libkmod-elf: resolve CRC if module is built with MODULE_REL_CRCS
Normally exported symbol's crc is stored as absolute (SHN_ABS)
value of special named symbol __crc_<symbol name>.
When the kernel and modules are built with the config option
CONFIG_MODULE_REL_CRCS, all the CRCs are put in a special section
and the __crc_<symbol name> symbols values are offsets in the
section. See patch description of the commit:
shared/util.c: assert_cc() can only be used inside functions
shared/macro.h has two versions of assert_cc, one that uses gcc
_Static_assert(), which requires recent enough gcc versions, and one
that uses a fake array to trigger a build error. The latter can only
work inside functions, so assert_cc() should only be used inside
functions.
Fixes the following build failure when building kmod with old gcc
versions such as gcc 4.3.x:
shared/util.c:52: error: expected identifier or '(' before 'do'
shared/util.c:52: error: expected identifier or '(' before 'while'
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Yauheni Kaliuta [Tue, 9 May 2017 19:09:24 +0000 (22:09 +0300)]
testsuite: add tests for external directory support
The following tests added:
- depmod_search_order_external_first -- checks if external module
is taken in use when it has higher priority;
- depmod_search_order_external_last -- checks if external module
is skipped when it has lower priority;
- test_modinfo_external -- checks if modinfo is able to look up
correct external module;
- modprobe_external -- checks if modprobe is able to look up
correct external module and loads it.
(here /lib/modules/$(uname -r)/kernel/module2.ko depends of
symbols, provided by /the/modules/dir/module1.ko and external has
higher priority).
modprobe and modinfo understand it out of box.
This is a pretty simple extention of existing logic, since now
depmod already is able to:
a) scan modules with full path from command line without -a
switch;
b) detects broken symbol dependencies and broken modversions,
what assumes, that modules are already are not built for the
existing kernel.
Lucas De Marchi [Fri, 2 Jun 2017 03:25:57 +0000 (20:25 -0700)]
test-depmod: avoid warning when we don't have zlib
testsuite/test-depmod.c:31:21: warning: ‘depmod_modules_order_for_compressed’ defined but not used [-Wunused-function]
static noreturn int depmod_modules_order_for_compressed(const struct test *t)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Yauheni Kaliuta [Tue, 9 May 2017 19:09:22 +0000 (22:09 +0300)]
depmod: rewrite depmod modules search with scratchbuf
The recursive search code used used pretty big, PATH_MAX,
automatic storage buffer for the module directory scanning. Some
time ago there was scratchbuf implemented, which dynamically
reallocates its buffer on demand. The patch takes it in use for
the scanning code also. The initial size is hardcoded to 256
bytes which sounds good enough for most usecases so there should
be not many reallocations.
Yauheni Kaliuta [Tue, 9 May 2017 19:09:21 +0000 (22:09 +0300)]
depmod: create depmod dir independent search function
Prepare to implement external directories support.
The patch splits depmod_modules_search() function to two
functions: depmod_modules_search(), called by the high level with
intention to search all possible modules, and
depmod_module_search_path(), which takes path as a parameter and
scans modules under the path only. Initially it is used to scan
the same depmod->cfg->dirname path only.
Yauheni Kaliuta [Mon, 20 Mar 2017 10:09:51 +0000 (12:09 +0200)]
depmod: fix errorpath memleaks in report cycles logic
The c7ce9f0c80f3d561078a78205a14c5ba7663cfdd commit (depmod:
handle nested loops) introduced a bunch of possible memory leaks
in error path. In the real world scenario it is not a problem,
since the utility quits if it detects any of the errors, but from
the programming point of view, it is not nice. So, add the
cleanups.
libkmod: modinfo: use own function for sig_key hex output
Refactor the code a bit to make it easier to extend for signature
output.
kmod_module_get_info() creats a hex string for the sig_key data
inplace. Separate it into own kmod_module_hex_to_string function
and handle the branch in the new kmod_module_info_append_hex,
keeping the same signature as the non-hex version.
For some reason the key for sig_id was set to "signature". The
length was calculated against the proper string, as the result in
the output it was truncated to "signat".
Pass the proper key to the kmod_module_info_append() call.
Lucas De Marchi [Fri, 24 Feb 2017 02:28:27 +0000 (18:28 -0800)]
man: make error message clearer for missing xstlproc
We can't do at configure phase since we actually ship the built man pages with
dist, so it's fine not having xsltproc if building from dist. If building from
the repository, it's better to have have a better message saying xsltproc was
not found than trying to execute the argument to xsltproc. Now message is:
XSLT depmod.d.5
/bin/sh: line 1: xsltproc: command not found
Instead of:
XSLT depmod.d.5
/bin/sh: --nonet: command not found
When a module is removed and re-inserted without unrefing, the
kmod_file is unconditionally re-opened. This results in a memory
and file descriptor leak.
