]> err.no Git - linux-2.6/log
linux-2.6
17 years ago[PATCH] i386: modpost apic related warning fixes
Vivek Goyal [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] i386: modpost apic related warning fixes

o Modpost generates warnings for i386 if compiled with CONFIG_RELOCATABLE=y

WARNING: vmlinux - Section mismatch: reference to .init.text:find_unisys_acpi_oem_table from .text between 'acpi_madt_oem_check' (at offset 0xc0101eda) and 'enable_apic_mode'
WARNING: vmlinux - Section mismatch: reference to .init.text:acpi_get_table_header_early from .text between 'acpi_madt_oem_check' (at offset 0xc0101ef0) and 'enable_apic_mode'
WARNING: vmlinux - Section mismatch: reference to .init.text:parse_unisys_oem from .text between 'acpi_madt_oem_check' (at offset 0xc0101f2e) and 'enable_apic_mode'
WARNING: vmlinux - Section mismatch: reference to .init.text:setup_unisys from .text between 'acpi_madt_oem_check' (at offset 0xc0101f37) and 'enable_apic_mode'WARNING: vmlinux - Section mismatch: reference to .init.text:parse_unisys_oem from .text between 'mps_oem_check' (at offset 0xc0101ec7) and 'acpi_madt_oem_check'
WARNING: vmlinux - Section mismatch: reference to .init.text:es7000_sw_apic from .text between 'enable_apic_mode' (at offset 0xc0101f48) and 'check_apicid_present'

o Some functions which are inline (acpi_madt_oem_check) are not inlined by
  compiler as these functions are accessed using function pointer. These
  functions are put in .text section and they in-turn access __init type
  functions hence modpost generates warnings.

o Do not iniline acpi_madt_oem_check, instead make it __init.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[PATCH] x86-64: Set HASHDIST_DEFAULT to 1 for x86_64 NUMA
Ravikiran G Thirumalai [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: Set HASHDIST_DEFAULT to 1 for x86_64 NUMA

Enable system hashtable memory to be distributed among nodes on x86_64 NUMA

Forcing the kernel to use node interleaved vmalloc instead of bootmem for
the system hashtable memory (alloc_large_system_hash) reduces the memory
imbalance on node 0 by around 40MB on a 8 node x86_64 NUMA box:

Before the following patch, on bootup of a 8 node box:

Node 0 MemTotal:      3407488 kB
Node 0 MemFree:       3206296 kB
Node 0 MemUsed:        201192 kB
Node 0 Active:           7012 kB
Node 0 Inactive:          512 kB
Node 0 Dirty:               0 kB
Node 0 Writeback:           0 kB
Node 0 FilePages:        1912 kB
Node 0 Mapped:            420 kB
Node 0 AnonPages:        5612 kB
Node 0 PageTables:        468 kB
Node 0 NFS_Unstable:        0 kB
Node 0 Bounce:              0 kB
Node 0 Slab:             5408 kB
Node 0 SReclaimable:      644 kB
Node 0 SUnreclaim:       4764 kB

After the patch (or using hashdist=1 on the kernel command line):

Node 0 MemTotal:      3407488 kB
Node 0 MemFree:       3247608 kB
Node 0 MemUsed:        159880 kB
Node 0 Active:           3012 kB
Node 0 Inactive:          616 kB
Node 0 Dirty:               0 kB
Node 0 Writeback:           0 kB
Node 0 FilePages:        2424 kB
Node 0 Mapped:            380 kB
Node 0 AnonPages:        1200 kB
Node 0 PageTables:        396 kB
Node 0 NFS_Unstable:        0 kB
Node 0 Bounce:              0 kB
Node 0 Slab:             6304 kB
Node 0 SReclaimable:     1596 kB
Node 0 SUnreclaim:       4708 kB

I guess it is a good idea to keep HASHDIST_DEFAULT "on" for x86_64 NUMA
since x86_64 has no dearth of vmalloc space?  Or maybe enable hash
distribution for all 64bit NUMA arches?  The following patch does it only
for x86_64.

I ran a HPC MPI benchmark -- 'Ansys wingsolid', which takes up quite a bit of
memory and uses up tlb entries.  This was on a 4 way, 2 socket
Tyan AMD box (non vsmp), with 8G total memory (4G pernode).

The results with and without hash distribution are:

1. Vanilla - runtime of 1188.000s
2. With hashdist=1 runtime of 1154.000s

Oprofile output for the duration of run is:

1. Vanilla:
PU: AMD64 processors, speed 2411.16 MHz (estimated)
Counted L1_AND_L2_DTLB_MISSES events (L1 and L2 DTLB misses) with a unit
mask of 0x00 (No unit mask) count 500
samples  %        app name                 symbol name
163054    6.5513  libansys1.so             MultiFront::decompose(int, int,
Elemset *, int *, int, int, int)
162061    6.5114  libansys3.so             blockSaxpy6L_fd
162042    6.5107  libansys3.so             blockInnerProduct6L_fd
156286    6.2794  libansys3.so             maxb33_
87879     3.5309  libansys1.so             elmatrixmultpcg_
84857     3.4095  libansys4.so             saxpy_pcg
58637     2.3560  libansys4.so             .st4560
46612     1.8728  libansys4.so             .st4282
43043     1.7294  vmlinux-t                copy_user_generic_string
41326     1.6604  libansys3.so             blockSaxpyBackSolve6L_fd
41288     1.6589  libansys3.so             blockInnerProductBackSolve6L_fd

2. With hashdist=1
CPU: AMD64 processors, speed 2411.13 MHz (estimated)
Counted L1_AND_L2_DTLB_MISSES events (L1 and L2 DTLB misses) with a unit
mask of 0x00 (No unit mask) count 500
samples  %        app name                 symbol name
162993    6.9814  libansys1.so             MultiFront::decompose(int, int,
Elemset *, int *, int, int, int)
160799    6.8874  libansys3.so             blockInnerProduct6L_fd
160459    6.8729  libansys3.so             blockSaxpy6L_fd
156018    6.6826  libansys3.so             maxb33_
84700     3.6279  libansys4.so             saxpy_pcg
83434     3.5737  libansys1.so             elmatrixmultpcg_
58074     2.4875  libansys4.so             .st4560
46000     1.9703  libansys4.so             .st4282
41166     1.7632  libansys3.so             blockSaxpyBackSolve6L_fd
41033     1.7575  libansys3.so             blockInnerProductBackSolve6L_fd
35762     1.5318  libansys1.so             inner_product_sub
35591     1.5245  libansys1.so             inner_product_sub2
28259     1.2104  libansys4.so             addVectors

Signed-off-by: Pravin B. Shelar <pravin.shelar@calsoftinc.com>
Signed-off-by: Ravikiran Thirumalai <kiran@scalex86.org>
Signed-off-by: Shai Fultheim <shai@scalex86.org>
Signed-off-by: Andi Kleen <ak@suse.de>
Acked-by: Christoph Lameter <clameter@engr.sgi.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[PATCH] x86-64: Minor white space cleanup in traps.c
Andi Kleen [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: Minor white space cleanup in traps.c

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Allow sys_uselib unconditionally
Andi Kleen [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: Allow sys_uselib unconditionally

Previously it wasn't enabled in the binfmt_aout is a module case.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Don't disable basic block reordering
Andi Kleen [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: Don't disable basic block reordering

When compiling with -Os (which is default) the compiler defaults to it
anyways. And with -O2 it probably generates somewhat better (although
also larger) code.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: fix x86_64-mm-sched-clock-share
Andrew Morton [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: fix x86_64-mm-sched-clock-share

Fix for the following patch. Provide dummy cpufreq functions when
CPUFREQ is not compiled in.

