GitHub/exynos8895/android_kernel_samsung_universal8895.git
19 years ago[IA64] Merge audit fix for fsyscalls with syscall-optimizations
David Mosberger-Tang [Thu, 5 May 2005 13:40:00 +0000 (06:40 -0700)]
[IA64] Merge audit fix for fsyscalls with syscall-optimizations

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] need r29=psr *after* rsm psr.i
David Mosberger-Tang [Thu, 28 Apr 2005 04:22:40 +0000 (21:22 -0700)]
[IA64] need r29=psr *after* rsm psr.i

Yanmin Zhang pointed out a sequence problem when saving the psr.  David
Mosberger provided this patch (which gave up a cycle).

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] use srlz.d instead of srlz.i in ia64_leave_kernel()
David Mosberger-Tang [Thu, 28 Apr 2005 04:22:08 +0000 (21:22 -0700)]
[IA64] use srlz.d instead of srlz.i in ia64_leave_kernel()

This patch switches the srlz.i in ia64_leave_kernel() to srlz.d.  As
per architecture manual, the former is needed only to ensure that the
clearing of PSR.IC is seen by the VHPT for subsequent instruction
fetches.  However, since the remainder of the code (up to and
including the RFI instruction) is mapped by a pinned TLB entry, there
is no chance of an iTLB miss and we don't care whether or not the VHPT
sees PSR.IC cleared.  Since srlz.d is substantially cheaper than
srlz.i, this should shave off a few cycles off the interrupt path
(unverified though; I'm not setup to measure this at the moment).

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Annotate fsys_bubble_down() with McKinley dispatch info.
David Mosberger-Tang [Thu, 28 Apr 2005 04:21:26 +0000 (21:21 -0700)]
[IA64] Annotate fsys_bubble_down() with McKinley dispatch info.

This patch changes comments & formatting only.  There is no code
change.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Reschedule fsys_bubble_down().
David Mosberger-Tang [Thu, 28 Apr 2005 04:20:51 +0000 (21:20 -0700)]
[IA64] Reschedule fsys_bubble_down().

Improvements come from eliminating srlz.i, not scheduling AR/CR-reads
too early (while there are others still pending), scheduling the
backing-store switch as well as possible, splitting the BBB bundle
into a MIB/MBB pair.

Why is it safe to eliminate the srlz.i?  Observe
that we used to clear bits ~PSR_PRESERVED_BITS in PSR.L.  Since
PSR_PRESERVED_BITS==PSR.{UP,MFL,MFH,PK,DT,PP,SP,RT,IC}, we
ended up clearing PSR.{BE,AC,I,DFL,DFH,DI,DB,SI,TB}.  However,

 PSR.BE : already is turned off in __kernel_syscall_via_epc()
 PSR.AC : don't care (kernel normally turns PSR.AC on)
 PSR.I  : already turned off by the time fsys_bubble_down gets invoked
 PSR.DFL: always 0 (kernel never turns it on)
 PSR.DFH: don't care --- kernel never touches f32-f127 on its own
  initiative
 PSR.DI : always 0 (kernel never turns it on)
 PSR.SI : always 0 (kernel never turns it on)
 PSR.DB : don't care --- kernel never enables kernel-level breakpoints
 PSR.TB : must be 0 already; if it wasn't zero on entry to
  __kernel_syscall_via_epc, the branch to fsys_bubble_down
  will trigger a taken branch; the taken-trap-handler then
  converts the syscall into a break-based system-call.

In other words: all the bits we're clearying are either 0 already or
are don't cares!  Thus, we don't have to write PSR.L at all and we
don't have to do a srlz.i either.

Good for another ~20 cycle improvement for EPC-based heavy-weight
syscalls.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Annotate __kernel_syscall_via_epc() with McKinley dispatch info.
David Mosberger-Tang [Thu, 28 Apr 2005 04:20:11 +0000 (21:20 -0700)]
[IA64] Annotate __kernel_syscall_via_epc() with McKinley dispatch info.

Two other very minor changes: use "mov.i" instead of "mov" for reading
ar.pfs (for clarity; doesn't affect the code at all).  Also, predicate
the load of r14 for consistency.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Reschedule __kernel_syscall_via_epc().
David Mosberger-Tang [Thu, 28 Apr 2005 04:19:37 +0000 (21:19 -0700)]
[IA64] Reschedule __kernel_syscall_via_epc().

Avoid some stalls, which is good for about 2 cycles when invoking a
light-weight handler.  When invoking a heavy-weight handler, this
helps by about 7 cycles, with most of the improvement coming from the
improved branch-prediction achieved by splitting the BBB bundle into
two MIB bundles.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Reschedule break_fault() for better performance.
David Mosberger-Tang [Thu, 28 Apr 2005 04:19:04 +0000 (21:19 -0700)]
[IA64] Reschedule break_fault() for better performance.

This patch reorganizes break_fault() to optimistically assume that a
system-call is being performed from user-space (which is almost always
the case).  If it turns out that (a) we're not being called due to a
system call or (b) we're being called from within the kernel, we fixup
the no-longer-valid assumptions in non_syscall() and .break_fixup(),
respectively.

With this approach, there are 3 major phases:

 - Phase 1: Read various control & application registers, in
    particular the current task pointer from AR.K6.
 - Phase 2: Do all memory loads (load system-call entry,
    load current_thread_info()->flags, prefetch
    kernel register-backing store) and switch
    to kernel register-stack.
 - Phase 3: Call ia64_syscall_setup() and invoke
    syscall-handler.

