Linux v2.5.41

Linux v2.5.41

[Linus Torvalds]

Tons of stuff. Mucho merges with the "A-Team" (Alan, Al, Alexey, Andrew,
Anton, Arjan, Arnaldo and Art), but the "M-Team" (Maksim, Marcel, Martin's
and Mike) is a close runner up. The J's are doing well too.

		Linus "Go L, go L" Torvalds

----

Summary of changes from v2.5.40 to v2.5.41
============================================

Alan Cox <alan@lxorguk.ukuu.org.uk>:
  o 2.5 trivial - MCA comments
  o disable GMX2000
  o PC110 pad docs are wrong
  o Forward port AMD random number generator
  o 2.5 Fix set_bit abuse in ATP driver
  o move tulip into ethernet 10,100
  o aacraid driver for 2.5
  o forward port toughbook fixes for maestro3
  o fix warning in longhaul.c
  o update docs to match maestro3 changes
  o flush the right thing in the rd cache
  o Clean up sf16fmi radio
  o Fix cs89x0 warnings
  o NCR5380 port to 2.5 first pass
  o Fix stupid scsi setup bug in 53c406, fix addressing
  o first pass at the ancient wd7000 crap
  o bring telephony in line with 2.4
  o add the mini 4x6 font from uclinux
  o make jffs/jffs2 work with signal changes
  o 6x4 font headers
  o sane minimum proc count

Alexander Viro <viro@math.psu.edu>:
  o pd switched to dynamic allocation
  o pd.c cleanups
  o mtd switched to dynamic allocation
  o md switched to dynamic allocation
  o old cdroms switched to dynamic allocation
  o loop.c switched to dynamic allocation
  o rd.c switched to dynamic allocation
  o hd.c switched to dynamic allocation
  o floppy.c switched to dynamic allocation
  o pcd switched to alloc_disk()
  o initrd fix (missing set_capacity)
  o umem switched to alloc_disk()
  o ps2esdi switched to alloc_disk()
  o xd switched to alloc_disk()
  o acorn mfm switched to alloc_disk()
  o i2o switched to alloc_disk()
  o stram/z2ram switched to alloc_disk()
  o nbd switched to alloc_disk()
  o dasd switched to alloc_disk()
  o ubd switched to alloc_disk()
  o swim* switched to alloc_disk()
  o jsflash switched to alloc_disk()
  o xpram switched to alloc_disk()
  o atari floppy switched to alloc_disk()
  o amiga floppy switched to alloc_disk()
  o acorn floppy switched to alloc_disk()
  o paride floppy switched to alloc_disk()
  o DAC960 switched to alloc_disk()
  o cciss.c switched to use of alloc_disk()
  o fix of bug in previous DAC960 patch
  o cpqarray switched to alloc_disk()
  o acsi switched to alloc_disk()

Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>:
  o net/ipv6/mcast.c: Handle IPV6_LEAVE_GROUP with ipv6mr_interface==0

Andrew Morton <akpm@digeo.com>:
  o misc (mainly documentation)
  o sys_ioperm atomicity fix
  o mprotect bugfix
  o remove bogus BUG in page_remove_rmap()
  o radix tree gang lookup
  o truncate/invalidate_inode_pages rewrite
  o add /proc/vmstat (start of /proc/stat cleanup)
  o add kswapd success accounting to /proc/vmstat
  o "io wait" process accounting
  o convert direct-io to use bio_add_page()
  o fix /proc/vmstat:pgpgout/pgpgin
  o hugetlb kmap fix
  o remove debug code from list_del()
  o distinguish between address span of a zone and the number
  o truncate fixes
  o O_DIRECT invalidation fix
  o mempool wakeup fix
  o separation of direct-reclaim and kswapd functions
  o fix reclaim for higher-order allocations
  o use bio_get_nr_vecs() hint for pagecache writeback
  o Documentation/filesystems/ext3.txt
  o use bio_get_nr_vecs() for sizing direct-io BIOs
  o remove write_mapping_buffers()
  o use buffer_boundary() for writeback scheduling  hints
  o remove page->virtual
  o stricter dirty memory clamping
  o clean up ll_rw_block()
  o smbfs compile fix

Anton Blanchard <anton@samba.org>:
  o 64-bit timer fix

Arjan van de Ven <arjanv@redhat.com>:
  o Remove sys_call_table export

Arnaldo Carvalho de Melo <acme@conectiva.com.br>:
  o LLC: remove unused mac_dev_peer
  o LLC: grab the skb in llc_conn_state_event, use llc_pdu_sn_hdr
  o LLC: kill llc_conn_free_ev, use plain kfree_skb instead
  o LLC: start using seq_file for proc stuff
  o LLC: now it only uses seq_file for proc stuff
  o IPX: use seq_file for proc stuff
  o X25: use seq_file for proc stuff
  o X25: fix permission bogosity in create_proc_entry usage
  o Appletalk: use seq_file for proc stuff

