Sam Ravnborg [Tue, 21 Mar 2006 06:22:35 +0000 (07:22 +0100)]
kbuild: fix make help & make *pkg
FORCE was not defined => error.
Use kbuild infrastructure to call down to the relevant
Makefile. This enables us to use the FORCE definition from kbuild.
Jan Beulich [Thu, 9 Mar 2006 13:47:46 +0000 (14:47 +0100)]
kconfig: fix time ordering of writes to .kconfig.d and include/linux/autoconf.h
Since .kconfig.d is used as a make dependency of include/linux/autoconf.h, it
should be written earlier than the header file, to avoid a subsequent rebuild
to consider the header outdated.
Signed-Off-By: Jan Beulich <jbeulich@novell.com> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Adrian Bunk [Wed, 8 Mar 2006 07:58:36 +0000 (23:58 -0800)]
Kconfig: remove the CONFIG_CC_ALIGN_* options
I don't see any use case for the CONFIG_CC_ALIGN_* options:
- they are only available if EMBEDDED
- people using EMBEDDED will most likely also enable
CC_OPTIMIZE_FOR_SIZE
- the default for -Os is to disable alignment
In case someone is doing performance comparisons and discovers that the
default settings gcc chooses aren't good, the only sane thing is to discuss
whether it makes sense to change this, not through offering options to change
this locally.
Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Sam Ravnborg [Sun, 12 Mar 2006 22:26:29 +0000 (23:26 +0100)]
kbuild: clean-up genksyms
o remove all inlines
o declare everything static which is only used by genksyms.c
o delete unused functions
o delete unused variables
o delete unused stuff in genksyms.h
o properly ident genksyms.h
Sam Ravnborg [Sun, 12 Mar 2006 21:54:34 +0000 (22:54 +0100)]
kbuild: fix genksyms build error
genksyms needs to know when a symbol must have a "_" prefex as is
true for a few architectures.
Pass $(ARCH) as commandline argument and hardcode what architectures that
needs this info.
Previous attemt to take it from elfconfig.h was br0ken since elfconfig.h
is a generated file.
Sam Ravnborg [Thu, 9 Mar 2006 23:23:32 +0000 (00:23 +0100)]
kbuild: in makefile.txt note that Makefile is preferred name for kbuild files
As noted by Roland Dreier <rdreier@cisco.com> makefiles.txt told
one to use the name 'Kbuild' as preferred name for kbuild files.
This is not yet true so let makefiles.txt reflect reality.
Luke Yang [Thu, 2 Mar 2006 10:35:31 +0000 (18:35 +0800)]
kbuild: Fix bug in crc symbol generating of kernel and modules
The scripts/genksyms/genksyms.c uses hardcoded "__crc_" prefix for
crc symbols in kernel and modules. The prefix should be replaced by
"MODULE_SYMBOL_PREFIX##__crc_" otherwise there will be warnings when
MODULE_SYMBOL_PREFIX is not NULL.
I am sorry my last patch for this issue is actually wrong. I revert
it in this patch.
Signed-off-by: Luke Yang <luke.adi@gmail.com> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Paul Smith [Sun, 5 Mar 2006 22:14:10 +0000 (17:14 -0500)]
kbuild: change kbuild to not rely on incorrect GNU make behavior
The kbuild system takes advantage of an incorrect behavior in GNU make.
Once this behavior is fixed, all files in the kernel rebuild every time,
even if nothing has changed. This patch ensures kbuild works with both
the incorrect and correct behaviors of GNU make.
Changes in this patch:
- Keep all targets that are to be marked .PHONY in a variable, PHONY.
- Add .PHONY: $(PHONY) to mark them properly.
- Remove any $(PHONY) files from the $? list when determining whether
targets are up-to-date or not.
Signed-off-by: Paul Smith <psmith@gnu.org> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Sam Ravnborg [Sun, 5 Mar 2006 11:10:58 +0000 (12:10 +0100)]
kbuild: fix make dir/file.xx when asm symlink is missing
Added a dependency so we do full preparation before trying to build single
file targets. This fixes a case where Andrew Morton did:
make kernel/sched.o
rm include/asm
make kernel/sched.o -> splat
Sam Ravnborg [Sun, 5 Mar 2006 11:02:46 +0000 (12:02 +0100)]
kbuild: in the section mismatch check try harder to find symbols
When searching for symbols the only check performed was if
offset equals st_value. Adding an additional check to see if st_name
points t a valid name made us sort out a few more false positives and
let us report more correct names in warnings.