Good for 26-30 cycles of improvement on break-based syscall-path.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] In ia64_leave_syscall(), fix comments and whitespace only.
David Mosberger-Tang [Thu, 28 Apr 2005 04:18:22 +0000 (21:18 -0700)]
[IA64] In ia64_leave_syscall(), fix comments and whitespace only.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Schedule ia64_leave_syscall() to read ar.bsp earlier
David Mosberger-Tang [Thu, 28 Apr 2005 04:17:44 +0000 (21:17 -0700)]
[IA64] Schedule ia64_leave_syscall() to read ar.bsp earlier

Reschedule code to read ar.bsp as early as possible.  To enable this,
don't bother clearing some of the registers when we're returning to
kernel stacks.  Also, instead of trying to support the pNonSys case
(which makes no sense), do a bugcheck instead (with break 0).  Finally,
remove a clear of r14 which is a left-over from the previous patch.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] In syscall-entry, use st8 instead of stf8 to clear pt_regs.r8
David Mosberger-Tang [Thu, 28 Apr 2005 04:17:03 +0000 (21:17 -0700)]
[IA64] In syscall-entry, use st8 instead of stf8 to clear pt_regs.r8

Using stf8 seemed like a clever idea at the time, but stf8 forces
the cache-line to be invalidated in the L1D (if it happens to be
there already).  This patch eliminates a guaranteed L1D cache-miss
and, by itself, is good for a 1-2 cycle improvement for heavy-weight
syscalls.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] On return from syscall, hint b7 with __kernel_syscall_via_epc().
David Mosberger-Tang [Thu, 28 Apr 2005 04:16:07 +0000 (21:16 -0700)]
[IA64] On return from syscall, hint b7 with __kernel_syscall_via_epc().

Why is this a good idea?  Clearing b7 to 0 is guaranteed to do us no
good and writing it with __kernel_syscall_via_epc() yields a 6 cycle
improvement _if_ the application performs another EPC-based system-
call without overwriting b7, which is not all that uncommon.  Well
worth the minimal cost of 1 bundle of code.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Schedule fp-clearing insns at least 6 cycles after reading ar.bsp.
David Mosberger-Tang [Thu, 28 Apr 2005 04:15:13 +0000 (21:15 -0700)]
[IA64] Schedule fp-clearing insns at least 6 cycles after reading ar.bsp.

Decreases syscall overhead by approximately 6 cycles.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Use dynamic prediction for RSE-clearing branches.
David Mosberger-Tang [Thu, 28 Apr 2005 04:13:33 +0000 (21:13 -0700)]
[IA64] Use dynamic prediction for RSE-clearing branches.

This by itself is good for a 1-2 cycle speed up.  Effect is bigger
when combined with the later patches.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] __ia64_syscall() is no longer used anywhere in the kernel. Remove it.
David Mosberger-Tang [Thu, 28 Apr 2005 04:10:45 +0000 (21:10 -0700)]
[IA64] __ia64_syscall() is no longer used anywhere in the kernel.  Remove it.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years agoAutomatic merge of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2...
Linus Torvalds [Wed, 27 Apr 2005 17:05:42 +0000 (10:05 -0700)]
Merge ... /linux/kernel/git/aegl/linux-2.6.git

19 years ago[PATCH] PC300 pci_enable_device fix
Marcelo Tosatti [Tue, 26 Apr 2005 20:09:35 +0000 (13:09 -0700)]
[PATCH] PC300 pci_enable_device fix

Call pci_enable_device() before looking at IRQ and resources,
and pci_disable_device() when shutting the interface down.

The driver requires this fix or the "pci=routeirq" workaround
on 2.6.10 and later kernels.

Reported and tested by Artur Lipowski.

From: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Marcelo Tosatti <marcelo.tosatti@cyclades.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc64: trivial user annotations
Al Viro [Tue, 26 Apr 2005 18:26:53 +0000 (11:26 -0700)]
[PATCH] ppc64: trivial user annotations

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] size_t portability fixes - drivers/usb
Al Viro [Tue, 26 Apr 2005 18:26:53 +0000 (11:26 -0700)]
[PATCH] size_t portability fixes - drivers/usb

size_t is zu, ssize_t is zd...

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[IA64] re-enable preempt before page allocation for pgtable quicklist
Martin Hicks [Tue, 26 Apr 2005 16:04:31 +0000 (09:04 -0700)]
[IA64] re-enable preempt before page allocation for pgtable quicklist

This is a fix to the pgtable_quicklist code.  There is a GFP_KERNEL
allocation in pgtable_quicklist_alloc(), which spews the usual warnings
if the kernel is under heavy VM pressure and the reclaim code is
invoked.  re-enable preempt before we allocate the new page.

This patch is against 2.6.12-rc2-mm2

Signed-off-by: Martin Hicks <mort@sgi.com>
Signed-off-by: Tony Luck <tony.luckintel.com>
19 years agoAutomatic merge of kernel.org:/home/rmk/linux-2.6-serial.git
Linus Torvalds [Tue, 26 Apr 2005 16:02:40 +0000 (09:02 -0700)]
Merge ... /home/rmk/linux-2.6-serial.git

19 years agoAutomatic merge of kernel.org:/home/rmk/linux-2.6-rmk.git
Linus Torvalds [Tue, 26 Apr 2005 15:58:22 +0000 (08:58 -0700)]
Merge ... kernel.org:/home/rmk/linux-2.6-rmk.git

19 years ago[PATCH] qla trivial iomem annotation
Al Viro [Tue, 26 Apr 2005 14:43:43 +0000 (07:43 -0700)]
[PATCH] qla trivial iomem annotation

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] oprofile trivial user annotations
Al Viro [Tue, 26 Apr 2005 14:43:42 +0000 (07:43 -0700)]
[PATCH] oprofile trivial user annotations

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] amd64 rt_sigframe user annotation
Al Viro [Tue, 26 Apr 2005 14:43:42 +0000 (07:43 -0700)]
[PATCH] amd64 rt_sigframe user annotation

->pretcode in struct rt_sigframe is a userland pointer (and already
treated as such by code using that field).

