This is basically about making it compile. I've not tried to tackle the
second problem of testing it all yet.
Linux 2.5.41-ac1
- Merge with Linus 2.5.41
- Drop S/390 drivers subtree for Linus
- Drop task queue fixes for schedule_work
- TODO: merge two sets of conflicting UML changes
- TODO: double check bluetooth merge
o Fix aacraid makefile (Mark Haverkamp)
o Fix ips compile (Paul Larson)
o Fix aha152x compile (Michel Eyckmans)
o Fix orinoco_cs compile (Wichert Akkerman, Martin Waitz)
o Fix i2o_core compiler (Gregoire Favre)
o Fix missing exports for netfilter
o Fix compile failure in jffs (me)
o Fix compile failure in jffs2 (me)
o Fix Divas_Mod compile (me)
o Fix hisax compile (me)
o Fix ipacx compile (me)
o Fix pcbit compile (me)
o Fix tpam compile (me)
o Fix i2o_lan build (me)
o Fix i2o_proc build (me)
o Fix ppa compile (me)
o Fix imm compile (me)
o Fix ipv6 compile (me)
Linux 2.5.40-ac6
o Cadet_wake can be static (me)
o Bluetooth configuration cleanups (Marcel Holtmann)
o Hardwired empty bar handling fix take two (Ivan Kokshaysky)
o Use kernel crc32 lib for bluetooth (Marcel Holtmann)
o Make scsi cdrom honour passed timeouts (Peter Osterlund)
o Make aironet4500_cs compile (me)
o Fix bugs where ibmtr unmapped the wrong address (me)
o Fix crash problem in oss dmabuf.c (me)
| Its still very broken but ALSA should replace it
o Fix opl3sa2 warnings (me)
o Make tcic compile again (me)
o Make i82365 also use del_timer_sync (me)
o Fix warnings in fpu emulator (me)
o Fix t128 for NCR5380 changes (me)
o Fix pas16 for NCR5380 changes (me)
o Fix dmx3191 for NCR5380 changes (me)
o First pass seagate st02 cleanups (me)
o Clean up de600 driver. Switch to spinlocks (me)
remove crud, formatting junk etc
| Still needs rewriting to use parport
o Remove extra unlock in wd7000 (Matthew Wilcox)
o First basic pass at qlogicgas (me)
o Clean up the fdomain isa scsi (me)
o Clean up max_thread setting limits (Matthew Wilcox)
o Ricoh cardbus performance fix (KOMURO)
* Switch appletalk to seq_file /proc (Arnaldo Carvalho de Melo)
o Switch X.25 to seq_file (Arnaldo Carvalho de Melo)
o Fix bugs in the above (Arnaldo Carvalho de Melo)
Linux 2.5.40-ac5
o Rework S/390 driver init sequences (Martin Schwidefsky)
o Swap immediate_bh for tasklets for s/390 3215 (Martin Schwidefsky)
o UML updates - crash fixes, driver cleanup (Jeff Dike)
pcap transport
* Switch fmi radio card to sleeping waits (me)
* Fixing missing printk \n in fmi radio (me)
o Update to newer uclinux patch (Greg Ungerer)
| Unresolved now:
| fs/exec.c kernel/fork.c procfs sysctl
| can nommu be folded in (Hch)
o Remove surplus irq_disable from mpt fusion (Carlos Gorges)
* Export gdt for APM (Carlos Gorges)
| Marked as _GPL because its deep internals stuff
o Merge the add/put disk gendisk changes for i2o (Al Viro)
* Switch NCR5380/g_NCR5380 to new_eh (me)
* Fix cs89x0 netdevice init as module (me)
o Change some of the wd7000 code to use
udelay and do other cleanups
o Switch wd7000 to new_eh (me)
o Serial driver updates (Russell King)
o Sync bluetooth with 2.4, fix SMP, hotplug (Maksim Krasnyanskiy)
support L2CAP, BNEP, HCI filter etc
o Move firmwareloading to hotplug for bluetooth (Maksim Krasnyanskiy)
* Pull hpfs out of shared struct superblock (Brian Gerst)
o Fix sleep with pre-empt disabled in (Manfred Spraul)
set_cpus_allowed
Linux 2.5.40-ac4
* Make ibm partition code compile again (Martin Schwidefsky)
* Remove unneeded config options on S/390 (Martin Schwidefsky)
o Update DASD drivers (Martin Schwidefsky)
o Update S/390 xpram driver (Martin Schwidefsky)
o Replace S/390 BH code by tasklets (Martin Schwidefsky)
* Fix S/390 bitops bugs (Martin Schwidefsky)
