Hmm.. Lots of threading work by Ingo Molnar, more merges with Andrew
Morton, and Rusty "trivial" Russell (because BK only holds one person
responsible the trivial patches get attributed to the original author, not
Rusty, but Rusty should get a special attribution for maintaining the set
of patches).
Oh, and Jeff Dike's UML is in.
ACPI, PPC, USB, Sparc etc updates.
Linus
Summary of changes from v2.5.34 to v2.5.35
============================================
<[email protected]>:
o designated initializer patches for fs_nfs
Andi Kleen <[email protected]>:
o Because x86-64 also always reserves the kbd region, we must not
call request_region() in i8042-io.h, like we don't for i386, alpha,
etc.
Andrew Morton <[email protected]>:
o writer throttling fix
o pass the correct flags to aops->releasepage()
o exact dirty state accounting
o discontigmem code cleanup #1
o discontigmem code cleanup #2
o reduce the default dirty memory thresholds
o buffer_head takedown for bighighmem machines
o rmap pte_chain speedup and space saving
o resurrect CONFIG_HIGHPTE
o readv/writev speedup
o Use a sync iocb for generic_file_read
<[email protected]>:
o typo: include_linux_pci_ids.h s_DEVIDE_DEVICE
<[email protected]>:
o Lexar USB CF Reader
<[email protected]>:
o drivers_net_pcmcia_fmvj18x_cs.c save_flags unsigned check
o drivers_net_arcnet_arcnet.c save_flags unsigned check
o drivers_net_hamradio_scc.c save_flags unsigned check
o drivers_net_3c505.c save_flags unsigned check
o drivers_net_pcmcia_3c574_cs.c save_flags unsigned check
o drivers_net_ni65.c save_flags unsigned check
o drivers_net_de600.c save_flags unsigned check
o drivers_net_pcmcia_aironet4500_cs.c save_flags unsigned check
o drivers_net_pcmcia_smc91c92_cs.c
o drivers_net_at1700.c save_flags unsigned check
<[email protected]>:
o net/ipv4/ip_options.c: IPOPT_END padding needs to increment optptr
<[email protected]>:
o Typo in do_syslog/__down_trylock lockup fix
<[email protected]>:
o USER_HZ & NTP problems
<[email protected]>:
o [PATCH] (repost) fix for big endian machines in scanner.c
<[email protected]>:
o Typos in drivers_s390_net_iucv.h
o drivers_scsi_aic7xxx_aic7xxx_core.c, typo: the the
<[email protected]>:
o Toshiba.c IRQ Patch (Christoph Hellwig eats people?)
Jeff Dike <[email protected]>:
o UML arch (user-mode Linux)
<[email protected]>:
o A small documentation update and a unused constant removal
<[email protected]>:
o usbmidi patch
<[email protected]>:
o 2.5.31_drivers_char_lp.c
<[email protected]>:
o [patch, 2.5] fix errorpath in apne.c
o [patch 2.5] at1700 trivial
<[email protected]>:
o HandyTech HandyLink patch
<[email protected]>:
o USB: se401 driver update
<[email protected]>:
o USB: pegasus driver patch
<[email protected]>:
o 2.5.X config: USB speedtouch driver
<[email protected]>:
o Few small fixes for Q40 keyboard support
<[email protected]>:
o ftape EXPORT_SYMBOL damage clean-up
o zftape: Cleanup zftape_syms.c
o drivers/char/Makefile: Remove pty.o from export-objs
<[email protected]>:
o 2.5.34 ufs/super.c
o Comment fix asm-i386_hardirq.h
o include/asm-sparc/hardirq.h: Fix comment
<[email protected]>:
o USB storage: abort bug fix
<[email protected]>:
o sleeping file locks
o Remove unused Config.help
o remove SERIAL_IO_GSC
<[email protected]>:
o IBM PCI Hotplug driver update
o IBM PCI Hotplug driver update for ISA based controllers
o IBM PCI Hotplug driver update for PCI based controllers
<[email protected]>:
o pci_free_consistent on ohci initialisation failure
Adam J. Richter <[email protected]>:
o The following patch shaves a six bytes from the loaded size
o Building list of drives in right order
Alexander Viro <[email protected]>:
o Missing IDE partition 3 of 3 on 2.5.34
Andries E. Brouwer <[email protected]>:
o Feiya 5-in-1 Card Reader
Andy Grover <[email protected]>:
o ACPI trivial cleanups (Kochi Takayoshi)
o By John Belmonte - improvements to Toshiba ACPI driver
o ACPI Config.in update by Christoph Hellwig
o Remove obsolete OSL functions (Kochi Takayoshi)
o ifdef some arch-specific ACPI code
o ACPI: Remove interpreter debugger and kdb directories. These
ultimately didn't prove useful enough to be used on a regular
basis.