Art Haas <ahaas@neosoft.com>:
  o C99 designated initializers for include/linux/isapnp.h
  o [IPV4/IPV6]: C99 designated initializers

Ben Collins <bcollins@debian.org>:
  o IEEE1394 updates to 2.5.40
  o More 1394 updates

Bjorn Andersson <bjorn.andersson@erc.ericsson.se>:
  o net/8021q/vlan_dev.c: Fix lockup when setting egress priority

Brian Gerst <bgerst@didntduck.org>:
  o struct super_block cleanup - hpfs
  o struct super_block cleanup - ext3
  o unistd.h cleanups

Christer Weinigel <christer@weinigel.se>:
  o Updated NatSemi SCx200 patches for Linux-2.5

Christoph Hellwig <hch@dhcp212.munich.sgi.com>:
  o XFS: temporarily switch to schedule_task for I/O completion
  o XFS: remove description of mount option not in mainline

Christoph Hellwig <hch@sgi.com>:
  o XFS updates for workqueues
  o workqueue flush on destroy
  o no more flush_workqueue in xfs
  o Remove some more devfs crap

Chuck Lever <cel@citi.umich.edu>:
  o add struct file* to ->direct_IO addr space op
  o remove NFS client internal dependence on page->index
  o initial support for NFS direct I/O for 2.5

Daisy Chang <daisy@teetime.dynamic.austin.ibm.com>:
  o sctp: Added the 'Unrecognized Parameter' handling
  o Remove excessive spaces
  o Remove more excessive spaces

Dave Kleikamp <shaggy@kleikamp.austin.ibm.com>:
  o JFS: Releasing LOGGC_LOCK too early

David Brownell <david-b@pacbell.net>:
  o USB: framework for testing usbcore
  o pci/pool.c less spinlock abuse

David S. Miller <davem@nuts.ninka.net>:
  o [EQL]: Rewrite to be SMP safe
  o net/sctp/inqueue.c: Convert to work queue
  o net/ipv6/route.c: Fix typo in previous change
  o net/ipv6/ipv6_sockglue.c: Support IPV6_ADDRFORM getsockopt
  o [SPARC64]: header cleanup, extern inline --> static inline
  o include/asm-sparc64/pstate.h: Kill asm routines, nobody uses them
  o [SPARC]: Update for dequeue_signal changes
  o [SPARC]: Uninline kmap atomic operations
  o arch/sparc64/defconfig: Update
  o [VLAN]: Accept zero vlan at unregister
  o arch/sparc64/mm/init.c: Initialize {min,max}_low_pfn and max_pfn
    properly
  o net/core/dev.c: Print lethal dev/protocol errors with KERN_CRIT
  o net/8021q/vlan.c: Unsigned value may never be < 0

David S. Miller <davem@redhat.com>:
  o USB: usbkbd fix

Florian Thiel <florian.thiel@gmx.net>:
  o Documentation/networking/tuntap.txt: Completely rework, this
    document was much outdated

GOTO Masanori <gotom@debian.or.jp>:
  o NinjaSCSI-32Bi/UDE PCI/Cardbus SCSI core driver
  o NinjaSCSI-32Bi/UDE PCI/Cardbus SCSI driver incidentals

Greg Kroah-Hartman <greg@kroah.com>:
  o USB: speedtouch driver fix due to ioctl function parameter change
  o hotplug: fix for non-pci and usb calls
  o USB: split the usb serial console code out into its own file
  o PCI: remove pcibios_find_class()
  o PCI: remove pci_find_device()
  o PCI: removed pcibios_present()
  o PCI: fixed remaining usages of pcibios_present() that I missed
    previously
  o PCI: remove usages of pcibios_find_class()
  o PCI: remove pcibios_find_device() from the 53c7,8xx.c SCSI driver

Hideaki Yoshifuji <yoshfuji@linux-ipv6.org>:
  o [IPv6]: Rework default router selection
  o [IPV4/IPV6]: General cleanups

Hugh Dickins <hugh@veritas.com>:
  o Oracle startup split_vma fix
  o tmpfs swapoff deadlock
  o cleanup of page->flags manipulations
  o shmem_rename() fixes
  o tpmfs: fake a non-zero size for directories
  o tmpfs: minor fixes
  o add shmem_vm_writeback()
  o shmem truncate race fix
  o shmem: remove info->sem
  o consolidate shmem_getpage and shmem_getpage_locked
  o shmem: avoid metadata leakiness
  o put shmem metadata in highmem
  o shmem accounting fixes
  o shmem: misc changes and cleanups
  o shmem whitespace cleanups

Ingo Molnar <mingo@elte.hu>:
  o Workqueue Abstraction
  o dump_stack() cleanup, BK-curr
  o futex-2.5.40-B5
  o sigfix-2.5.40-D6
  o timer-2.5.40-F7
  o workqueue lossage (fwd)