o S/390x 31bit emulation fixes (Martin Schwidefsky)
* Update S/390 link scripts (Martin Schwidefsky)
* Add S/390 pre-empt support (Martin Schwidefsky)
* Inline some S/390 old compilers couldnt handle (Martin Schwidefsky)
* Use diag 44 for S/390x spinlocks (Martin Schwidefsky)
* Better S/390 timer handling (Martin Schwidefsky)
* S/390 code cleanups (Martin Schwidefsky)
* Clean up S/390 fpu load/stores (Martin Schwidefsky)
o DECnet updates for testing (Steve Whitehouse)
* Add console shutdown handling to S/390 (Martin Schwidefsky)
* Remove some bogus S/390 sanity checks (Martin Schwidefsky)
* Clean up S/390 process irq (Martin Schwidefsky)
o Fix/simplify chpids handling on S/390 (Martin Schwidefsky)
* No /proc/interrupts on S/390 (Martin Schwidefsky)
o Remove now unneeded S/390 hack in init/main.c (Martin Schwidefsky)
o Clean up all the S/390 ptrace handling (Martin Schwidefsky)
o Fix build with local apic enabled (James Bottomley)
o Initial i2o_block merge of 2.4/2.5 code (me)
| Not yet functional
o Initial i2o_scsi merge of 2.4/2.5 code (me)
| Needs dma mapping, 64bit, be and new_eh
o Revert Ivan's pci change (breaks serverworks)
* PCI serial oops fix (William Irwin)
* Remove dead wood from unistd.h (Brian Gerst)
o Fix bug in capget (Chris Wright)
o Switch qnxfs to new style initializers (Art Haas)
o Recongize qnx v6 file systems (Anders Larsen)
* Kill off remaining pcibios_ users (Greg "Ninja Turtle" Kroah-Hartmann)
o Fix scsi debug for scsi scan changes (Mike Anderson)
o Fix some bugs in scsi error handling (Mike Andersen)
o Forward port RMK's 2.4 scsi fixes (Mike Andersen)
o Allow longer settle times for scsi reset (Mike Andersen)
o Hopefully improve error policies a bit (Mike Andersen)
Linux 2.5.40-ac3
* Resync telephony drivers with 2.4 (me)
| Forward port security and other minor fixes
o Fix aironet4500 build for tq changes (me)
o Fix keyspan USB warnings with gcc 3 (me)
o Switch to the newer 2.4 depca driver (me)
o Re-merge depca fixes from 2.5.0->2.5.40]
o Fix depca spinning waiting for irq probe (me)
o Fix depca copy with interrupts off (me)
o Fix depca clash with other ALIGN macros (me)
* Initial port of NCR5380/g_NCR5380 to new locks (me)
| This still needs new_eh, further clean up
| and possibly making NCR5380_main a thread
* Initial locking rework for the wd7000 scsi (me)
| Still needs new_eh
* Update jffs to the dequeue_signal changes (me)
* Update jffs2 to the dequeue_signal changes (me)
* Fix shpnt misuse in NCR53c406a, wrong free_irq (me)
* Update NCR53c406a to new style sglist (me)
| Still needs new_eh
* Architecture updates for S/390 (Martin Schwidefsky)
o Include updates for S/390 (Martin Schwidefsky)
o Base S/390 driver updates (Martin Schwidefsky)
o Add the new syscalls to S/390 (Martin Schwidefsky)
o Fix sleeping with locks in sound_core (Jaroslav Kysela)
o Fix oops on shutdown of cs4281 (Suresh Siddha)
o Fix cdrom paths in devfs (Jordan Breeding)
o Fix missing cache tag entry in intel cpu table (Jean Delvare)
* Remove old 2.2 compatibility pci functions (Greg Kroah-Hartmann)
o Clean up some dead devfs bits (Greg Kroah-Hartmann)
* Fix an oops in the hugetblpage stuff (Andrew Morton)
| Its still a stupid idea but now it doesnt oops
o Handle read only BARs with type bits set (Ivan Kokshaysky)
Linux 2.5.40-ac2
* Fix a cut and paste error in the amd rng docs (Troels Hansen)
* Forward port OSS maestro3 fixes for toughbook
o Forward port ramdisk cache coherency
o RTL8150 USB updates (Petko Manalov)
o Fix corega USB ident (Petko Manalov)
o USB keyboard driver fix (Dave Miller)
o USB prototype fix (Luc Vanoostenryck)
o USB string fixes ([email protected])