o ACPI: Do not do certain bits of APIC config if CONFIG_ACPI_HT_ONLY
is set
o ACPI: Do not compile functions not used in HT_ONLY mode
o ACPI: Fix possible sleeping at interrupt context (Matthew Wilcox)
Anton Blanchard <[email protected]>:
o ppc64: remove some unnecessary sign extensions
o ppc64: remove ancient stat syscalls
o ppc64: add mmap64 support
o ppc64: add sendfile64 support and restore ioperm syscall
o ppc64: Dont force O_LARGEFILE on for 32 bit apps. From sparc64
o ppc64: merge in changes from x86 irq balance code
o ppc64: Update the fake pci read code to handle a return of all 1s
o ppc64: Fix sys32_readahead wrapper to obey ABI wrt passing long
longs
o ppc64: remove status, no longer used
o ppc64: Remove use of <asm/smplock.h>
o ppc64: remove some old code
o ppc64: clean up syscall table, making it obvious which are obsolete
and which are 32 bit only
o ppc64: Remove old keyboard code
o ppc64: fixes for 2.5.32
o hvc_console: stop HVC console while xmon is running
o ppc64: make udelay a barrier, fixes problem with input layer
keyboard probing
o ppc64: defconfig update
o ppc64: config.in cleanup
o ppc64: Add security and AIO syscalls ppc64: copy FE0 and FE1 bits
into MSR when ptracing ppc64: warn when registering duplicate
ioctls
o ppc64: Compile in LLC, needed for token ring
o ppc64: turn off token ring for the moment, it oopses
o ppc64: add arg to do_fork and fix ELF_AUX entries as done in ppc32
o ppc64: INIT_SIGNALS fix
o ppc64: add rwlock_is_locked
o ppc64: xtime.tv_nsec fixes
o ppc64: DISCONTIGMEM updates, rework to be like x86 version
o ppc64: add in_atomic
o ppc64: updates from Rochester
o ppc64: EEH update from Todd Inglett
o ppc64: Allocate RTAS above OF, from Peter Bergner
o ppc64: new pci config methods, from Todd Inglett
o ppc64: updates from Rochester
o ppc64: UP compile fixes
Arnaldo Carvalho de Melo <[email protected]>:
o LLC: small cleanups, leave debug on for a while
o LLC: tcpfying the beast
o LLC: sys_listen already checks for backlog > SOMAXCONN
o LLC: llc_build_and_send_pkt
o LLC: kill llc_prim_data and LLC_PRIM_DATA for sap->ind() and
sap->conf()
o [LLC] split llc_pdu_router into llc_{station,sap,conn}_rcv
o [LLC] llc_ui_wait_for_data and socket locking fixes
o [LLC] use llc_mac_{match,null} in more places
o [LLC] turn tons of simple pdu functions into returning void
o [LLC] use just one struct sock per connection
o LLC: llc_lookup_listener
o [LLC] llc_establish_connection & LLC_CONN_PRIM bits the bucket
o [LLC] llc_send_disc & LLC_DISC_PRIM bites the dust
o [LLC] add missing kfree_skb in llc_conn_state_process
o [LLC] remove unsupported flowcontrol prim bits
o [LLC] save sockaddr_llc info in connection packets
o [LLC] kill sap->req()
o [LLC] remove all tmr ev structs & fix psnap and p8022 wrt ui
sending
Brad Hards <[email protected]>:
o header cleanup - drivers_char_dz.c
o header cleanup - drivers_char_serial_tx3912.c
o Re: header cleanup - drivers_ieee1394_sbp2.c
o Change "D: Drivers=" to "H: Handlers=" in /proc/bus/input/devices
Chris Wright <[email protected]>:
o 2.5.34 kernel-api DocBook fix
Dave Kleikamp <[email protected]>:
o Bump up JFS_LINK_MAX from 64K to 4G
o Extended attribute fixes for JFS
o JFS: add permission checks before getting or setting xattrs
o JFS: cleanup -- Remove excessive typedefs
David Brownell <[email protected]>:
o uhci, doc + cleanup
o Re: [patch 2.5.31-bk5] uhci, misc
o ehci locking