Ivan Kokshaysky <ink@jurassic.park.msu.ru>:
  o alpha compile fixes

James Bottomley <jejb@mulgrave.(none)>:
  o first cut at fixing unable to requeue with no outstanding commands
  o add cache synchronisation to sd
  o remove mid-layer assumption that devices must be able to queue at
    least one command
  o [PATCH SCSI] make BUSY status stall the device for a while
  o [SCSI] remove debugging from zero depth queue handling
  o [SCSI] remove comment that every host is expected to be able to
    queue at least one command
  o [SCSI] tidy up sd synchronize cache messages into a single line
  o [SCSI] sd moved synchronisation from release to detach

Jaroslav Kysela <perex@suse.cz>:
  o ALSA update [1-12] - 2002/08/09-2002/10/01
  o ALSA fixes
  o fix sgalaxy.c driver cli/sti code

Jeff Garzik <jgarzik@mandrakesoft.com>:
  o Fix natsemi net drvr build, s/KERN_WARN/KERN_WARNING/
  o airo wireless netdrvr: s/routine/func/ to fix build (wq-related
    breakage)

Jens Axboe <axboe@suse.de>:
  o deadline updates
  o ide-cd updates
  o ide config.in
  o cleanup taskfile submit
  o remove _P/_p delaying iops
  o ide low level driver updates
  o pass elevator type by reference, not value
  o ide io port types

Jochen Karrer <cip307@cip.physik.uni-wuerzburg.de>:
  o USB: string query fix

Jon Grimm <jgrimm@touki.austin.ibm.com>:
  o sctp:  mark functions needed by testsuite as SCTP_STATIC
  o sctp:  Fix GFP_KERNEL allocation with lock held
  o sctp: Fix GFP_KERNEL allocation with lock held
  o sctp:  Fix bug where we were erroneously throwing away packets >
    frag_point. (jgrimm)
  o sctp:  Cleanup 'sacked' queue upon teardown.  (jgrimm)

Kai Germaschewski <kai@tp1.ruhr-uni-bochum.de>:
  o NET: Do not use dev->hard_header_len in eth_header()
  o NET: Do not use dev->hard_header_len in eth_type_trans()
  o ISDN: Use a skb queue instead of open coded solution in isdn_ppp.c
  o ISDN: More moving of per-channel stuff into isdn_net_dev
  o ISDN: More sorting out of members for isdn_net_local / isdn_net_dev
  o ISDN: adapt to task queue changes
  o ISDN: Use list.h list for list of online channels
  o Cset exclude:
    kai@tp1.ruhr-uni-bochum.de|ChangeSet|20020929194514|33195
  o kbuild: Small cleanups
  o kbuild: Remove xfs vpath hack
  o kbuild: Standardize ACPI Makefiles
  o kbuild: Small quirks for separate obj / src trees
  o kbuild: Add some bug traps
  o kbuild: Handle $(core-y) the same way as $(init-y), $(drivers-y)
    etc
  o kbuild: Use $(core-y) and friends directly
  o kbuild: Always build helpers in script/
  o kbuild: Don't cd into subdirs during build
  o kbuild: include arch-Makefile in common place
  o kbuild: Adapt mrproper targets
  o ISDN: Alloc isdn_net_dev and struct net_device separately
  o ISDN: Use generic eth_type_trans()
  o ISDN: Separate hard_start_xmit() for different types of ISDN net
    devices
  o ISDN: Make hard_start_xmit() device type specific
  o kbuild: Fix build with modversions
  o kbuild: small fixes
  o kbuild: Nicer warnings
  o kbuild: Don't descend into arch/i386/boot
  o kbuild: Put .bss back to the end of vmlinux
  o ISDN: New file for net interface config and basic setup
  o ISDN: Convert remaining users of the old slave list
  o ISDN: split isdn_net state machine
  o kbuild: fix make -jN warnings
  o ISDN: Reuse the dial_timer for idle hangup
  o kbuild: Fix arch/i386/boot clean targets
  o ISDN: Make idle timeout and retry wait parts of the state machine
  o Cset exclude:
    kai@tp1.ruhr-uni-bochum.de|ChangeSet|20021005215705|12351
  o kbuild: Fix kallsyms build
  o kbuild: Fix make clean in scripts/lxdialog
  o ISDN: tidy up isdn_net_log_skb()
  o ISDN: Replace rx_netdev, st_netdev by a single field
  o ISDN: Separate state machine actions into single functions
  o ISDN: Move call control to isdn_net_lib.c
  o ISDN: Make the state machine explicit
  o ISDN: Use a tasklet for the Eicon driver
  o ISDN: Extend state machine
  o ISDN: Allow for return values in the state machine

Kai Makisara <Kai.Makisara@kolumbus.fi>:
  o SCSI tape devfs & driverfs fix

Lennert Buytenhek <buytenh@gnu.org>:
  o [NET]: Remove net_call_rx_atomic
  o [BRIDGE]: Skip the LISTENING_STP state if STP is disabled
  o [BRIDGE]: take BR_NETPROTO_LOCK for unlinking bridge device slaves