o USB test driver (David Brownell)
o Speedtouch USB driver fixes (Greg Kroah-Hartmann)
* Clean environment for hotplug (Greg Kroah-Hartmann)
* Fix mprotect oops (Hugh Dickins)
o NUMA-Q cleanups (Martin Dobson)
o Split timers into one x86 timer type per file (John Stultz)
o Cyclone timer support for x440 etc (John Stultz)
* Fix sleeping from illegal context for ioperm (Andrew Morton)
o Fix imm compile ([email protected])
o Fix irda for tq changes (Carlos Gorges)
o Fix xjack telephony build (Carlos Gorges)
o Fix ppa compile (Carlos Gorges)
o Fix aha152x compile for tq changes (Carlos Gorges)
o Fix hamradio drivers for tq changes (Carlos Gorges)
o Fix plip driver for tq changes (Carlos Gorges)
o Fix mpt fusion for tq changes (Carlos Gorges)
o Fix isdn for tq changes (Carlos Gorges)
o Fix ieee1394 for tq changes (Carlos Gorges)
o Fix new timer code to build with cpufreq on (me)
o Fix capi build for new tq_ code (me)
| ISDN still needs moving to real locks
| this just cleans up one item
o Fix missing header in mtdblock_ro (Carlos Gorges)
o Fix a typo and other header (me)
o Fix up ixj_pcmcia for 2.5 (me)
| Note for janitors - it looks like a lot of the pcmcia release
| code people "fixed" should be using del_timer_sync not del_timer
* Fix missing header in longhaul cpu speed driver (me)
* Pipe read/write cleanup (Manfred Spraul)
* Make IDE PCI config text clearer (Andrzej Krzysztofowicz)
Linux 2.5.40-ac1
+ Initial port of aacraid driver to 2.5 (me)
* vfat corruption fix (Petr Vandrovec)
+ Clean up firestream warnings (Francois Romieu)
+ Voyager support (James Bottomley)
* Fix split_vma (Hugh Dickins)
+ Fix config in video subdirectory (John Levon)
+ Update olympic driver to 2.5 (Mike Phillips)
* Fix sg init error (Mike Anderson)
+ Fix Rules.make
o Merge most of ucLinux stuff (Greg Ungerer)
| It needs putting somewhere so we can pick over the
| hard bits left
| Q: Wouldn't drivers/char/mem-nommu.c be better
| Q: How to do the procfs stuff tidily
| Q: Wouldn't it be nicer to move all mm or mmnommu specific ksyms
| int the relevant mm/*.c file area instead of kernel/ksyms
| Q: Why ifdef out overcommit - its even easier to account on
| MMUless and useful info
* Stick tulip back under 10/100 ethernet (me)
* Correct docs for IBM touchpad back to how (me)
they were before
o Fix abuse of set_bit in winbond-840 (me)
* Fix abuse of set_bit in atp (me)
On Mon, 7 Oct 2002, Alan Cox wrote:
>....
> Linux 2.5.40-ac1
>...
> + Voyager support (James Bottomley)
>...
It seems at least one file is missing, the following compile error occurs
when trying to compile 2.5.41-ac1 with both CONFIG_VOYAGER and CONFIG_SMP
enabled:
<-- snip -->
...
make[1]: *** No rule to make target `arch/i386/mach-voyager/trampoline.o',
needed by `arch/i386/mach-voyager/built-in.o'. Stop.
make: *** [arch/i386/mach-voyager] 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
[email protected] said:
> make[1]: *** No rule to make target `arch/i386/mach-voyager/
> trampoline.o', needed by `arch/i386/mach-voyager/built-in.o'. Stop.
> make: *** [arch/i386/mach-voyager] Error 2
That one's pulled in from ../kernel by the vpath in mach-voyager (or should
be). It builds for me, so it could be the version of make you are using?
James
On Tue, 8 Oct 2002, James Bottomley wrote:
> [email protected] said:
> > make[1]: *** No rule to make target `arch/i386/mach-voyager/
> > trampoline.o', needed by `arch/i386/mach-voyager/built-in.o'. Stop.
> > make: *** [arch/i386/mach-voyager] Error 2
>
> That one's pulled in from ../kernel by the vpath in mach-voyager (or should
> be). It builds for me, so it could be the version of make you are using?