Cc: Andi Kleen <ak@suse.de>
Cc: Dave Jones <davej@codemonkey.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Move cpu verification code to common file
Vivek Goyal [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: Move cpu verification code to common file

o This patch moves the code to verify long mode and SSE to a common file.
  This code is now shared by trampoline.S, wakeup.S, boot/setup.S and
  boot/compressed/head.S

o So far we used to do very limited check in trampoline.S, wakeup.S and
  in 32bit entry point. Now all the entry paths are forced to do the
  exhaustive check, including SSE because verify_cpu is shared.

o I am keeping this patch as last in the x86 relocatable series because
  previous patches have got quite some amount of testing done and don't want
  to distrub that. So that if there is problem introduced by this patch, at
  least it can be easily isolated.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Extend bzImage protocol for relocatable bzImage
Vivek Goyal [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: Extend bzImage protocol for relocatable bzImage

o Extend the bzImage protocol (same as i386) to allow bzImage loaders to
  load the protected mode kernel at non-1MB address. Now protected mode
  component is relocatable and can be loaded at non-1MB addresses.

o As of today kdump uses it to run a second kernel from a reserved memory
  area.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: build-time checking
Vivek Goyal [Wed, 2 May 2007 17:27:08 +0000 (19:27 +0200)]
[PATCH] x86-64: build-time checking

o X86_64 kernel should run from 2MB aligned address for two reasons.
- Performance.
- For relocatable kernels, page tables are updated based on difference
  between compile time address and load time physical address.
  This difference should be multiple of 2MB as kernel text and data
  is mapped using 2MB pages and PMD should be pointing to a 2MB
  aligned address. Life is simpler if both compile time and load time
  kernel addresses are 2MB aligned.

o Flag the error at compile time if one is trying to build a kernel which
  does not meet alignment restrictions.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[PATCH] x86-64: Relocatable Kernel Support
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: Relocatable Kernel Support

This patch modifies the x86_64 kernel so that it can be loaded and run
at any 2M aligned address, below 512G.  The technique used is to
compile the decompressor with -fPIC and modify it so the decompressor
is fully relocatable.  For the main kernel the page tables are
modified so the kernel remains at the same virtual address.  In
addition a variable phys_base is kept that holds the physical address
the kernel is loaded at.  __pa_symbol is modified to add that when
we take the address of a kernel symbol.

When loaded with a normal bootloader the decompressor will decompress
the kernel to 2M and it will run there.  This both ensures the
relocation code is always working, and makes it easier to use 2M
pages for the kernel and the cpu.

AK: changed to not make RELOCATABLE default in Kconfig

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: __pa and __pa_symbol address space separation
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86: __pa and __pa_symbol address space separation

Currently __pa_symbol is for use with symbols in the kernel address
map and __pa is for use with pointers into the physical memory map.
But the code is implemented so you can usually interchange the two.

__pa which is much more common can be implemented much more cheaply
if it is it doesn't have to worry about any other kernel address
spaces.  This is especially true with a relocatable kernel as
__pa_symbol needs to peform an extra variable read to resolve
the address.

There is a third macro that is added for the vsyscall data
__pa_vsymbol for finding the physical addesses of vsyscall pages.

Most of this patch is simply sorting through the references to
__pa or __pa_symbol and using the proper one.  A little of
it is continuing to use a physical address when we have it
instead of recalculating it several times.

swapper_pgd is now NULL.  leave_mm now uses init_mm.pgd
and init_mm.pgd is initialized at boot (instead of compile time)
to the physmem virtual mapping of init_level4_pgd.  The
physical address changed.

Except for the for EMPTY_ZERO page all of the remaining references
to __pa_symbol appear to be during kernel initialization.  So this
should reduce the cost of __pa in the common case, even on a relocated
kernel.

As this is technically a semantic change we need to be on the lookout
for anything I missed.  But it works for me (tm).

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: do not use virt_to_page on kernel data address
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: do not use virt_to_page on kernel data address

o virt_to_page() call should be used on kernel linear addresses and not
  on kernel text and data addresses. Swsusp code uses it on kernel data
  (statically allocated swsusp_header).

o Allocate swsusp_header dynamically so that virt_to_page() can be used
  safely.

o I am changing this because in next few patches, __pa() on x86_64 will
  no longer support kernel text and data addresses and hibernation breaks.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: Move swsusp __pa() dependent code to arch portion
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86: Move swsusp __pa() dependent code to arch portion

o __pa() should be used only on kernel linearly mapped virtual addresses
  and not on kernel text and data addresses.

o Hibernation code needs to determine the physical address associated
  with kernel symbol to mark a section boundary which contains pages which
  don't have to be saved and restored during hibernate/resume operation.

o Move this piece of code in arch dependent section. So that architectures
  which don't have kernel text/data mapped into kernel linearly mapped
  region can come up with their own ways of determining physical addresses
  associated with a kernel text.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Remove the identity mapping as early as possible
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: Remove the identity mapping as early as possible

With the rewrite of the SMP trampoline and the early page
allocator there is nothing that needs identity mapped pages,
once we start executing C code.

So add zap_identity_mappings into head64.c and remove
zap_low_mappings() from much later in the code.  The functions
 are subtly different thus the name change.

This also kills boot_level4_pgt which was from an earlier
attempt to move the identity mappings as early as possible,
and is now no longer needed.  Essentially I have replaced
boot_level4_pgt with trampoline_level4_pgt in trampoline.S

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Modify discover_ebda to use virtual addresses
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: Modify discover_ebda to use virtual addresses

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: 64bit ACPI wakeup trampoline
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: 64bit ACPI wakeup trampoline

o Moved wakeup_level4_pgt into the wakeup routine so we can
  run the kernel above 4G.

o Now we first go to 64bit mode and continue to run from trampoline and
  then then start accessing kernel symbols and restore processor context.
  This enables us to resume even in relocatable kernel context when
  kernel might not be loaded at physical addr it has been compiled for.

o Removed the need for modifying any existing kernel page table.

o Increased the size of the wakeup routine to 8K. This is required as
  wake page tables are on trampoline itself and they got to be at 4K
  boundary, hence one page is not sufficient.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: wakeup.S misc cleanups
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: wakeup.S misc cleanups

o Various cleanups. One of the main purpose of cleanups is that make
  wakeup.S as close as possible to trampoline.S.

o Following are the changes
- Indentations for comments.
- Changed the gdt table to compact form and to resemble the
  one in trampoline.S
- Take the jump to 32bit from real mode using ljmpl. Makes code
  more readable.
- After enabling long mode, directly take a long jump for 64bit
  mode. No need to take an extra jump to "reach_comaptibility_mode"
- Stack is not used after real mode. So don't load stack in
    32 bit mode.
- No need to enable PGE here.
- No need to do extra EFER read, anyway we trash the read contents.
- No need to enable system call (EFER_SCE). Anyway it will be
  enabled when original EFER is restored.
- No need to set MP, ET, NE, WP, AM bits in cr0. Very soon we will
     reload the original cr0 while restroing the processor state.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: wakeup.S rename registers to reflect right names
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: wakeup.S rename registers to reflect right names

o Use appropriate names for 64bit regsiters.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Get rid of dead code in suspend resume
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: Get rid of dead code in suspend resume

o Get rid of dead code in wakeup.S

o We never restore from saved_gdt, saved_idt, saved_ltd, saved_tss, saved_cr3,
  saved_cr4, saved_cr0, real_save_gdt, saved_efer, saved_efer2. Get rid
  of of associated code.

o Get rid of bogus_magic, bogus_31_magic and bogus_magic2. No longer being
  used.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: 64bit PIC SMP trampoline
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: 64bit PIC SMP trampoline

This modifies the SMP trampoline and all of the associated code so
it can jump to a 64bit kernel loaded at an arbitrary address.

The dependencies on having an idenetity mapped page in the kernel
page tables for SMP bootup have all been removed.