o Re: updated ehci patch
o ohci-hcd endpoint scheduling, driverfs
o usbnet, add YOPY device IDs
o [PATCH 2.5.33+] ohci and iso-in
o usbnet, Epson client
o ehci misc fixes
o ehci, async idle timout
David S. Miller <[email protected]>:
o [TIGON3]: Merge to version 1.1
o [TIGON3] Initial TCP segmentation offload support
o [TIGON3] Fix typos in TSO changes
o [TIGON3]: Force use of PCI config space reg writes when loading
firmare
o [TIGON3]: Disable TSO for now, tso firmware can hang tx cpu
o [TCP]: Delay tstamp state commit in input fast path until we verify
csum
o [TIGON3]: Do not reference vlgrp unless TG3_VLAN_TAG_USED is set
o [TIGON3]: Fix slight perf regression from TSO changes
o [VLAN] Use unregister_netdevice to prevent rtnl double-lock
o [TIGON3]: New way to flush posted writes of GRC_MISC_CFG
o [NAPI]: Do not check netif_running() in netif_rx_schedule_prep
o [NAPI]: Set SCHED before dev->open, clear if fails. Restore
netif_running check to netif_rx_schedule_prep
o [TIGON3]: Use spin_lock_irqsave in tg3_interrupt, fixes SMP hang
o [TIGON3]: Add 5704 support
o [TIGON3]: GRC_MISC_CFG_BOARD_ID_5704CIOBE is wrong
o kernel/signal.c: Not all systems have SIGSTKFLT
o [SPARC]: Catchup with signal infrastructure changes
o [SPARC]: pcibios_enable_device has new mask argument
o [SPARC64]: timespecs now have tv_nsec in place of tv_usec
o [SPARC64]: Delete do_gettimeofday asm
o [SPARC]: Update ide headers. WARNING: this is known broken, fixes
coming from Jens Axboe
o [SPARC64]: Add rwlock_is_locked and in_atomic
o arch/sparc64/defconfig: Update
o arch/sparc/kernel/check_asm.sh: Handle output from newer versions
of GCC
o [SPARC]: Add rwlock_is_locked
o [SPARC]: Add is_atomic
o [SPARC]: Update for tv_nsec in xtime
o [SPARC]: Add irqs_disabled
o [SPARC]: Add kmap_atomic_to_page
o [SPARC]: Add sys_exit_group syscall entries
o [LLC]: Fix build bustage
[email protected] <[email protected]>:
o I needed this small patch if i8042.c is built as a module. Franz
Greg Kroah-Hartman <[email protected]>:
o USB: remove __NO_VERSION__
o USB: clean up the error path in create_special_files() for usbfs
o Compaq PCI Hotplug driver: fixed __FUNCTION__ usages
o PCI: hotplug core cleanup to get pci hotplug working again
o PCI: export pci_scan_bus() as the IBM PCI Hotplug driver needs it
o PCI Hotplug: remove pci_*_nodev() prototypes as the functions are
gone
o Compaq PCI Hotplug driver: changed calls to pci_*_nodev() to
pci_bus_*()
o IBM PCI Hotplug driver: changed calls to pci_*_nodev() to
pci_bus_*()
o USB: compile time fix for previous kaweth patch
Ingo Molnar <[email protected]>:
o Re: do_syslog/__down_trylock lockup in current BK
o exit.c compilation warning fix
o sys_exit_group(), threading, 2.5.34
o Thread deadlock fix
o ptrace-fix-2.5.34-A2, BK-curr
o sys_exit() threading improvements, BK-curr
o NMI watchdog SMP fix
o thread exit deadlock bug
o signal failures in nightly LTP test
o hide-threads-2.5.34-C1
o wait4-fix-2.5.34-A0, BK-curr
o clone-fix-2.5.34-A0, BK-curr
o detached-fix-2.5.34-A0, BK-curr
o exit-thread-2.5.34-A0, BK-curr
o wait4-fix-2.5.34-B2, BK-curr
o exit-fix-2.5.34-C0, BK-curr
o thread-exec-2.5.34-B1, BK-curr
o thread exec fix, BK-curr
Ivan Kokshaysky <[email protected]>:
o alpha update
Jens Axboe <[email protected]>:
o Move around IDE files to match 2.4.20-pre5-ac4 layout. Do this
before applying patches, for clarity and for keeping bk revision
history.
