If a fault in the kernel leads to an unexpected protection fault whilst in
a code path which holds mmap_sem we will deadlock in do_page_fault() while
trying to classify the fault.  By carefully testing the source of the fault
we can detect and OOPS on the vast majority of these, greatly enhancing
diagnosis of such bugs.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

We're now putting 24-byte structures into .init.setup via __setup.  But
x86_64's compiler is emitting a `.align 16' in there, so they end up on
32-byte boundaries and do_early_param()'s pointer arithmetic goes wrong.

Fix that up by forcing the compiler to align these structures to sizeof(long).

Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

Trivial patch to quiet messages about corrected errors.
Signed-off-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

This is a revised version of Kishore's patch to set message age appropriately
in STP configuration packets.
Signed-off-by: Kishore A K <KishoreAK@myw.ltindia.com>
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@redhat.com>

I'm hitting this problem because the module I want to load was
compiled with -g, so the filesize is absolutely huge.  kernel/module.c
does a vmalloc() to inhale the entire file, and then two calls to
module_alloc() (which calls vmalloc() on ia64) to load the 'init' and
'core' sections. The 'init' is small and slips into a gap early in the
vmalloc playspace, while the 'core' area is allocated after the huge
area that was allocated for the inhaled copy of the whole file.

I made a one coding change to Jean-Marc's version, adding a check to
see whether the init/core sections are close enough together for the
PCREL21B to reach (they almost always are).

I've kept the test that there are no jumps from core to init (though
the message that is printed is almost as unhelpful as the one that you
get when you statically link a module into the kernel that has calls
to the discarded .exit section :-)

This patch stalled out before on the question of whether a PLT was
overkill for a section-to-section branch, and whether it would be
better to use relaxation techniques to extend the range.  Doing that
might be more elegant in some way, but it needs a whole heap of extra
code (to count how many relaxed branches need to be added and
allocated them as needed).  This seems a lot of extra code that is
only ever going to be exercised by maniacs like me with 49MB .ko
files.

I fixed up the comments to be a little more heplful.

Signed-off-by: <tony.luck@intel.com>
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

into kroah.com:/home/greg/linux/BK/usb-2.6

The Altix console driver has been replaced in Andrew's tree by a new
one that uses the serial core API.  Update the early_console_setup
routine to use the new config option so that it's ready when the
driver hits Linus' tree.  This will also fix a build bug since the old
driver hasn't been updated.
Signed-off-by: Jesse Barnes <jbarnes@sgi.com>
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

The packet scheduler simulates 64 bit jiffies on 32 bit platforms by running
a timer keeping a mark and and offset.  Since there is no locking and this is
racy and doesn't handle jiffie wrap real well.

We can use get_jiffies_64 on 2.6 do get what is needed.
The downside is the overhead of a function call, and a cache miss in
get_jiffies_64.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@redhat.com>

The last argument in the PSCHED_TDIFF_SAFE is no longer used;
only usage eliminated by previous patch. It gets rid of a bad macro
usage.

Also, can use the standard min_t macro which also eliminates the
macro problem of double evaluation of bound.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@redhat.com>

This code in the csz scheduler, is just plain broken.  The TDIFF_SAFE 
effectively expands to:
	unsigned long delay = now - q->t_c;
	if (delay > 0) {
		delay = 0;
		goto do_reset;
	}
	if (delay >> q->delta_log)

So delay is always 0!  I assume that what was originally intended
is the to keep delay bounded to 1<<q->delta_log. 
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@redhat.com>

The patch to move the current pointer for init_task from region 7 to
region 5 inadvertendly caused ar.k4 (CURRENT_STACK) to be initialized
improperly for the bootstrap processor.  Fix this by initializing it
to -1.
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

I don't think we're in K&R any more, Toto.

If you want a NULL pointer, use NULL. Don't use an integer.

Most of the users really didn't seem to know the proper type.

This fixes the issue where the Generic driver would bind to all usb-serial
devices, so the disconnect would not properly go to the real driver that
controlled the device.  This was very bad when unloading the module with
the device still connected.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

The packet scheduling code has some ugly define's which were to deal with
configuration possibilities and the old style module exports.  With the current
2.6 method, this is unnecessary.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@redhat.com>

When CONFIG_NET_CLS_ACT is not set 'result' variable in prio_classify is
unused. Also I was looking over the rest of the module and had hard time
understanding the logic in prio_enqueue - I rearranged it a bit for better
readability. Plus there are some formatting changes.
Signed-off-by: Dmitry Torokhov <dtor_core@ameritech.net>
Signed-off-by: David S. Miller <davem@redhat.com>

The code allocated all 96 stacked registers, for no apparent reason.