Sam Ravnborg [Sat, 4 Mar 2006 23:16:26 +0000 (00:16 +0100)]
kbuild: kill false positives from section mismatch warnings for powerpc
Building an allmodconfig kernel for ppc64 revealed a number of false
positives - originally reported by Andrew Morton.
This patch removes most if not all false positives for ppc64:
Section .opd
The .opd section contains function descriptors at least for ppc64.
So ignore it for .init.text (was ignored for .exit.text).
See description of function descriptors here:
http://www.linuxbase.org/spec/ELF/ppc64/PPC-elf64abi-1.7.html
Section .toc1
ppc64 places some static variables in .toc1 - ignore the.
Section __bug_tabe
BUG() and friends uses __bug_table. Ignore warnings from that section.
Module parameters are placed in .data.rel for ppc64, for adjust pattern to
match on section named .data*
Jesper Juhl [Sat, 25 Feb 2006 20:52:50 +0000 (21:52 +0100)]
kbuild: small update of allnoconfig description
'allnoconfig' is described by 'make help' as a "minimal config", that's not
strictly correct. To be pedantic, a minimal config would be one where
EMBEDDED was set to Y and most things therein disabled etc. Simply
answering 'no' to all options does not give a minimal config.
A better description of allnoconfig is that it answers all options with 'no'.
This patch updates the description.
Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Aaron Brooks [Mon, 27 Feb 2006 03:25:46 +0000 (22:25 -0500)]
kbuild: make namespace.pl CROSS_COMPILE happy
Using the fixed path to /usr/bin/{nm,objdump} does not allow
CROSS_COMPILE environments to use namespace.pl. This patch causes
namespace.pl to use $NM and $OBJDUMP if defined or fall back to the nm
and objdump found in the path.
Signed-off-by: Aaron Brooks <aaron.brooks@sicortex.com> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Sam Ravnborg [Sun, 26 Feb 2006 21:18:11 +0000 (22:18 +0100)]
kbuild: whitelist false section mismatch warnings
In several cases the section mismatch check triggered false warnings.
Following patch introduce a whitelist to 'false positives' are not warned of.
Two types of patterns are recognised:
1) Typical case when a module parameter is _initdata
2) When a function pointer is assigned to a driver structure
In both patterns we rely on the actual name of the variable assigned
Jan Beulich [Wed, 22 Feb 2006 12:29:04 +0000 (13:29 +0100)]
kbuild: version.h should depend on .kernelrelease
Rebuilding a previously built tree while using make's -j options from time to
time results in the version.h check running at the same time as the updating
of .kernelrelease, resulting in UTS_RELEASE remaining an empty string (and as
a side effect causing the entire kernel to be rebuilt).
Signed-Off-By: Jan Beulich <jbeulich@novell.com> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Sam Ravnborg [Wed, 22 Feb 2006 20:24:50 +0000 (21:24 +0100)]
kbuild: do not warn when unwind sections references .init/.exit sections
Andrew Morton reported a number of false positives for ia64 - like these:
WARNING: drivers/acpi/button.o - Section mismatch: reference to .init.text: from .IA_64.unwind.init.text after '' (at offset 0x0)
WARNING: drivers/acpi/button.o - Section mismatch: reference to .exit.text: from .IA_64.unwind.exit.text after '' (at offset 0x0)
WARNING: drivers/acpi/processor.o - Section mismatch: reference to .init.text: from .IA_64.unwind after '' (at offset 0x1e8)
They are all false positives - or at least the .c code looks OK.
It is not known why sometimes a section name is appended and sometimes not.
Fix is to accept references from all sections that includes "unwind." in the name.
akpm@osdl.org [Wed, 22 Feb 2006 11:19:54 +0000 (03:19 -0800)]
kbuild: fix modpost compile with older gcc
The kernel now requires that CC be 3.1.0 or higher. But we shouldn't place
that requirement upon HOSTCC unless we really need to. Fixes my ia64 problem.
Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Sam Ravnborg [Sat, 18 Feb 2006 23:49:37 +0000 (00:49 +0100)]
kbuild: do not segfault in modpost if MODVERDIR is not defined
A combination of calling modpost with option -a and MODVERDIR undefined
caused segmentation fault. So provide a default value and accept the
error messages it generates instead.
