sfrench/cifs-2.6.git
11 years agokconfig: use long options in conf
Sam Ravnborg [Sat, 31 Jul 2010 21:35:26 +0000 (23:35 +0200)]
kconfig: use long options in conf

The list of options supported by conf is growing
and their abbreviation did not resemble anything usefull.

So drop the single letter options in favour of long options.

The long options are named equal to what we know from
the make target.
The internal implmentation was changed to match this,
resulting in much more readable code.

Support for short options is dropped - no one is supposed
to call this program direct anyway.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: fix MODULES-related bug in case of no .config
Ulf Magnusson [Tue, 27 Jul 2010 19:57:43 +0000 (21:57 +0200)]
kconfig: fix MODULES-related bug in case of no .config

There seems to be a kconfig bug due to MODULES not always being
evaluated if no .config is found. Take the following Kconfig as an
example:

config MODULES
def_bool y

config FOO
def_tristate m

With no .config, the following configuration is generated:

CONFIG_MODULES=y
CONFIG_FOO=y

With an empty .config, the following:

CONFIG_MODULES=y
CONFIG_FOO=m

Tristate choice statements can also exhibit the problem, due to having an
implicit rev_dep (select) containing "m".

The problem is that MODULES is never evaluted in conf_read_simple() unless
there's a .config. The following patch fixes this.

Signed-off-by: Ulf Magnusson <ulfalizer.lkml@gmail.com>
Reviewed-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: make randconfig fair for booleans
Peter Korsgaard [Thu, 22 Jul 2010 12:24:57 +0000 (14:24 +0200)]
kconfig: make randconfig fair for booleans

Give boolean symbols a 50% chance of getting enabled, rather than 67%.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: Don't write invisible choice values
Jan Beulich [Wed, 30 Jun 2010 12:11:01 +0000 (13:11 +0100)]
kconfig: Don't write invisible choice values

This makes it so "make oldconfig" really prompts for any choice where
options that previously weren't visible just became so. Previously one
would have to remember to go over all choice values and check whether
some that previously couldn't be selected now can be.

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: Warn on selecting symbols with unmet direct dependencies
Catalin Marinas [Tue, 8 Jun 2010 16:25:57 +0000 (17:25 +0100)]
kbuild: Warn on selecting symbols with unmet direct dependencies

The "select" statement in Kconfig files allows the enabling of options
even if they have unmet direct dependencies (i.e. "depends on" expands
to "no"). Currently, the "depends on" clauses are used in calculating
the visibility but they do not affect the reverse dependencies in any
way.

The patch introduces additional tracking of the "depends on" statements
and prints a warning on selecting an option if its direct dependencies
are not met.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts:conf.c Fix warning: variable 'type' set but not used
Justin P. Mattock [Fri, 11 Jun 2010 20:41:04 +0000 (13:41 -0700)]
scripts:conf.c Fix warning: variable 'type' set but not used

Not sure if this is correct or not, but with
make menuconfig
  HOSTCC  scripts/kconfig/conf.o
scripts/kconfig/conf.c: In function 'conf_sym':
scripts/kconfig/conf.c:159:6: warning: variable 'type' set but not used
scripts/kconfig/conf.c: In function 'conf_choice':
scripts/kconfig/conf.c:231:6: warning: variable 'type' set but not used
  HOSTLD  scripts/kconfig/mconf

I get this using gcc 4.6.0 the below change fixes this form me.

Signed-off-by: Justin P. Mattock <justinmattock@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomenuconfig: truncate list items
Li Zefan [Thu, 3 Jun 2010 07:24:57 +0000 (15:24 +0800)]
menuconfig: truncate list items

Truncate list items to fit in a single line, otherwise those items
which have long prompts will cover some other items.

This follows the behavior of menubox.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomenuconfig: fix to center checklist correctly in a corner case
Li Zefan [Thu, 3 Jun 2010 07:24:39 +0000 (15:24 +0800)]
menuconfig: fix to center checklist correctly in a corner case

Run:
make ARCH=arm menuconfig

And then select "System Type" -> "ARM system type". The kconfig
"choice" menu at this point looks empty.

It's because config ARCH_S3C2410 has a long prompt:

config ARCH_S3C2410
        bool "Samsung S3C2410, S3C2412, S3C2413, S3C2416, S3C2440, S3C2442, S3C2443, S3C2450"
...

menuconfig centers the checklist according to this prompt without
considering the width of the list, and then things get wrong.

Reported-by: Nobin Mathew <nobin.mathew@gmail.com>
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoxconfig: add support to show hidden options which have prompts
Li Zefan [Mon, 10 May 2010 08:33:41 +0000 (16:33 +0800)]
xconfig: add support to show hidden options which have prompts

This feature has been supported in menuconfig and gconfig, so
here add it to xconfig.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoxconfig: remove unused function
Li Zefan [Fri, 7 May 2010 05:58:21 +0000 (13:58 +0800)]
xconfig: remove unused function

Remove ConfigInfoView::setSource().

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoxconfig: clean up
Li Zefan [Fri, 7 May 2010 05:58:04 +0000 (13:58 +0800)]
xconfig: clean up

@ok is a pointer to a bool var, so we should check the value of
*ok. But actually we don't need to check it, so just remove the
if statement.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agogconfig: fix null pointer warning
Li Zefan [Fri, 7 May 2010 05:57:49 +0000 (13:57 +0800)]
gconfig: fix null pointer warning

In gconfig if you enable "Show all options", you'll see some "(null)"
config options, and clicking those options triggers a warning:

(gconf:9368): Gtk-CRITICAL **: gtk_text_buffer_insert_with_tags: assertion `text != NULL' failed

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agogconfig: fix to tag NEW symbols correctly
Li Zefan [Fri, 7 May 2010 05:57:35 +0000 (13:57 +0800)]
gconfig: fix to tag NEW symbols correctly

The logic should be reversed.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomenuconfig: improive help text a bit
Li Zefan [Fri, 7 May 2010 05:57:22 +0000 (13:57 +0800)]
menuconfig: improive help text a bit

Suggested-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: fix to tag NEW symbols correctly
Li Zefan [Fri, 7 May 2010 05:57:07 +0000 (13:57 +0800)]
kconfig: fix to tag NEW symbols correctly

Those configs are not new:

  $ cat .config
  ...
  CONFIG_NAMESPACES=y
  ...
  CONFIG_BLOCK=y
  ...

