k3 kernel: [unmap_vma+113/128] unmap_vma+0x71/0x80
May 6 14:23:30 k3 kernel: [unmap_vma+113/128] unmap_vma+0x71/0x80
May 6 14:23:30 k3 kernel: [sys_socket+28/80] sys_socket+0x1c/0x50
May 6 14:23:30 k3 kernel: [sys_socketcall+125/448] sys_socketcall+0x7d/0x1c0
May 6 14:23:30 k3 kernel: [smp_apic_timer_interrupt+43/224] smp_apic_timer_interrupt+0x2b/0xe0
May 6 14:23:30 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:23:30 k3 kernel:
May 6 14:23:30 k3 kernel: Code: 0f 0b 1c 01 f2 cd 8b f8 8d 74 26 00 ff 05 00 ef 8b f8 0f b7
May 6 14:23:36 k3 kernel: ------------[ cut here ]------------
May 6 14:23:36 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:23:36 k3 kernel: invalid operand: 0000 [#6]
May 6 14:23:36 k3 kernel: CPU: 0
May 6 14:23:36 k3 kernel: EIP: 0060:[_end+944664612/1069725248] Not tainted
May 6 14:23:36 k3 kernel: EFLAGS: 00210246
May 6 14:23:36 k3 kernel: EIP is at unix_create1+0x84/0x160 [unix]
May 6 14:23:36 k3 kernel: eax: 00000000 ebx: f5cf8040 ecx: 00000001 edx: f88bee80
May 6 14:23:36 k3 kernel: esi: 00000000 edi: f5cf8b80 ebp: f0acdea4 esp: f0acde94
May 6 14:23:36 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:23:36 k3 kernel: Process gkrellm (pid: 1722, threadinfo=f0acc000 task=f5d74660)
May 6 14:23:36 k3 kernel: Stack: 00000000 f0acdefc f5cf8b80 f88ba340 f0acded4 f88ba398 00000000 f6f3b824
May 6 14:23:36 k3 kernel: f0acdefc 0000006e 0000001a 7fffffff 00000000 00000000 fffffff4 f6f3b824
May 6 14:23:36 k3 kernel: f0acdf7c c0230d1f f6f3b824 f0acdefc 0000006e 00000002 bfffed04 f0acdfa4
May 6 14:23:36 k3 kernel: Call Trace:
May 6 14:23:36 k3 kernel: [_end+944667008/1069725248] unix_stream_connect+0x30/0x470 [unix]
May 6 14:23:36 k3 kernel: [_end+944667096/1069725248] unix_stream_connect+0x88/0x470 [unix]
May 6 14:23:36 k3 kernel: [sys_connect+95/144] sys_connect+0x5f/0x90
May 6 14:23:36 k3 kernel: [sys_socketcall+157/448] sys_socketcall+0x9d/0x1c0
May 6 14:23:36 k3 kernel: [do_syscall_trace+52/96] do_syscall_trace+0x34/0x60
May 6 14:23:36 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:23:36 k3 kernel:
May 6 14:23:36 k3 kernel: Code: 0f 0b 1c 01 f2 cd 8b f8 8d 74 26 00 ff 05 00 ef 8b f8 0f b7
May 6 14:23:37 k3 kernel: ------------[ cut here ]------------
May 6 14:23:37 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:23:37 k3 kernel: invalid operand: 0000 [#7]
May 6 14:23:37 k3 kernel: CPU: 0
May 6 14:23:37 k3 kernel: EIP: 0060:[_end+944664612/1069725248] Not tainted
May 6 14:23:37 k3 kernel: EFLAGS: 00210246
May 6 14:23:37 k3 kernel: EIP is at unix_create1+0x84/0x160 [unix]
May 6 14:23:37 k3 kernel: eax: 00000000 ebx: f5cf8280 ecx: 00000001 edx: f88bee80
May 6 14:23:37 k3 kernel: esi: 00000000 edi: f6008040 ebp: f0c33ea4 esp: f0c33e94
May 6 14:23:37 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:23:37 k3 kernel: Process galeon-bin (pid: 1742, threadinfo=f0c32000 task=f0c82000)