In addition the trampoline has been modified to verify
that long mode is supported.  Asking if long mode is implemented is
down right silly but we have traditionally had some of these checks,
and they can't hurt anything.  So when the totally ludicrous happens
we just might handle it correctly.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Add EFER to the register set saved by save_processor_state
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: Add EFER to the register set saved by save_processor_state

EFER varies like %cr4 depending on the cpu capabilities, and which cpu
capabilities we want to make use of.  So save/restore it make certain
we have the same EFER value when we are done.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: cleanup segments
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: cleanup segments

Move __KERNEL32_CS up into the unused gdt entry.  __KERNEL32_CS is
used when entering the kernel so putting it first is useful when
trying to keep boot gdt sizes to a minimum.

Set the accessed bit on all gdt entries.  We don't care
so there is no need for the cpu to burn the extra cycles,
and it potentially allows the pages to be immutable.  Plus
it is confusing when debugging and your gdt entries mysteriously
change.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: modify copy_bootdata to use virtual addresses
Vivek Goyal [Wed, 2 May 2007 17:27:07 +0000 (19:27 +0200)]
[PATCH] x86-64: modify copy_bootdata to use virtual addresses

Use virtual addresses instead of physical addresses
in copy bootdata.  In addition fix the implementation
of the old bootloader convention.  Everything is
at real_mode_data always.  It is just that sometimes
real_mode_data was relocated by setup.S to not sit at
0x90000.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Fix early printk to use standard ISA mapping
Vivek Goyal [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: Fix early printk to use standard ISA mapping

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Clean up the early boot page table
Vivek Goyal [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: Clean up the early boot page table

- Merge physmem_pgt and ident_pgt, removing physmem_pgt.  The merge
  is broken as soon as mm/init.c:init_memory_mapping is run.
- As physmem_pgt is gone don't export it in pgtable.h.
- Use defines from pgtable.h for page permissions.
- Fix the physical memory identity mapping so it is at the correct
  address.
- Remove the physical memory mapping from wakeup_level4_pgt it
  is at the wrong address so we can't possibly be usinging it.
- Simply NEXT_PAGE the work to calculate the phys_ alias
  of the labels was very cool.  Unfortuantely it was a brittle
  special purpose hack that makes maitenance more difficult.
  Instead just use label - __START_KERNEL_map like we do
  everywhere else in assembly.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Kill temp boot pmds
Vivek Goyal [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: Kill temp boot pmds

Early in the boot process we need the ability to set
up temporary mappings, before our normal mechanisms are
initialized.  Currently this is used to map pages that
are part of the page tables we are building and pages
during the dmi scan.

The core problem is that we are using the user portion of
the page tables to implement this.  Which means that while
this mechanism is active we cannot catch NULL pointer dereferences
and we deviate from the normal ways of handling things.

In this patch I modify early_ioremap to map pages into
the kernel portion of address space, roughly where
we will later put modules, and I make the discovery of
which addresses we can use dynamic which removes all
kinds of static limits and remove the dependencies
on implementation details between different parts of the code.

Now alloc_low_page() and unmap_low_page() use
early_iomap() and early_iounmap() to allocate/map and
unmap a page.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Assembly safe page.h and pgtable.h
Vivek Goyal [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: Assembly safe page.h and pgtable.h

This patch makes pgtable.h and page.h safe to include
in assembly files like head.S.  Allowing us to use
symbolic constants instead of hard coded numbers when
refering to the page tables.

This patch copies asm-sparc64/const.h to asm-x86_64 to
get a definition of _AC() a very convinient macro that
allows us to force the type when we are compiling the
code in C and to drop all of the type information when
we are using the constant in assembly.  Previously this
was done with multiple definition of the same constant.
const.h was modified slightly so that it works when given
CONFIG options as arguments.

This patch adds #ifndef __ASSEMBLY__ ... #endif
and _AC(1,UL) where appropriate so the assembler won't
choke on the header files.  Otherwise nothing
should have changed.

AK: added const.h to exported headers to fix headers_check

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: dma_ops as const
Stephen Hemminger [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: dma_ops as const

The dma_ops structure can be const since it never changes
after boot.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Simplify smp_call_function*() by using common implementation
Jeremy Fitzhardinge [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] i386: Simplify smp_call_function*() by using common implementation

smp_call_function and smp_call_function_single are almost complete
duplicates of the same logic.  This patch combines them by
implementing them in terms of the more general
smp_call_function_mask().

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Stephane Eranian <eranian@hpl.hp.com>
Cc: Andrew Morton <akpm@osdl.org>
Cc: Andi Kleen <ak@suse.de>
Cc: "Randy.Dunlap" <rdunlap@xenotime.net>
Cc: Ingo Molnar <mingo@elte.hu>
17 years ago[PATCH] x86-64: fix cpu MHz reporting on constant_tsc cpus
Joerg Roedel [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: fix cpu MHz reporting on constant_tsc cpus

This patch fixes the reporting of cpu_mhz in /proc/cpuinfo on CPUs with
a constant TSC rate and a kernel with disabled cpufreq.

Signed-off-by: Mark Langsdorf <mark.langsdorf@amd.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Andi Kleen <ak@suse.de>
 arch/x86_64/kernel/apic.c     |    2 -
 arch/x86_64/kernel/time.c     |   58 +++++++++++++++++++++++++++++++++++++++---
 arch/x86_64/kernel/tsc.c      |   12 +++++---
 arch/x86_64/kernel/tsc_sync.c |    2 -
 include/asm-x86_64/proto.h    |    1
 5 files changed, 65 insertions(+), 10 deletions(-)

17 years ago[PATCH] x86-64: Remove duplicated code for reading control registers
Glauber de Oliveira Costa [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: Remove duplicated code for reading control registers

On Tue, Mar 13, 2007 at 05:33:09AM -0700, Randy.Dunlap wrote:
> On Tue, 13 Mar 2007, Glauber de Oliveira Costa wrote:
>
> > Tiny cleanup:
> >
> > In x86_64, the same functions for reading cr3 and writing cr{3,4} are
> > defined in tlbflush.h and system.h, whith just a name change.
> > The only difference is the clobbering of memory, which seems a safe, and
> > even needed change for the write_cr4. This patch removes the duplicate.
> > write_cr3() is moved to system.h for consistency.
>
> missing patch.....
>
thanks. Attached now

--
Glauber de Oliveira Costa
Red Hat Inc.
"Free as in Freedom"

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Remove unused set_seg_base
Rusty Russell [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86-64: Remove unused set_seg_base

The set_seg_base function isn't used anywhere (2.6.21-rc3-git1)

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Fix usage of -mtune when X86_GENERIC=y or CONFIG_MCORE2=y
Lasse Collin [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] i386: Fix usage of -mtune when X86_GENERIC=y or CONFIG_MCORE2=y

Hi!

I sent this simple patch to lkml about two weeks ago and also cc'ed
to Linus, but seems that the patch got ignored. I decided to write to
you, because you have modified the relevant file most recently.

Below is a copy of the mail that is also available at
<http://lkml.org/lkml/2007/2/28/230>.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: clean up mach_reboot_fixups
Jeremy Fitzhardinge [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] i386: clean up mach_reboot_fixups

The reboot_fixups stuff seems to be a bit of a mess, specifically the
header is in linux/ when its a purely i386-specific piece of code.  I'm
not sure why it has its config option; its only currently needed for
"geode-gx1/cs5530a", so perhaps whatever config option controls that
hardware should enable this?

