Pull localmodconfig updates from Steven Rostedt:
"Improve localmodconfig to remove even more unused module configs.
These changes drastically improve the amount of module configs removed
from a config file. It also adds some debug that I can have users
easily enable if things do not work for them."
* tag 'localmodconfig-v3.6' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-kconfig:
localmodconfig: Add debug environment variable LOCALMODCONFIG_DEBUG
localmodconfig: Check if configs are already set for selects
localmodconfig: Read in orig config file to avoid extra processing
localmodconfig: Comments and cleanup for streamline_config.pl
The make target 'oldnoconfig' is a misnomer. It doesn't set new symbols
to 'n', but instead sets it to their default values.
Unfortunately, assuming that it actually did this, broke ktest in some
of its tests. For example, the tests to create a minimum config and even
a config bisect, depends on removing various configs and using
oldnoconfig to get rid of other configs that may have depended on it.
But because some configs that it was trying to disable, were in fact
default enabled, this caused those configs to re-enable and corrupt the
test.
I thought about fixing oldnoconfig, but I'm afraid that people are
already dependent on its current behavior. Instead, I'm just updating
the documentation to state that it sets the new symbols to their default
values and not to 'n'.
Ideally, this would be called, 'olddefconfig' and we have an
'oldnoconfig' that actually disables the new symbols. But it's useless
for me now. If it changed, ktest would need to be consistent between
each version, and that would be to difficult to detect. I'll handle this
issue with ktest with other means.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
They function just like they do in less(1).
Also correct some discrepancy between the help text and the code wrt
function keys.
Signed-off-by: Benjamin Poirier <bpoirier@suse.de>
Signed-off-by: Michal Marek <mmarek@suse.cz>
For some config options (CONFIG_EXTRA_FIRMWARE, for example), the length
of a config file line can exceed the 1024 byte buffer.
Switch from fgets to compat_getline to fix. compat_getline is an
internally implimented getline work-alike for portability purposes.
Signed-off-by: Cody Schafer <cody@linux.vnet.ibm.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Commit 5a6f8d2bd9 ("kconfig: nuke
LKC_DIRECT_LINK cruft") removed all traces of lkc_defs.h from the tree.
Remove its entries in dontdiff and kconfig's .gitignore file too.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Various schemes exist to allow parallel installations of multiple major
versions of Qt (4.x with the previous 3.x and/or the upcoming 5.x).
QtCore.pc includes a moc_location variable which should be a more reliable
way to find moc.
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Import libraries on Cygwin and MinGW/MSYS use the .dll.a suffix, so
checking this suffix is necessary to make sure ncurses will still be
found when built without static libraries.
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Commit 8c41e5e363 added a check for
ncursesw/curses.h for the case where ncurses and ncursesw are build
separately but only one is installed. But if both are installed,
the headers ncurses/curses.h and ncursesw/curses.h differ, and since
libncursesw will be found first, so should ncursesw/curses.h.
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Signed-off-by: Michal Marek <mmarek@suse.cz>
ESCDELAY is a global variable which is replaced by getter and setter
functions with NCURSES_REENTRANT. This fixes the following error:
nconf.c: In function ‘main’:
nconf.c:1506:2: error: lvalue required as left operand of assignment
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Signed-off-by: Michal Marek <mmarek@suse.cz>
If the environment variable LOCALMODCONFIG_DEBUG is set, then debug output
will appear in the make localmodconfig. This will simplify debugging what
people get with their output, as I can just tell people to do:
LOCALMODCONFIG_DEBUG=1 make localmodconfig 2>out.txt
and have them send me the out.txt. I'll be able to see why things are not
working as they think it should be.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
There are some cases that a required module does not have a prompt
and needs to have another module enabled that selects it to be set.
As localmodconfig is conservative and tries to make the minimum config
without breaking the user's kernel, or keeping the user from using
devices that were loaded when the lsmod was done, all modules that
select this module will also be enabled.
If you needed module A, but module A did not have a prompt but needed
module B to be selected, localmodconfig would make sure B was still
enabled. If not only B selected A, but C, D, E, F, and G also
selected A, then all of those would also be included, as well as the
modules they depend on. This ballooned the number of configs that
localmodconfig would keep.
The fix here is to process the depends first, and then record those
configs that did not have a prompt and needed to be selected.
