Andrey Panin <pazke@orbita1.ru>: drivers_net_eepro100: missing __devinit:
  This patch adds missing __devinit modifiers for speedo_found1() and
  do_eeprom_cmd() functions. Patch against 2.5.8. Compiles, but untested.

Andrew Morton <akpm@zip.com.au>: fix a warning in i386_pci-pc.c:

Andrey Panin <pazke@orbita1.ru>: drivers_net_dl2k.c: missing __devinit's:
  This patch adds missing __devinit modifiers for read_eeprom() and
  parse_eeprom() functions. Patch against 2.5.8. Compiles, but untested.

 - sane retention policy for /proc/<pid>/fd/* - ->d_revalidate() says
   "kill it" if descriptor is not opened anymore (in addition to checks
   for task being dead) and we allow dentries of /proc/<pid>/fd/<n> to
   stay around.

 - preparation for sane policy for /proc/<pid>/fd/* - don't store
   struct file * in these inodes anymore.

 - sane dentry retention.  Namely, we don't kill /proc/<pid> dentries at the
   first opportunity (as the current tree does).  Instead we do the following:
	* ->d_delete() kills it only if process is already dead.
	* all ->lookup() in proc/base.c end with checking if process is still
	  alive and unhash if it isn't.
	* proc_pid_lookup() (lookup for /proc/<pid>) caches reference to dentry
	  in task_struct.  It's _not_ counted in ->d_count.
	* ->d_iput() resets said reference to NULL.
	* release_task() (burying a zombie) checks if there is a cached
	  reference and if there is - shrinks the subtree.
	* tasklist_lock is used for exclusion.
   That way we are guaranteed that after release_task() all dentries in
   /proc/<pid> will go away as soon as possible; OTOH, before release_task()
   we have normal retention policy - they go away under memory pressure with
   the same rules as for dentries on any other fs.

 - new helper in fs/proc/base.c - name_to_int(dentry) returns ~0U if name
   doesn't match 0|[1-9][0-9]* or is too large.  Otherwise it returns
   numeric value of name.  proc_pid_lookup() and proc_lookupfd() converted.

 - take unhash_process() into sched.c, move zeroing ->pid into it (and
   under tasklist_lock)

Fix buffer-head leak on truncate race.

The reason why it hadn't been triggered left right and center is that we
step on that path only when partial truncate blows away an indirect
block shared between truncated and remaining areas _and_ races with
ext2_get_block() in the beginning of said area.  IOW, we need

	* indirect block with hole in the beginning
	* get_block starting to read that block
	* truncate killing everything past the hole (and thus freeing indirect
	  block in question)
	* get_block completing the read and noticing that block is actually
	  gone.

OK...  IMO we should do that brelse() outside of the loop.

 - fixed off-by-PAGE_OFFSET in populate_pmd() (alpha, again)

 - alpha/mm/init.c forgot to set max_pfn

 - (dumb) default_idle() provided (alpha)

 - s/p_pptr/parent/, s/p_opptr/real_parent/ done for alpha

 - missing defines/fields/includes for alpha (accumulated since 2.5.0)

 - bogus return value type for ->open() and ->release() on nfsd
   (generic, caught on alpha since there ssize_t != int)

into home.transmeta.com:/home/torvalds/v2.5/linux

into flint.arm.linux.org.uk:/usr/src/linux-bk-2.5/linux-2.5-rmk

Make cpu type message consistent with cache type message.

into home.transmeta.com:/home/torvalds/v2.5/linux

Make the PCMCIA client drivers depend on hisax.o, 
use the right CONFIG_ variable for hisax submodules,
handle the case of built-in hisax, modular hisax submodules.

Remove the repeated information on what a module is.

into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.isdn

into tp1.ruhr-uni-bochum.de:/home/kai/kernel/v2.5/linux-2.5.make

This patch brings 2.5.8 in sync with the x86-64 2.4 development tree again
(excluding device drivers)

It has lots of bug fixes and enhancements. It only touches architecture
specific files.

- Sync with 2.5.8
- SMP/APIC supported now.
- Module loading works now.
- Time keeping bugs fixed.
- entry.S streamlined and some bugs fixed.
- modify_ldt works now
- mostly rewritten FPU support (including FXRSTOR for initial FPU
  initialization based on the initial state)
- 32bit emulation enhanced and bugs fixed.
- rewrote mm initialization and lots of cleanups in the page table handling
  __PAGE_OFFSET is now moved to 0x10000000000 and some vmalloc/ioremap
  problems have been fixed. They have an own PML4 slot now.
- WCHAN reporting support for RIP (but not RSP)
- Lots of various other bug fixes and cleanups.