Fix it by checking if the file is already open in
kmod_module_insert_module().
Lucas De Marchi [Mon, 23 Jan 2017 19:59:31 +0000 (11:59 -0800)]
testsuite: add test for kernel cmdline with quotes
Add some tests in which we quotes in kernel cmdline and also spaces
inside quotes. This doesn't yet cover the case in which quotes are used
for module name, wihch should be forbidden.
Lucas De Marchi [Fri, 11 Nov 2016 01:37:04 +0000 (23:37 -0200)]
build: fix build with disabled test modules
install: cannot stat 'testsuite/module-playground/mod-loop-f.ko': No
such file or directory
Makefile:2881: recipe for target 'rootfs' failed
make[1]: *** [rootfs] Error 1
make[1]: *** Waiting for unfinished jobs....
Makefile:2101: recipe for target 'check-recursive' failed
We need to ship pre-compiled binaries so it's possible to run
"make check" on servers without kernel headers.
depmod: ignore related modules in depmod_report_cycles
Only print actual cyclic dependencies. Print count of all the modules
in cyclic dependency at the end of the function so that dependent
modules which are not in cyclic chain can be ignored.
Printing dependent modules which are not in cyclic chain causes buffer
overflow as m->modnamesz is not included in buffer size calculations
(loop == m is never true). This buffer overflow causes kmod to crash.
Update depmod test to reflect the change as well.
Reported-by: Andreas Färber <afaerber@suse.de> Signed-off-by: Mian Yousaf Kaukab <yousaf.kaukab@suse.com>
The array elements in the tests are strings, what means "char *"
in С. The comparation funtion takes pointers to the elements, so
the arguments become "char **". It means, that strcmp() cannot be
used directrly.
The patch creates a wrapper on strcmp() which perfoms
dereferencing of the "char **" to supply the actual strings to
strcmp(), and uses the wrapper as a comparation function for the
qsort() call.
Lucas De Marchi [Wed, 10 Aug 2016 17:20:32 +0000 (14:20 -0300)]
Add scratchbuf implementation
This should fill the requirements for "we need to loop over a lot of
strings that usually are small enough to remain on stack, but we want to
protect ourselves against huge strings not fitting in the static
buffer we estimated as sufficient"
Lucas De Marchi [Wed, 10 Aug 2016 15:42:12 +0000 (12:42 -0300)]
util: fix warning of equal values on logical OR
shared/util.c: In function ‘read_str_safe’:
shared/util.c:211:24: warning: logical ‘or’ of equal expressions [-Wlogical-op]
if (errno == EAGAIN || errno == EWOULDBLOCK ||
^~
shared/util.c: In function ‘write_str_safe’:
shared/util.c:237:24: warning: logical ‘or’ of equal expressions [-Wlogical-op]
if (errno == EAGAIN || errno == EWOULDBLOCK ||
^~
This is because EAGAIN and EWOULDBLOCK have the same value. Prefer
EAGAIN, but add a static assert to catch if it's not the same in another
architecture.
Lucas De Marchi [Tue, 28 Jun 2016 15:14:38 +0000 (12:14 -0300)]
libkmod: fix integration with gtk-doc
It was failing to generate doc with recent version of gtk-doc
[kmod]$ ./bootstrap
libkmod/docs/gtk-doc.make:33: error: EXTRA_DIST must be set with '=' before using '+='
libkmod/docs/Makefile.am:29: 'libkmod/docs/gtk-doc.make' included from here
autoreconf: automake failed with exit status: 1
Michal Marek [Fri, 17 Jun 2016 14:04:15 +0000 (16:04 +0200)]
libkmod: Handle long lines in /proc/modules
kmod_module_new_from_loaded() calls fgets with a 4k buffer. When a
module such as usbcore is used by too many modules, the rest of the line
is considered a beginning of another lines and we eventually get errors
like these from lsmod:
libkmod: kmod_module_get_holders: could not open '/sys/module/100,/holders': No such file or directory
together with bogus entries in the output. In kmod_module_get_size, the
problem does not affect functionality, but the line numbers in error
messages will be wrong.
kmod: compiling with old sed version (!ERE support)
Makefile.am uses `sed -E', which it is found on BSD sed; however a
replacement on GNU sed would be `sed -r'. Both intend to use extended
regular expressions (ERE). However I have a system that does not support
those, in benefit for portability could you consider replacing ERE by BRE.