But are tagged as NEW:

  $ yes "" | make config > myconf
  $ cat myconf | grep '(NEW)'
  Namespaces support (NAMESPACES) [Y/?] (NEW) y
  ...
  Enable the block layer (BLOCK) [Y/?] (NEW) y
  ...

You can also notice this bug when using gconfig/xconfig.

It's because the SYMBOL_DEF_USER bit of an invisible symbol is cleared
when the config file is read:

int conf_read(const char *name)
{
...
for_all_symbols(i, sym) {
if (sym_has_value(sym) && !sym_is_choice_value(sym)) {
/* Reset values of generates values, so they'll appear
 * as new, if they should become visible, but that
 * doesn't quite work if the Kconfig and the saved
 * configuration disagree.
 */
if (sym->visible == no && !conf_unsaved)
sym->flags &= ~SYMBOL_DEF_USER;
...
}

But a menu item which represents an invisible symbol is still
visible, if it's sub-menu is visible, so its SYMBOL_DEF_USER
bit should be set to indicate it's not NEW.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: print the range of integer/hex symbol in help text
Li Zefan [Fri, 7 May 2010 05:56:50 +0000 (13:56 +0800)]
kconfig: print the range of integer/hex symbol in help text

Without this patch, one has to refer to the Kconfig file to find
out the range of an integer/hex symbol.

  │ Symbol: NR_CPUS [=4]
  │ Type  : integer
  │ Range : [2 8]
  │ Prompt: Maximum number of CPUs
  │   Defined at arch/x86/Kconfig:761
  │   Depends on: SMP [=y] && !MAXSMP [=n]
  │   Location:
  │     -> Processor type and features

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: print symbol type in help text
Li Zefan [Fri, 7 May 2010 05:56:33 +0000 (13:56 +0800)]
kconfig: print symbol type in help text

Randy suggested to print out the symbol type in gconfig.

Note this change does more than Randy's suggestion, that it also
affects menuconfig and "make config".

  │ Symbol: BLOCK [=y]
  │ Type  : boolean
  │ Prompt: Enable the block layer
  │   Defined at block/Kconfig:4
  │   Depends on: EMBEDDED [=n]

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: introduce nonint_oldconfig and loose_nonint_oldconfig
Aristeu Rozanski [Thu, 6 May 2010 16:48:34 +0000 (12:48 -0400)]
kconfig: introduce nonint_oldconfig and loose_nonint_oldconfig

This patch has been around for a long time in Fedora and Red Hat Enterprise
Linux kernels and it may be useful for others. The nonint_oldconfig target
will fail and print the unset config options while loose_nonint_oldconfig will
simply let the config option unset. They're useful in distro kernel packages
where the config files are built using a combination of smaller config files.

Arjan van de Ven wrote the initial nonint_config and Roland McGrath added the
loose_nonint_oldconfig.

Signed-off-by: Arjan van de Ven <arjan@redhat.com> [defunct email]
Whatevered-by: Kyle McMartin <kyle@redhat.com>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Aristeu Rozanski <aris@redhat.com>
[mmarek: whitespace fixes]
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: Revert part of e8d400a to resolve a conflict
Michal Marek [Thu, 27 May 2010 14:07:37 +0000 (16:07 +0200)]
kbuild: Revert part of e8d400a to resolve a conflict

A more complete patch in the kernel-doc tree also contains this change.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: Fix checking of scm-identifier variable
Greg Thelen [Wed, 5 May 2010 17:41:44 +0000 (10:41 -0700)]
kbuild: Fix checking of scm-identifier variable

I'm looking Makefile in the -mm branch (dated 2010-04-28-16-53) and
seeing what looks like a bug in the checking of scm-identifier.  The
"ifneq ($scm-identifier)" seems to always execute "ifeq
($(LOCALVERSION,)) ...".  This patch fixes the checking of
scm-identifier.

Signed-off-by: Greg Thelen <gthelen@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agogconfig: add support to show hidden options that have prompts
Li Zefan [Wed, 14 Apr 2010 03:46:24 +0000 (11:46 +0800)]
gconfig: add support to show hidden options that have prompts

There's a button in gconfig to "Show all options", but I think
normally we are not interested in those configs which have no
prompt and thus can't be changed, so here I add a new button to
show hidden options which have prompts.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomenuconfig: add support to show hidden options which have prompts
Li Zefan [Wed, 14 Apr 2010 03:46:02 +0000 (11:46 +0800)]
menuconfig: add support to show hidden options which have prompts

Usage:
  Press <Z> to show all config symbols which have prompts.

Quote Tim Bird:

| I've been bitten by this numerous times.  I most often
| use ftrace on ARM, but when I go back to x86, I almost
| always go through a sequence of searching for the
| function graph tracer in the menus, then realizing it's
| completely missing until I disable CC_OPTIMIZE_FOR_SIZE.
|
| Is there any way to have the menu item appear, but be
| unsettable unless the SIZE option is disabled?  I'm
| not a Kconfig guru...

I myself found this useful too. For example, I need to test
ftrace/tracing and want to be sure all the tracing features are
enabled, so I  enter the "Tracers" menu, and press <Z> to
see if there is any config hidden.