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Update __copy_to_user_inatomic linuxdoc description
Aneesh Kumar K.V [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] i386: Update __copy_to_user_inatomic linuxdoc description

Explicity specify that the caller should pin the user memory
otherwise the function will sleep

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@gmail.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: Drop cc-options call for all options supported in gcc 3.2+
Andi Kleen [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] x86: Drop cc-options call for all options supported in gcc 3.2+

The kernel only supports gcc 3.2+ now so it doesn't make sense
anymore to explicitely check for options this compiler version
already has.

This actually fixes a bug. The -mprefered-stack-boundary check
never worked because gcc rightly complains

  CC      arch/i386/kernel/asm-offsets.s
cc1: -mpreferred-stack-boundary=2 is not between 4 and 12

We just never saw the error because of cc-options.
I changed it to 4 to actually work.

Tested by compiling i386 and x86-64 defconfig with gcc 3.2.

Should speed up the build time a tiny bit and improve
stack usage on i386 slightly.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Support Oprofile for AMD Family 10 CPUs
Andi Kleen [Wed, 2 May 2007 17:27:06 +0000 (19:27 +0200)]
[PATCH] i386: Support Oprofile for AMD Family 10 CPUs

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Correct max number of CPUs in Kconfig
Andi Kleen [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] x86-64: Correct max number of CPUs in Kconfig

Pointed out by Adrian Bunk

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Change sysenter_setup to __cpuinit & improve __INIT, __INITDATA
Prarit Bhargava [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: Change sysenter_setup to __cpuinit & improve __INIT, __INITDATA

Change sysenter_setup to __cpuinit.
Change __INIT & __INITDATA to be cpu hotplug aware.

Resolve MODPOST warnings similar to:

WARNING: vmlinux - Section mismatch: reference to .init.text:sysenter_setup from
 .text between 'identify_cpu' (at offset 0xc040a380) and 'detect_ht'

and

WARNING: vmlinux - Section mismatch: reference to .init.data:vsyscall_int80_end
from .text between 'sysenter_setup' (at offset 0xc041a269) and 'enable_sep_cpu'
WARNING: vmlinux - Section mismatch: reference to
.init.data:vsyscall_int80_start from .text between 'sysenter_setup' (at offset
0xc041a26e) and 'enable_sep_cpu'
WARNING: vmlinux - Section mismatch: reference to
.init.data:vsyscall_sysenter_end from .text between 'sysenter_setup' (at offset
0xc041a275) and 'enable_sep_cpu'
WARNING: vmlinux - Section mismatch: reference to
.init.data:vsyscall_sysenter_start from .text between 'sysenter_setup' (at
offset 0xc041a27a) and 'enable_sep_cpu'

Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Add __init to probe_bigsmp
Prarit Bhargava [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: Add __init to probe_bigsmp

Add __init to probe_bigsmp.  All callers are __init and data being examined
is __initdata.

Resolves MODPOST warning similar to:

WARNING: vmlinux - Section mismatch: reference to .init.data: from .text between 'probe_bigsmp' (at offset 0xc0401e56) and 'init_apic_ldr'

Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: x86_64 make NMI use PERFCTR1 for architectural perfmon (take 2)
Stephane Eranian [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] x86-64: x86_64 make NMI use PERFCTR1 for architectural perfmon (take 2)

Hello,

This patch against 2.6.20-git14 makes the NMI watchdog use PERFSEL1/PERFCTR1
instead of PERFSEL0/PERFCTR0 on processors supporting Intel architectural
perfmon, such as Intel Core 2. Although all PMU events can work on
both counters, the Precise Event-Based Sampling (PEBS) requires that the
event be in PERFCTR0 to work correctly (see section 18.14.4.1 in the
IA32 SDM Vol 3b). This versions has 3 chunks compared to previous where
we had missed on check.

Changelog:
        - make the x86-64 NMI watchdog use PERFSEL1/PERFCTR1 instead of PERFSEL0/PERFCTR0
          on processors supporting the Intel architectural perfmon (e.g. Core 2 Duo).
          This allows PEBS to work when the NMI watchdog is active.

signed-off-by: stephane eranian <eranian@hpl.hp.com>

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: i386 make NMI use PERFCTR1 for architectural perfmon (take 2)
Stephane Eranian [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: i386 make NMI use PERFCTR1 for architectural perfmon (take 2)

Hello,

This patch against 2.6.20-git14 makes the NMI watchdog use PERFSEL1/PERFCTR1
instead of PERFSEL0/PERFCTR0 on processors supporting Intel architectural
perfmon, such as Intel Core 2. Although all PMU events can work on
both counters, the Precise Event-Based Sampling (PEBS) requires that the
event be in PERFCTR0 to work correctly (see section 18.14.4.1 in the
IA32 SDM Vol 3b).

A similar patch for x86-64 is to follow.

Changelog:
        - make the i386 NMI watchdog use PERFSEL1/PERFCTR1 instead of PERFSEL0/PERFCTR0
          on processors supporting the Intel architectural perfmon (e.g. Core 2 Duo).
          This allows PEBS to work when the NMI watchdog is active.

signed-off-by: stephane eranian <eranian@hpl.hp.com>

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Some cleanup in time.c
Andi Kleen [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] x86-64: Some cleanup in time.c

Move prototypes into header files
Remove unneeded includes.

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: Fix i386 and x86_64 fault information pollution
Andi Kleen [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] x86: Fix i386 and x86_64 fault information pollution

a userspace fault or a kernelspace fault which will result in the
immediate death of the process.  They should not be filled in as a
result of a kernelspace fault which can be fixed up.

Otherwise, if the process is handling SIGSEGV and examining the fault
information, this can result in the kernel space fault trashing the
previously stored fault information if it arrives between the
userspace fault happening and the SIGSEGV being delivered to the process.

Signed-off-by: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Acked-by: Jan Beulich <jbeulich@novell.com>
--
 arch/i386/kernel/traps.c   |   24 ++++++++++++++++++------
 arch/x86_64/kernel/traps.c |   30 +++++++++++++++++++++++-------
 2 files changed, 41 insertions(+), 13 deletions(-)

17 years ago[PATCH] i386: Add dwarf2 annotations to *_user and checksum functions
Jan Beulich [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: Add dwarf2 annotations to *_user and checksum functions

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: a few missing entry.S annotations
Jan Beulich [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] x86-64: a few missing entry.S annotations

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: probe_roms() cleanup
Rene Herman [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: probe_roms() cleanup

Remove the assumption that if the first page of a legacy ROM is mapped,
it'll all be mapped. This'll also stop people reading this code from
wondering if they're looking at a bug...

Signed-off-by: Rene Herman <rene.herman@gmail.com>
Signed-off-by: Martin Murray <murrayma@citi.umich.edu>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: Zachary Amsden <zach@vmware.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[PATCH] i386: Add an option for the VIA C7 which sets appropriate L1 cache
Simon Arlott [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: Add an option for the VIA C7 which sets appropriate L1 cache

The VIA C7 is a 686 (with TSC) that supports MMX, SSE and SSE2, it also has
a cache line length of 64 according to
http://www.digit-life.com/articles2/cpu/rmma-via-c7.html.  This patch sets
gcc to -march=686 and select s the correct cache shift.

Signed-off-by: Simon Arlott <simon@fire.lp0.eu>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Cc: Dave Jones <davej@codemonkey.org.uk>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[PATCH] i386: pit_latch_buggy has no effect
takada [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: pit_latch_buggy has no effect

Eliminated the arch/i386/kernel/timers in 2.6.18, use clocksoures instead.
pit_latch_buggy was referred in timers/timer_tsc.c, and currently removed.
Therefore nobody refer it.

Until 2.6.17, MediaGX's TSC works correctly.  after 2.6.18, warned "TSC
appears to be running slowly.  Marking it as unstable".  So marked unstable
TSC when CS55x0.

Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: mtrr range check correction
Jan Beulich [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: mtrr range check correction

Whether a region is below 1Mb is determined by its start rather than
its end.