Lucas De Marchi [Mon, 11 Jan 2016 12:48:33 +0000 (10:48 -0200)]
travis: workaround bug in environment setup
Travis is poluting the environment, particularly PYTHON_CFLAGS which
makes the build to fail. Just unset the variable since we don't want to
override these cflags.
The -f switch is accepted by insmod, but silently ignored. This
causes the user to wonder why things don't work. As insmod is
most often used with "evil" modules, -f is almost default and
thus needs to work.
Signed-off-by: Philippe De Swert <philippe.deswert@jollamobile.com>
Laura Abbott [Mon, 28 Sep 2015 22:39:14 +0000 (15:39 -0700)]
depmod: Don't fall back to uname on bad version
Currently, if a value that doesn't match a kernel version
("%u.%u") is passed in, depmod silently falls back to
using uname. Rather than try and work around the caller passing
bad data, just exit out instead.
Laura Abbott [Mon, 28 Sep 2015 22:39:13 +0000 (15:39 -0700)]
depmod: Remove unprinted debug messages
In between the start of the program and the call to log_setup_kmod_log,
the only messages that will be printed are the ones at or above the
global default level. Debug messages in this range will never be printed
so remove them.
Laura Abbott [Mon, 28 Sep 2015 22:39:12 +0000 (15:39 -0700)]
Change default log level
The default log level is currently LOG_ERR. Tools can override this
default but there is a non-trivial amount of setup that needs to
happen before the log level can be changed. Since tools may want to
use the warn level for things such as deprecated flags, change the
default to LOG_WARNING to ensure messages get printed.
This is fairly cryptic and not at all obvious there is a problem
unless the error code is checked or verbose flags are used.
Update the error message to indicate a problem and print out the
directory that failed.
Natanael Copa [Mon, 22 Jun 2015 12:09:17 +0000 (14:09 +0200)]
build: let sed use posix ERE instead of GNU extension
Use POSIX Extended Regular Expression (ERE) instead of the GNU extension
\| in the install-exec-hook. This makes it create the symlink properly
with busybox sed built with musl libc. It will silently create a broken
symlink otherwise.
Lucas De Marchi [Sat, 13 Jun 2015 21:29:47 +0000 (18:29 -0300)]
libkmod-module: fix return code in error path
ENOSYS is the wrong errno to return when we don't find a module in
kmod_module_insert_module(). Why is it there in the first place? This
goes back to kmod v1 when we couldn't load modules by names, but we
should give a path instead.
708624a ("ELF: initial support for modinfo and strip of modversions and
vermagic.") changed that so we do a lazy-search by the module path in
this function. Later f304afe ("Change error message to reflect
reality") fixed the log message but the return coded remained the same.
Lucas De Marchi [Tue, 9 Jun 2015 05:01:11 +0000 (02:01 -0300)]
build: silence stderr on coverity rules
We try to execute git in order to get the dependencies for the coverity
rules. And it gets executed even when we are not calling that specific
rule. Later we may want to improve it, but for now let's just silence
the errors of not being a git repository when executing this on a
packaged version.
Lucas De Marchi [Sun, 7 Jun 2015 05:44:02 +0000 (02:44 -0300)]
build: add relevant features to KMOD_FEATURES in config.h
Not all the features (i.e. those available in --enable-* or --with-*)
are really relevant to the final user. Create a KMOD_FEATURES definition
in config.h containing these features.
Lucas De Marchi [Sun, 7 Jun 2015 04:15:30 +0000 (01:15 -0300)]
build: add function to build string of features
Add a CC_FEATURE_APPEND function that we can use to append the features
that are enabled/disabled. This will generate a single string in the
form "+FEATURE1 +FEATURE2 -FEATURE3".
Lucas De Marchi [Sun, 31 May 2015 18:38:47 +0000 (15:38 -0300)]
build: cache modules from playground
Now that we are able to build our own test modules, also allow to use
cached modules so a) kernel headers are not required and b) distro
maintainers are happy. It's still need a "--disable-test-modules" in
the configure since the default is enabled.
There's no license problems anymore since all modules come from our own
repository, we ship the sources and the modules can be easily rebuilt.
testsuite: remove .gitignore files when populating rootfs
Usually this file is added to keep a directory existing in the
repository but without any real content. In rootfs this can be
problematic if a directory will have all its files inspected. This
happens for kmod_module_get_holders().