I also noticed gconfig and xconfig have a button "Show all options",
but that's a bit too much, and I think normally what we are not
interested in those configs which have no prompt thus can't be
changed by users.

Exmaple:

      --- Tracers
      -*-   Kernel Function Tracer
      - -     Kernel Function Graph Tracer
      [*]   Interrupts-off Latency Tracer
      - -   Preemption-off Latency Tracer
      [*]   Sysprof Tracer

Here you can see 2 tracers are not selectable, and then can find
out how to make them selectable.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agogconfig: remove show_debug option
Li Zefan [Wed, 14 Apr 2010 03:44:51 +0000 (11:44 +0800)]
gconfig: remove show_debug option

This option is a no-op, so remove it.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agogconfig: remove dbg_print_ptype() and dbg_print_stype()
Li Zefan [Wed, 14 Apr 2010 03:44:34 +0000 (11:44 +0800)]
gconfig: remove dbg_print_ptype() and dbg_print_stype()

Just use sym_get_type() and prop_get_type_name().

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: fix zconfdump()
Li Zefan [Wed, 14 Apr 2010 03:44:20 +0000 (11:44 +0800)]
kconfig: fix zconfdump()

zconfdump(), which is used for debugging, can't recognize P_SELECT,
P_RANGE and P_MENU (if associated with a symbol, aka "menuconfig"),
and output something like this:

config X86
  boolean
  default y
  unknown prop 6!
  unknown prop 6!
  unknown prop 6!
  ...

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: some small fixes
Li Zefan [Wed, 14 Apr 2010 03:44:05 +0000 (11:44 +0800)]
kconfig: some small fixes

- fix a typo in documentation
- fix a typo in a printk on error
- fix comments in dialog_inputbox()

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoadd random binaries to .gitignore
Jan III Sobieski [Sun, 28 Mar 2010 13:38:31 +0000 (15:38 +0200)]
add random binaries to .gitignore

Signed-off-by: Jan III Sobieski <jan3sobi3ski@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: Include gen_initramfs_list.sh and the file list in the .d file
Jason Gunthorpe [Mon, 22 Mar 2010 22:49:32 +0000 (16:49 -0600)]
kbuild: Include gen_initramfs_list.sh and the file list in the .d file

Expand the dependency set used for the initrd to include the
CONFIG_INITRAMFS_SOURCE file and the generator script itself.
Otherwise changing the initramfs file list does not rebuild the CPIO.

Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: recalc symbol value before showing search results
Li Zefan [Fri, 19 Mar 2010 06:57:47 +0000 (14:57 +0800)]
kconfig: recalc symbol value before showing search results

A symbol's value won't be recalc-ed until we save config file or
enter the menu where the symbol sits.

So If I enable OPTIMIZE_FOR_SIZE, and search FUNCTION_GRAPH_TRACER:

  Symbol: FUNCTION_GRAPH_TRACER [=y]
  Prompt: Kernel Function Graph Tracer
    Defined at kernel/trace/Kconfig:140
    Depends on: ... [=y] && (!X86_32 [=y] || !CC_OPTIMIZE_FOR_SIZE [=y])
    ...

From the dependency it should result in FUNCTION_GRAPH_TRACER=n,
but it still shows FUNCTION_GRAPH_TRACER=y.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years ago.gitignore: ignore *.lzo files
Philipp Kohlbecher [Wed, 17 Mar 2010 18:52:12 +0000 (19:52 +0100)]
.gitignore: ignore *.lzo files

Ignore files compressed with lzop.

Signed-off-by: Philipp Kohlbecher <xt28@gmx.de>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoheaderdep: perlcritic warning
Stephen Hemminger [Mon, 8 Mar 2010 16:24:27 +0000 (08:24 -0800)]
headerdep: perlcritic warning

Minor perlcritic warning:
headerdep.pl: "return" statement with explicit "undef" at line 84, column 2.  See page 199 of PBP.  (Severity: 5)

The rationale according to PBP is that an explicit return of undef
(contrary to most people's expectations) doesn't
always evaluate as false. It has to with the fact that perl return value
depends on context the function is called. If function is used in
list context, the appropriate return value for false is an empty list;
whereas in scalar context the return value for false is undefined.
By just using a "return" both cases are handled.

In the context of a trivial script this doesn't matter. But one script
may be cut-paste into later code (most people like me only know 50%
of perl), that is why perlcritic always complains

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts/Makefile.lib: Align the output of LZO
Wu Zhangjin [Thu, 11 Mar 2010 09:42:14 +0000 (17:42 +0800)]
scripts/Makefile.lib: Align the output of LZO

The output of LZO is not aligned with the other output:
  ...
  CC      drivers/usb/mon/usbmon.mod.o
  LZO    arch/mips/boot/compressed/vmlinux.lzo
  ...

This patch fixes it.

Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: Generate modules.builtin in make modules_install
Michal Marek [Wed, 10 Mar 2010 11:28:58 +0000 (12:28 +0100)]
kbuild: Generate modules.builtin in make modules_install

The previous approach didn't work if one did
make modules && make modules_install
Add modules.builtin as dependency of _modinst_, which is the target that
actually needs the file.

Reported-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRevert "kbuild: specify absolute paths for cscope"
Michal Marek [Mon, 8 Mar 2010 09:26:22 +0000 (10:26 +0100)]
Revert "kbuild: specify absolute paths for cscope"

This reverts commit eb8f844c0a41c4529a7d06b7801296eca9ae67aa. Ian
Campbell writes:
> I keep my kernel source tree on a more powerful build box where I run my
> builds etc (including "make cscope") but run my editor from my
> workstation with an NFS mount to the source. This worked fine for me
> using relative paths for cscope. Using absolute paths in cscope breaks
> this previously working setup because the root path is not the same on
> both systems. I guess this is similar to moving the source tree around.
>
> Without wanting to start a flamewar it really sounds to me like we are
> working around a vim (or cscope) bug here, emacs with cscope bindings
> works fine in this configuration.