Linus Torvalds <torvalds@home.transmeta.com>:
  o Update to DRI CVS tree
  o Fix sigio process lookup handling
  o Undo due to weird behaviour on various boxes Cset exclude:
    ink@jurassic.park.msu.ru|ChangeSet|20021003201553|58706
  o Use dump_stack() for the USB storage buffer size checking, to make
    it possible to track down.
  o Increase the delay in waiting for pcmcia drivers to register
  o Make wildcard dependency filenames be relative, not absolute
  o Don't add the $(obj) prefix twice
  o bitmap_member() => DECLARE_BITMAP()
  o Add <linux/linkage.h> include to get FASTCALL() define
  o IO scheduler is a subsystem, not a driver. Initialize it as such
  o Oops, it's 'xxx_initcall()', not 'xxx_init()' (except for the
    legacy module_init(), just to confuse people).
  o Export the gdt table GPL-only for APM
  o Linux v2.5.41

Luc Van Oostenryck <luc.vanoostenryck@easynet.be>:
  o 2.5.40: warning fix for drivers/usb/core/usb.c

Maksim Krasnyanskiy <maxk@viper.(none)>:
  o Sync up Bluetooth core with 2.4.x
  o Bluetooth USB driver update
  o Syncup HCI UART driver with 2.4.x
  o BNEP (Bluetooth Network Encapsulation Protocol) support
  o RFCOMM protocol support
  o Fix designated initializers in RFCOMM TTY layer

Manfred Spraul <manfred@colorfullife.com>:
  o drivers/net/natsemi.c:  create a function for rx refill
  o drivers/net/natsemi.c: combine drain_ring and init_ring
  o drivers/net/natsemi.c: OOM handling
  o drivers/net/natsemi.c: stop abusing netdev_device_{de,a}ttach
  o drivers/net/natsemi.c: write MAC address back to the chip
  o drivers/net/natsemi.c: lengthen EEPROM timeout, and always warn
    about all timeouts
  o drivers/net/natsemi.c: comments update
  o pipe bugfix /cleanup

Marcel Holtmann <marcel@holtmann.org>:
  o Bluetooth kbuild fix and config cleanup

Martin Devera <devik@cdi.cz>:
  o net/sched/sch_htb.c: Check that node is really leaf before
    modifying cl->un.leaf

Martin Schwidefsky <schwidefsky@de.ibm.com>:
  o s390 update (1/27): arch
  o s390 update (2/27): include
  o s390 update (3/27): drivers
  o s390 update (4/27): syscalls
  o s390 update (5/27): ibm partition
  o s390 update (6/27): config
  o s390 update (7/27): dasd driver
  o s390 update (8/27): xpram driver
  o s390 update (9/27): bottom half removal
  o s390 update (10/27): bitops bug
  o s390 update (11/27): 31 bit emulation
  o s390 update (12/27): linker scripts
  o s390 update (13/27): preemption support
  o s390 update (14/27): inline optimizations
  o s390 update (15/27): 64 bit spinlocks
  o s390 update (16/27): timer interrupts
  o s390 update (17/27): beautification
  o s390 update (18/27): fpu registers
  o s390 update (19/27): ptrace cleanup
  o s390 update (20/27): signal quiesce
  o s390 update (21/27): sync i/o bug
  o s390 update (22/27): s390_process_IRQ
  o s390 update (23/27): channel paths
  o s390 update (24/27): boot sequence
  o s390 update (25/27): init call
  o s390 update (26/27): /proc/interrupts
  o s390 update (27/27): control characters
  o s390 dasd driver update

Matthew Wilcox <willy@debian.org>:
  o [NET]: Move common ioctl code up a layer
  o Remove another for_each_process loop

Mike Anderson <andmike@us.ibm.com>:
  o sg might_sleep fixes

Paul Mackerras <paulus@au1.ibm.com>:
  o pcmcia resource allocation fix

Pavel Machek <pavel@ucw.cz>:
  o Swsusp updates, do not thrash ide disk on suspend

Pete Zaitcev <zaitcev@redhat.com>:
  o mm/highmem.c: Include asm/tlbflush.h
  o arch/sparc/kernel/sun4d_irq.c: init_timers --> sparc_init_timers

Petko Manolov <petkan@users.sourceforge.net>:
  o USB: rtl8150 update
  o USB: pegasus update

Petr Vandrovec <vandrove@vc.cvut.cz>:
  o FAT/VFAT memory corruption during mount()

Richard Henderson <rth@twiddle.net>:
  o alpha strncpy fix

Russell King <rmk@flint.arm.linux.org.uk>:
  o [SERIAL] Allow PCMCIA serial cards to work again
  o [SERIAL] Fix serial includes for modversions/modules

Steven Whitehouse <steve@gw.chygwyn.com>:
  o Trivial fix to aio.c:__aio_get_req()

Tim Hockin <thockin@freakshow.cobalt.com>:
  o Add dp83816 support to drivers/net/natsemi.c
  o drivers/net/natsemi.c: janitorial - whitespace, wrap, and indenting
    cleanup
  o drivers/net/natsemi.c: stop tx/rx and reinit_ring on a PHY reset
  o drivers/net/natsemi.c: cleanup version string, fix compile error
  o drivers/net/natsemi.c: boost some printk() levels to WARN

Trond Myklebust <trond.myklebust@fys.uio.no>:
  o NFS: readdir reply truncated
  o NFS: readdir reply truncated!