o Add Makefile's for the new arm/ legacy/ pci/ pci/ directories
o update IDE to 2.4.20-pre5-ac4
o ide_map_buffer() and ide_unmap_buffer() could cause imbalanced
calls to bio_kmap/kunmap_irq(), which would screw the preemption
count. pass in rq to ide_unmap_buffer() as well to make the right
decision.
o PCI individual resource handling
o blk_fs_request()
o IDE pci ids
o hdreg command updates etc
Juan Quintela <[email protected]>:
o Grammatical fixes
Kai Germaschewski <[email protected]>:
o kbuild: Fix up non-verbose mode
o kbuild: Fix copying of shipped files
o kbuild: Use normal rule for preprocessing vmlinux.lds.S
Linus Torvalds <[email protected]>:
o Get Intel model name from the CPU
o atari_rootsec.h moved to fs/partitions/atari.h, but somehow the
version in include/linux didn't get deleted.
o The scheduler should complain not just about interrupts, but also
about being called whenever we're holding any
o Oops, lost ID in 2.4.x merge
o Missing <linux/version.h>, yet testing the kernel version
o Make sure MTRR setting is atomic on SMP, since
o Use CLONE_KERNEL for the common kernel thread flags
o Never _ever_ BUG() if you don't have to Cset exclude:
[email protected]|ChangeSet|20020905153320|19047
o Allocate system call numbers: 250 and 251 for hugetlb, with 252 for
exit_group
Matthew Dharm <[email protected]>:
o usb-storage: fix software eject
[email protected] <[email protected]>:
o Domsch zip code change
Mikael Pettersson <[email protected]>:
o 2.5.34 floppy driver init/exit fixes
o undo 2.5.34 ftape damage
Neil Brown <[email protected]>:
o md - 1 of 3 - Remove BUG in md.c that change in 2.5.33 triggers
o md - 2 of 3 - Fix bug in raid5 AGAIN
o md - 3 of 3 - Fix compile errors when tracing enabled in MD
o kNFSd 1: New structure initialisers for lockd
o kNFSd 2: Lockd to shutdown without engaging with nfsd
o kNFSd 3: Increase separation between lockd and nfsd
o kNFSd 4: Discard svc_uidmap structure
o kNFSd 5: Get rid of ex_parent from svc_export
o kNFSd 6: Expose anon uid and gid in /proc/fs/nfs/exports
o kNFSd 7: Discard cl_idlen
o kNFSd 8: Don't store path in exports table
o kNFSd 9: Discard cl_addr
o kNFSd 10: Discard ex_dev and ex_ino from svc_export
o kNFSd 11: Remove problematic "security" checks when NFS exporting
o kNFSd 12: Change exp_parent to talk directory tree, not hash table
o kNFSd 13: Separate out the multiple keys in the export hash table
o kNFSd 14: Filehandle lookup makes use of new export table structure
o kNFSd 15: Unite per-client export key hash tables
o kNFSd 16: Remove per-client list of exports
o md - Fix problems with freeing gendisk in md.c
o cset 1.497.59.25 breaks MD autodetect
Oliver Neukum <[email protected]>:
o fix for error handling in microtek
o new ids for hpusbscsi
o open/close fix for kaweth
o fixes for races in kaweth probe
[email protected] <[email protected]>:
o two byte offset for kaweth
Patrick Mochel <[email protected]>:
o Re: Performance issue in 2.5.32+
o Reorganize the mtrr init sequence a bit. All mtrr init now happens
during the initcall sequence, after all CPUs have been brought up.
Paul Mackerras <[email protected]>:
o PPC32: Use vunmap rather than vfree in iounmap
o PPC32: Update the PCI config-space access functions for PReP
o PPC32: rearrange includes in arch/ppc/kernel/irq.c to fix a compile
error
o PPC32: extra argument for pcibios_enable_resources/device
o PPC32: add argument to INIT_SIGNALS use in
arch/ppc/kernel/process.c
o PPC32: convert xtime usage from timeval to timespec
o PPC32: define atomic_add_negative
o PPC32: allocate syscall #s for alloc/free_hugepages and exit_group
and add exit_group to the syscall table.