Given that absolute paths can be forced by make O=. cscope, change the
default back to relative paths.

Ian Campbell <ijc@hellion.org.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: Do not unnecessarily regenerate modules.builtin
Michal Marek [Mon, 8 Mar 2010 09:07:12 +0000 (10:07 +0100)]
kbuild: Do not unnecessarily regenerate modules.builtin

Only regenerate it if the configuration has changed. Also, do this after
the modules build to fix errors with some weird Makefiles that are
generated during build.

Reported-by: Eric Miao <eric.y.miao@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoheaders_install: use local file handles
Stephen Hemminger [Mon, 22 Feb 2010 23:17:26 +0000 (15:17 -0800)]
headers_install: use local file handles

Better practice to use 3 arg open and local file handles.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: WANG Cong <amwang@redhat.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoheaders_check: fix perl warnings
Stephen Hemminger [Mon, 22 Feb 2010 23:17:24 +0000 (15:17 -0800)]
headers_check: fix perl warnings

According to PBP; best way practice is to use local reference for file
handle and three argument open. Also perl prototypes are a mistake.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: WANG Cong <amwang@redhat.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoexport_report: fix perl warnings
Stephen Hemminger [Mon, 22 Feb 2010 23:17:22 +0000 (15:17 -0800)]
export_report: fix perl warnings

Use local file handles, use three argument open.
Don't modify arguments in perl grep (use sed instead)

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: WANG Cong <amwang@redhat.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoprofile2linkerlist: fix perl warnings
Stephen Hemminger [Mon, 22 Feb 2010 23:17:20 +0000 (15:17 -0800)]
profile2linkerlist: fix perl warnings

Turn on strict checking.
Simplify code by using "unless" statement.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: WANG Cong <amwang@redhat.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agonamespace: perlcritic warnings
Stephen Hemminger [Mon, 22 Feb 2010 23:17:18 +0000 (15:17 -0800)]
namespace: perlcritic warnings

Use local file handle not global.
Make loop and other variables local in scope.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Cc: Hui Zhu <teawater@gmail.com>
Cc: Cong Wang <amwang@redhat.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agocheckversion: perl cleanup
Stephen Hemminger [Mon, 22 Feb 2010 23:17:14 +0000 (15:17 -0800)]
checkversion: perl cleanup

Turn on strict checking.
Use three arguement open
Standard practice in perl is to use undef not zero for false

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Cc: Cong Wang <amwang@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agocheckincludes: fix perlcritic warnings
Stephen Hemminger [Mon, 22 Feb 2010 23:17:12 +0000 (15:17 -0800)]
checkincludes: fix perlcritic warnings

Turn on strict checking.
Use local file handles.
Use three argument open.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Cc: Cong Wang <amwang@redhat.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts: improve checkstack
Stephen Hemminger [Mon, 22 Feb 2010 23:17:09 +0000 (15:17 -0800)]
scripts: improve checkstack

Cleanup checkstack script:
  * Turn on strict checking
  * Fix resulting error message because the declaration syntax
    was incorrect.
  * Remove incorrect and misleading use of prototype
     - prototype not required for this type of sort function
       because $a and $b are being used in this contex
     - if prototype was being used it should be for both arguments
  * Use closure for sort function

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Cong Wang <amwang@redhat.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: deb-pkg md5sums
FEJES Jozsef [Fri, 5 Mar 2010 17:19:36 +0000 (18:19 +0100)]
kbuild: deb-pkg md5sums

This patch creates the standard md5sums file for 'make deb-pkg' just
like the dh_md5sums debhelper script.

Signed-off-by: Jozsef Fejes <fejes@joco.name>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .text.start to .text..start.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:57 +0000 (01:03 +0100)]
Rename .text.start to .text..start.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename special text sections in arch/frv from .text.XXX to .text..XXX.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:56 +0000 (01:03 +0100)]
Rename special text sections in arch/frv from .text.XXX to .text..XXX.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.lock_aligned to .data..lock_aligned.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:55 +0000 (01:03 +0100)]
Rename .data.lock_aligned to .data..lock_aligned.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.initvect to .data..initvect.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:54 +0000 (01:03 +0100)]
Rename .data.initvect to .data..initvect.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.init to .data..init.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:53 +0000 (01:03 +0100)]
Rename .data.init to .data..init.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.nosave to .data..nosave.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:52 +0000 (01:03 +0100)]
Rename .data.nosave to .data..nosave.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .text.startup to .text..startup.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:51 +0000 (01:03 +0100)]
Rename .text.startup to .text..startup.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .text.page_aligned to .text..page_aligned.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:49 +0000 (01:03 +0100)]
Rename .text.page_aligned to .text..page_aligned.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .text.lock to .text..lock.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:48 +0000 (01:03 +0100)]
Rename .text.lock to .text..lock.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .text.ivt to .text..ivt.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:47 +0000 (01:03 +0100)]
Rename .text.ivt to .text..ivt.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .rodata.compressed to .rodata..compressed.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:46 +0000 (01:03 +0100)]
Rename .rodata.compressed to .rodata..compressed.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.vmpages and .data.vm0.XXX to .data..vmpages and .data..vm0.XXX.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:45 +0000 (01:03 +0100)]
Rename .data.vmpages and .data.vm0.XXX to .data..vmpages and .data..vm0.XXX.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.read_mostly to .data..read_mostly.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:44 +0000 (01:03 +0100)]
Rename .data.read_mostly to .data..read_mostly.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data[.percpu][.XXX] to .data[..percpu][..XXX].
Denys Vlasenko [Sat, 20 Feb 2010 00:03:43 +0000 (01:03 +0100)]
Rename .data[.percpu][.XXX] to .data[..percpu][..XXX].