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] drivers/scsi NULL noise removal
Al Viro [Tue, 26 Apr 2005 14:43:42 +0000 (07:43 -0700)]
[PATCH] drivers/scsi NULL noise removal

NULL noise in sbus scsi drivers got missed

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] imsttfb missing iomem annotations
Al Viro [Tue, 26 Apr 2005 14:43:42 +0000 (07:43 -0700)]
[PATCH] imsttfb missing iomem annotations

write_reg_le32() and read_reg_le32() expect iomem pointers...

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc-opc NULL noise removal
Al Viro [Tue, 26 Apr 2005 14:43:41 +0000 (07:43 -0700)]
[PATCH] ppc-opc NULL noise removal

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] savagefb iomem annotations
Al Viro [Tue, 26 Apr 2005 14:43:41 +0000 (07:43 -0700)]
[PATCH] savagefb iomem annotations

trivial iomem annotations + memset() replaced with memset_io() in a
place that deals with ioremapped area.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] pegasus NULL noise removal
Al Viro [Tue, 26 Apr 2005 14:43:41 +0000 (07:43 -0700)]
[PATCH] pegasus NULL noise removal

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] tpm 64bit fixes (size_t)
Al Viro [Tue, 26 Apr 2005 14:43:41 +0000 (07:43 -0700)]
[PATCH] tpm 64bit fixes (size_t)

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] fix SND_POWERMAC dependencies
Al Viro [Tue, 26 Apr 2005 14:43:40 +0000 (07:43 -0700)]
[PATCH] fix SND_POWERMAC dependencies

This beast is pmac-only; moreover, it won't build on other
subarchitectures.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] Serial: Convert 8250_hp300 to use serial8250_{un,}register_port
Russell King [Tue, 26 Apr 2005 14:37:45 +0000 (15:37 +0100)]
[PATCH] Serial: Convert 8250_hp300 to use serial8250_{un,}register_port

Convert 8250_hp300 to use serial8250_register_port() and
serial8250_unregister_port().

Tested by Kars de Jong, 4/4/2005.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] Serial: Move error path processing inline
Russell King [Tue, 26 Apr 2005 14:32:00 +0000 (15:32 +0100)]
[PATCH] Serial: Move error path processing inline

With unlikely() there's no need for the error path to
use gotos.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] Serial: Ensure error paths are marked with unlikely()
Russell King [Tue, 26 Apr 2005 14:29:44 +0000 (15:29 +0100)]
[PATCH] Serial: Ensure error paths are marked with unlikely()

Ensure ARM serial driver error paths are marked with the
unlikely() compiler hint.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years agoAutomatic merge of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2...
Linus Torvalds [Tue, 26 Apr 2005 14:22:55 +0000 (07:22 -0700)]
Merge ... /linux/kernel/git/gregkh/usb-2.6.git/

19 years ago[PATCH] ARM: make entry*.S includes more logical
Russell King [Tue, 26 Apr 2005 14:21:02 +0000 (15:21 +0100)]
[PATCH] ARM: make entry*.S includes more logical

Move common includes to entry-header, and file specific includes
to the relevant file.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: Remove single-use user save/restore macros
Russell King [Tue, 26 Apr 2005 14:20:34 +0000 (15:20 +0100)]
[PATCH] ARM: Remove single-use user save/restore macros

Assembly macros are pointless if they're only used once.  Move
them inline.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: remove PT_TRACESYS
Russell King [Tue, 26 Apr 2005 14:20:12 +0000 (15:20 +0100)]
[PATCH] ARM: remove PT_TRACESYS

PT_TRACESYS is unused, remove it.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: Remove SVC_MODE definition
Russell King [Tue, 26 Apr 2005 14:19:48 +0000 (15:19 +0100)]
[PATCH] ARM: Remove SVC_MODE definition

SVC_MODE reflects the MODE_SVC definition in asm/ptrace.h.  Use
the asm/ptrace.h definition instead, and remove SVC_MODE.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: Use __NR_SYSCALL_BASE and __ARM_NR_BASE in asm code
Russell King [Tue, 26 Apr 2005 14:19:24 +0000 (15:19 +0100)]
[PATCH] ARM: Use __NR_SYSCALL_BASE and __ARM_NR_BASE in asm code

Don't define our own local constants, but use those already defined
in asm/unistd.h instead.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years agoAutomatic merge of rsync://rsync.kernel.org/pub/scm/linux/kernel/git/davem/net-2...
Linus Torvalds [Tue, 26 Apr 2005 14:19:05 +0000 (07:19 -0700)]
Merge ... /linux/kernel/git/davem/net-2.6.git

19 years ago[PATCH] ARM: pt_regs offsets
Russell King [Tue, 26 Apr 2005 14:18:59 +0000 (15:18 +0100)]
[PATCH] ARM: pt_regs offsets

Generate pt_regs S_xx offsets from the structure itself instead
of #defining them.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: Remove argument for disable_irq/enable_irq
Russell King [Tue, 26 Apr 2005 14:18:26 +0000 (15:18 +0100)]
[PATCH] ARM: Remove argument for disable_irq/enable_irq

Since we do not require a register for these operations, we can
remove this unnecessary argument.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[PATCH] ARM: remove some entry initialisation asm code
Russell King [Tue, 26 Apr 2005 14:17:42 +0000 (15:17 +0100)]
[PATCH] ARM: remove some entry initialisation asm code

Convert the trivial vector entry initialisation code to C code.