o PPC32: remove the ppc32-specific ide_fix_driveid
o PPC32: define kmap_atomic_to_page
o PPC32: remove unused IDE functions from include/asm-ppc/ide.h
o PPC32: define rwlock_is_locked()
Pete Zaitcev <[email protected]>:
o arch/sparc/config.in: Add missing parts for modern fashion configs
o arch/sparc/defconfig: Supply working defconfig to show what is
working, what is not
o [SPARC]: Kill remaining remnants of kgdb support
o [SPARC64]: Cleanup serial_console declarations
o [SPARC]: Get 2.5.x building once more
o drivers/serial/sunzilog.c: Fix build of sparc32 probing code
o [SPARC] sparc 2.5.x again
Peter Samuelson <[email protected]>:
o remove duplicated AGP Config.in
Petr Vandrovec <[email protected]>:
o 2.5.34: recalc_sigpending missing for modules
o 2.5.34-bk fcntl lockup
Randy Dunlap <[email protected]>:
o 2.5.31 spell_typo fix
Russell King <[email protected]>:
o 2.5.32-usb
Rusty Russell <[email protected]>:
o Designated initializers for shm
o Designated initializers for cs46xx drivers
Stephen Rothwell <[email protected]>:
o cdrom.c is the only file to include asm/fcntl.h
Vojtech Pavlik <[email protected]>:
o This fixes problems in serport.c found by Russell King
FYI:
Since 2.5.34 the compilation of JFFS2 fails with a compile error similar
to the one in JFFS:
<-- snip -->
...
gcc -Wp,-MD,./.background.o.d -D__KERNEL__
-I/home/bunk/linux/kernel-2.5/linux-2.5.35/include -Wall
-Wstrict-prototypes -Wno-trigraphs -O2 -fomit-frame-pointer
-fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2
-march=k6 -nostdinc -iwithprefix include -DKBUILD_BASENAME=background
-c -o background.o background.c
background.c: In function `jffs2_garbage_collect_thread':
background.c:118: warning: passing arg 1 of `dequeue_signal' from
incompatible pointer type
background.c:118: warning: passing arg 2 of `dequeue_signal' from
incompatible pointer type
background.c:118: too few arguments to function `dequeue_signal'
make[2]: *** [background.o] Error 1
make[2]: Leaving directory `/home/bunk/linux/kernel-2.5/linux-2.5.35/fs/jffs2'
<-- 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
[email protected] said:
> Since 2.5.34 the compilation of JFFS2 fails with a compile error
> similar to the one in JFFS:
Yeah. Somebody changed the prototype of dequeue_signal() and even though
there are few users evidently didn't bother to fix them accordingly.
Current JFFS2 code depends on some fairly trivial rbtree updates which Linus
keeps ignoring without comment. Once those are in, I can update the JFFS2
code in 2.5 too, and will take a look if whoever broke it hasn't got around
to testing (or even compiling) their changes by then.
--
dwmw2
On Sun, 15 Sep 2002, Linus Torvalds wrote:
>...
> <[email protected]>:
> o USER_HZ & NTP problems
>...
This change results in a compile error in the ATM drivers:
<-- snip -->
...
gcc -Wp,-MD,./.eni.o.d -D__KERNEL__
-I/home/bunk/linux/kernel-2.5/linux-2.5.35/include -Wall
-Wstrict-prototypes -Wno-trigraphs -O2 -fomit-frame-pointer
-fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2
-march=k6 -nostdinc -iwithprefix include -g -DKBUILD_BASENAME=eni -c
-o eni.o eni.c
eni.c: In function `get_service':
eni.c:705: incompatible types in assignment
make[2]: *** [eni.o] Error 1
make[2]: Leaving directory `/home/bunk/linux/kernel-2.5/linux-2.5.35/drivers/atm'
<-- 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
Hi all,
I can not compile 2.5.35
#make install
...
...
...
make[1]: Leaving directory `/usr/src/linux-2.5.33/arch/i386/pci'
gcc -E -Wp,-MD,arch/i386/.vmlinux.lds.s.d_ -D__ASSEMBLY__ -D__KERNEL__ -I/usr/src/linux-2.5.33/include -nostdinc -iwithprefix include -P -C -Ui386 -o arch/i386/vmlinux.lds.s arch/i386/vmlinux.lds.S
Generating build number
make[1]: Entering directory `/usr/src/linux-2.5.33/init'
Generating /usr/src/linux-2.5.33/include/linux/compile.h (unchanged)
make[1]: Leaving directory `/usr/src/linux-2.5.33/init'
ld -m elf_i386 -T arch/i386/vmlinux.lds.s -e stext arch/i386/kernel/head.o arch/i386/kernel/init_task.o init/init.o --start-group arch/i386/kernel/kernel.o arch/i386/mm/mm.o kernel/kernel.o mm/mm.o fs/fs.o ipc/ipc.o security/built-in.o /usr/src/linux-2.5.33/arch/i386/lib/lib.a lib/lib.a /usr/src/linux-2.5.33/arch/i386/lib/lib.a drivers/built-in.o sound/sound.o arch/i386/pci/pci.o net/network.o --end-group -o vmlinux
fs/fs.o: In function `flush_old_exec':
fs/fs.o(.text+0xa227): undefined reference to `wait_task_inactive'
make: *** [vmlinux] Error 1
Any hints?