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data..patch.XXX to .data..patch.XXX.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:42 +0000 (01:03 +0100)]
Rename .data..patch.XXX to .data..patch.XXX.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.init_irqstack to .data..init_irqstack.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:41 +0000 (01:03 +0100)]
Rename .data.init_irqstack to .data..init_irqstack.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.gate to .data..gate.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:40 +0000 (01:03 +0100)]
Rename .data.gate to .data..gate.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .bss.stack to .bss..stack.
Denys Vlasenko [Sat, 20 Feb 2010 00:03:39 +0000 (01:03 +0100)]
Rename .bss.stack to .bss..stack.

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .bss.page_aligned to .bss..page_aligned.
Tim Abbott [Sat, 20 Feb 2010 00:03:38 +0000 (01:03 +0100)]
Rename .bss.page_aligned to .bss..page_aligned.

Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.page_aligned to .data..page_aligned.
Tim Abbott [Sat, 20 Feb 2010 00:03:37 +0000 (01:03 +0100)]
Rename .data.page_aligned to .data..page_aligned.

Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agopowerpc: remove unused __page_aligned definition.
Tim Abbott [Sat, 20 Feb 2010 00:03:36 +0000 (01:03 +0100)]
powerpc: remove unused __page_aligned definition.

There is already an architecture-independent __page_aligned_data macro
for this purpose, so removing the powerpc-specific macro should be
harmless.

Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.init_task to .data..init_task.
Tim Abbott [Sat, 20 Feb 2010 00:03:35 +0000 (01:03 +0100)]
Rename .data.init_task to .data..init_task.

Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoRename .data.cacheline_aligned to .data..cacheline_aligned.
Tim Abbott [Sat, 20 Feb 2010 00:03:34 +0000 (01:03 +0100)]
Rename .data.cacheline_aligned to .data..cacheline_aligned.

Signed-off-by: Tim Abbott <tabbott@ksplice.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agotags: Add the ability to make tags for all archs using "all"
John Kacur [Tue, 2 Mar 2010 15:57:52 +0000 (16:57 +0100)]
tags: Add the ability to make tags for all archs using "all"

make ALLSOURCE_ARCHS=all tags
- Document this in kbuild.txt
Without this change you have to type each arch separately.

Signed-off-by: John Kacur <jkacur@redhat.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agotags: Fix spelling error in comment (is->if)
John Kacur [Tue, 2 Mar 2010 14:38:10 +0000 (15:38 +0100)]
tags: Fix spelling error in comment (is->if)

Signed-off-by: John Kacur <jkacur@redhat.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agotags: Use $SRCARCH
Michal Marek [Fri, 19 Feb 2010 15:18:41 +0000 (16:18 +0100)]
tags: Use $SRCARCH

$ make mrproper
$ make tags
  GEN     tags
  find: `arch/x86_64/': No such file or directory

Caused by commit f81b1be (tags: include headers before source files)

Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: fix a couple of typos in Documentation
Kirill Smelkov [Wed, 17 Feb 2010 08:45:33 +0000 (11:45 +0300)]
kbuild: fix a couple of typos in Documentation

Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agotags: include headers before source files
Guennadi Liakhovetski [Sun, 7 Feb 2010 23:25:59 +0000 (00:25 +0100)]
tags: include headers before source files

Currently looking up a structure definition in TAGS / tags takes one to
one of multiple "static struct X" definitions in arch sources, which makes
it for many structs practically impossible to get to the required header.
This patch changes the order of sources being tagged to first scan
architecture includes, then the top-level include/ directory, and only
then the rest. It also takes into account, that many architectures have
more than one include directory, i.e., not only arch/$ARCH/include, but
also arch/$ARCH/mach-X/include etc.

Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Reviewed-by: WANG Cong <xiyou.wangcong@gmail.com>
[mmarek@suse.cz: fix 'var+=text' bashism]
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: move -fno-dwarf2-cfi-asm to powerpc only
Andi Kleen [Tue, 2 Feb 2010 22:40:02 +0000 (14:40 -0800)]
kbuild: move -fno-dwarf2-cfi-asm to powerpc only

Better dwarf2 unwind information is a good thing, it allows better
debugging with kgdb and crash and helps systemtap.

Commit 003086497f07f7f1e67c0c295e261740f822b377 ("Build with
-fno-dwarf2-cfi-asm") disabled some CFI information globally to work
around a module loader bug on powerpc.

But this disables the better unwind tables for all architectures, not just
powerpc.  Move the workaround to powerpc and also add a suitable comment
that's it really a workaround.

This improves dwarf2 unwind tables on x86 at least.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomarkup_oops.pl: minor fixes
Hui Zhu [Mon, 1 Feb 2010 05:41:22 +0000 (13:41 +0800)]
markup_oops.pl: minor fixes

1. Fix a little format issue.
2. Check the return of "Getopt::Long::GetOptions".  Output usage and
   exit if it get error.
3. Change $ARGV[$#ARGV] to $ARGV[0].
4. Change the code which get $modulefile from modinfo.  Replace the
   pipeline with `modinfo -F filename $module`.
4. Change usage from "Specify the module directory name" to "Specify the
   module filename".

Signed-off-by: Hui Zhu <teawater@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomarkup_oops.pl: add options to improve cross-sompilation environments
Hui Zhu [Tue, 26 Jan 2010 09:13:07 +0000 (17:13 +0800)]
markup_oops.pl: add options to improve cross-sompilation environments

The markup_oops.pl have 3 troubles to support cross-compiler environment:
1.  It use objdump directly.
2.  It use modinfo to get the message of module.
3.  It use hex function that cannot support 64-bit number in 32-bit arch.