May 6 14:23:37 k3 kernel: Stack: 00000000 f0c33efc f6008040 f88ba340 f0c33ed4 f88ba398 00000000 f66a162c
May 6 14:23:37 k3 kernel: f0c33efc 0000006e 00000018 7fffffff 00000000 00000000 fffffff4 f66a162c
May 6 14:23:37 k3 kernel: f0c33f7c c0230d1f f66a162c f0c33efc 0000006e 00000002 bffff074 f0c33fa4
May 6 14:23:37 k3 kernel: Call Trace:
May 6 14:23:37 k3 kernel: [_end+944667008/1069725248] unix_stream_connect+0x30/0x470 [unix]
May 6 14:23:37 k3 kernel: [_end+944667096/1069725248] unix_stream_connect+0x88/0x470 [unix]
May 6 14:23:37 k3 kernel: [sys_connect+95/144] sys_connect+0x5f/0x90
May 6 14:23:37 k3 kernel: [sys_socket+46/80] sys_socket+0x2e/0x50
May 6 14:23:37 k3 kernel: [sys_socketcall+157/448] sys_socketcall+0x9d/0x1c0
May 6 14:23:37 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:23:37 k3 kernel:
May 6 14:23:37 k3 kernel: Code: 0f 0b 1c 01 f2 cd 8b f8 8d 74 26 00 ff 05 00 ef 8b f8 0f b7
May 6 14:24:09 k3 kernel: ------------[ cut here ]------------
May 6 14:24:09 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:24:09 k3 kernel: invalid operand: 0000 [#8]
May 6 14:24:09 k3 kernel: CPU: 0
May 6 14:24:09 k3 kernel: EIP: 0060:[_end+944664612/1069725248] Not tainted
May 6 14:24:09 k3 kernel: EFLAGS: 00010246
May 6 14:24:09 k3 kernel: EIP is at unix_create1+0x84/0x160 [unix]
May 6 14:24:09 k3 kernel: eax: 00000000 ebx: f6008700 ecx: 00000001 edx: f88bee80
May 6 14:24:09 k3 kernel: esi: f637823c edi: c03cc7e0 ebp: f0ce5f10 esp: f0ce5f00
May 6 14:24:09 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:24:09 k3 kernel: Process zsh (pid: 1846, threadinfo=f0ce4000 task=f0c83980)
May 6 14:24:09 k3 kernel: Stack: f637823c 00000004 c03cc7e0 00000001 f0ce5f1c f88b9b1d f637823c f0ce5f5c
May 6 14:24:09 k3 kernel: c0230847 f637823c 00000000 00000001 f0ce5fa4 00000000 ffffff9f 00030002
May 6 14:24:09 k3 kernel: 00001000 f78b4e08 00000000 00000000 f0ce5f98 c0152d01 f650bdc0 f0ce5f7c
May 6 14:24:09 k3 kernel: Call Trace:
May 6 14:24:09 k3 kernel: [_end+944664925/1069725248] unix_create+0x5d/0x80 [unix]
May 6 14:24:09 k3 kernel: [sock_create+279/448] sock_create+0x117/0x1c0
May 6 14:24:09 k3 kernel: [do_brk+257/480] do_brk+0x101/0x1e0
May 6 14:24:09 k3 kernel: [sys_socket+28/80] sys_socket+0x1c/0x50
May 6 14:24:09 k3 kernel: [sys_socketcall+125/448] sys_socketcall+0x7d/0x1c0
May 6 14:24:09 k3 kernel: [error_code+45/56] error_code+0x2d/0x38
May 6 14:24:09 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:24:09 k3 kernel:
May 6 14:24:09 k3 kernel: Code: 0f 0b 1c 01 f2 cd 8b f8 8d 74 26 00 ff 05 00 ef 8b f8 0f b7
May 6 14:24:10 k3 kernel: ------------[ cut here ]------------
May 6 14:24:10 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:24:10 k3 kernel: invalid operand: 0000 [#9]
May 6 14:24:10 k3 kernel: CPU: 0
May 6 14:24:10 k3 kernel: EIP: 0060:[_end+944664612/1069725248] Not tainted