Jan Beulich [Mon, 30 Jan 2006 09:05:09 +0000 (10:05 +0100)]
kbuild: fix mkmakefile
With the current way of generating the Makefile in the output directory
for builds outside of the source tree, specifying real targets (rather
than phony ones) doesn't work in an already (partially) built tree, as
the stub Makefile doesn't have any dependency information available.
Thus, all targets where files may actually exist must be listed
explicitly and, due to what I'd call a make misbehavior, directory
targets must then also be special cased.
Signed-Off-By: Jan Beulich <jbeulich@novell.com> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Jan Beulich [Mon, 30 Jan 2006 09:04:27 +0000 (10:04 +0100)]
kbuild: consolidate command line escaping
While the recent change to also escape # symbols when storing C-file
compilation command lines was helpful, it should be in effect for all
command lines, as much as the dollar escaping should be in effect for
C-source compilation commands. Additionally, for better readability and
maintenance, consolidating all the escaping (single quotes, dollars,
and now sharps) was also desirable.
Signed-Off-By: Jan Beulich <jbeulich@novell.com> Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Sam Ravnborg [Sun, 22 Jan 2006 12:34:15 +0000 (13:34 +0100)]
kbuild: make cc-version available in kbuild files
Move $(CC) support functions to Kbuild.include so they are available
in the kbuild files.
In addition the following was done:
o as-option documented in Documentation/kbuild/makefiles.txt
o Moved documentation to new section to match
new scope of functions
o added cc-ifversion used to conditionally select a text string
dependent on actual $(CC) version
o documented cc-ifversion
o change so Kbuild.include is read before the kbuild file
Sam Ravnborg [Fri, 17 Feb 2006 21:42:02 +0000 (22:42 +0100)]
kbuild: check for section mismatch during modpost stage
Section mismatch is identified as references to .init*
sections from non .init sections. And likewise references
to .exit.* sections outside .exit sections.
.init.* sections are discarded after a module is initialized
and references to .init.* sections are oops candidates.
.exit.* sections are discarded when a module is built-in and
thus references to .exit are also oops candidates.
The checks were possible to do using 'make buildcheck' which
called the two perl scripts: reference_discarded.pl and
reference_init.pl. This patch just moves the same functionality
inside modpost and the scripts are then obsoleted.
They will though be kept for a while so users can do double
checks - but note that some .o files are skipped by the perl scripts
so result is not 1:1.
All credit for the concept goes to Keith Owens who implemented
the original perl scrips - this patch just moves it to modpost.
Compared to the perl script the implmentation in modpost will be run
for each kernel build - thus catching the error much sooner, but
the downside is that the individual .o file are not always identified.
Sam Ravnborg [Sat, 28 Jan 2006 21:22:33 +0000 (22:22 +0100)]
kbuild: warn about duplicate exported symbols
In modpost introduce a check for symbols exported twice.
This check caught only one victim (inet_bind_bucket_create) for
which a patch is already sent to netdev.
Sam Ravnborg [Sat, 28 Jan 2006 21:15:55 +0000 (22:15 +0100)]
kbuild: improved modversioning support for external modules
With following patch a second option is enabled to obtain
symbol information from a second external module when a
external module is build.
The recommended approach is to use a common kbuild file but
that may be impractical in certain cases.
With this patch one can copy over a Module.symvers from one
external module to make symbols (and symbol versions) available
for another external module.
Updated documentation in Documentation/kbuild/modules.txt
Sam Ravnborg [Wed, 25 Jan 2006 06:13:18 +0000 (07:13 +0100)]
kbuild: support building individual files for external modules
Support building individual files when dealing with separate modules.
So say you have a module named "foo" which consist of two .o files bar.o
and fun.o.
You can then do:
make -C $KERNELSRC M=`pwd` bar.o
make -C $KERNELSRC M=`pwd` bar.lst
make -C $KERNELSRC M=`pwd` bar.i
make -C $KERNELSRC M=`pwd` / <= will build all .o files
and link foo.o
make -C $KERNELSRC M=`pwd` foo.ko <= will build the module
and do the modpost step
to create foo.ko
The above will also work if the external module is placed in a
subdirectory using a hirachy of kbuild files.
Thanks to Andreas Gruenbacher <agruen@suse.de> for initial feature
request / bug report.
acpi_rs_get_list_length() needs to account for all the vendor-defined data
bytes. Failing to include these causes buffers to be sized too small,
which causes slab corruption when we later convert AML to resources and run
off the end of the buffer.
This causes slab corruption on machines that use ACPI vendor-defined
resources. All HP ia64 machines do, and I'm told that some NEC machines
may as well.