Signed-off-by: Russell King <rmk@arm.linux.org.uk>
19 years ago[NET]: kill gratitious includes of major.h
Al Viro [Tue, 26 Apr 2005 04:40:39 +0000 (21:40 -0700)]
[NET]: kill gratitious includes of major.h

A lot of places in there are including major.h for no reason
whatsoever.  Removed.  And yes, it still builds.

The history of that stuff is often amusing.  E.g. for net/core/sock.c
the story looks so, as far as I've been able to reconstruct it: we used to
need major.h in net/socket.c circa 1.1.early.  In 1.1.13 that need had
disappeared, along with register_chrdev(SOCKET_MAJOR, "socket", &net_fops)
in sock_init().  Include had not.  When 1.2 -> 1.3 reorg of net/* had moved
a lot of stuff from net/socket.c to net/core/sock.c, this crap had followed...

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TCP]: Trivial tcp_data_queue() cleanup
James Morris [Tue, 26 Apr 2005 04:39:29 +0000 (21:39 -0700)]
[TCP]: Trivial tcp_data_queue() cleanup

This patch removes a superfluous intialization from tcp_data_queue().

Signed-off-by: James Morris <jmorris@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[PATCH] kill gratitious includes of major.h under net/*
Al Viro [Tue, 26 Apr 2005 01:32:13 +0000 (18:32 -0700)]
[PATCH] kill gratitious includes of major.h under net/*

A lot of places in there are including major.h for no reason whatsoever.
Removed.  And yes, it still builds.

The history of that stuff is often amusing.  E.g.  for net/core/sock.c
the story looks so, as far as I've been able to reconstruct it: we used
to need major.h in net/socket.c circa 1.1.early.  In 1.1.13 that need
had disappeared, along with register_chrdev(SOCKET_MAJOR, "socket",
&net_fops) in sock_init().  Include had not.  When 1.2 -> 1.3 reorg of
net/* had moved a lot of stuff from net/socket.c to net/core/sock.c,
this crap had followed...

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] isofs includes sanitized
Al Viro [Tue, 26 Apr 2005 01:32:12 +0000 (18:32 -0700)]
[PATCH] isofs includes sanitized

fs/isofs includes trimmed down to something resembling sanity.

Kernel-only parts of linux/iso_fs.h and entire linux/iso_fs_{sb,i}.h
moved to fs/isofs/isofs.h.

A lot of useless #include in fs/isofs/*.c killed.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc annotations: i2c-mpc
Al Viro [Tue, 26 Apr 2005 01:32:12 +0000 (18:32 -0700)]
[PATCH] ppc annotations: i2c-mpc

Usual iomem annotations and NULL noise removal.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc annotations: mpsc
Al Viro [Tue, 26 Apr 2005 01:32:12 +0000 (18:32 -0700)]
[PATCH] ppc annotations: mpsc

Usual iomem annotations + NULL noise removal.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years agoAutomated merge of kernel.org:/home/rmk/linux-2.6-rmk.git
Linus Torvalds [Mon, 25 Apr 2005 23:19:32 +0000 (16:19 -0700)]
Automated merge of /home/rmk/linux-2.6-rmk.git

19 years ago[PATCH] ARM: 2653/1: Fix memset and memzero macro double-reference of parameters
Deepak Saxena [Mon, 25 Apr 2005 22:40:05 +0000 (23:40 +0100)]
[PATCH] ARM: 2653/1: Fix memset and memzero macro double-reference of parameters

Patch from Deepak Saxena

The current memset() and memzero() macros on ARM reference the
incoming parameters more than once and this can cause uninted
side-effects. The issue was found while debugging SCTP protocol
and with the specific usage of memzero(skb_put(skb,size),size).
This call would call skb_put(skb,size) twice leading to badness.
The fixed version copies the incoming parameters into local
variables and uses those instead.

Signed-off-by: Deepak Saxena
Signed-off-by: Russell King
19 years ago[PATCH] ARM: 2650/1: PXA27x sleep - workaround Errata 39 & 50 (Patch 2667)
Jeff Lackey [Mon, 25 Apr 2005 22:38:55 +0000 (23:38 +0100)]
[PATCH] ARM: 2650/1: PXA27x sleep - workaround Errata 39 & 50 (Patch 2667)

Patch from Jeff Lackey

This patch updates arch/arm/mach-pxa/sleep.S to support
the PXA270 CPU.  It works around Errata 39 & 50 from the
Intel(R) PXA27x Processor Family Specification Update.

Signed-off-by: Jeff Lackey
Signed-off-by: Russell King
19 years ago[TG3]: Fix tg3_restart_ints()
Michael Chan [Mon, 25 Apr 2005 22:17:17 +0000 (15:17 -0700)]
[TG3]: Fix tg3_restart_ints()

tg3_restart_ints() is called to re-enable interrupts after tg3_poll()
has finished all the work. It calls tg3_cond_int() to force an interrupt
if the status block updated bit is set. The updated bit will be set if
there is a new status block update sometime during tg3_poll() and it can
be very often. The worst part is that even if all the work has been
processed, the updated bit remains set and an interrupt will be forced
unnecessarily.

The fix is to call tg3_has_work() instead to determine if new work is
posted before forcing an interrupt. The way to force an interrupt is
also changed to use "coalesce_now" instead of "SETINT". The former is
generally a safer way to force the interrupt.

