From: Jan Hubicka <jh@suse.cz>

GCC now converts sprintf (a,"%s",b) to strcpy.  This lose on kernel as
strcpy is not inlined and not present in library, so one gets linker
failure.  It seems to make sense to apply this optimization by hand.

From: Jan Hubicka <jh@suse.cz>

GCC never inline extern inline function redefined by new body (because it is
not clear what body one should choose)

parport contains such duplicated functions for apparently no good reasons.
Both copies differ slightly, not sure whether it is intentional or just
garbage.

include/asm/unistd.h:397: warning: conflicting types for built-in function '_exit'


Just remove the dang thing - nobody uses _exit().

Except for the vmware stub, which breaks.  So I kindly exported do_exit() to
kernel modules, just for vmware...

drivers/char/keyboard.c:205: warning: use of conditional expressions as lvalues is deprecated

include/linux/elevator.h:106: sorry, unimplemented: inlining failed in call to 'elv_try_last_merge': function body not available

include/linux/ide.h:1424: sorry, unimplemented: inlining failed in call to 'SELECT_MASK': function body not available

include/linux/reiserfs_fs.h:1837: sorry, unimplemented: inlining failed in call to 'decrement_bcount': function body not available

sound/core/pcm_misc.c:557: warning: use of cast expressions as lvalues is deprecated
sound/core/pcm_misc.c:568: warning: use of cast expressions as lvalues is deprecated
sound/core/pcm_misc.c:569: warning: use of cast expressions as lvalues is deprecated
sound/core/pcm_misc.c:570: warning: use of cast expressions as lvalues is deprecated
sound/core/pcm_misc.c:586: warning: use of cast expressions as lvalues is deprecated
sound/core/pcm_misc.c:596: warning: use of cast expressions as lvalues is deprecated

fs/binfmt_elf.c:171: warning: use of cast expressions as lvalues is deprecated

arch/i386/kernel/msr.c:1:10: warning: extra tokens at end of #ident directive
arch/i386/kernel/cpuid.c:1:10: warning: extra tokens at end of #ident directive

include/linux/bio.h:234: sorry, unimplemented: inlining failed in call to 'bio_phys_segments': function body not available

From: Andi Kleen <ak@muc.de>

gcc 3.4 optimizes sprintf(foo,"%s",string) into strcpy.  Unfortunately that
isn't seen by the inliner and linux/i386 has no out-of-line strcpy so you
end up with a linker error.

This patch adds out of line copies for most string functions to avoid this.
Actually it doesn't export them to modules yet, that would be the next
step.

BTW In my opinion we shouldn't use inline string functions at all.  The
__builtin_str* in modern gcc are better (I used them very successfully on
x86-64) and for the bigger functions like strrchr,strtok et.al.  it just
doesn't make any sense to inline them or even code them in assembler.

Also fix the bcopy prototype gcc was complaining about.

From: Andi Kleen <ak@muc.de>

Just many more warning fixes for a gcc 3.4 snapshot.

It warns for a lot of things now, e.g.  for ?: and ({ ...  }) and casts as
lvalues.  And for functions marked inline in headers, but no body.

Actually there are more warnings, i stopped fixing at some point.  Some of
the warnings seem to be dubious (e.g.  the binfmt_elf.c one, which looks
more like a compiler bug to me)

I also fixed the _exit() prototype to be void because gcc was complaining
about this.

From: Thomas Winischhofer <thomas@winischhofer.net>

sisfb is simply broken in current 2.6.x.  This patch updates sisfb to the
current development version which no less than 11 months ahead of the version
in the kernel.

Updated includes

- many fixes (duh)

- support for new chipsets (661, 741, 760)

- support for new video bridges (301C, 302ELV)

- removal of all offending fp code (as discussed earlier this month)

- a lot of code clean-up (which is the main reason for its size)

From: Sam Ravnborg <sam@ravnborg.org>

When utilising the make O=...  option the include options for gcc were
mangled even when absolute paths was used.  Also remove duplication of
CPPFLAGS.  They were assigned twice.  [It is still possible for
architectures to modify CPPFLAGS].

This patch allows xconfig to be build with make O=...  xconfig.It will also
help development of external modules with absolute paths for their -I
options.

Note: As a side effect a full recompile of the kernel takes place due to
changes in number of gcc options.

From: "Nakajima, Jun" <jun.nakajima@intel.com>

To get the vector-based interrupt handling work, we need to give the vector
number to device drivers instead of the IRQ if IRQ < 16.  It was not
happening for SCI, and the patch fixes it.  In many cases, the IRQ for SCI is
9, and the problem was not detected, but one particular machine exposed the
bug.