May 6 14:24:10 k3 kernel: EFLAGS: 00010246
May 6 14:24:10 k3 kernel: EIP is at unix_create1+0x84/0x160 [unix]
May 6 14:24:10 k3 kernel: eax: 00000000 ebx: f6f3ab80 ecx: 00000001 edx: f88bee80
May 6 14:24:10 k3 kernel: esi: f5c18044 edi: c03cc7e0 ebp: f0d17f10 esp: f0d17f00
May 6 14:24:10 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:24:10 k3 kernel: Process local (pid: 1848, threadinfo=f0d16000 task=f5fed980)
May 6 14:24:10 k3 kernel: Stack: f5c18044 00000004 c03cc7e0 00000001 f0d17f1c f88b9b1d f5c18044 f0d17f5c
May 6 14:24:10 k3 kernel: c0230847 f5c18044 00000000 00000001 f0d17fa4 00000000 ffffff9f 00030002
May 6 14:24:10 k3 kernel: 00001000 f2341778 00000000 00000000 f0d17f98 c0152d01 f73f34c0 f0d17f7c
May 6 14:24:10 k3 kernel: Call Trace:
May 6 14:24:10 k3 kernel: [_end+944664925/1069725248] unix_create+0x5d/0x80 [unix]
May 6 14:24:10 k3 kernel: [sock_create+279/448] sock_create+0x117/0x1c0
May 6 14:24:10 k3 kernel: [do_brk+257/480] do_brk+0x101/0x1e0
May 6 14:24:10 k3 kernel: [sys_socket+28/80] sys_socket+0x1c/0x50
May 6 14:24:10 k3 kernel: [sys_socketcall+125/448] sys_socketcall+0x7d/0x1c0
May 6 14:24:10 k3 kernel: [error_code+45/56] error_code+0x2d/0x38
May 6 14:24:10 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:24:10 k3 kernel:
May 6 14:24:10 k3 kernel: Code: 0f 0b 1c 01 f2 cd 8b f8 8d 74 26 00 ff 05 00 ef 8b f8 0f b7
May 6 14:24:10 k3 kernel: ------------[ cut here ]------------
May 6 14:24:10 k3 kernel: kernel BUG at include/linux/module.h:284!
May 6 14:24:10 k3 kernel: invalid operand: 0000 [#10]
May 6 14:24:10 k3 kernel: CPU: 0
May 6 14:24:10 k3 kernel: EIP: 0060:[sys_accept+113/336] Not tainted
May 6 14:24:10 k3 kernel: EFLAGS: 00010246
May 6 14:24:10 k3 kernel: EIP is at sys_accept+0x71/0x150
May 6 14:24:10 k3 kernel: eax: 00000000 ebx: f88bee40 ecx: 00000001 edx: f88bee80
May 6 14:24:10 k3 kernel: esi: f5c18434 edi: f66a123c ebp: f646df7c esp: f646dee8
May 6 14:24:10 k3 kernel: ds: 007b es: 007b ss: 0068
May 6 14:24:10 k3 kernel: Process medusa-idled (pid: 1183, threadinfo=f646c000 task=f6688cc0)
May 6 14:24:10 k3 kernel: Stack: bffff7c4 f646dfa4 00000004 f646df94 00000000 f646df60 f646df60 bffff668
May 6 14:24:10 k3 kernel: f646c000 f646dfb4 f7fff080 f7210808 f646c000 c03ad300 f646c01c 00000000
May 6 14:24:10 k3 kernel: 00000002 00000000 00003198 00000001 00000000 f6688cc0 c011bfd0 00000000
May 6 14:24:10 k3 kernel: Call Trace:
May 6 14:24:10 k3 kernel: [default_wake_function+0/32] default_wake_function+0x0/0x20
May 6 14:24:10 k3 kernel: [default_wake_function+0/32] default_wake_function+0x0/0x20
May 6 14:24:10 k3 kernel: [sys_socketcall+187/448] sys_socketcall+0xbb/0x1c0
May 6 14:24:10 k3 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
May 6 14:24:10 k3 kernel:
May 6 14:24:10 k3 kernel: Code: 0f 0b 1c 01 4d 36 2e c0 8d b4 26 00 00 00 00 ff 83 c0 00 00