Currently broken:
- ACPI
- MTRR

It needs some other bugfixes outside architecture specific code. I sent
them all in separate mail.

When vt_init happens to run before the keyboard is initialized (the order is random
because they're both different initcalls) then set_leds raises an not yet enabled
tasklet. This causes an endless loop on the first schedule() call because the tasklet
handling cannot handle raised but disabled tasklets.

This patch just does not do set_leds in virtual terminal initialization to avoid that.
It is done later anyways.

Minor patch for the x86-64 32bit emulation. Originally proposed by hpa and it
makes autofs 32bit/64bit clean for x86-64

(a minor fix to the previous version - 1096/2)
The enclosed patch includes trizeps-specific PCMCIA code. All
modifications are enclosed in
#ifdef CONFIG_SA1100_TRIZEPS
#endif
blocks or are linked as
+sa1100_cs-objs-        += sa1100_trizeps.o
so, other platforms remain unaffected. The patch is built against
2.5.6-rmk1.

Initialize the saved FPU/XMM state in the task struct and fall through
to restore_fpu() to make sure that all state is fully initialized.

This means that old SSE/SSE2 information cannot ever leak into newly
created processes. 

 - (tcq, general) Remove the 'attempt to keep queue full option'. It worked
   on some IBM models, but failed miserably on others. Also removes some
   uglies in ide_queue_commands()

 - (tcq0 Change default depth back to 32.

 - (general) Add isr for no-dataphase taskfile, like task_no_data_intr but
   doesn't complain about failure. This is handy for commands what we _know_
   will fail, such as WIN_NOP.

 - (general) ide_cmd_type_parser() must set a handler to WIN_NOP... Otherwise
   we will just hang the ide system issuing a nop.

 - (general) HWGROUP(drive)->rq->special -> IDE_CUR_AR(drive)

 - (general) Have ide_raw_taskfile() copy back the taskfile after execution,
   otherwise we cannot use the info that ide_end_drive_cmd() puts in
   there.

 - (tcq) Use nIEN bit correctly in ide-tcq

 - (tcq) Small ide_tcq_wait_altstat() changes. Do initial 400ns delay (1us
   here), then 10us each successive run.

 - (tcq) Add beginning for 'nop auto poll' support check.

 - (tcq) Arm handler before GET_STAT() service check in
   ide_dma_queued_start, WD seemed to trigger interrupt before that.
   Makes WD Expert drives work with tcq.

Use the (well, one) standard way to build a multi-part module. 
Standard except for that we build aix7xxx_mod.o but then rename it
at modules_install time (just as it was before).

(The O_TARGET is used only internally when linking, the name of the 
module is unchanged).

I'm currently working on a node affine scheduler extension for NUMA
machines and the load balancer behaves a bit different from the original.
So after a few boot failures with those slowly booting 16 CPU IA64
machines I thought there must be a simpler solution than synchronizing and
waiting for the load balancer: just let migration_CPU0 do what it is
designed for. So my proposal is:
   - start all migration threads on CPU#0
   - initialize migration_CPU0 (trivial, reliable, as it already is on
     the right CPU)
   - let all other migration threads use set_cpus_allowed() to get to the
     right place

The only synchronization needed is the non-zero migration threads waiting
for migration_CPU0 to start working, which it will, as it is already on
the right CPU. This saves quite some lines of code.

I first posted this to LKML on March 6th (BTW, the fix #1, too) and since
then it was tested on several big NUMA platforms: 16 CPU NEC AzusA (IA64)
(also known as HP rx....), up to 32 CPU SGI IA64, 16 CPU IBM NUMA-Q
(IA32). No more lock-ups at boot since then. So I consider it working.

There is another good reason for this approach: the integration of the CPU
hotplug patch with the new scheduler becomes easier. One just needs to
create the new migration thread, it will move itself to the right CPU
without any additional magic (which you otherwise need because of the
synchronizations which won't be there at hotplug). Kimi Suganuma in the
neighboring cube is fiddling this out currently.

Attached is a patch that disables interrupts while holding the rq_lock.
This is certainly needed to prevent a race against the timer tick, as
Erich Focht pointed out.