This patch add 3 options to markup_oops.pl:
1. -c CROSS_COMPILE Specify the prefix used for toolchain.
2. -m MODULE_DIRNAME Specify the module directory name.
3. Change hex function to Math::BigInt->from_hex.

After this patch, parse the x8664 oops in x86, we can:
cat amd64m | perl ~/kernel/tmp/m.pl -c /home/teawater/kernel/bin/x8664- -m ./e.ko vmlinux

Thanks,
Hui

Signed-off-by: Hui Zhu <teawater@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: ozan@pardus.org.tr
Cc: Matthew Wilcox <willy@linux.intel.com>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoMerge branch 'modpost' of git://git.pengutronix.de/git/ukl/linux-2.6 into kbuild...
Michal Marek [Tue, 2 Feb 2010 15:21:08 +0000 (16:21 +0100)]
Merge branch 'modpost' of git://git.pengutronix.de/git/ukl/linux-2.6 into kbuild/for-next

Acked-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoMAINTAINERS: add a few more patterns to kbuild
Michal Marek [Fri, 29 Jan 2010 13:22:43 +0000 (14:22 +0100)]
MAINTAINERS: add a few more patterns to kbuild

Also, add a note that "unmaintained" files below scripts/ should go via
the kbuild tree (best current practice).

Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts: use %_tmppath in "make rpm-pkg"
John Saalwaechter [Mon, 1 Feb 2010 00:18:58 +0000 (16:18 -0800)]
scripts: use %_tmppath in "make rpm-pkg"

The mkspec script hardcodes "/var/tmp" into the generated rpm spec file's
BuildRoot. The user, however, may have a custom setting for %_tmppath,
which should be used in BuildRoot.  This patch changes mkspec's
BuildRoot output to appropriately use %_tmppath.

Signed-off-by: John Saalwaechter <saalwaechter@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomarkup_oops.pl: fix for faulting instruction in the first line of a range
Hui Zhu [Thu, 28 Jan 2010 06:58:02 +0000 (06:58 +0000)]
markup_oops.pl: fix for faulting instruction in the first line of a range

I got a "No matching code found" when I use markup_oops.pl parse a error
in a x86_64 module.

cat e.c

int init_module(void)
{
char *buf = 0;

buf[0] = 3;

return 0;
}

void cleanup_module(void)
{
//char *buf = 0;

//buf[0] = 3;
}

MODULE_AUTHOR("Hui Zhu");
MODULE_LICENSE("GPL");

0000000000000000 <init_module>:
init_module():
/home/teawater/study/kernel/stack2core/example/e.c:10
   0: c6 04 25 00 00 00 00  movb   $0x3,0x0
   7: 03
/home/teawater/study/kernel/stack2core/example/e.c:13
   8: 31 c0                 xor    %eax,%eax
   a: c3                    retq
   b: 0f 1f 44 00 00        nopl   0x0(%rax,%rax,1)

0000000000000010 <cleanup_module>:
cleanup_module():
/home/teawater/study/kernel/stack2core/example/e.c:20
  10: f3 c3                 repz retq
  12: 90                    nop
  13: 90                    nop
Disassembly of section .modinfo:

This is because the faulting instruction "movb   $0x3,0x0" is the first
line of the range.

In the markup_oops.pl:
main::(./scripts/markup_oops.pl:245):
245: if (InRange($1, $target)) {
  DB<2> p $line
ffffffffa001b000: c6 04 25 00 00 00 00  movb   $0x3,0x0
  DB<3> p $counter
0

It just set $center in next loop. So it cannot get the $center.

And even if $center is set to the right value 0.
if ($center == 0) {
print "No matching code found \n";
exit;
}
The first line $center will be 0, so I change the default value to -1.

Signed-off-by: Hui Zhu <teawater@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts: change scripts to use system python instead of env
Don Zickus [Tue, 26 Jan 2010 22:20:41 +0000 (22:20 +0000)]
scripts: change scripts to use system python instead of env

Just a small change to a couple of scripts to go from

 #!/usr/bin/env python

to

 #!/usr/bin/python

This shouldn't effect anyone, unless they don't install python there.

In preparation for python3, Fedora is doing a big push to change the scripts
to use the system python.  This allows developers to put the python3 in
their path without fear of breaking existing scripts.

Now I am pretty sure anyone using python3 for testing purposes will probably
not run any of the scripts I changed, but Fedora has this automated tool
that checks for this stuff so I thought I would try to push it upstream.

Signed-off-by: Don Zickus <dzickus@redhat.com>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Acked-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts/kallsyms: suppress build warning
Himanshu Chauhan [Thu, 28 Jan 2010 00:53:20 +0000 (16:53 -0800)]
scripts/kallsyms: suppress build warning

Suppress a warn_unused_result warning.

fgets is called as a part of error handling.  It is called just to drop a
line and return immediately.  read_map is reading the file in a loop and
read_symbol reads line by line.  So I think there is no point in using
return value for useful checking.  Other checks like 3 items were returned
or !EOF have already been done.

Signed-off-by: Himanshu Chauhan <hschauhan@nulltrace.org>
Cc: WANG Cong <xiyou.wangcong@gmail.com>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts/mkcompile_h: don't test for hardcoded paths
Michal Marek [Wed, 27 Jan 2010 08:46:23 +0000 (09:46 +0100)]
scripts/mkcompile_h: don't test for hardcoded paths

Don't test for /bin/{dnsdomainname,domainname}, simply try to execute
the command and check if it returned something.