Ah, then Kai's changes to the build system in 2.5.41 broke it.
Kai, what's the recommended way to get this working again?
> James
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
On Tue, 8 Oct 2002, Adrian Bunk wrote:
> On Tue, 8 Oct 2002, James Bottomley wrote:
>
> > [email protected] said:
> > > make[1]: *** No rule to make target `arch/i386/mach-voyager/
> > > trampoline.o', needed by `arch/i386/mach-voyager/built-in.o'. Stop.
> > > make: *** [arch/i386/mach-voyager] Error 2
> >
> > That one's pulled in from ../kernel by the vpath in mach-voyager (or should
> > be). It builds for me, so it could be the version of make you are using?
>
> Ah, then Kai's changes to the build system in 2.5.41 broke it.
>
> Kai, what's the recommended way to get this working again?
Well, don't use vpath in the kernel Makefiles ;)
To restore the previous state, just do
obj-y += ... ../kernel/trampoline.o
However, why don't you just set CONFIG_X86_SMP when CONFIG_VOYAGER is
selected, which would be cleaner anyway?
Compiling the same file from different subdirs depending on .config is
surely a recipe for confusion.
--Kai
On Tue, 8 Oct 2002, Kai Germaschewski wrote:
>...
> To restore the previous state, just do
>
> obj-y += ... ../kernel/trampoline.o
>...
There seems to be a bug in kbuild that makes your suggestion impossible:
<-- snip -->
...
make -f arch/i386/mach-voyager/Makefile
ld -m elf_i386 -r -o arch/i386/mach-voyager/built-in.o
arch/i386/mach-voyager/setup.o arch/i386/mach-voyager/voyager_basic.o
arch/i386/mach-voyager/voyager_thread.o
arch/i386/mach-voyager/voyager_smp.o arch/i386/mach-voyager/voyager_cat.o
arch/i386/mach-voyager/../kernel/trampoline.o
ld: cannot open arch/i386/mach-voyager/../kernel/trampoline.o: No such
file or directory
make[1]: *** [arch/i386/mach-voyager/built-in.o] Error 1
make: *** [arch/i386/mach-voyager] Error 2
<-- snip -->
> --Kai
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
Perhaps a better option is to add a single config option that compiles the
trampoline (CONFIG_X86_TRAMPOLINE) then I can build the object for both
voyager and smp x86.
James
On Tue, 8 Oct 2002, Adrian Bunk wrote:
> On Tue, 8 Oct 2002, Kai Germaschewski wrote:
>
> >...
> > To restore the previous state, just do
> >
> > obj-y += ... ../kernel/trampoline.o
> >...
>
> There seems to be a bug in kbuild that makes your suggestion impossible:
Oh well, actually, I only made it possible to have multipart objects pull
in objects from different dirs (but I still discourage that except for
special cases), not for obj-y.
So you have to do sth like
obj-y += .. trampoline.o
trampoline-objs := ../kernel/trampoline.o
And you do not really want to do this, you rather want CONFIG_VOYAGER to
define_bool CONFIG_X86_SMP y and be done without ugly hacks in the
Makefiles.
--Kai
[email protected] said:
> And you do not really want to do this, you rather want CONFIG_VOYAGER
> to define_bool CONFIG_X86_SMP y and be done without ugly hacks in the
> Makefiles.
I'm afraid voyager can't do that. CONFIG_SMP was split from CONFIG_X86_SMP
for precisely this reason. CONFIG_X86_SMP is for generic intel SMP
architectures (i.e. APIC based). The voyager has it's own rather wierd VIC
based SMP architecture.
However, how about the attached patch, which fixes the problem for me (I can
only compile, not test, because my voyager system is at home and I'm away on
business)
James
On Tue, 8 Oct 2002, James Bottomley wrote:
> [email protected] said:
> > And you do not really want to do this, you rather want CONFIG_VOYAGER
> > to define_bool CONFIG_X86_SMP y and be done without ugly hacks in the
> > Makefiles.
>
> I'm afraid voyager can't do that. CONFIG_SMP was split from CONFIG_X86_SMP
> for precisely this reason. CONFIG_X86_SMP is for generic intel SMP
> architectures (i.e. APIC based). The voyager has it's own rather wierd VIC
> based SMP architecture.
>
> However, how about the attached patch, which fixes the problem for me (I can
> only compile, not test, because my voyager system is at home and I'm away on
> business)
The patch does exactly what I would have suggested next, so I am very
happy with it ;)
--Kai