Also, fix ia64_pal_call_static() to switch RSE to enforced-lazy, LE
mode before the PAL-call.
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

Another patchlet after turning on preempt and rerunning tests.
- against dev.c is a genuine bug - i wish preempt would have
given a more sane hint. Strange on how its a non issue when preempt is
off. 
- against act_api.c is just an annoyance. I removed the kmod because it
will only make sense to list actions if policies exist for that action.
And if policies exist, the module would be loaded.
Signed-off-by: Jamal Hadi Salim <hadi@zynx.com>
Signed-off-by: David S. Miller <davem@redhat.com>

Change ia64_switch_mode_phys() and ia64_switch_mode_virt() to preserve
the virtual sp/bsp and update the call-sites accordingly.  This avoids
problems with the init_task pointer which lives in region 5 now.
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

Gcc 3.4 has a new DFA scheduler for ia64. Unfortunately, it is not
well tested for Itanium 1. When I tried to use gcc 3.4 to compile
2.6 kernel configured for Merced, gcc failed in many places:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15598
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15653
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15655
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16130
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16142
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16143

Some of them have been fixed. But I am afraid many more remain. I don't
know how many Itanium 1 machines left in production use. If people want
to optimize kernel for Itanium 1, they can stick with the older gccs.
Or they should upgrade their machines. I think the gcc developer's time
should be better spent on somewhere else. I have been using this patch
for a while, kernel seems to work OK.
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

Signed-off-by: David Mosberger <davidm@hpl.hp.com>

Define cpu_logical_id() even when !SMP.  I added uses of this in some
iosapic printk's, which broke the UP build.
Signed-off-by: David Mosberger <davidm@hpl.hp.com>

into nuts.davemloft.net:/disk1/BK/net-2.6

On Sat, 26 Jun 2004, Edward C. Bailey wrote:

> Hello,
>
>     I was downloading some pictures from my Digital Wallet to my Fedora
> Core 2 system at the same time I happened to be watching
> /var/log/messages.  Here's what I saw:
>
> Jun 26 12:16:02 raptor kernel: usb 1-1.2: new full speed USB device using address 6
> Jun 26 12:16:02 raptor kernel: usb-storage: This device (097a,0001,0001 S 06 P 01) has unneeded SubClass and Protocol entries in unusual_devs.h
> Jun 26 12:16:02 raptor kernel:    Please send a copy of this message to <linux-usb-devel@lists.sourceforge.net>
>
> Given that Minds@Work are out of business, you might not care about this
> information, but I thought I'd pass it on, just in case you do... :-)

What the heck, we'll use it anyway.  Thanks for sending this in.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

I strongly recommend that this patch be applied.

Without this patch, running tcpdump on an SMP machine with the
pegasus adapter locks up 9 times out of 10.  This is what happens:

CPU0				CPU1
pegasus_start_xmit
	netif_stop_queue
				pegasus_set_multicast
					netif_stop_queue
					netif_wake_queue

Which crashes if another packet is sent to pegasus_start_xmit before
the first one is finished.

If there are other drivers doing this, please fix them too.

PS I submitted this to Petkan years ago but he probably lost it.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

For some reason, gcc-2.95.4 dies horridly on those asmlinkage
declarations.

I was unable to work out _why_ those functions have asmlinkage, as there
seem to be no instances of them - nobody calls
pwc_register_decompressor().
 What's up with that?
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

took me a while to get my external usb drive running under linux-2.6.6,
appears to need the appended fix. according to the header in unusual_devs.h
one should send them to you, so here you go.

the funny thing is that it works fine unter linux-2.6.0-test11 on different
hardware, but then i'm out of my depth concerning what might have been
changed in the usb driver in the meantime. i don't really follow the
changes. i didn't find any real documentation about what these flags do,
only a couple of "my usb drive worked with 2.4 but doesn't work any more
with 2.6" messages on the web, and some of them pointed to unusual_devs.h. i
do not claim to know what i've done with this fix, but i'd like to see it
officially included. :-)


T:  Bus=04 Lev=01 Prnt=01 Port=02 Cnt=01 Dev#=  2 Spd=480 MxCh= 0
D:  Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs=  1
P:  Vendor=0402 ProdID=5621 Rev= 1.03
S:  Product=USB 2.0 Storage Device
S:  SerialNumber=00042222200000113608
C:* #Ifs= 1 Cfg#= 1 Atr=c0 MxPwr=  0mA
I:  If#= 0 Alt= 0 #EPs= 2 Cls=08(stor.) Sub=06 Prot=50 Driver=usb-storage
E:  Ad=81(I) Atr=02(Bulk) MxPS= 512 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS= 512 Ivl=0ms
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

I noticed that fs/buffer.c::drop_buffers() contains some code that
AFAICS doesn't actually do anything other than waste cpu cycles so here
is patch to remove it...  The local variable was_uptodate is being
messed with but it is not being read anywhere so it seems entirely
pointless.

I assume this must be a remainder from old code which mucked around with
the page uptodateness but which has since been (re-)moved.
Signed-off-by: Anton Altaparmakov <aia21@cantab.net>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

efi.h declares a function type, and then uses that as an argument to
another function, and expects the compiler to magically demote the
function to a function pointer.

Even a gcc person (rth) was surprised that this was legal, and it
doesn't match any other use of a function pointer in the kernel, and
sparse doesn't like the implicit type-conversion.

So make the type sane in the first place, instead of depending on
a very weird corner case of the C language.

This fixes a bug in the ppc64 memset where the code that gets the
destination address aligned (or is supposed to) was looking at the
bottom 3 bits of the count rather than the destination address.  The
result of this was that the kernel wouldn't boot on POWER3 machines. 
The patch also removes an unnecessary duplicate instruction.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

into nuts.davemloft.net:/disk1/BK/net-2.6

Thanks to Gordon Elam <gbe@shoremicro.com> for the information needed
to do this.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Verify at least a full header is there before derefencing
length struct member.

into ppc970.osdl.org:/home/torvalds/v2.6/linux