David Gibson [Fri, 17 Feb 2006 21:52:56 +0000 (13:52 -0800)]
[PATCH] powerpc: Fix accidentally-working typo in __pud_free_tlb
One of the parameters to the __pud_free_tlb() macro for powerpc is
incorrect (see patch) . We get away with it by accident, because the one
place the macro is called, the second parameter is a variable named "pud".
Signed-off-by: David Gibson <dwg@au1.ibm.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Johannes Berg [Fri, 17 Feb 2006 21:52:54 +0000 (13:52 -0800)]
[PATCH] allow windfarm_pm112 module to load
The windfarm_pm112 module relies on smu_sat_get_sdb_partition which is in
windfarm_smu_sat.c but is not exported to modules, so despite Kconfig
having the option to build the pm112 as modules, this can never be loaded.
This patch fixes that by exporting smu_sat_get_sdb_partition with
EXPORT_SYMBOL_GPL
Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Restore the compatibility with the older code and make it possible to
suspend if the kernel command line doesn't contain the "resume=" argument
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Cc: Pavel Machek <pavel@ucw.cz> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Heiko Carstens [Fri, 17 Feb 2006 21:52:48 +0000 (13:52 -0800)]
[PATCH] s390: smp initialization speed
The last changes that introduced the additional_cpus command line parameter
also introduced a regression regarding smp initialization speed. In
smp_setup_cpu_possible_map() cpu_present_map is set to the same value as
cpu_possible_map. Especially that means that bits in the present map will be
set for cpus that are not present. This will cause a slow down in the initial
cpu_up() loop in smp_init() since trying to take cpus online that aren't
present takes a while.
Fix this by setting only bits for present cpus in cpu_present_map and set
cpu_present_map to cpu_possible_map in smp_cpus_done().
Heiko Carstens [Fri, 17 Feb 2006 21:52:47 +0000 (13:52 -0800)]
[PATCH] s390: possible_cpus parameter
Introduce possible_cpus command line option. Hard sets the number of bits set
in cpu_possible_map. Unlike the additional_cpus parameter this one guarantees
that num_possible_cpus() will stay constant even if the system gets rebooted
and a different number of cpus are present at startup.
Heiko Carstens [Fri, 17 Feb 2006 21:52:46 +0000 (13:52 -0800)]
[PATCH] s390: additional_cpus parameter
Introduce additional_cpus command line option. By default no additional cpu
can be attached to the system anymore. Only the cpus present at IPL time can
be switched on/off. If it is desired that additional cpus can be attached to
the system the maximum number of additional cpus needs to be specified with
this option.
This change is necessary in order to limit the waste of per_cpu data
structures.
Heiko Carstens [Fri, 17 Feb 2006 21:52:46 +0000 (13:52 -0800)]
[PATCH] s390: fix preempt_count of idle thread with cpu hotplug
Set preempt_count of idle_thread to zero before switching off cpu. Otherwise
the preempt_count will be wrong if the cpu is switched on again since the
thread will be reused.
Cornelia Huck [Fri, 17 Feb 2006 21:52:45 +0000 (13:52 -0800)]
[PATCH] s390: ccw device disbanding
If __ccw_device_disband_start() fails to initiate disbanding, it should finish
with ccw_device_disband_done() (which leaves the device in offline state)
instead of ccw_device_verify_done() (which leaves the device in online state).
Ingo Molnar [Fri, 17 Feb 2006 21:52:44 +0000 (13:52 -0800)]
[PATCH] Introduce CONFIG_DEFAULT_MIGRATION_COST
Heiko Carstens <heiko.carstens@de.ibm.com> wrote:
The boot sequence on s390 sometimes takes ages and we spend a very long
time (up to one or two minutes) in calibrate_migration_costs. The time
spent there differs from boot to boot. Also the calculated costs differ
a lot. I've seen differences by up to a factor of 15 (yes, factor not
percent). Also I doubt that making these measurements make much sense on
a completely virtualized architecture where you cannot tell how much cpu
time you will get anyway.
So introduce the CONFIG_DEFAULT_MIGRATION_COST method for an architecture
to set the scheduler migration costs. This turns off automatic detection
of migration costs. Makes sense on virtual platforms, where migration
costs are hard to measure accurately.