Reported-by: Glenn Sommer <glemsom@gmail.com>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Tested-by: Glenn Sommer <glemsom@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoImprove kconfig symbol hashing
Andi Kleen [Wed, 13 Jan 2010 16:02:44 +0000 (17:02 +0100)]
Improve kconfig symbol hashing

While looking for something else I noticed that the symbol
hash function used by kconfig is quite poor. It doesn't
use any of the standard hash techniques but simply
adds up the string and then uses power of two masking,
which is both known to perform poorly.

The current x86 kconfig has over 7000 symbols.

When I instrumented it showed that the minimum hash chain
length was 16 and a significant number of them was over
30.

It didn't help that the hash table size was only 256 buckets.

This patch increases the hash table size to a larger prime
and switches to a FNV32 hash. I played around with a couple of hash
functions, but that one seemed to perform best with reasonable
hash table sizes.

Increasing the hash table size even further didn't
seem like a good idea, because there are a couple of global
walks which walk the complete hash table.

I also moved the unnamed bucket to 0. It's still the longest
of all the buckets (44 entries), but hopefully it's not
often hit except for the global walk which doesn't care.

The result is a much nicer distribution:
(first column bucket length, second number of buckets with that length)

1: 3505
2: 1236
3: 294
4: 52
5: 3
47: 1 <--- this is the unnamed symbols bucket

There are still some 5+ buckets, but increasing the hash table
even more would be likely not worth it.

This also cleans up the code slightly by removing hard coded
magic numbers.

I didn't notice a big performance difference either way
on my Nehalem system, but I presume it'll help somewhat
on slower systems.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoMakefile: Document ability to make file.lst and file.S
Joe Perches [Wed, 13 Jan 2010 17:31:44 +0000 (09:31 -0800)]
Makefile: Document ability to make file.lst and file.S

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: improve version string logic
David Rientjes [Wed, 13 Jan 2010 21:01:05 +0000 (13:01 -0800)]
kbuild: improve version string logic

The LOCALVERSION= string passed to "make" will now always be appended to
the kernel version after CONFIG_LOCALVERSION, if it exists, regardless of
whether CONFIG_LOCALVERSION_AUTO is set or not.  This allows users to
uniquely identify their kernel builds with a string.

If CONFIG_LOCALVERSION_AUTO is enabled, the unique SCM tag reported by
setlocalversion (or .scmversion) is appended to the kernel version, if it
exists.  When CONFIG_LOCALVERSION_AUTO is not enabled, a `+' is appended
to the kernel version to represent that the kernel has been revised since
the last release unless "make LOCALVERSION=" was used to uniquely identify
the build.

The end result is this:

 - when LOCALVERSION= is passed to "make", it is appended to the kernel
   version,

 - when CONFIG_LOCALVERSION_AUTO is enabled, a unique SCM identifier is
   appended if the respository has been revised beyond a tagged commit,
   and

 - when CONFIG_LOCALVERSION_AUTO is disabled, a `+' is appended if the
   repository has been revised beyond a tagged commit and LOCALVERSION=
   was not passed to "make".

Examples:

With CONFIG_LOCALVERSION_AUTO: "make" results in
v2.6.32-rc4-00149-ga3ccf63.  If there are uncommited changes to the
respository, it results in v2.6.32-rc4-00149-ga3ccf63-dirty.  If
"make LOCALVERSION=kbuild" were used, it results in
v2.6.32-rc4-kbuild-00149-ga3ccf63-dirty.

Without CONFIG_LOCALVERSION_AUTO, "make" results in v2.6.32-rc4+
unless the repository is at the Linux v2.6.32-rc4 commit (in which
case the version would be v2.6.32-rc4).  If "make LOCALVERSION=kbuild"
were used, it results in v2.6.32-rc4-kbuild.

Also renames variables such as localver-auto and _localver-auto to more
accurately describe what they represent: localver-extra and
scm-identifier, respectively.

Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agonconfig: minor fix
Nir Tzachar [Wed, 13 Jan 2010 05:32:35 +0000 (07:32 +0200)]
nconfig: minor fix

This patch fixes two problems reported by Jan Engelhardt:
1) Border is now properly placed, to always be visible
2) Long menu items are properly displayed

Reported-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Nir Tzachar <nir.tzachar@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agonconfig: mark local functions as such
Michal Marek [Thu, 7 Jan 2010 12:59:57 +0000 (13:59 +0100)]
nconfig: mark local functions as such

scripts/kconfig/nconf.gui.c:23: warning: no previous prototype for 'set_normal_colors'
scripts/kconfig/nconf.gui.c:68: warning: no previous prototype for 'normal_color_theme'
scripts/kconfig/nconf.gui.c:100: warning: no previous prototype for 'no_colors_theme'
scripts/kconfig/nconf.c:455: warning: no previous prototype for 'process_special_keys'
scripts/kconfig/nconf.c:487: warning: no previous prototype for 'get_next_hot'
scripts/kconfig/nconf.c:506: warning: no previous prototype for 'canbhot'
scripts/kconfig/nconf.c:514: warning: no previous prototype for 'is_hot'
scripts/kconfig/nconf.c:522: warning: no previous prototype for 'make_hot'
scripts/kconfig/nconf.c:582: warning: no previous prototype for 'item_make'
scripts/kconfig/nconf.c:626: warning: no previous prototype for 'item_add_str'
scripts/kconfig/nconf.c:656: warning: no previous prototype for 'item_tag'
scripts/kconfig/nconf.c:668: warning: no previous prototype for 'curses_item_index'
scripts/kconfig/nconf.c:673: warning: no previous prototype for 'item_data'
scripts/kconfig/nconf.c:684: warning: no previous prototype for 'item_is_tag'
scripts/kconfig/nconf.c:691: warning: no previous prototype for 'set_config_filename'

Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig: new configuration interface (nconfig)
nir.tzachar@gmail.com [Wed, 25 Nov 2009 10:28:43 +0000 (12:28 +0200)]
kconfig: new configuration interface (nconfig)

This patch was inspired by the kernel projects page, where an ncurses
replacement for menuconfig was mentioned (by Sam Ravnborg).