Re: Linux v2.5.41

[Linus Torvalds]

On Mon, 7 Oct 2002, Linus Torvalds wrote:
> 
> Tons of stuff. Mucho merges with the "A-Team" (Alan, Al, Alexey, Andrew,
> Anton, Arjan, Arnaldo and Art), but the "M-Team" (Maksim, Marcel, Martin's
> and Mike) is a close runner up. The J's are doing well too.

Oh, and I totally forgot .. 

The merge from Andrew merged in various VM counter updates, which change 
/proc accounting in ways that make some of the system monitoring stuff 
unhappy. In particular, you'll find "vmstat" dumping core, and "top" has a 
few glitches too.

You can fix all of this by just upgrading to a newer "procps" package. Rik 
maintains a procps-2.0.9 (many of the stats were his) at

	http://surriel.com/procps/

I don't think there are ready-made binary packages, but maybe some
enterprising and helpful - and trustworthy - soul can do that and get
vendors to add it to their upgrade list (or at least make it available in
contrib).

		Linus

Re: Linux v2.5.41

[Robert Love]

On Mon, 2002-10-07 at 15:22, Linus Torvalds wrote:

> I don't think there are ready-made binary packages, but maybe some
> enterprising and helpful - and trustworthy - soul can do that and get
> vendors to add it to their upgrade list (or at least make it available in
> contrib).

I am enterprising but do not know about trustworthy :)

I have RPM packages available here:

	http://tech9.net/rml/procps

for the 2.0.9 tree and a CVS dump of a day or two ago as well as
tarballs of the source tree for 2.0.9 and CVS.

I try to keep the CVS dumps fairly up-to-date as we break stuff in
2.5-mm.  Rik and I have also added quite a bit of interesting new
features, so I would recommend everyone try out the new procps.

	Robert "my last name is Love can I be L-Team too ?" Love

Re: Linux v2.5.41 -> don't compil

[Gregoire Favre]

Hello,

don't compil:

make -f drivers/message/i2o/Makefile 
  gcc -Wp,-MD,drivers/message/i2o/.i2o_pci.o.d -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fomit-frame-pointer -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686 -Iarch/i386/mach-generic -nostdinc -iwithprefix include -DMODULE   -DKBUILD_BASENAME=i2o_pci   -c -o drivers/message/i2o/i2o_pci.o drivers/message/i2o/i2o_pci.c
drivers/message/i2o/i2o_pci.c: In function `i2o_pci_core_attach':
drivers/message/i2o/i2o_pci.c:371: warning: implicit declaration of function `i2o_sys_init'
  gcc -Wp,-MD,drivers/message/i2o/.i2o_core.o.d -D__KERNEL__ -Iinclude -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fomit-frame-pointer -fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2 -march=i686 -Iarch/i386/mach-generic -nostdinc -iwithprefix include -DMODULE   -DKBUILD_BASENAME=i2o_core -DEXPORT_SYMTAB  -c -o drivers/message/i2o/i2o_core.o drivers/message/i2o/i2o_core.c
drivers/message/i2o/i2o_core.c:45:26: linux/tqueue.h: No such file or directory
In file included from drivers/message/i2o/i2o_core.c:54:
drivers/message/i2o/i2o_lan.h:139: field `i2o_batch_send_task' has incomplete type
make[3]: *** [drivers/message/i2o/i2o_core.o] Error 1
make[2]: *** [drivers/message/i2o] Error 2
make[1]: *** [drivers/message] Error 2
make: *** [drivers] Error 2

	Grégoire
________________________________________________________________
http://ulima.unil.ch/greg ICQ:16624071 mailto:greg@ulima.unil.ch

[PATCH] Fix IPv6 [was Re: Linux v2.5.41]

[Jasper Spaans]

On Mon, Oct 07, 2002 at 12:02:59PM -0700, Linus Torvalds wrote:

>   o [IPV4/IPV6]: General cleanups

Some part of this makes my tree noncompilable:

net/ipv6/addrconf.c: In function `ipv6_addr_type':
net/ipv6/addrconf.c:155: case label does not reduce to an integer constant
net/ipv6/addrconf.c:159: case label does not reduce to an integer constant
etc.

Please apply this (undo-) patch... 