Also deleted the first parameter to tg3_has_work() which is unused.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TG3]: Refresh hw index in tg3_rx()
Michael Chan [Mon, 25 Apr 2005 22:14:32 +0000 (15:14 -0700)]
[TG3]: Refresh hw index in tg3_rx()

This patch refreshes the hw rx producer in tg3_rx() so that additional
work posted by the hardware can be processed.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[TG3]: Fix bug in tg3_rx()
Michael Chan [Mon, 25 Apr 2005 22:14:03 +0000 (15:14 -0700)]
[TG3]: Fix bug in tg3_rx()

This patch fixes a bug that causes tg3_has_work() to always return 1.

rx work is determined by comparing tp->rx_rcb_ptr with the current hw
producer index. The hw producer index is modulo the ring size, but tp-
>rx_rcb_ptr is a free running counter that goes up beyond the ring size.
After the ring wraps around once, tg3_has_work() will always return 1.

The fix is to always do modulo arithmetic on tp->rx_rcb_ptr.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[PATCH] ARM: 2645/1: Adds IIS definitions for the S3C2400
Lucas Correia Villa Real [Mon, 25 Apr 2005 22:13:15 +0000 (23:13 +0100)]
[PATCH] ARM: 2645/1: Adds IIS definitions for the S3C2400

Patch from Lucas Correia Villa Real

Adds IISFCON definitions for the S3C2400 at
include/asm-arm/arch-s3c2400/regs-iis.h.

Signed-off-by: Lucas Correia Villa Real
Signed-off-by: Ben Dooks
Signed-off-by: Russell King
19 years ago[PATCH] ARM: 2644/1: Adds S3C2400 support to uncompress.h
Lucas Correia Villa Real [Mon, 25 Apr 2005 22:12:50 +0000 (23:12 +0100)]
[PATCH] ARM: 2644/1: Adds S3C2400 support to uncompress.h

Patch from Lucas Correia Villa Real

The S3C2400 doesn't have a cpuid information stored anywhere. This patch adds
support to the S3C2400 at include/asm-arm/arch-s3c2400/uncompress.h by
initializing the cpuid variable to the S3C2410, as they share the same
routine. The GSTATUS1 pin is then used only if not compiling for the S3C2400.

Signed-off-by: Lucas Correia Villa Real
Signed-off-by: Ben Dooks
Signed-off-by: Russell King
19 years agoFrom: jbarnes@sgi.com
Jesse Barnes [Mon, 25 Apr 2005 20:31:04 +0000 (13:31 -0700)]
From: jbarnes@sgi.com

[IA64] fix ia64 Kconfig to allow CONFIG_PM on sn2

This probably should have been fixed when I fixed up the generic build for
discontig+numa machines, but oh well.

CONFIG_PM is allowable for generic builds but not for sn2 builds, which
doesn't make much sense, and in fact breaks the build if recent ACPI bits are
added to the tree.  It looks like the only arch that needs to prevent
CONFIG_PM stuff is the ski simulator (though those options could probably use
some cleanup as well), so remove the big conditional and replace it with a
simple test for IA64_HP_SIM instead.

Signed-off-by: Jesse Barnes <jbarnes@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] snsc_event.c new file
Greg Howard [Mon, 25 Apr 2005 20:29:46 +0000 (13:29 -0700)]
[IA64-SGI] snsc_event.c new file

Forgot the "bk new" to add this file. Part of the patch
from Greg Howard

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Altix system controller event handling
Greg Howard [Mon, 25 Apr 2005 20:28:52 +0000 (13:28 -0700)]
[IA64] Altix system controller event handling

The following is an update of the patch I sent yesterday
(3/9/05) incorporating suggestions from Christoph Hellwig and
Andreas Schwab.  It allows Altix and Altix-like systems to
handle environmental events generated by the system controllers,
and should apply on top of Jack Steiner's patch of 3/1/05 ("New
chipset support for SN platform") and Mark Goodwin's patch of
3/8/05 ("Altix SN topology support for new chipsets and pci
topology").

Signed-off-by: Greg Howard <ghoward@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] iosapic.c: typo ... s/spin_unlock_irq/spin_unlock/
Kenji Kaneshige [Mon, 25 Apr 2005 20:27:48 +0000 (13:27 -0700)]
[IA64] iosapic.c: typo ... s/spin_unlock_irq/spin_unlock/

vector sharing patch had a typo ... mismatched spin_lock() with
a spin_unlock_irq().  Fix from Kenji Kaneshige.

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] print "siblings" before {physical,core,thread} id
Tony Luck [Mon, 25 Apr 2005 20:27:12 +0000 (13:27 -0700)]
[IA64] print "siblings" before {physical,core,thread} id

Rohit and Suresh changed their mind about the order to print things
in /proc/cpuinfo, but didn't include the change in the version of
the patch they sent to me.

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] vector sharing (Large I/O system support)
Kenji Kaneshige [Mon, 25 Apr 2005 20:26:23 +0000 (13:26 -0700)]
[IA64] vector sharing (Large I/O system support)

Current ia64 linux cannot handle greater than 184 interrupt sources
because of the lack of vectors. The following patch enables ia64 linux
to handle greater than 184 interrupt sources by allowing the same
vector number to be shared by multiple IOSAPIC's RTEs. The design of
this patch is besed on "Intel(R) Itanium(R) Processor Family Interrupt
Architecture Guide".

Even if you don't have a large I/O system, you can see the behavior of
vector sharing by changing IOSAPIC_LAST_DEVICE_VECTOR to fewer value.

Signed-off-by: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] multi-core/multi-thread identification
Suresh Siddha [Mon, 25 Apr 2005 20:25:06 +0000 (13:25 -0700)]
[IA64] multi-core/multi-thread identification

Version 3 - rediffed to apply on top of Ashok's hotplug cpu
patch.  /proc/cpuinfo output in step with x86.

This is an updated MC/MT identification patch based on the
previous discussions on list.