Paolo
--
Get your free email from http://www.linuxmail.org
Powered by Outblaze
On Monday 16 September 2002 13:52, Adrian Bunk wrote:
>
> This change results in a compile error in the ATM drivers:
>
Yep, There are even many other places where xtime is used which may result in
the same kind of compilation problems. Two possible solutions come to mind:
* Maintain both the ordinary xtime (timeval) and an xtime_nsec (timespec).
xtime may then be a timeval shadow value of xtime_nsec.
* Just fix it everywhere.
I cannot tell fore sure which option Linus prefers, but I think ...
Some places may even have other problems as well (e.g. ATM drivers): xtime is
a complex data type which needs read_lock / read_unlock, but there's no
locking in many places.
Adrian Bunk wrote:
> This change results in a compile error in the ATM drivers:
Yes, this was a bug in the ATM drivers... they really should have been using
do_gettimeofday() instead of touching xtime anyway.
Here's a patch that should fix up all these ATM compile errors. Linus,
please apply
-Mitch (deadbeat ATM maintainer)
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/ambassador.c linux-2.5.35/drivers/atm/ambassador.c
--- linux-2.5.35-VIRGIN/drivers/atm/ambassador.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/ambassador.c 2002-09-16 21:03:56.000000000 -0700
@@ -516,7 +516,7 @@
// VC layer stats
atomic_inc(&atm_vcc->stats->rx);
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
// end of our responsability
atm_vcc->push (atm_vcc, skb);
return;
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/atmtcp.c linux-2.5.35/drivers/atm/atmtcp.c
--- linux-2.5.35-VIRGIN/drivers/atm/atmtcp.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/atmtcp.c 2002-09-16 21:04:30.000000000 -0700
@@ -275,7 +275,7 @@
result = -ENOBUFS;
goto done;
}
- new_skb->stamp = xtime;
+ do_gettimeofday(&new_skb->stamp);
memcpy(skb_put(new_skb,skb->len),skb->data,skb->len);
out_vcc->push(out_vcc,new_skb);
atomic_inc(&vcc->stats->tx);
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/eni.c linux-2.5.35/drivers/atm/eni.c
--- linux-2.5.35-VIRGIN/drivers/atm/eni.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/eni.c 2002-09-16 21:04:45.000000000 -0700
@@ -702,7 +702,7 @@
DPRINTK("Grr, servicing VCC %ld twice\n",vci);
continue;
}
- ENI_VCC(vcc)->timestamp = xtime;
+ do_gettimeofday(&ENI_VCC(vcc)->timestamp);
ENI_VCC(vcc)->next = NULL;
if (vcc->qos.rxtp.traffic_class == ATM_CBR) {
if (eni_dev->fast)
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/firestream.c linux-2.5.35/drivers/atm/firestream.c
--- linux-2.5.35-VIRGIN/drivers/atm/firestream.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/firestream.c 2002-09-16 21:04:55.000000000 -0700
@@ -814,7 +814,7 @@
skb_put (skb, qe->p1 & 0xffff);
ATM_SKB(skb)->vcc = atm_vcc;
atomic_inc(&atm_vcc->stats->rx);
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
fs_dprintk (FS_DEBUG_ALLOC, "Free rec-skb: %p (pushed)\n", skb);
atm_vcc->push (atm_vcc, skb);
fs_dprintk (FS_DEBUG_ALLOC, "Free rec-d: %p\n", pe);
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/fore200e.c linux-2.5.35/drivers/atm/fore200e.c
--- linux-2.5.35-VIRGIN/drivers/atm/fore200e.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/fore200e.c 2002-09-16 21:05:24.000000000 -0700
@@ -1134,7 +1134,8 @@
return;
}
- skb->stamp = vcc->timestamp = xtime;
+ do_gettimeofday(&skb->stamp);
+ vcc->timestamp = skb->stamp;
#ifdef FORE200E_52BYTE_AAL0_SDU
if (cell_header) {
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/horizon.c linux-2.5.35/drivers/atm/horizon.c
--- linux-2.5.35-VIRGIN/drivers/atm/horizon.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/horizon.c 2002-09-16 21:05:36.000000000 -0700
@@ -1049,7 +1049,7 @@