(using gcc 2.95.4 this does compile)

Regards,

Jasper

--- linux-2.5.40/net/ipv6/addrconf.c~	2002-10-07 22:26:37.000000000 +0200
+++ linux-2.5.40/net/ipv6/addrconf.c	2002-10-07 22:27:43.000000000 +0200
@@ -151,16 +151,16 @@
 	if ((st & htonl(0xFF000000)) == htonl(0xFF000000)) {
 		int type = IPV6_ADDR_MULTICAST;
 
-		switch((st & htonl(0x00FF0000))) {
-			case htonl(0x00010000):
+		switch((st & __constant_htonl(0x00FF0000))) {
+			case __constant_htonl(0x00010000):
 				type |= IPV6_ADDR_LOOPBACK;
 				break;
 
-			case htonl(0x00020000):
+			case __constant_htonl(0x00020000):
 				type |= IPV6_ADDR_LINKLOCAL;
 				break;
 
-			case htonl(0x00050000):
+			case __constant_htonl(0x00050000):
 				type |= IPV6_ADDR_SITELOCAL;
 				break;
 		};


-- 
Jasper Spaans
http://jsp.ds9a.nl/contact/
Tel/Fax: +31-84-8749842
``Got no clue? Too bad for you.''

Re: [PATCH] Fix IPv6

[YOSHIFUJI Hideaki / 吉藤英明]

In article <20021007210343.GA1486@mayo.ds9a.tudelft.nl> (at Mon, 7 Oct 2002 23:03:43 +0200), Jasper Spaans <jasper@spaans.ds9a.nl> says:

> On Mon, Oct 07, 2002 at 12:02:59PM -0700, Linus Torvalds wrote:
> 
> >   o [IPV4/IPV6]: General cleanups
> 
> Some part of this makes my tree noncompilable:
> 
> net/ipv6/addrconf.c: In function `ipv6_addr_type':
> net/ipv6/addrconf.c:155: case label does not reduce to an integer constant
> net/ipv6/addrconf.c:159: case label does not reduce to an integer constant
> etc.

Agreed.

BTW, I repleced __constant_{hton,ntoh}{l,s}() with {hton,htoh}{l,s}(),
but I believe I did NOT do it for "case" in my patch...

-- 
Hideaki YOSHIFUJI @ USAGI Project <yoshfuji@linux-ipv6.org>
GPG FP: 9022 65EB 1ECF 3AD1 0BDF  80D8 4807 F894 E062 0EEA

Re: [PATCH] Fix IPv6

[kuznet]

Hello!

> Agreed.

Me too.

Sigh... that's why code is full of __constant_* in the most unexpected places.
For my straight brains it is much easier to use __constant_* each time when
I know forward that it is a constant instead of keeping in mind all
the gcc bugs. Well, the beast which does not eliminate empty loops,
is unlikely to eliminate if (is_constant_p()) too, right? :-) 

Alexey

Re: Linux v2.5.41

[Adrian Bunk]

On Mon, 7 Oct 2002, Linus Torvalds wrote:

>...
> Summary of changes from v2.5.40 to v2.5.41
> ============================================
>...
> Pavel Machek <pavel@ucw.cz>:
>   o Swsusp updates, do not thrash ide disk on suspend
>...

This change causes the following compile error with CONFIG_DISCONTIGMEM
enabled:

<--  snip  -->

...
  gcc -Wp,-MD,kernel/.suspend.o.d -D__KERNEL__ -Iinclude -Wall
-Wstrict-prototypes -Wno-trigraphs -O2 -fomit-frame-pointer
-fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2
-march=k6 -Iarch/i386/mach-generic -nostdinc -iwithprefix include
-DKBUILD_BASENAME=suspend   -c -o kernel/suspend.o kernel/suspend.c
...
kernel/suspend.c: In function `count_and_copy_data_pages':
kernel/suspend.c:479: `max_mapnr' undeclared (first use in this function)
kernel/suspend.c:479: (Each undeclared identifier is reported only once
kernel/suspend.c:479: for each function it appears in.)
make[1]: *** [kernel/suspend.o] Error 1
make: *** [kernel] Error 2

<--  snip  -->


cu
Adrian

-- 

You only think this is a free country. Like the US the UK spends a lot of
time explaining its a free country because its a police state.
								Alan Cox

Re: Linux v2.5.41

[William Lee Irwin III]

On Mon, 7 Oct 2002, Linus Torvalds wrote:
>> Pavel Machek <pavel@ucw.cz>:
>>   o Swsusp updates, do not thrash ide disk on suspend

On Tue, Oct 08, 2002 at 02:02:50PM +0200, Adrian Bunk wrote:
> This change causes the following compile error with CONFIG_DISCONTIGMEM
> enabled:
> kernel/suspend.c: In function `count_and_copy_data_pages':
> kernel/suspend.c:479: `max_mapnr' undeclared (first use in this function)
> kernel/suspend.c:479: (Each undeclared identifier is reported only once
> kernel/suspend.c:479: for each function it appears in.)
> make[1]: *** [kernel/suspend.o] Error 1
> make: *** [kernel] Error 2

max_mapnr must die. It's mostly buggy largely because it's not what
people think it is. Most of the time people want max_pfn, and the rest
they don't want it at all.