Add the Multi-core and Multi-threading detection for IPF.
  - Add new core and threading related fields in /proc/cpuinfo.
Physical id
Core id
Thread id
Siblings
  - setup the cpu_core_map and cpu_sibling_map appropriately
  - Handles Hot plug CPU

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Gordon Jin <gordon.jin@intel.com>
Signed-off-by: Rohit Seth <rohit.seth@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] __copy_user breaks on unaligned src
Keith Owens [Mon, 25 Apr 2005 20:23:47 +0000 (13:23 -0700)]
[IA64] __copy_user breaks on unaligned src

memcpy_mck.S::__copy_user breaks in the prefetch code under these conditions :-

* src is unaligned and
* dst is near the end of a page and
* the page after dst is unmapped.

Signed-off-by: Keith Owens <kaos@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Need to handle lfetch in "no_context" case.
Tony Luck [Mon, 25 Apr 2005 20:22:44 +0000 (13:22 -0700)]
[IA64] Need to handle lfetch in "no_context" case.

Thanks to Mark for tracking down this one.  Users of __copy_from_user_inatomic()
will be sad if we don't handle lfetch faults for the "no_context" case.

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Altix SN add support for slots in geoid_t locator
Mark Goodwin [Mon, 25 Apr 2005 20:21:54 +0000 (13:21 -0700)]
[IA64-SGI] Altix SN add support for slots in geoid_t locator

This patch against ia64-test-2.6.12 is needed for forthcoming
Altix chipsets. It renames geoid_any_t to geoid_common_t and
splits the 8bit 'slab' field into two 4bit fields for 'slab'
and 'slot'. Similar changes in the Altix SAL will retain backward
compatibility for old kernels.

Signed-off-by: Mark Goodwin <markgw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] fix syscall-optimization goof
David Mosberger-Tang [Mon, 25 Apr 2005 20:20:38 +0000 (13:20 -0700)]
[IA64] fix syscall-optimization goof

Sadly, I goofed in this syscall-tuning patch:

ChangeSet 1.1966.1.40 2005/01/22 13:31:05 davidm@hpl.hp.com
  [IA64] Improve ia64_leave_syscall() for McKinley-type cores.

  Optimize ia64_leave_syscall() a bit better for McKinley-type cores.
  The patch looks big, but that's mostly due to renaming r16/r17 to r2/r3.
  Good for a 13 cycle improvement.

The problem is that the size of the physical stacked registers was
loaded into the wrong register (r3 instead of r17).  Since r17 by
coincidence always had the value 1, this had the effect of turning
rse_clear_invalid into a no-op.  That poses the risk of leaking kernel
state back to user-land and is hence not acceptable.

The fix below is simple, but unfortunately it costs us about 28 cycles
in syscall overhead. ;-(

Unfortunately, there isn't much we can do about that since those
registers have to be cleared one way or another.

--david

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Shub2 BTE support - BTE recovery code
Russ Anderson [Mon, 25 Apr 2005 20:19:52 +0000 (13:19 -0700)]
[IA64-SGI] Shub2 BTE support - BTE recovery code

patch 2:
Shub2 BTE recovery code will be implemented in SAL.
Define the SAL interface.
Modify bte_error to call SAL for shub2.

Signed-off-by: Russ Anderson <rja@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Add new MMR definitions/Modify BTE initialiation&copy.
Russ Anderson [Mon, 25 Apr 2005 20:19:11 +0000 (13:19 -0700)]
[IA64-SGI] Add new MMR definitions/Modify BTE initialiation&copy.

patch 1:
Add new MMR definitions.
Modify BTE initialiation.
Modify BTE copy.

Signed-off-by: Russ Anderson <rja@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] disable TIOCA GART TLB prefetching
Mark Maule [Mon, 25 Apr 2005 20:18:02 +0000 (13:18 -0700)]
[IA64-SGI] disable TIOCA GART TLB prefetching

Patch to disable SGI TIOCA GART TLB prefetching due to hw bug.

Signed-off-by: Mark Maule <maule@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] MAX_PGT_FREES_PER_PASS must be 'L' to avoid warning
Tony Luck [Mon, 25 Apr 2005 20:16:59 +0000 (13:16 -0700)]
[IA64] MAX_PGT_FREES_PER_PASS must be 'L' to avoid warning

'min' is very picky about types of arguments, make it happy

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] fix: warning: `ql_size' might be used uninitialized
Tony Luck [Mon, 25 Apr 2005 20:16:16 +0000 (13:16 -0700)]
[IA64] fix: warning: `ql_size' might be used uninitialized

Oops.  Should have caught this before I checked it in.

Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] sba_iommu bug fixes
Alex Williamson [Mon, 25 Apr 2005 20:14:36 +0000 (13:14 -0700)]
[IA64] sba_iommu bug fixes

   This fixes a couple of bugs in the zx1/sx1000 sba_iommu.  These are
all pretty low likelihood of hitting.  The first problem is a simple off
by one, deep in the sba_alloc_range() error path.  Surrounding that was
a lock ordering problem that could have potentially deadlocked with the
order the locks are grabbed in sba_unmap_single().  I moved the resource
locking into sba_search_bitmap() to prevent this.  Finally, there's a
potential race between unmapping pdir entries and marking incoming DMA
pages clean.  If you see any oddities, please let me know, but I've
tested it pretty thoroughly here.  Tony, please apply.  Thanks,

BTW, many of the options in this driver not on by default are becoming
more and more broken.  I'll be working on some patches to clean them
out, but I wanted to get this bug fix out first.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] Percpu quicklist for combined allocator for pgd/pmd/pte.
Robin Holt [Mon, 25 Apr 2005 20:13:16 +0000 (13:13 -0700)]
[IA64] Percpu quicklist for combined allocator for pgd/pmd/pte.