struct atm_vcc * vcc = ATM_SKB(skb)->vcc;
// VC layer stats
atomic_inc(&vcc->stats->rx);
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
// end of our responsability
vcc->push (vcc, skb);
}
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/idt77252.c linux-2.5.35/drivers/atm/idt77252.c
--- linux-2.5.35-VIRGIN/drivers/atm/idt77252.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/idt77252.c 2002-09-16 21:06:40.000000000 -0700
@@ -1099,7 +1099,7 @@
cell, ATM_CELL_PAYLOAD);
ATM_SKB(sb)->vcc = vcc;
- sb->stamp = xtime;
+ do_gettimeofday(&sb->stamp);
vcc->push(vcc, sb);
atomic_inc(&vcc->stats->rx);
@@ -1177,7 +1177,7 @@
skb_trim(skb, len);
ATM_SKB(skb)->vcc = vcc;
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
vcc->push(vcc, skb);
atomic_inc(&vcc->stats->rx);
@@ -1199,7 +1199,7 @@
skb_trim(skb, len);
ATM_SKB(skb)->vcc = vcc;
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
vcc->push(vcc, skb);
atomic_inc(&vcc->stats->rx);
@@ -1337,7 +1337,7 @@
ATM_CELL_PAYLOAD);
ATM_SKB(sb)->vcc = vcc;
- sb->stamp = xtime;
+ do_gettimeofday(&sb->stamp);
vcc->push(vcc, sb);
atomic_inc(&vcc->stats->rx);
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/lanai.c linux-2.5.35/drivers/atm/lanai.c
--- linux-2.5.35-VIRGIN/drivers/atm/lanai.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/lanai.c 2002-09-16 21:06:54.000000000 -0700
@@ -1606,7 +1606,7 @@
}
skb_put(skb, size);
ATM_SKB(skb)->vcc = lvcc->rx.atmvcc;
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
vcc_rx_memcpy(skb->data, lvcc, size);
lvcc->rx.atmvcc->push(lvcc->rx.atmvcc, skb);
atomic_inc(&lvcc->rx.atmvcc->stats->rx);
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/nicstar.c linux-2.5.35/drivers/atm/nicstar.c
--- linux-2.5.35-VIRGIN/drivers/atm/nicstar.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/nicstar.c 2002-09-16 21:15:15.000000000 -0700
@@ -2264,7 +2264,7 @@
memcpy(sb->tail, cell, ATM_CELL_PAYLOAD);
skb_put(sb, ATM_CELL_PAYLOAD);
ATM_SKB(sb)->vcc = vcc;
- sb->stamp = xtime;
+ do_gettimeofday(&sb->stamp);
vcc->push(vcc, sb);
atomic_inc(&vcc->stats->rx);
cell += ATM_CELL_PAYLOAD;
@@ -2395,7 +2395,7 @@
skb->destructor = ns_sb_destructor;
#endif /* NS_USE_DESTRUCTORS */
ATM_SKB(skb)->vcc = vcc;
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
vcc->push(vcc, skb);
atomic_inc(&vcc->stats->rx);
}
@@ -2422,7 +2422,7 @@
sb->destructor = ns_sb_destructor;
#endif /* NS_USE_DESTRUCTORS */
ATM_SKB(sb)->vcc = vcc;
- sb->stamp = xtime;
+ do_gettimeofday(&sb->stamp);
vcc->push(vcc, sb);
atomic_inc(&vcc->stats->rx);
}
@@ -2448,7 +2448,7 @@
memcpy(skb->data, sb->data, NS_SMBUFSIZE);
skb_put(skb, len - NS_SMBUFSIZE);
ATM_SKB(skb)->vcc = vcc;
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
vcc->push(vcc, skb);
atomic_inc(&vcc->stats->rx);
}
@@ -2554,7 +2554,7 @@
#ifdef NS_USE_DESTRUCTORS
hb->destructor = ns_hb_destructor;
#endif /* NS_USE_DESTRUCTORS */
- hb->stamp = xtime;
+ do_gettimeofday(&hb->stamp);
vcc->push(vcc, hb);
atomic_inc(&vcc->stats->rx);
}
diff -u -r linux-2.5.35-VIRGIN/drivers/atm/zatm.c linux-2.5.35/drivers/atm/zatm.c
--- linux-2.5.35-VIRGIN/drivers/atm/zatm.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/zatm.c 2002-09-16 21:08:40.000000000 -0700
@@ -584,7 +584,7 @@
#ifdef CONFIG_ATM_ZATM_EXACT_TS
skb->stamp = exact_time(zatm_dev,here[1]);
#else
- skb->stamp = xtime;
+ do_gettimeofday(&skb->stamp);
#endif
#if 0
printk("[-3..0] 0x%08lx 0x%08lx 0x%08lx 0x%08lx\n",((unsigned *) skb->data)[-3],
On Mon, 16 Sep 2002, Mitchell Blank Jr wrote:
> Adrian Bunk wrote:
> > This change results in a compile error in the ATM drivers:
>
> Yes, this was a bug in the ATM drivers... they really should have been using
> do_gettimeofday() instead of touching xtime anyway.