This hunk got erroneously dropped from a previous patch.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
17 years ago[PATCH] i386: No need to use -traditional for processing asm in i386/kernel/
Jeremy Fitzhardinge [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] i386: No need to use -traditional for processing asm in i386/kernel/

No need to use -traditional for processing asm in i386/kernel/

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: consolidate smp_send_stop()
Jan Beulich [Wed, 2 May 2007 17:27:05 +0000 (19:27 +0200)]
[PATCH] x86: consolidate smp_send_stop()

Synchronize i386's smp_send_stop() with x86-64's in only try-locking
the call lock to prevent deadlocks when called from panic().
In both version, disable interrupts before clearing the CPU off the
online map to eliminate races with IRQ handlers inspecting this map.
Also in both versions, save/restore interrupts rather than disabling/
enabling them.
On x86-64, eliminate one function used here by folding it into its
single caller, convert to static, and rename for consistency with i386
(lkcd may like this).

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: adjust inclusion of asm/vsyscall32.h
Jan Beulich [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86-64: adjust inclusion of asm/vsyscall32.h

Avoid including asm/vsyscall32.h in virtually every source file.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: adjust inclusion of asm/fixmap.h
Jan Beulich [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86: adjust inclusion of asm/fixmap.h

Move inclusion of asm/fixmap.h to where it is really used rather than
where it may have been used long ago (requires a few other adjustments
to includes due to previous implicit dependencies).

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: adjustments to page table dump during oops (v4)
Jan Beulich [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] i386: adjustments to page table dump during oops (v4)

- make the page table contents printing PAE capable
- make sure the address stored in current->thread.cr2 is unmodified
  from what was read from CR2
- don't call oops_may_print() multiple times, when one time suffices
- print pte even in highpte case, as long as the pte page isn't in
  actually in high memory (which is specifically the case for all page
  tables covering kernel space)

(Changes to v3: Use sizeof()*2 rather than the suggested sizeof()*4 for
printing width, use fixed 16-nibble width for PAE, and also apply the
max_low_pfn range check to the middle level lookup on PAE.)

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86: default to physical mode on hotplug CPU kernels
Ingo Molnar [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86: default to physical mode on hotplug CPU kernels

Default to physical mode on hotplug CPU kernels.  Furher simplify and clean up
the APIC initialization code.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Li, Shaohua" <shaohua.li@intel.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86-64: remove clustered APIC mode
Ingo Molnar [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86-64: remove clustered APIC mode

Remove now unused clustered APIC mode code.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Li, Shaohua" <shaohua.li@intel.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86-64: always use physical delivery mode on > 8 CPUs
Ingo Molnar [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86-64: always use physical delivery mode on > 8 CPUs

Remove clustered APIC mode.  There's little point in the use of clustered APIC
mode, broadcasting is limited to within the cluster only, and chipsets have
bugs in this area as well.  So default to physical APIC mode when the CPU
count is large, and default to logical APIC mode when the CPU count is 8 or
smaller.

(this patch only removes the use of genapic_cluster and cleans up the
resulting genapic.c file - removal of all remaining traces of clustered
mode will be done by another patch.)

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Li, Shaohua" <shaohua.li@intel.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86-64: optimize & fix APIC mode setup
Ingo Molnar [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86-64: optimize & fix APIC mode setup

Fix a couple of inconsistencies/problems I found while reviewing the x86_64
genapic code (when I was chasing mysterious eth0 timeouts that would only
trigger if CPU_HOTPLUG is enabled):

 - AMD systems defaulted to the slower flat-physical mode instead
   of the flat-logical mode. The only restriction on AMD systems
   is that they should not use clustered APIC mode.

 - removed the CPU hotplug hacks, switching the default for small
   systems back from phys-flat to logical-flat. The switching to logical
   flat mode on small systems fixed sporadic ethernet driver timeouts i
   was getting on a dual-core Athlon64 system:

    NETDEV WATCHDOG: eth0: transmit timed out
    eth0: Transmit timeout, status 0c 0005 c07f media 80.
    eth0: Tx queue start entry 32  dirty entry 28.
    eth0:  Tx descriptor 0 is 0008a04a. (queue head)
    eth0:  Tx descriptor 1 is 0008a04a.
    eth0:  Tx descriptor 2 is 0008a04a.
    eth0:  Tx descriptor 3 is 0008a04a.
    eth0: link up, 100Mbps, full-duplex, lpa 0xC5E1

 - The use of '<= 8' was a bug by itself (the valid APIC ids
   for logical flat mode go from 0 to 7, not 0 to 8). The new logic
   is to use logical flat mode on both AMD and Intel systems, and
   to only switch to physical mode when logical mode cannot be used.
   If CPU hotplug is racy wrt. APIC shutdown then CPU hotplug needs
   fixing, not the whole IRQ system be made inconsistent and slowed
   down.

 - minor cleanups: simplified some code constructs

build & booted on a couple of AMD and Intel SMP systems.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andi Kleen <ak@suse.de>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Li, Shaohua" <shaohua.li@intel.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
17 years ago[PATCH] x86: revert x86_64-mm-fix-the-irqbalance-quirk-for-e7320-e7520-e7525
Andrew Morton [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86: revert x86_64-mm-fix-the-irqbalance-quirk-for-e7320-e7520-e7525

Obsoleted by Ingo's genapic stuff.

Cc: Ingo Molnar <mingo@elte.hu>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Li, Shaohua" <shaohua.li@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: revert x86_64-mm-add-genapic_force
Andrew Morton [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86-64: revert x86_64-mm-add-genapic_force

This is obsoleted by new Ingo genapic patches.

Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Li, Shaohua" <shaohua.li@intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: revert i386-fix-the-verify_quirk_intel_irqbalance
Andrew Morton [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] i386: revert i386-fix-the-verify_quirk_intel_irqbalance

This is unneeded with Ingo's genapic rework.

Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Andi Kleen <ak@suse.de>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] i386: Update defconfig
Andi Kleen [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] i386: Update defconfig

Signed-off-by: Andi Kleen <ak@suse.de>
17 years ago[PATCH] x86-64: Update defconfig
Andi Kleen [Wed, 2 May 2007 17:27:04 +0000 (19:27 +0200)]
[PATCH] x86-64: Update defconfig

Signed-off-by: Andi Kleen <ak@suse.de>
17 years agolibata: honour host controllers that want just one host
Linus Torvalds [Tue, 1 May 2007 00:43:48 +0000 (17:43 -0700)]
libata: honour host controllers that want just one host

The Marvell IDE interface on my machine would hit a BUG_ON() in
lib/iomem.c because it was calling ata_pci_init_one() specifying just a
single port on the host, but that would actually end up trying to
initialize two ports, the second one with bogus information.

This fixes "ata_pci_init_one()" so that it actually passes down the
n_ports variable that it got from the low-level driver to the host
allocation routine ("ata_host_alloc_pinfo()"), which results in the ATA
layer actually having the correct port number information.

And in order to make it all work, I also needed to fix a few places that
had incorrectly hard-coded the fact that a host always had exactly two
ports (both ata_pci_init_bmdma() and ata_request_legacy_irqs() would
just always iterate over both ports).

Acked-by: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopm: include EIO from errno-base.h
David Rientjes [Mon, 30 Apr 2007 22:09:56 +0000 (15:09 -0700)]
pm: include EIO from errno-base.h

For backwards compatibility, call_platform_enable_wakeup() can return 0
instead of -EIO since we aren't guaranteed to have errno defined.

Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: David Rientjes <rientjes@google.com>
Cc: "Randy.Dunlap" <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoAdd kvasprintf()
Jeremy Fitzhardinge [Mon, 30 Apr 2007 22:09:56 +0000 (15:09 -0700)]
Add kvasprintf()

Add a kvasprintf() function to complement kasprintf().