Peter Staubach [Fri, 17 Feb 2006 21:52:36 +0000 (13:52 -0800)]
[PATCH] fix deadlock in ext2
Fix a deadlock possible in the ext2 file system implementation. This
deadlock occurs when a file is removed from an ext2 file system which was
mounted with the "sync" mount option.
The problem is that ext2_xattr_delete_inode() was invoking the routine,
sync_dirty_buffer(), using a buffer head which was previously locked via
lock_buffer(). The first thing that sync_dirty_buffer() does is to lock
the buffer head that it was passed. It does this via lock_buffer(). Oops.
The solution is to unlock the buffer head in ext2_xattr_delete_inode()
before invoking sync_dirty_buffer(). This makes the code in
ext2_xattr_delete_inode() obey the same locking rules as all other callers
of sync_dirty_buffer() in the ext2 file system implementation.
Signed-off-by: Peter Staubach <staubach@redhat.com> Cc: <stable@kernel.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Linus Torvalds [Fri, 17 Feb 2006 19:38:21 +0000 (20:38 +0100)]
[PATCH] Handle holes in node mask in node fallback list setup
Change the find_next_best_node algorithm to correctly skip
over holes in the node online mask. Previously it would not handle
missing nodes correctly and cause crashes at boot.
Jay Vosburgh [Wed, 8 Feb 2006 05:17:22 +0000 (21:17 -0800)]
[PATCH] bonding: fix a locking bug in bond_release
bond_release returns EINVAL without releasing the bond lock if the
slave device is not being bonded by the bond. The following patch
ensures that the lock is released in this case.
Signed-off-by: Stephen J. Bevan <stephen@dino.dnsalias.com> Acked-by: Jay Vosburgh <fubar@us.ibm.com> Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Frank Pavlic [Tue, 7 Feb 2006 16:04:38 +0000 (17:04 +0100)]
[PATCH] s390: some qeth driver fixes
[patch 2/2] s390: some qeth driver fixes
From: Frank Pavlic <fpavlic@de.ibm.com>
- fixed kernel panic when using EDDP support in Layer 2 mode
- NULL pointer exception in qeth_set_offline fixed.
- setting EDDP in Layer 2 mode did not set NETIF_F_(SG/TSO)
flags when device became online.
- use sscanf for parsing and converting IPv4 addresses
from string to __u8 values.
- qeth_string_to_ipaddr6 fixed. in case of double colon
the converted IPv6 address out from the string was not correct
in previous implementation.
Frank Pavlic [Tue, 7 Feb 2006 16:04:36 +0000 (17:04 +0100)]
[PATCH] s390: lcs performance enhancements
[patch 1/2] s390: lcs performance enhancements
From: Klaus Wacker <kdwacker@de.ibm.com>
- When flood pinging (with large packet size) an LCS device,
about 90 % of all packets are dropped by driver.
- increased number of lcs IO buffers to 32.
- use netif_stop_queue/netif_wake_queue in lcs_start_xmit routine
- don't lock the whole xmit routine but just the piece of code where
tx_buffer is touched.
Andrew Morton [Fri, 10 Feb 2006 10:00:43 +0000 (02:00 -0800)]
[PATCH] smctr warning fix
drivers/net/tokenring/smctr.c: In function `smctr_load_firmware':
drivers/net/tokenring/smctr.c:2981: warning: assignment discards qualifiers from pointer target type
Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Users report problems w/ auto-negotiation disabled and the link set
to 100/Half or 10/Half. Problems range from poor performance to no
link at all.
The current sky2 code does not set things properly on link up if
autonegotiation is disabled. Plus it does not contemplate a 10Mbit
setting at all. This patch corrects that.
Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Stephen Hemminger <shemminger@osdl.org> Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
This is a clone of John Linville's fixed for speed setting on sky2 driver.
The skge driver has the same code (and bug). It would not allow manually forcing
100 and 10 mbit.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org> Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
The sk98lin driver was changed a while ago to remove support for the
D-Link 530T card because that hardware has no working VPD data. The help
text for Kconfig was not updated.
Signed-off-by: Stephen Hemminger <shemminger@osdl.org> Signed-off-by: Jeff Garzik <jgarzik@pobox.com>
Joshua Kinard [Fri, 17 Feb 2006 03:52:25 +0000 (03:52 +0000)]
[PATCH] Fix SGI O2 compile error in drivers/video/gbefb.c
A sysfs function call uses the wrong parameter, and thus breaks a build on
SGI O2.