Pavel, you might also want to make config options conflict instead of
panicking.

It's mostly an expression of opinion, but here's a patch.
Totally untested, ignoring all but x86's arch code, vs. 2.5.41:


diff -urN linux-2.5.41/arch/i386/config.in swsusp-2.5.41/arch/i386/config.in
--- linux-2.5.41/arch/i386/config.in	2002-10-07 11:24:02.000000000 -0700
+++ swsusp-2.5.41/arch/i386/config.in	2002-10-08 06:17:11.000000000 -0700
@@ -441,7 +441,7 @@
 
 mainmenu_option next_comment
 comment 'Kernel hacking'
-if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
+if [ "$CONFIG_EXPERIMENTAL" = "y" -a "$CONFIG_DISCONTIGMEM" != "y" ]; then
    dep_bool 'Software Suspend (EXPERIMENTAL)' CONFIG_SOFTWARE_SUSPEND $CONFIG_PM
 fi
 
diff -urN linux-2.5.41/arch/i386/mm/discontig.c swsusp-2.5.41/arch/i386/mm/discontig.c
--- linux-2.5.41/arch/i386/mm/discontig.c	2002-10-07 11:24:39.000000000 -0700
+++ swsusp-2.5.41/arch/i386/mm/discontig.c	2002-10-08 06:18:19.000000000 -0700
@@ -271,13 +271,3 @@
 	totalram_pages += totalhigh_pages;
 #endif
 }
-
-void __init set_max_mapnr_init(void)
-{
-#ifdef CONFIG_HIGHMEM
-	highmem_start_page = NODE_DATA(0)->node_zones[ZONE_HIGHMEM].zone_mem_map;
-	num_physpages = highend_pfn;
-#else
-	num_physpages = max_low_pfn;
-#endif
-}
diff -urN linux-2.5.41/arch/i386/mm/init.c swsusp-2.5.41/arch/i386/mm/init.c
--- linux-2.5.41/arch/i386/mm/init.c	2002-10-07 11:23:35.000000000 -0700
+++ swsusp-2.5.41/arch/i386/mm/init.c	2002-10-08 06:18:07.000000000 -0700
@@ -411,19 +411,9 @@
 }
 
 #ifndef CONFIG_DISCONTIGMEM
-static void __init set_max_mapnr_init(void)
-{
-#ifdef CONFIG_HIGHMEM
-	highmem_start_page = pfn_to_page(highstart_pfn);
-	max_mapnr = num_physpages = highend_pfn;
-#else
-	max_mapnr = num_physpages = max_low_pfn;
-#endif
-}
 #define __free_all_bootmem() free_all_bootmem()
 #else
 #define __free_all_bootmem() free_all_bootmem_node(NODE_DATA(0))
-extern void set_max_mapnr_init(void);
 #endif /* !CONFIG_DISCONTIGMEM */
 
 #ifdef CONFIG_HUGETLB_PAGE
@@ -447,8 +437,6 @@
 	
 	bad_ppro = ppro_with_ram_bug();
 
-	set_max_mapnr_init();
-
 	high_memory = (void *) __va(max_low_pfn * PAGE_SIZE);
 
 	/* clear the zero-page */
diff -urN linux-2.5.41/include/asm-i386/page.h swsusp-2.5.41/include/asm-i386/page.h
--- linux-2.5.41/include/asm-i386/page.h	2002-10-07 11:23:32.000000000 -0700
+++ swsusp-2.5.41/include/asm-i386/page.h	2002-10-08 06:17:43.000000000 -0700
@@ -145,7 +145,7 @@
 #ifndef CONFIG_DISCONTIGMEM
 #define pfn_to_page(pfn)	(mem_map + (pfn))
 #define page_to_pfn(page)	((unsigned long)((page) - mem_map))
-#define pfn_valid(pfn)		((pfn) < max_mapnr)
+#define pfn_valid(pfn)		((pfn) < max_pfn)
 #endif /* !CONFIG_DISCONTIGMEM */
 #define virt_to_page(kaddr)	pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)
 
diff -urN linux-2.5.41/include/linux/mm.h swsusp-2.5.41/include/linux/mm.h
--- linux-2.5.41/include/linux/mm.h	2002-10-07 11:23:22.000000000 -0700
+++ swsusp-2.5.41/include/linux/mm.h	2002-10-08 06:17:31.000000000 -0700
@@ -15,10 +15,6 @@
 #include <linux/rbtree.h>
 #include <linux/fs.h>
 