No in-tree users yet, but I have some coming up.

[akpm@linux-foundation.org: EXPORT it]
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Keir Fraser <keir@xensource.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopower management: force pm_ops.valid callback to be assigned
Johannes Berg [Mon, 30 Apr 2007 22:09:55 +0000 (15:09 -0700)]
power management: force pm_ops.valid callback to be assigned

This patch changes the docs and behaviour from "all states valid" to "no
states valid" if no .valid callback is assigned.  Users of pm_ops that only
need mem sleep can assign pm_valid_only_mem without any overhead, others
will require more elaborate callbacks.

Now that all users of pm_ops have a .valid callback this is a safe thing to
do and prevents things from getting messy again as they were before.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: Pavel Machek <pavel@ucw.cz>
Looks-okay-to: Rafael J. Wysocki <rjw@sisk.pl>
Cc: <linux-pm@lists.linux-foundation.org>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopower management: implement pm_ops.valid for everybody
Johannes Berg [Mon, 30 Apr 2007 22:09:54 +0000 (15:09 -0700)]
power management: implement pm_ops.valid for everybody

Almost all users of pm_ops only support mem sleep, don't check in .valid and
don't reject any others in .prepare so users can be confused if they check
/sys/power/state, especially when new states are added (these would then
result in s-t-r although they're supposed to be something different).

This patch implements a generic pm_valid_only_mem function that is then
exported for users and puts it to use in almost all existing pm_ops.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: David Brownell <david-b@pacbell.net>
Acked-by: Pavel Machek <pavel@ucw.cz>
Cc: linux-pm@lists.linux-foundation.org
Cc: Len Brown <lenb@kernel.org>
Acked-by: Russell King <rmk@arm.linux.org.uk>
Cc: Greg KH <greg@kroah.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agopower management: remove firmware disk mode
Johannes Berg [Mon, 30 Apr 2007 22:09:53 +0000 (15:09 -0700)]
power management: remove firmware disk mode

This patch removes the firmware disk suspend mode which is the wrong approach,
it is supposed to be used for implementing firmware-based disk suspend but
cannot actually be used for that.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: Pavel Machek <pavel@ucw.cz>
Cc: <linux-pm@lists.linux-foundation.org>
Cc: David Brownell <david-b@pacbell.net>
Cc: Len Brown <lenb@kernel.org>
Acked-by: Russell King <rmk@arm.linux.org.uk>
Cc: Greg KH <greg@kroah.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agorework pm_ops pm_disk_mode, kill misuse
Johannes Berg [Mon, 30 Apr 2007 22:09:51 +0000 (15:09 -0700)]
rework pm_ops pm_disk_mode, kill misuse

This patch series cleans up some misconceptions about pm_ops.  Some users of
the pm_ops structure attempt to use it to stop the user from entering suspend
to disk, this, however, is not possible since the user can always use
"shutdown" in /sys/power/disk and then the pm_ops are never invoked.  Also,
platforms that don't support suspend to disk simply should not allow
configuring SOFTWARE_SUSPEND (read the help text on it, it only selects
suspend to disk and nothing else, all the other stuff depends on PM).

The pm_ops structure is actually intended to provide a way to enter
platform-defined sleep states (currently supported states are "standby" and
"mem" (suspend to ram)) and additionally (if SOFTWARE_SUSPEND is configured)
allows a platform to support a platform specific way to enter low-power mode
once everything has been saved to disk.  This is currently only used by ACPI
(S4).

This patch:

The pm_ops.pm_disk_mode is used in totally bogus ways since nobody really
seems to understand what it actually does.

This patch clarifies the pm_disk_mode description.

It also removes all the arm and sh users that think they can veto suspend to
disk via pm_ops; not so since the user can always do echo shutdown >
/sys/power/disk, they need to find a better way involving Kconfig or such.

ACPI is the only user left with a non-zero pm_disk_mode.

The patch also sets the default mode to shutdown again, but when a new pm_ops
is registered its pm_disk_mode is selected as default, that way the default
stays for ACPI where it is apparently required.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: David Brownell <david-b@pacbell.net>
Acked-by: Pavel Machek <pavel@ucw.cz>
Cc: <linux-pm@lists.linux-foundation.org>
Cc: Len Brown <lenb@kernel.org>
Acked-by: Russell King <rmk@arm.linux.org.uk>
Cc: Greg KH <greg@kroah.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoreiserfs: suppress lockdep warning
Jeff Mahoney [Mon, 30 Apr 2007 22:09:50 +0000 (15:09 -0700)]
reiserfs: suppress lockdep warning

We're getting lockdep warnings due to a post-2.6.21-rc7 bugfix.

The xattr_sem can never be taken in the manner described. Internal inodes
are protected by I_PRIVATE.  Add the appropriate annotation.

Cc: <stable@kernel.org>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years agoExtend print_symbol capability
Robert Peterson [Mon, 30 Apr 2007 22:09:48 +0000 (15:09 -0700)]
Extend print_symbol capability

Today's print_symbol function dumps a kernel symbol with printk.  This
patch extends the functionality of kallsyms.c so that the symbol lookup
function may be used without the printk.  This is useful for modules that
want to dump symbols elsewhere, for example, to debugfs.  I intend to use
the new function call in the GFS2 file system (which will be a separate
patch).

[akpm@linux-foundation.org: build fix]
[clameter@sgi.com: sprint_symbol should return length of string like sprintf]
Signed-off-by: Robert Peterson <rpeterso@redhat.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Roman Zippel <zippel@linux-m68k.org>
Cc: "Randy.Dunlap" <rdunlap@xenotime.net>
Cc: Sam Ravnborg <sam@ravnborg.org>
Acked-by: Paulo Marques <pmarques@grupopie.com>
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
17 years ago[UDP]: Do not allow specific bind when wildcard bind exists.
David S. Miller [Mon, 30 Apr 2007 21:51:58 +0000 (14:51 -0700)]
[UDP]: Do not allow specific bind when wildcard bind exists.

When allocating local ports, do not allow a bind to a port
with a specific local address when a bind to that port with
a wildcard local address already exists.

Noticed by Linus.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4] UDP: Fix endianness bugs in hashing changes.
David S. Miller [Mon, 30 Apr 2007 20:35:29 +0000 (13:35 -0700)]
[IPV4] UDP: Fix endianness bugs in hashing changes.

I accidently applied an earlier version of Eric Dumazet's patch, from
March 21st.  His version from March 30th didn't have these bugs, so
this just interdiffs to the correct patch.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394...
Linus Torvalds [Mon, 30 Apr 2007 15:59:57 +0000 (08:59 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6: (56 commits)
  ieee1394: remove garbage from Kconfig
  ieee1394: more help in Kconfig
  ieee1394: ohci1394: Fix mistake in printk message.
  ieee1394: ohci1394: remove unnecessary rcvPhyPkt bit flipping in LinkControl register
  ieee1394: ohci1394: fix cosmetic problem in error logging
  ieee1394: eth1394: send async streams at S100 on 1394b buses
  ieee1394: eth1394: fix error path in module_init
  ieee1394: eth1394: correct return codes in hard_start_xmit
  ieee1394: eth1394: hard_start_xmit is called in atomic context
  ieee1394: eth1394: some conditions are unlikely
  ieee1394: eth1394: clean up fragment_overlap
  ieee1394: eth1394: don't use alloc_etherdev
  ieee1394: eth1394: omit useless set_mac_address callback
  ieee1394: eth1394: CONFIG_INET is always defined
  ieee1394: eth1394: allow MTU bigger than 1500
  ieee1394: unexport highlevel_host_reset
  ieee1394: eth1394: contain host reset
  ieee1394: eth1394: shorter error messages
  ieee1394: eth1394: correct a memset argument
  ieee1394: eth1394: refactor .probe and .update
  ...