CC drivers/video/gbefb.o
drivers/video/gbefb.c: In function ‘gbefb_remove’:
drivers/video/gbefb.c:1246: error: ‘dev’ undeclared (first use in this function)
drivers/video/gbefb.c:1246: error: (Each undeclared identifier is reported only once
drivers/video/gbefb.c:1246: error: for each function it appears in.)
make[2]: *** [drivers/video/gbefb.o] Error 1
Paul Mackerras [Thu, 16 Feb 2006 23:30:23 +0000 (10:30 +1100)]
[PATCH] Provide an interface for getting the current tick length
This provides an interface for arch code to find out how many
nanoseconds are going to be added on to xtime by the next call to
do_timer. The value returned is a fixed-point number in 52.12 format
in nanoseconds. The reason for this format is that it gives the
full precision that the timekeeping code is using internally.
The motivation for this is to fix a problem that has arisen on 32-bit
powerpc in that the value returned by do_gettimeofday drifts apart
from xtime if NTP is being used. PowerPC is now using a lockless
do_gettimeofday based on reading the timebase register and performing
some simple arithmetic. (This method of getting the time is also
exported to userspace via the VDSO.) However, the factor and offset
it uses were calculated based on the nominal tick length and weren't
being adjusted when NTP varied the tick length.
Note that 64-bit powerpc has had the lockless do_gettimeofday for a
long time now. It also had an extremely hairy routine that got called
from the 32-bit compat routine for adjtimex, which adjusted the
factor and offset according to what it thought the timekeeping code
was going to do. Not only was this only called if a 32-bit task did
adjtimex (i.e. not if a 64-bit task did adjtimex), it was also
duplicating computations from kernel/timer.c and it wasn't clear that
it was (still) correct.
The simple solution is to ask the timekeeping code how long the
current jiffy will be on each timer interrupt, after calling
do_timer. If this jiffy will be a different length from the last one,
we then need to compute new values for the factor and offset used in
the lockless do_gettimeofday. In this way we can keep xtime and
do_gettimeofday in sync, even when NTP is varying the tick length.
Note that when adjtimex varies the tick length, it almost always
introduces the variation from the next tick on. The only case I could
see where adjtimex would vary the length of the current tick is when
an old-style adjtime adjustment is being cancelled. (It's not clear
to me why the adjustment has to be cancelled immediately rather than
from the next tick on.) Thus I don't see any real need for a hook in
adjtimex; the rare case of an old-style adjustment being cancelled can
be fixed up at the next tick.
Signed-off-by: Paul Mackerras <paulus@samba.org> Acked-by: john stultz <johnstul@us.ibm.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Andi Kleen [Fri, 17 Feb 2006 00:39:16 +0000 (01:39 +0100)]
[PATCH] Handle all and empty zones when setting up custom zonelists for mbind
The memory allocator doesn't like empty zones (which have an
uninitialized freelist), so a x86-64 system with a node fully
in GFP_DMA32 only would crash on mbind.
Fix that up by putting all possible zones as fallback into the zonelist
and skipping the empty ones.
In fact the code always enough allocated space for all zones,
but only used it for the highest. This change just uses all the
memory that was allocated before.
This should work fine for now, but whoever implements node hot removal
needs to fix this somewhere else too (or make sure zone datastructures
by itself never go away, only their memory)
Andi Kleen [Thu, 16 Feb 2006 22:42:16 +0000 (23:42 +0100)]
[PATCH] x86_64: Always pass full number of nodes to NUMA hash computation
Previously the numa hash code would be confused by holes in the node space
and stop early. This is the first part of the fix for the non boot issue
with empty nodes on Opterons.
Andi Kleen [Thu, 16 Feb 2006 22:42:04 +0000 (23:42 +0100)]
[PATCH] x86_64: Don't enable ATI apicmaintimer workaround when the machine has C2 or C3
Many laptops have problems with ticking the local APIC timer in C2/C3.
The code added earlier to use it by default on ATI didn't really work
for them. Don't enable it when the system supports C2/C3.
This doesn't fix the problem fully, but at least it's not worse than before.
Jan Beulich [Thu, 16 Feb 2006 22:41:55 +0000 (23:41 +0100)]
[PATCH] x86_64: make touch_nmi_watchdog() not touch impossible cpus' private data
Along with that, also suppress the memory touching altogether when the
watchdog is not running, to eliminate needless crosstalk. Plus ad a call
to it to make things consistent (one could also consider removing the call
in enable_timer_nmi_watchdog()).
projects / linux-2.6-omap-h63xx.git / log