>
> Here's a patch that should fix up all these ATM compile errors. Linus,
> please apply
Thanks for this patch, the next compile error is that compilation of
firestream.c fails at all occurences of func_enter:
<-- snip -->
...
gcc -Wp,-MD,./.firestream.o.d -D__KERNEL__
-I/home/bunk/linux/kernel-2.5/linux-2.5.35/include -Wall
-Wstrict-prototypes -Wno-trigraphs -O2 -fomit-frame-pointer
-fno-strict-aliasing -fno-common -pipe -mpreferred-stack-boundary=2
-march=k6 -nostdinc -iwithprefix include -g -DKBUILD_BASENAME=firestream
-c -o firestream.o firestream.c
firestream.c: In function `fs_open':
firestream.c:870: called object is not a function
firestream.c:870: parse error before string constant
...
make[2]: *** [firestream.o] Error 1
make[2]: Leaving directory `/home/bunk/linux/kernel-2.5/linux-2.5.35/drivers/atm'
<-- snip -->
> -Mitch (deadbeat ATM maintainer)
I didn't Cc you because you are only listed as PPP OVER ATM (RFC 2364)
maintainer in MAINTAINERS. Do you now maintain the complete ATM subsystem?
If yes, could you send a patch for MAINTAINERS?
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
Adrian Bunk wrote:
> Thanks for this patch, the next compile error is that compilation of
> firestream.c fails at all occurences of func_enter:
This is just the breakage that newer gcc's provoke when you try to use
__FUNCTION__ as a string constant. Easy fix (included at end of message,
Linus please apply)
> > -Mitch (deadbeat ATM maintainer)
>
> I didn't Cc you because you are only listed as PPP OVER ATM (RFC 2364)
> maintainer in MAINTAINERS. Do you now maintain the complete ATM subsystem?
I'm sort of the maintainer, if there is one at all. Werner handed me the
torch a year and a half ago since he has other projects and I was one of
the few people still hacking on the core ATM code. At the time I was doing
some related work so it fit naturally. Unfortunately, I'm not working
in that field at the moment and my other commitments are taking about 110%
of my time. I've been trying to keep it sort of maintained in the meantime
(keeping it compiling, basic mailing list admin) but sadly that's been about
it from me lately. :-(
-Mitch
--- linux-2.5.35-VIRGIN/drivers/atm/firestream.c 2002-08-24 00:08:21.000000000 -0700
+++ linux-2.5.35/drivers/atm/firestream.c 2002-09-17 14:34:57.000000000 -0700
@@ -330,8 +330,8 @@
#define FS_DEBUG_QSIZE 0x00001000
-#define func_enter() fs_dprintk (FS_DEBUG_FLOW, "fs: enter " __FUNCTION__ "\n")-#define func_exit() fs_dprintk (FS_DEBUG_FLOW, "fs: exit " __FUNCTION__ "\n")+#define func_enter() fs_dprintk (FS_DEBUG_FLOW, "fs: enter %s\n", __FUNCTION__)+#define func_exit() fs_dprintk (FS_DEBUG_FLOW, "fs: exit %s\n", __FUNCTION__)
struct fs_dev *fs_boards = NULL;