17 years agoMerge branch 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Mon, 30 Apr 2007 15:58:21 +0000 (08:58 -0700)]
Merge branch 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jikos/hid

* 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/jikos/hid: (21 commits)
  USB HID: don't warn on idVendor == 0
  USB HID: add 'quirks' module parameter
  USB HID: add support for dynamically-created quirks
  USB HID: clarify static quirk handling as squirks
  USB HID: encapsulate quirk handling into hid-quirks.c
  USB HID: EMS USBII device needs HID_QUIRK_MULTI_INPUT
  HID: update copyright and authorship macro
  HID: introduce proper zeroing of unused bits in output reports
  USB HID: add support for WiseGroup MP-8800 Quad Joypad
  USB HID: add FF support for Logitech Force 3D Pro Joystick
  USB HID: numlock quirk for dell W7658 keyboard
  USB HID: Logitech MX3000 keyboard needs report descriptor quirk
  USB HID: extend quirk for Logitech S510 keyboard
  USB HID: usbkbd/usbmouse - handle errors when registering devices
  USB HID: add QUIRK_HIDDEV for Belkin Flip KVM
  HID: enable dead keys on a belkin wireless keyboard
  USB HID: Thustmaster firestorm dual power v1 support
  USB HID: specify explicit size for hid_blacklist.quirks
  USB HID: fix retry & reset logic
  USB HID: consolidate vendor/product ids
  ...

17 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 30 Apr 2007 15:14:42 +0000 (08:14 -0700)]
Merge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6: (21 commits)
  [IPV4] SNMP: Support OutMcastPkts and OutBcastPkts
  [IPV4] SNMP: Support InMcastPkts and InBcastPkts
  [IPV4] SNMP: Support InTruncatedPkts
  [IPV4] SNMP: Support InNoRoutes
  [SNMP]: Add definitions for {In,Out}BcastPkts
  [TCP] FRTO: RFC4138 allows Nagle override when new data must be sent
  [TCP] FRTO: Delay skb available check until it's mandatory
  [XFRM]: Restrict upper layer information by bundle.
  [TCP]: Catch skb with S+L bugs earlier
  [PATCH] INET : IPV4 UDP lookups converted to a 2 pass algo
  [L2TP]: Add the ability to autoload a pppox protocol module.
  [SKB]: Introduce skb_queue_walk_safe()
  [AF_IUCV/IUCV]: smp_call_function deadlock
  [IPV6]: Fix slab corruption running ip6sic
  [TCP]: Update references in two old comments
  [XFRM]: Export SPD info
  [IPV6]: Track device renames in snmp6.
  [SCTP]: Fix sctp_getsockopt_local_addrs_old() to use local storage.
  [NET]: Remove NETIF_F_INTERNAL_STATS, default to internal stats.
  [NETPOLL]: Remove CONFIG_NETPOLL_RX
  ...

17 years agoMerge branch 'for-linus' of git://git.kernel.dk/data/git/linux-2.6-block
Linus Torvalds [Mon, 30 Apr 2007 15:12:39 +0000 (08:12 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/data/git/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/data/git/linux-2.6-block:
  [PATCH] elevator: elv_list_lock does not need irq disabling
  [BLOCK] Don't pin lots of memory in mempools
  cfq-iosched: speedup cic rb lookup
  ll_rw_blk: add io_context private pointer
  cfq-iosched: get rid of cfqq hash
  cfq-iosched: tighten queue request overlap condition
  cfq-iosched: improve sync vs async workloads
  cfq-iosched: never allow an async queue idling
  cfq-iosched: get rid of ->dispatch_slice
  cfq-iosched: don't pass unused preemption variable around
  cfq-iosched: get rid of ->cur_rr and ->cfq_list
  cfq-iosched: slice offset should take ioprio into account
  [PATCH] cfq-iosched: style cleanups and comments
  cfq-iosched: sort IDLE queues into the rbtree
  cfq-iosched: sort RT queues into the rbtree
  [PATCH] cfq-iosched: speed up rbtree handling
  cfq-iosched: rework the whole round-robin list concept
  cfq-iosched: minor updates
  cfq-iosched: development update
  cfq-iosched: improve preemption for cooperating tasks

17 years agoMerge branch 'for-2.6.22' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus...
Linus Torvalds [Mon, 30 Apr 2007 15:10:12 +0000 (08:10 -0700)]
Merge branch 'for-2.6.22' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc

* 'for-2.6.22' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc: (255 commits)
  [POWERPC] Remove dev_dbg redefinition in drivers/ps3/vuart.c
  [POWERPC] remove kernel module option for booke wdt
  [POWERPC] Avoid putting cpu node twice
  [POWERPC] Spinlock initializer cleanup
  [POWERPC] ppc4xx_sgdma needs dma-mapping.h
  [POWERPC] arch/powerpc/sysdev/timer.c build fix
  [POWERPC] get_property cleanups
  [POWERPC] Remove the unused HTDMSOUND driver
  [POWERPC] cell: cbe_cpufreq cleanup and crash fix
  [POWERPC] Declare enable_kernel_spe in a header
  [POWERPC] Add dt_xlate_addr() to bootwrapper
  [POWERPC] bootwrapper: CONFIG_ -> CONFIG_DEVICE_TREE
  [POWERPC] Don't define a custom bd_t for Xilixn Virtex based boards.
  [POWERPC] Add sane defaults for Xilinx EDK generated xparameters files
  [POWERPC] Add uartlite boot console driver for the zImage wrapper
  [POWERPC] Stop using ppc_sys for Xilinx Virtex boards
  [POWERPC] New registration for common Xilinx Virtex ppc405 platform devices
  [POWERPC] Merge common virtex header files
  [POWERPC] Rework Kconfig dependancies for Xilinx Virtex ppc405 platform
  [POWERPC] Clean up cpufreq Kconfig dependencies
  ...

17 years ago[IPV4] SNMP: Support OutMcastPkts and OutBcastPkts
Mitsuru Chinen [Mon, 30 Apr 2007 07:48:20 +0000 (00:48 -0700)]
[IPV4] SNMP: Support OutMcastPkts and OutBcastPkts

A transmitted IP multicast datagram should be counted as OutMcastPkts.
By the same token, a transmitted IP broadcast datagram should be
counted as OutBcastPkts.

Signed-off-by: Mitsuru Chinen <mitch@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4] SNMP: Support InMcastPkts and InBcastPkts
Mitsuru Chinen [Mon, 30 Apr 2007 07:48:10 +0000 (00:48 -0700)]
[IPV4] SNMP: Support InMcastPkts and InBcastPkts

A received IP multicast datagram should be counted as InMcastPkts.
By the same token, a received IP broadcast datagram should be
counted as InBcastPkts.

Signed-off-by: Mitsuru Chinen <mitch@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4] SNMP: Support InTruncatedPkts
Mitsuru Chinen [Mon, 30 Apr 2007 07:46:30 +0000 (00:46 -0700)]
[IPV4] SNMP: Support InTruncatedPkts

An IP datagram which is being discarded because the datagram frame
didn't carry enough data should be counted as InTruncatedPkts.

Signed-off-by: Mitsuru Chinen <mitch@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4] SNMP: Support InNoRoutes
Mitsuru Chinen [Mon, 30 Apr 2007 07:45:49 +0000 (00:45 -0700)]
[IPV4] SNMP: Support InNoRoutes

An IP datagram which is being discarded because of no routes in the
forwarding path should be counted as InNoRoutes.

Signed-off-by: Mitsuru Chinen <mitch@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SNMP]: Add definitions for {In,Out}BcastPkts
Mitsuru Chinen [Mon, 30 Apr 2007 07:45:02 +0000 (00:45 -0700)]
[SNMP]: Add definitions for {In,Out}BcastPkts

The updated IP-MIB RFC (RFC4293) specifys new objects, InBcastPkts
and OutBcastPkts. This adds definitions for them.