After the depends are done, check what configs are needed to select
the configs in the list, and if a config that selects it is already
set, then we don't need to do anything else.
If no config that selects the config is set, then just pick one and
try again.
This change brought down the number of selected modules from 290
to 67! Both before and after were run against a config that had 3095
modules enabled.
Tested-by: John David Yost <johnyost@ptd.net> # AlleyTrotter
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Read in the entire config file. If there's a config that we depend on
that happens to be in the core set (not a module) then we do not need
to process it as a module.
Currently, we follow the entire depend and selects even if they
are enabled as core and not modules. By checking to make sure that we
only look at modules we can drop the count a little.
From one of my tests, localmodconfig went from taking 3095 set modules
down to 356 before this patch, and down to 290 modules after the change.
Tested-by: John David Yost <johnyost@ptd.net> # AlleyTrotter
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Added some more comments and cleaned up part of the the code to use
a named variable instead of one of the special $1 perl variables.
No functional changes.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Pull kconfig changes from Michal Marek:
- Error handling for make KCONFIG_ALLCONFIG=<...> all*config plus a fix
for a bug that was exposed by this
- Fix for the script/config utility.
* 'kconfig' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
scripts/config: properly report and set string options
kbuild: all{no,yes,mod,def,rand}config only read files when instructed to.
kconfig: Add error handling to KCONFIG_ALLCONFIG
Hardware with MCA bus is limited to 386 and 486 class machines
that are now 20+ years old and typically with less than 32MB
of memory. A quick search on the internet, and you see that
even the MCA hobbyist/enthusiast community has lost interest
in the early 2000 era and never really even moved ahead from
the 2.4 kernels to the 2.6 series.
This deletes anything remaining related to CONFIG_MCA from core
kernel code and from the x86 architecture. There is no point in
carrying this any further into the future.
One complication to watch for is inadvertently scooping up
stuff relating to machine check, since there is overlap in
the TLA name space (e.g. arch/x86/boot/mca.c).
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: x86@kernel.org
Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Prevent subtle surprises to both people working on the kconfig code
and people using make allnoconfig allyesconfig allmoconfig and
randconfig by only attempting to read a config file if
KCONFIG_ALLCONFIG is set.
Common sense suggests attempting to read the extra config files does
not make sense unless requested. The documentation says the code
won't attempt to read the extra config files unless requested.
Current usage does not appear to include people depending on the code
reading the config files without the variable being set So do the
simple thing and stop reading config files when passed
all{no,yes,mod,def,rand}config unless KCONFIG_ALLCONFIG environment
variable is set.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Michal Marek <mmarek@suse.cz>
- Only try to read the file specified if KCONFIG_ALL_CONFIG is set to
something other than the empty string or "1".
- Don't use stat to check the name passed to conf_read_simple so that
zconf_fopen can find the file in the current directory or in SRCTREE
removing a extremely source of confusing failure, where KCONFIG_ALL_CONFIG
was not interpreted with respect to the directory make was called in.
- If conf_read_simple fails complain clearly and stop processing.
Allowing the simple debugging of typos.
- Clearly document the behavior so it is clear to users which
values are treated as flags and which values are treated as
filenames.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
We've now fixed IS_ENABLED() and friends to not require any special
"__enabled_" prefixed versions of the normal Kconfig options, so delete
the last traces of them being generated.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This reverts commit 953742c8fe.
Dumping two lines into autoconf.h for all existing Kconfig options
results in a giant file (~16k lines) we have to process each time we
compile something. We've weaned IS_ENABLED() and similar off of
requiring the __enabled_ definitions so now we can revert the change
which caused all the extra lines.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Provide a -r option to display when fragments contain redundant
options. This is really useful when breaking apart a config into
fragments, as well as cleaning up older fragments.
Signed-off-by: John Stultz <john.stultz@linaro.org>
Acked-by: Darren Hart <dvhart@linux.intel.com>
Acked-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Take the first config fragment and use it verbatim as the initial config
set. This avoids running the verification loop for the first file, as
nothing has actually been merged at this point. This significantly
increases performance for large config fragments.
Signed-off-by: Josh Boyer <jwboyer@redhat.com>
Acked-by: John Stultz <john.stultz@linaro.org>
Acked-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