-#ifndef CONFIG_DISCONTIGMEM          /* Don't use mapnrs, do it properly */
-extern unsigned long max_mapnr;
-#endif
-
 extern unsigned long num_physpages;
 extern void * high_memory;
 extern int page_cluster;
diff -urN linux-2.5.41/kernel/ksyms.c swsusp-2.5.41/kernel/ksyms.c
--- linux-2.5.41/kernel/ksyms.c	2002-10-07 11:22:56.000000000 -0700
+++ swsusp-2.5.41/kernel/ksyms.c	2002-10-08 06:16:37.000000000 -0700
@@ -117,7 +117,6 @@
 #ifndef CONFIG_DISCONTIGMEM
 EXPORT_SYMBOL(contig_page_data);
 EXPORT_SYMBOL(mem_map);
-EXPORT_SYMBOL(max_mapnr);
 #endif
 EXPORT_SYMBOL(high_memory);
 EXPORT_SYMBOL(vmtruncate);
diff -urN linux-2.5.41/kernel/suspend.c swsusp-2.5.41/kernel/suspend.c
--- linux-2.5.41/kernel/suspend.c	2002-10-07 11:23:37.000000000 -0700
+++ swsusp-2.5.41/kernel/suspend.c	2002-10-08 06:16:54.000000000 -0700
@@ -474,9 +474,9 @@
 #ifdef CONFIG_DISCONTIGMEM
 	panic("Discontingmem not supported");
 #else
-	BUG_ON (max_mapnr != num_physpages);
+	BUG_ON (max_pfn != num_physpages);
 #endif
-	for (pfn = 0; pfn < max_mapnr; pfn++) {
+	for (pfn = 0; pfn < max_pfn; pfn++) {
 		page = pfn_to_page(pfn);
 		if (PageHighMem(page))
 			panic("Swsusp not supported on highmem boxes. Send 1GB of RAM to <pavel@ucw.cz> and try again ;-).");
diff -urN linux-2.5.41/mm/memory.c swsusp-2.5.41/mm/memory.c
--- linux-2.5.41/mm/memory.c	2002-10-07 11:24:12.000000000 -0700
+++ swsusp-2.5.41/mm/memory.c	2002-10-08 06:17:18.000000000 -0700
@@ -55,7 +55,6 @@
 
 #ifndef CONFIG_DISCONTIGMEM
 /* use the per-pgdat data instead for discontigmem - mbligh */
-unsigned long max_mapnr;
 struct page *mem_map;
 #endif
 

Re: [PATCH] Fix IPv6

[David S. Miller]

   From: YOSHIFUJI Hideaki / 吉藤英明 <yoshfuji@linux-ipv6.org>
   Date: Tue, 08 Oct 2002 10:08:53 +0900 (JST)
   
   BTW, I repleced __constant_{hton,ntoh}{l,s}() with {hton,htoh}{l,s}(),
   but I believe I did NOT do it for "case" in my patch...

It is error I introduced into the 2.5.x version of
your patch since I had to make some parts by hand.

Fixed in my tree now.

Re: Linux v2.5.41

[Pavel Machek]

Hi!

> > This change causes the following compile error with CONFIG_DISCONTIGMEM
> > enabled:
> > kernel/suspend.c: In function `count_and_copy_data_pages':
> > kernel/suspend.c:479: `max_mapnr' undeclared (first use in this function)
> > kernel/suspend.c:479: (Each undeclared identifier is reported only once
> > kernel/suspend.c:479: for each function it appears in.)
> > make[1]: *** [kernel/suspend.o] Error 1
> > make: *** [kernel] Error 2
> 
> max_mapnr must die. It's mostly buggy largely because it's not what
> people think it is. Most of the time people want max_pfn, and the rest
> they don't want it at all.

It seems more logical now.

> Pavel, you might also want to make config options conflict instead of
> panicking.

Actually, I think I prefer panic(). That way it will be less broken
when someone will try to fix discontigmem.

> diff -urN linux-2.5.41/kernel/suspend.c swsusp-2.5.41/kernel/suspend.c
> --- linux-2.5.41/kernel/suspend.c	2002-10-07 11:23:37.000000000 -0700
> +++ swsusp-2.5.41/kernel/suspend.c	2002-10-08 06:16:54.000000000 -0700
> @@ -474,9 +474,9 @@
>  #ifdef CONFIG_DISCONTIGMEM
>  	panic("Discontingmem not supported");
>  #else
> -	BUG_ON (max_mapnr != num_physpages);
> +	BUG_ON (max_pfn != num_physpages);
>  #endif
> -	for (pfn = 0; pfn < max_mapnr; pfn++) {
> +	for (pfn = 0; pfn < max_pfn; pfn++) {
>  		page = pfn_to_page(pfn);
>  		if (PageHighMem(page))
>  			panic("Swsusp not supported on highmem boxes. Send 1GB of RAM to <pavel@ucw.cz> and try again ;-).");

Applied.

								Pavel
-- 
When do you have heart between your knees?