Signed-off-by: Mitsuru Chinen <mitch@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP] FRTO: RFC4138 allows Nagle override when new data must be sent
Ilpo Järvinen [Mon, 30 Apr 2007 07:42:20 +0000 (00:42 -0700)]
[TCP] FRTO: RFC4138 allows Nagle override when new data must be sent

This is a corner case where less than MSS sized new data thingie
is awaiting in the send queue. For F-RTO to work correctly, a
new data segment must be sent at certain point or F-RTO cannot
be used at all. RFC4138 allows overriding of Nagle at that
point.

Implementation uses frto_counter states 2 and 3 to distinguish
when Nagle override is needed.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP] FRTO: Delay skb available check until it's mandatory
Ilpo Järvinen [Mon, 30 Apr 2007 07:39:55 +0000 (00:39 -0700)]
[TCP] FRTO: Delay skb available check until it's mandatory

No new data is needed until the first ACK comes, so no need to check
for application limitedness until then.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[XFRM]: Restrict upper layer information by bundle.
Masahide NAKAMURA [Mon, 30 Apr 2007 07:33:35 +0000 (00:33 -0700)]
[XFRM]: Restrict upper layer information by bundle.

On MIPv6 usage, XFRM sub policy is enabled.
When main (IPsec) and sub (MIPv6) policy selectors have the same
address set but different upper layer information (i.e. protocol
number and its ports or type/code), multiple bundle should be created.
However, currently we have issue to use the same bundle created for
the first time with all flows covered by the case.

It is useful for the bundle to have the upper layer information
to be restructured correctly if it does not match with the flow.

1. Bundle was created by two policies
Selector from another policy is added to xfrm_dst.
If the flow does not match the selector, it goes to slow path to
restructure new bundle by single policy.

2. Bundle was created by one policy
Flow cache is added to xfrm_dst as originated one. If the flow does
not match the cache, it goes to slow path to try searching another
policy.

Signed-off-by: Masahide NAKAMURA <nakam@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Catch skb with S+L bugs earlier
Ilpo Järvinen [Mon, 30 Apr 2007 07:57:33 +0000 (00:57 -0700)]
[TCP]: Catch skb with S+L bugs earlier

SACKED_ACKED and LOST are mutually exclusive with SACK, thus
having their sum larger than packets_out is bug with SACK.
Eventually these bugs trigger traps in the tcp_clean_rtx_queue
with SACK but it's much more informative to do this here.

Non-SACK TCP, however, could get more than packets_out duplicate
ACKs which each increment sacked_out, so it makes sense to do
this kind of limitting for non-SACK TCP but not for SACK enabled
one. Perhaps the author had the opposite in mind but did the
logic accidently wrong way around? Anyway, the sacked_out
incrementer code for non-SACK already deals this issue before
calling sync_left_out so this trapping can be done
unconditionally.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PATCH] INET : IPV4 UDP lookups converted to a 2 pass algo
Eric Dumazet [Mon, 30 Apr 2007 07:26:00 +0000 (00:26 -0700)]
[PATCH] INET : IPV4 UDP lookups converted to a 2 pass algo

Some people want to have many UDP sockets, binded to a single port but
many different addresses. We currently hash all those sockets into a
single chain.  Processing of incoming packets is very expensive,
because the whole chain must be examined to find the best match.

I chose in this patch to hash UDP sockets with a hash function that
take into account both their port number and address : This has a
drawback because we need two lookups : one with a given address, one
with a wildcard (null) address.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[L2TP]: Add the ability to autoload a pppox protocol module.
James Chapman [Mon, 30 Apr 2007 07:21:02 +0000 (00:21 -0700)]
[L2TP]: Add the ability to autoload a pppox protocol module.

This patch allows a name "pppox-proto-nnn" to be used in modprobe.conf
to autoload a PPPoX protocol nnn.

Signed-off-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agoMerge branch 'cfq' into for-linus
Jens Axboe [Mon, 30 Apr 2007 07:09:27 +0000 (09:09 +0200)]
Merge branch 'cfq' into for-linus

17 years ago[PATCH] elevator: elv_list_lock does not need irq disabling
Jens Axboe [Thu, 26 Apr 2007 12:41:53 +0000 (14:41 +0200)]
[PATCH] elevator: elv_list_lock does not need irq disabling

It's never grabbed from irq context, so just make it plain spin_lock().

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
17 years ago[BLOCK] Don't pin lots of memory in mempools
Jens Axboe [Mon, 2 Apr 2007 08:06:42 +0000 (10:06 +0200)]
[BLOCK] Don't pin lots of memory in mempools

Currently we scale the mempool sizes depending on memory installed
in the machine, except for the bio pool itself which sits at a fixed
256 entry pre-allocation.

There's really no point in "optimizing" this OOM path, we just need
enough preallocated to make progress. A single unit is enough, lets
scale it down to 2 just to be on the safe side.

This patch saves ~150kb of pinned kernel memory on a 32-bit box.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
17 years ago[SKB]: Introduce skb_queue_walk_safe()
James Chapman [Mon, 30 Apr 2007 07:07:31 +0000 (00:07 -0700)]
[SKB]: Introduce skb_queue_walk_safe()

This patch provides a method for walking skb lists while inserting or
removing skbs from the list.

Signed-off-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years agocfq-iosched: speedup cic rb lookup
Jens Axboe [Tue, 24 Apr 2007 19:23:53 +0000 (21:23 +0200)]
cfq-iosched: speedup cic rb lookup

We often lookup the same queue many times in succession, so cache
the last looked up queue to avoid browsing the rbtree.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
17 years agoll_rw_blk: add io_context private pointer
Jens Axboe [Tue, 24 Apr 2007 19:17:33 +0000 (21:17 +0200)]
ll_rw_blk: add io_context private pointer

To be used by as/cfq as they see fit.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
17 years agocfq-iosched: get rid of cfqq hash
Vasily Tarasov [Wed, 25 Apr 2007 10:29:51 +0000 (12:29 +0200)]
cfq-iosched: get rid of cfqq hash

cfq hash is no more necessary.  We always can get cfqq from io context.
cfq_get_io_context_noalloc() function is introduced, because we don't
want to allocate cic on merging and checking may_queue.  In order to
identify sync queue we've used hash key = CFQ_KEY_ASYNC. Since hash is
eliminated we need to use other criterion: sync flag for queue is added.
In all places where we dig in rb_tree we're in current context, so no
additional locking is required.

Advantages of this patch: no additional memory for hash, no seeking in
hash, code is cleaner. But it is necessary now to seek cic in per-ioc
rbtree, but it is faster:
- most processes work only with few devices
- most systems have only few block devices
- it is a rb-tree

Signed-off-by: Vasily Tarasov <vtaras@openvz.org>
Changes by me:

- Merge into CFQ devel branch
- Get rid of cfq_get_io_context_noalloc()
- Fix various bugs with dereferencing cic->cfqq[] with offset other
  than 0 or 1.
- Fix bug in cfqq setup, is_sync condition was reversed.
- Fix bug where only bio_sync() is used, we need to check for a READ too

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
17 years agocfq-iosched: tighten queue request overlap condition
Jens Axboe [Fri, 20 Apr 2007 18:45:39 +0000 (20:45 +0200)]
cfq-iosched: tighten queue request overlap condition

For tagged devices, allow overlap of requests if the idle window
isn't enabled on the current active queue.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
17 years agocfq-iosched: improve sync vs async workloads
Jens Axboe [Mon, 23 Apr 2007 06:33:33 +0000 (08:33 +0200)]
cfq-iosched: improve sync vs async workloads

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>