Building on menuconfig, this patch implements a more modern look
interface using ncurses and ncurses' satellite libraries (menu, panel,
form). The implementation does not depend on lxdialog, which is
currently distributed with the kernel.

Signed-off-by: Nir Tzachar <nir.tzachar@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agogenksyms: close ref_file after use
Alexander Beregalov [Thu, 7 Jan 2010 02:22:41 +0000 (05:22 +0300)]
genksyms: close ref_file after use

It is the last place when the file is read, so close it.

Signed-off-by: Alexander Beregalov <a.beregalov@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomenuconfig: wrap long help lines
Vadim Bendebury (вб) [Sun, 20 Dec 2009 08:29:49 +0000 (00:29 -0800)]
menuconfig: wrap long help lines

Help text for certain config options is very extensive (the text
includes the names of all  other options the option in question depends
on). Long lines are not wrapped, making it impossible to see the list
without scrolling horizontally.

This patch adds some logic which wraps help screen lines at word
boundaries to prevent truncating.

Tested by running

  ARCH=powerpc make menuconfig O=/tmp/build

which shows that the long lines are now wrapped, and

 ARCH=powerpc make xconfig O=/tmp/build

to demonstrate that it still compiles and operates as expected.

Signed-off-by: Vadim Bendebury <vbendeb@google.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoscripts: add ARM support to decodecode
Rabin Vincent [Tue, 5 Jan 2010 14:57:58 +0000 (20:27 +0530)]
scripts: add ARM support to decodecode

This patch adds support for decoding ARM oopses to scripts/decodecode.
The following things are handled:

 - ARCH and CROSS_COMPILE environment variables are respected.

 - The Code: in x86 oopses is in bytes, while it is in either words (4
   bytes) or halfwords for ARM.

 - Some versions of ARM objdump refuse to disassemble instructions
   generated by literal constants (".word 0x...").  The workaround is to
   strip the object file first.

 - The faulting instruction is marked (liked so) in ARM, but <like so>
   in x86.

 - ARM mnemonics may include characters such as [] which need to be
   escaped before being passed to sed for the "<- trapping instruction"
   substitution.

Signed-off-by: Rabin Vincent <rabin@rab.in>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokconfig CROSS_COMPILE option
Roland McGrath [Tue, 22 Dec 2009 00:24:06 +0000 (16:24 -0800)]
kconfig CROSS_COMPILE option

This adds CROSS_COMPILE as a kconfig string so you can store it in
.config.  Then you can use plain "make" in the configured kernel build
directory to do the right cross compilation without setting the
command-line or environment variable every time.

With this, you can set up different build directories for different kernel
configurations, whether native or cross-builds, and then use the simple:

make -C /build/dir M=module-source-dir

idiom to build modules for any given target kernel, indicating which one
by nothing but the build directory chosen.

I tried a version that defaults the string with env="CROSS_COMPILE" so
that in a "make oldconfig" with CROSS_COMPILE in the environment you can
just hit return to store the way you're building it.  But the kconfig
prompt for strings doesn't give you any way to say you want an empty
string instead of the default, so I punted that.

Signed-off-by: Roland McGrath <roland@redhat.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Anibal Monsalve Salazar <anibal@debian.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Michal Marek <mmarek@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agokbuild: Fix linking error built-in.o no such file or directory
Jiafu He [Fri, 18 Dec 2009 01:22:13 +0000 (18:22 -0700)]
kbuild: Fix linking error built-in.o no such file or directory

This patch fixes the link error "built-in.o: no such file or directory".
The problem happens if "dirx/Makefile" contains only "obj-m += diry/
dirz/" and the empty "dirx/built-in.o" is missing. Adding $(subdir-m)
into check for builtin-target fixes this error.

Signed-off-by: Jiafu He <jay@goldhive.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agomodpost: don't allow *driver to reference .init.*
Uwe Kleine-König [Sat, 30 Jan 2010 20:14:23 +0000 (21:14 +0100)]
modpost: don't allow *driver to reference .init.*

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
11 years agomodpost: make symbol white list a per mismatch type variable
Uwe Kleine-König [Sat, 30 Jan 2010 19:52:50 +0000 (20:52 +0100)]
modpost: make symbol white list a per mismatch type variable

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
11 years agomodpost: remove now unused NO_MISMATCH constant
Uwe Kleine-König [Sat, 30 Jan 2010 15:57:48 +0000 (16:57 +0100)]
modpost: remove now unused NO_MISMATCH constant

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
11 years agomodpost: pass around const struct sectioncheck * instead of enum mismatch
Uwe Kleine-König [Sat, 30 Jan 2010 15:56:20 +0000 (16:56 +0100)]
modpost: pass around const struct sectioncheck * instead of enum mismatch

This prepares having a per-check whitelist of symbol names.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
11 years agomodpost: give most mismatch constants a better name
Uwe Kleine-König [Sat, 30 Jan 2010 15:35:47 +0000 (16:35 +0100)]
modpost: give most mismatch constants a better name

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
11 years agoLinux 2.6.33-rc6 v2.6.33-rc6
Linus Torvalds [Fri, 29 Jan 2010 21:57:50 +0000 (13:57 -0800)]
Linux 2.6.33-rc6

11 years agomodpost: define ALL_XXX{IN,EX}IT_SECTIONS
Uwe Kleine-König [Fri, 29 Jan 2010 11:04:26 +0000 (12:04 +0100)]
modpost: define ALL_XXX{IN,EX}IT_SECTIONS

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>