From: Rusty Russell <rusty@rustcorp.com.au>

The cpucontrol mutex is not required when no cpus can go up and down.
Andrew wrote a wrapper for it to avoid #ifdefs, this expands that to only
be defined for CONFIG_HOTPLUG_CPU, and uses it everywhere.

The only downside is that the cpucontrol lock was overloaded by my recent
patch to net/core/flow.c to protect it from reentrance, so this
reintroduces the local flow_flush_sem.  This code isn't speed critical, so
taking two locks when CONFIG_HOTPLUG_CPU=y is not really an issue.

From: Benjamin Herrenschmidt <benh@kernel.crashing.org>

- Make sure that all console operations are approriately protected under
  console_sem.

- Adds checks to make sure that people are taking console_sem when it is
  expected to be held.

From: Jes Sorensen <jes@trained-monkey.org>

I'd like to propose the following for 2.6.1-mm/2.6.2. On systems with a
large number of CPUs the number of printk's flowing by for each CPU
booting starts becoming a real console hog.

The following patch eliminates a couple of them (already sent a patch to
David for the ia64 specific ones) as well as changes the
"Building zonelist : X" in "Built Y zonelists". IMHO it doesn't make any
sense to print for each zonelist since it's run in a for loop running
from 0 to Y-1 anyway.

The patch nukes a few new printk's that were introduced with the
scheduler changes to the NUMA code in -mm3, if these are still needed
then I won't fight for that part of the patch.

From: Anton Blanchard <anton@samba.org>

We have IO BARs on ppc64 machines that begin at address 0. The current
pci probe code will ignore anything that starts at 0. Remove these checks.

From: David Gibson <david@gibson.dropbear.id.au>

In the current ppc64 code the function __ste_allocate() in
arch/ppc64/mm/stab.c takes a context parameter which is never used.  This
patch removes it.

From: Tom Rini <trini@kernel.crashing.org>

The following has been approved by Paul, FWIW.

The following patch makes PPC601_SYNC_FIX depend on PPC_PMAC || PPC_PREP.
It used to depend on ALL_PPC I believe, but this was (at least
semi-intentionally I gather) backed out.  The last time this was discussed,
the only platforms old enough to have a 601 to support were pmacs and
preps, so it makes some sense to only ask for these machines.

From: Tom Rini <trini@kernel.crashing.org>

Various fixes for the PPC_GEN550 backend.

- Move PPC_GEN550 bool into a more appropriate spot.

- Add PPC_GEN550 support to the MCPN765 platform.

- Allow for SERIAL_TEXT_DEBUG on PPC_GEN550.

- Add missing headers to arch/ppc/syslib/gen550_dbg.c

- Clean-up the KGDB interface such that we allow for a kgdb_map_scc call,
  but do not require one.

- Add gen550 prototypes to <asm/kgdb.h>

- PPC_GEN550 backend code doesn't depend on 8250_SERIAL.

From: Jes Sorensen <jes@trained-monkey.org>

The following patch sets the IOMMU bounce limit on the SN2 which is require
to avoid some ISA checks in ll_rw_blk.c causing a BUG_ON().

From: Armin <armin@melware.de>

flush_scheduled_work() may not be called from irq-context and it is not
necessary for the diva_os_cancel_soft_isr() function anyway.

From: Martin Schwidefsky <schwidefsky@de.ibm.com>

Fix a race on sleep_avg in sched_exit().

The symptom I saw on 64-bit s390 has been a fixpoint divide exception
because sleep_avg had a value > NS_MAX_SLEEP_AVG.  I tracked it down and
the problem is sched_exit which recalculates the parents sleep average
without taking the runqueue lock.  schedule() subtracts run_time from
sleep_avg of the previous process.  This can turn out negative and is
corrected shortly after the subtraction but that is already too late.
sched_exit() already read the negative value an miscalculated the parents
sleep_avg -> bang.

I fixed this by adding task_rq_lock/task_rq_unlock to sched_exit().

See http://www.cse.unsw.edu.au/~gernot/persona/hobbies/dunnart.html
for more information about Dunnarts, in case you've never heard of
them before. Courtesy of Gernot Heiser.

the size to "vsnprintf()".

That's a pretty clear case of overflow.

into home.osdl.org:/home/torvalds/v2.5/linux

into home.osdl.org:/home/torvalds/v2.5/linux

into kernel.bkbits.net:/home/davem/tg3-2.6

  http://bugzilla.kernel.org/show_bug.cgi?id=1766

into kernel.bkbits.net:/home/davem/tg3-2.6

sparc_do_fault assumes that pmd_val(*pmd) = pmd_val(*pmd_k) will
duplicate a PMD entry.  With large soft-PMDs this doesn't work.

Align the pkmap area on a 4MB region so that a single
large pagetable can be used.  Fixes random killing of
processes which used high memory.