2003-05-06 14:07:35

by David Miller

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Tue, 2003-05-06 at 06:39, Roger Luethi wrote:
> I'm seeing "kernel BUG at include/linux/module.h:284!" with 2.5.69.
>
> I first suspected the early summer in Europe made my hardware flaky, but I
> can't reproduce with 2.5.68.

Arnaldo, it's the socket module stuff. He's using AF_UNIX
as a module.

--
David S. Miller <[email protected]>

2003-05-06 18:12:06

by Arnaldo Carvalho de Melo

[permalink] [raw]
Subject: Re: Linux 2.5.69

Em Tue, May 06, 2003 at 07:08:52AM -0700, David S. Miller escreveu:
> On Tue, 2003-05-06 at 06:39, Roger Luethi wrote:
> > I'm seeing "kernel BUG at include/linux/module.h:284!" with 2.5.69.
> >
> > I first suspected the early summer in Europe made my hardware flaky, but I
> > can't reproduce with 2.5.68.
>
> Arnaldo, it's the socket module stuff. He's using AF_UNIX
> as a module.

I'm reading this thread now, will study this.

2003-05-07 02:18:55

by Rusty Russell

[permalink] [raw]
Subject: Re: Linux 2.5.69

In message <[email protected]> you write:
> > On Tue, 2003-05-06 at 06:39, Roger Luethi wrote:
> > > I'm seeing "kernel BUG at include/linux/module.h:284!" with 2.5.69.

Clearly, this statement in sys_accept is false:

/*
* We don't need try_module_get here, as the listening socket (sock)
* has the protocol module (sock->ops->owner) held.
*/
__module_get(sock->ops->owner);

Now, the question is, when is this not true? sock_create certainly
seems to grab a reference count to sock->ops->owner. Maybe a
refcounting bug elsewhere?

Rusty.
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.

2003-05-07 19:36:21

by Geert Uytterhoeven

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Sun, 4 May 2003, Linus Torvalds wrote:
> Summary of changes from v2.5.68 to v2.5.69
> ============================================
>
> Ben Collins:
> o [VIDEO]: Revert cfbimgblt.c back to a working state on 64-bit
> o [VIDEO]: Revert atyfb back to known working clean base

For future changes, could you please run these `reversals' through
linux-fbdev-devel, instead of silently passing them behind our backs? Thanks!

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

2003-05-07 19:53:33

by Ben Collins

[permalink] [raw]
Subject: Re: Linux 2.5.69

On Wed, May 07, 2003 at 09:44:08PM +0200, Geert Uytterhoeven wrote:
> On Sun, 4 May 2003, Linus Torvalds wrote:
> > Summary of changes from v2.5.68 to v2.5.69
> > ============================================
> >
> > Ben Collins:
> > o [VIDEO]: Revert cfbimgblt.c back to a working state on 64-bit
> > o [VIDEO]: Revert atyfb back to known working clean base
>
> For future changes, could you please run these `reversals' through
> linux-fbdev-devel, instead of silently passing them behind our backs? Thanks!

This was far from silent. These were discussed with James, DaveM and
Linus, and agreed to by James.

--
Debian - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
Deqo - http://www.deqo.com/

2003-05-07 21:56:08

by Linus Torvalds

[permalink] [raw]
Subject: Re: Linux 2.5.69


On Wed, 7 May 2003, Geert Uytterhoeven wrote:
> On Sun, 4 May 2003, Linus Torvalds wrote:
> > Summary of changes from v2.5.68 to v2.5.69
> > ============================================
> >
> > Ben Collins:
> > o [VIDEO]: Revert cfbimgblt.c back to a working state on 64-bit
> > o [VIDEO]: Revert atyfb back to known working clean base
>
> For future changes, could you please run these `reversals' through
> linux-fbdev-devel, instead of silently passing them behind our backs? Thanks!

As mentioned already, this was done by several people, including the
maintainer.

But even if it wasn't, the fact is that new additions that break major
architectures _will_ be reverted. No ifs, buts or maybes about it. If new
code shows itself to be broken, it's going to get reverted, and the sooner
the better. And in this case, the code was not just slightly broken, it
was totally non-working, in ways that made it clear that it had NEVER
worked as intended.

Linus