This patch introduces using the quicklists for pgd, pmd, and pte levels
by combining the alloc and free functions into a common set of routines.
This greatly simplifies the reading of this header file.

This patch is simple but necessary for large numa configurations.
It simply ensures that only pages from the local node are added to a
cpus quicklist.  This prevents the trapping of pages on a remote nodes
quicklist by starting a process, touching a large number of pages to
fill pmd and pte entries, migrating to another node, and then unmapping
or exiting.  With those conditions, the pages get trapped and if the
machine has more than 100 nodes of the same size, the calculation of
the pgtable high water mark will be larger than any single node so page
table cache flushing will never occur.

I ran lmbench lat_proc fork and lat_proc exec on a zx1 with and without
this patch and did not notice any change.

On an sn2 machine, there was a slight improvement which is possibly
due to pages from other nodes trapped on the test node before starting
the run.  I did not investigate further.

This patch shrinks the quicklist based upon free memory on the node
instead of the high/low water marks.  I have written it to enable
preemption periodically and recalculate the amount to shrink every time
we have freed enough pages that the quicklist size should have grown.
I rescan the nodes zones each pass because other processess may be
draining node memory at the same time as we are adding.

Signed-off-by: Robin Holt <holt@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI]
Bruce Losure [Mon, 25 Apr 2005 20:12:02 +0000 (13:12 -0700)]
[IA64-SGI]

Missed the "bk new" for this file in the last commit.

Signed-off-by: Bruce Losure <blosure@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Altix: enable poweroff
Aaron J Young [Mon, 25 Apr 2005 20:11:14 +0000 (13:11 -0700)]
[IA64-SGI] Altix: enable poweroff

This patch adds the necessary "hook" to allow SGI/SN
machines to perform a system power off upon a
'init 0', 'halt -p', 'poweroff' or 'shutdown -h'.

The "hook" is to set the pm_power_off callback
to ia64_sn_power_down(). pm_power_off is checked
in machine_power_off()/do_poweroff() and, if set, is executed.
ia64_sn_power_down() is a function already present (but not
used currently) in the sn kernel.
ia64_sn_power_down() makes a SAL call to execute the
power off.

Signed-off-by: Aaron J Young <ayoung@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Bus driver for the CX port of SGI's TIO chip.
Bruce Losure [Mon, 25 Apr 2005 20:09:41 +0000 (13:09 -0700)]
[IA64-SGI] Bus driver for the CX port of SGI's TIO chip.

This patch is to provide CX port infrastructure for SGI TIO-based
h/w.   Also a 'core services' driver for SGI FPGA-based h/w.

Signed-off-by: Bruce Losure <blosure@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] perfmon: make pfm_sysctl a global, and other cleanup
Stephane Eranian [Mon, 25 Apr 2005 20:08:30 +0000 (13:08 -0700)]
[IA64] perfmon: make pfm_sysctl a global, and other cleanup

- make pfm_sysctl a global such that it is possible
  to enable/disable debug printk in sampling formats
  using PFM_DEBUG.

- remove unused pfm_debug_var variable

- fix a bug in pfm_handle_work where an BUG_ON() could
  be triggered. There is a path where pfm_handle_work()
  can be called with interrupts enabled, i.e., when
  TIF_NEED_RESCHED is set. The fix correct the masking
  and unmasking of interrupts in pfm_handle_work() such
  that we restore the interrupt mask as it was upon entry.

signed-off-by: stephane eranian <eranian@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] support variable length nasids in shub2
Colin Ngam [Mon, 25 Apr 2005 20:07:00 +0000 (13:07 -0700)]
[IA64-SGI] support variable length nasids in shub2

This patch enables our TIO IO chipset to support variable length nasids in
Shub2 chipset.

Signed-off-by: Colin Ngam <cngam@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Shub2 provides an addition of 2 External Interrupt events.
Colin Ngam [Mon, 25 Apr 2005 20:06:28 +0000 (13:06 -0700)]
[IA64-SGI] Shub2 provides an addition of 2 External Interrupt events.

Signed-off-by: Colin Ngam <cngam@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Altix SN topology fix potential infinite loop
Mark Goodwin [Mon, 25 Apr 2005 20:05:08 +0000 (13:05 -0700)]
[IA64-SGI] Altix SN topology fix potential infinite loop

Fix infinite loop if sn_hwperf_location_to_bpos() fails.

Signed-off-by: Mark Goodwin <markgw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Altix SN topology support for new chipsets and pci topology
Mark Goodwin [Mon, 25 Apr 2005 20:04:22 +0000 (13:04 -0700)]
[IA64-SGI] Altix SN topology support for new chipsets and pci topology

please accept this patch to the Altix SN platform topology export
interface to support new chipsets and to export PCI topology.

This follows on top of Jack Steiner's patch dated March 1st
("New chipset support for SN platform").

Signed-off-by: Mark Goodwin <markgw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] speed up syscall path a bit more
David Mosberger-Tang [Mon, 25 Apr 2005 20:03:16 +0000 (13:03 -0700)]
[IA64] speed up syscall path a bit more

Recently I noticed that clearing ar.ssd/ar.csd right before srlz.d is
causing significant stalling in the syscall path.  The patch below
fixes that by moving the register-writes after srlz.d.  On a Madison,
this drops break-based getpid() from 241 to 226 cycles (-15 cycles).

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[PKT_SCHED]: Eliminate unnecessary includes in simple.c
David S. Miller [Mon, 25 Apr 2005 19:15:01 +0000 (12:15 -0700)]
[PKT_SCHED]: Eliminate unnecessary includes in simple.c

Noted by Al Viro.

Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[NETFILTER]: Drop conntrack reference when packet leaves IP
Patrick McHardy [Mon, 25 Apr 2005 19:01:07 +0000 (12:01 -0700)]
[NETFILTER]: Drop conntrack reference when packet leaves IP

In the event a raw socket is created for sending purposes only, the creator
never bothers to check the socket's receive queue.  But we continue to
add skbs to its queue until it fills up.

Unfortunately, if ip_conntrack is loaded on the box, each skb we add to the
queue potentially holds a reference to a conntrack.  If the user attempts
to unload ip_conntrack, we will spin around forever since the queued skbs
are pinned.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[NETFILTER]: Fix truncated sequence numbers in FTP helper
Yasuyuki KOZAKAI [Mon, 25 Apr 2005 19:00:04 +0000 (12:00 -0700)]
[NETFILTER]: Fix truncated sequence numbers in FTP helper

Signed-off-by: Yasuyuki KOZAKAI <yasuyuki.kozkaai@toshiba.co.jp>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
19 years ago[IA64] Tighten up unw_unwind_to_user check
Keith Owens [Mon, 25 Apr 2005 18:45:26 +0000 (11:45 -0700)]
[IA64] Tighten up unw_unwind_to_user check

Detect user space by the unwind frame with predicate PRED_USER_STACK
set, instead of a user space IP.  Tighten up the last ditch check for
running off the top of the kernel stack.

Based on a suggestion by David Mosberger, reworked to fit the current
tree.  This survives my stress test which used to break 2.6.9 kernels.
Unlike 2.6.11, the stress test now unwinds to the correct point, so
gdb can get the user space registers.

Signed-off-by: Keith Owens <kaos@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64] add missing cpu_relax() in ITC syncing code
David Mosberger-Tang [Mon, 25 Apr 2005 18:44:02 +0000 (11:44 -0700)]
[IA64] add missing cpu_relax() in ITC syncing code

Call cpu_relax() in busy-waiting loops of the ITC-syncing code.

Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] Change SAL call request code for SN systems
Jack Steiner [Mon, 25 Apr 2005 18:42:39 +0000 (11:42 -0700)]
[IA64-SGI] Change SAL call request code for SN systems

Change the value of the SAL call number for a new SAL request. The
initial implementation in the PROM did not match what the OS expected.
Since the OS can run on PROMs that do not implement the new call,
changing the call number avoids the issue. New PROMs will implement
the new call number. (This avoids problems with the 4.05 PROM).

Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] altix: tioca chip driver (agp)
Mark Maule [Mon, 25 Apr 2005 18:35:54 +0000 (11:35 -0700)]
[IA64-SGI] altix: tioca chip driver (agp)

Provide a driver for the altix TIOCA AGP chipset.  An agpgart backend will
be provided as a separate patch.

Signed-off-by: Mark Maule <maule@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] sn2-move-pci-headers.patch
Mark Maule [Mon, 25 Apr 2005 18:32:16 +0000 (11:32 -0700)]
[IA64-SGI] sn2-move-pci-headers.patch

Move a couple of headers out of arch/ia64/sn/include/pci and into
include/asm-ia64/sn.

Signed-off-by: Mark Maule <maule@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[IA64-SGI] sn2-pci-dma-abstraction.patch
Mark Maule [Mon, 25 Apr 2005 18:26:03 +0000 (11:26 -0700)]
[IA64-SGI] sn2-pci-dma-abstraction.patch

Provide an abstraction of the altix pci dma runtime layer so that multiple
pci-based bridges can be supported.

Signed-off-by: Mark Maule <maule@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
19 years ago[PATCH] ARM: 2641/1: Adds S3C2400_SPPIN_nCS to arch-s3c2410/regs-spi.h
Lucas Correia Villa Real [Mon, 25 Apr 2005 17:40:31 +0000 (18:40 +0100)]
[PATCH] ARM: 2641/1: Adds S3C2400_SPPIN_nCS to arch-s3c2410/regs-spi.h

Patch from Lucas Correia Villa Real

Adds a define to the SPI Card Select bit on the S3C2400 into
include/asm-arm/arch-s3c2410/regs-spi.h

Signed-off-by: Lucas Correia Villa Real
Signed-off-by: Russell King
19 years ago[PATCH] ARM: 2640/1: Adds DRAM/BANKSIZE_MASK definitions for S3C2400
Lucas Correia Villa Real [Mon, 25 Apr 2005 17:36:50 +0000 (18:36 +0100)]
[PATCH] ARM: 2640/1: Adds DRAM/BANKSIZE_MASK definitions for S3C2400

Patch from Lucas Correia Villa Real

Adds DRAM refresh definitions and sets the BANKSIZE_MASK for the S3C2400 on
include/asm-arm/arch-s3c2400/regs-mem.h

Signed-off-by: Lucas Correia Villa Real
Signed-off-by: Ben Dooks
Signed-off-by: Russell King
19 years ago[PATCH] fs/aio.c: make some code static
Adrian Bunk [Mon, 25 Apr 2005 15:18:14 +0000 (08:18 -0700)]
[PATCH] fs/aio.c: make some code static

This patch makes some needlessly global code static.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Benjamin LaHaise <bcrl@kvack.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc user annotations: debug_setconetext(2)
Al Viro [Mon, 25 Apr 2005 14:55:59 +0000 (07:55 -0700)]
[PATCH] ppc user annotations: debug_setconetext(2)

3rd argument of sys_debug_setcontext() is also a userland pointer.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
19 years ago[PATCH] ppc iomem annotations: mv643xx_eth
Al Viro [Mon, 25 Apr 2005 14:55:58 +0000 (07:55 -0700)]
[PATCH] ppc iomem annotations: mv643xx_eth

void * __iomem replaced with intended void __iomem *.

Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>