Side-note: the 'test-loaded.c' is hit by this problem but doesn't
"notice" because the invalid module returned by get_holders() is not
checked. The modules in its loop are only used to get the name and
generate an output, and NULL was a valid value to generate the name.
Lucas De Marchi [Fri, 27 Feb 2015 18:39:36 +0000 (15:39 -0300)]
modprobe: stop checking initstate for builtin modules
builtin modules are handled in libkmod by looking at the modules.builtin
index. There's no need to check again for the module's initstate to
decide if it's builtin.
Lucas De Marchi [Fri, 27 Feb 2015 04:51:45 +0000 (01:51 -0300)]
testsuite: test builtin state
The second test, that creates the module by name and then retrieves the
initstate was broken before b95fa91 ('Fix race while loading modules').
We would check /sys and return either builtin (if the module has
parameters) or give an error because we don't find the module (even if
it's in the modules.builtin index)
Harish Jenny K N [Sun, 22 Feb 2015 18:41:07 +0000 (15:41 -0300)]
Fix race while loading modules
usecase: two sd cards are being mounted in parallel at same time on
dual core. example modules which are getting loaded is nls_cp437.
While one module is being loaded , it starts creating sysfs files.
meanwhile on other core, modprobe might return saying the module
is KMOD_MODULE_BUILTIN, which might result in not mounting sd card.
Experiments done to prove the issue in kmod.
Added sleep in kernel module.c at the place of creation of sysfs files.
Then tried `modprobe nls_cp437` from two different shells.
While the first was still waiting for its completion ,
the second one returned saying the module is built-in.
[ Lucas:
The problem is that the creation of /sys/module/<name> and
/sys/module/<name>/initstate are not atomic. There's a small window in
which the directory exists but the initstate file was still not
created.
Built-in modules can be handled by searching the modules.builtin file.
We actually lose some "modules" that create entries in /sys/modules
(e.g. vt) and are not in modules.builtin file: only those that can be
compiled as module are present in this file.
We enforce mod->builtin to always be up-to-date when
kmod_module_get_initstate() is called. This way if the directory
exists but the initstate doesn't, we can be sure this is because the
module is in the "coming" state, i.e. kernel didn't create the file
yet, but since builtin modules were already handled by checking our
index the only reason for that to happen is that we hit the race
condition.
I also added some tweaks to the patch, so we don't repeat the code for builtin
lookup. ]
Lucas De Marchi [Thu, 26 Feb 2015 16:02:04 +0000 (13:02 -0300)]
shared: avoid checking for __STDC_VERSION__
Since we pass -std=gnu99 to the compiler __STDC_VERSION__ is defined to
199901. Even in this case we can use _Noreturn. Delegate the testing to
the configure phase with a compile test.
In future we want to stop using AC_PROG_CC_99 and start using the new
AC_PROG_CC from autoconf 2.70, which tries to choose the most recent
standard.
Lucas De Marchi [Thu, 26 Feb 2015 05:56:04 +0000 (02:56 -0300)]
build: simplify and organize configure
- There's no need for the various AC_PROG_CC* macros
- No need for AC_C_BIGENDIAN: the only place we care about endianness,
we use __BYTE_ORDER
- Organize calls to be in similar order as systemd
- Disable --disable-largefile. We already had possible bugs in the
index implementation if this were passed. It's now fixed, but it's
never tested. Disable it.
- Bump autoconf requirement to 2.64
Lucas De Marchi [Thu, 26 Feb 2015 06:38:52 +0000 (03:38 -0300)]
depmod: fix leak in case of malloc(0)
malloc(0) can return != NULL. We need to pass the pointer to free().
This happens if index__haschildren(node) returned true, but
child_count is set to 0.
Lucas De Marchi [Wed, 25 Feb 2015 15:06:44 +0000 (12:06 -0300)]
depmod: use cleanup attribute to simplify free on exit
Reusing the root variable was a bad idea. Doing so we could call free()
on a variable that was not allocated. For example: "depmod -b / -h".
Since we would jump to cmdline_failed, root would not be duplicated.
Instead of fighting the order in the options, just used the cleanup
attribute and remove the calls to free() on "config_paths" and "root".
David Herrmann [Fri, 13 Sep 2013 10:42:12 +0000 (12:42 +0200)]
from systemd: build: check for build/link flags harder
Use AC_LINK_IFELSE instead of AC_COMPILE_IFELSE to test for flags that
might succeed during compilation but not during linking. An example is gcc
compiled with libssp support but gnu-ld without it. In this case
-fstack-protector works fine during compilation but fails during linking
as several internal helpers are missing.