| Age | Commit message (Collapse) | Author |
|---|
|
Someone hacked the http://hambedded site or it was moved and some
links to that site in the BB manual had been hijacked to point to
an entry portal for a pornography site. Replaced the link with an
archived version that restores the integrity of the links.
Signed-off-by: Scott Rifenbark <srifenbark@gmail.com>
|
|
self.appendlist is a dict and as such unordered. This can lead to cases
where appends with different names (e.g. x_%.bbappend vs. x_123.bbappend)
can be reordered in application which in turn reorders the variables
that those bbappend files might touch. Reorderd variables changes the sstate
cache signatures causing real world issues.
To avoid this, use a list for the append files instead.
This patch is conservative and just adds a new data structure alongside
the existing one and uses it to resolve the core issue. Later patches
(post release) can handle some of the wider but less problematic ones
(e.g. issues in bitbake-layers flatten).
[YOCTO #7511]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently, I'm not sure how the prserver managed to shut down cleanly. These
issues may explain some of the hangs people have reported.
This change:
* Ensures the connection acceptance thread monitors self.quit
* We wait for the thread to exit before exitting
* We sync the database when the thread exits
* We do what the comment mentions, timeout after 30s and sync the database
if needed. Previously, there was no timeout (the 0.5 applies to sockets,
not the Queue object)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
(cherry picked from commit 0926492295d485813d8a4f6b77c7b152e4c5b4c4)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If the PR server or indeed any other child process takes some time to
exit (which it sometimes does when saving its database), it can end up
holding bitbake.lock after the UI exits, which led to errors if you ran
bitbake commands successively - we saw this when running the PR server
oe-selftest tests in OE-Core. The recent attempt to fix this wasn't
quite right and ended up breaking memory resident bitbake. This time we
close the lock file when cooker shuts down (inside the UI process)
instead of unlocking it, and this is done in the cooker code rather than
the actual UI code so it doesn't matter which UI is in use. Additionally
we report that we're waiting for the lock to be released, using lsof or
fuser if available to list the processes with the lock open.
The 'magic' in the locking is due to all spawned subprocesses of bitbake
holding an open file descriptor to the bitbake.lock. It is automatically
unlocked when all those fds close the file (as all the processes terminate).
We close the UI copy of the lock explicitly, then close the server process
copy, any remaining open copy is therefore some proess exiting.
(The reproducer for the problem is to set PRSERV_HOST = "localhost:0"
and add a call to time.sleep(20) after self.server_close() in
lib/prserv/serv.py, then run "bitbake -p; bitbake -p" ).
Cleanup work done by Paul Eggleton <paul.eggleton@linux.intel.com>.
This reverts bitbake commit 69ecd15aece54753154950c55d7af42f85ad8606 and
e97a9f1528d77503b5c93e48e3de9933fbb9f3cd.
(Bitbake rev: a29780bd43f74b7326fe788dbd65177b86806fcf)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Conflicts:
bitbake/lib/bb/cooker.py
bitbake/lib/bb/main.py
bitbake/lib/bb/tinfoil.py
bitbake/lib/bb/ui/knotty.py
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
don't exist
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Add some very basic safeguard against recursively deleting paths such
as / and /home in the event of bugs or user mistakes.
Addresses [YOCTO #7620].
(Bitbake master rev: 56cddeb9e1e4d249f84ccd6ef65db245636e38ea)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If the subpath parameter to the git fetcher ends with a trailing '/',
bb.utils.prunedir() will be called on '/'...
Fixes [YOCTO #7620].
(Bitbake master rev: 380a3fb372c8b0a53dd7528562e6e7a222dc76ef)
Signed-off-by: Anders Darander <anders@chargestorm.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The _remove operator isn't working correctly when used with a variable that
expands to several items, so add a test case to exercise this path.
(Bitbake rev: cb2a62a5fbffb358528a85b46c1fc6383286cb9d)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Given these assignments:
TEST="a b c d"
TEST_remove = "b d"
TEST evaluates to "a c". However, if the _remove override is given as a
variable:
TEST="a b c d"
FOO = "b d"
TEST_remove = "${FOO}
TEST evaluates to "a b c d", because when FOO is expanded it isn't split into a
list.
Solve this by splitting all members of removeactive once they've been expanded.
[ YOCTO #7272 ]
(Bitbake rev: 207013b6dde82f9654f9be996695c8335b95a288)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Running bitbake inside make results in the exported environment variable
MAKEOVERRIDES="${-*-command-variables-*-}", which the shell chokes on
when trying to expand it. But of course, it probably shouldn't have been
trying to expand it in the first place -- so just escape the dollar
sign.
(Bitbake rev: 18cd0ce6a55c9065c3f1bf223b47d817b5efcd8f)
Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The recent inotify changes are causing a 100% cpu usage issue in the
idle handlers. To avoid this, we update the idle functions to optionally
report a float value which is the delay before the function needs to be
called again. 1 second is fine for the inotify handler, in reality its
more like 0.1s due to the default idle function sleep.
This reverts performance regressions of 1.5 minutes on a kernel build
and ~5-6 minutes on a image from scratch.
(Bitbake rev: 0e0ba408c2dce14a0fabd3fdf61d8465a031495b)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The inotify facility monitoring changes to the config files
could be overwhelmed by massive changes to the watched files
while server is running.
This patch adds verification the notification watches to the
server idle functions, in addition to the cooker updateCache
command which executes only infrequently, thus preventing
overflowing the notification buffer.
[YOCTO #7316]
(Bitbake rev: 996e663fd5c254292f44eca46f5fdc95af897f98)
Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Benchmarks show that the introduction of pyinotify regressed
performance. This patch ensures we only call the add_watch() function
for new entries, not ones we've already processed which does improve
performance as measured by "time bitbake -p".
This doesn't completely remove the overhead but it does substantially
reduce it.
(Bitbake rev: 493361f35f6cc332d4ea359a2695622c2c91a9c2)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We currently add crazy numbers of watches on files. The per user limit is 8192
by default and on a system handling multiple builds, this can be an issue.
We don't need to watch all files individually, we can watch the directory containing
the file instead. This gives better resource utilisation and better performance
further reverting some of the performance regression seen with the introduction
of pyinotify.
(Bitbake rev: a2d441237916a99405b800c1a3dc39f860100a8c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We try and add watches for files that don't exist but if they did, would influence
the parser. The parent directory of these files may not exist, in which case we need
to watch any parent that does exist for changes. This change implements that fallback
handling.
(Bitbake rev: 979ddbe4b7340d7cf2f432f6b1eba1c58d55ff42)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Memory resident bitbake has one current flaw, changes in the base configuration
are not noticed by bitbake. The parsing cache is also refreshed on each invocation
of bitbake (although the mtime cache is not cleared so its pointless).
This change adds in pyinotify support and adds two different watchers, one
for the base configuration and one for the parsed recipes.
Changes in the latter will trigger a reparse (and an update of the mtime cache).
The former will trigger a complete reload of the configuration.
Note that this code will also correctly handle creation of new configuration files
since the __depends and __base_depends variables already track these for cache
correctness purposes.
We could be a little more clever about parsing cache invalidation, right now we just
invalidate the whole thing and recheck. For now, its better than what we have and doesn't
seem to perform that badly though.
For education and QA purposes I can document a workflow that illustrates this:
$ source oe-init-build-env-memres
$ time bitbake bash
[base configuration is loaded, recipes are parsed, bash builds]
$ time bitbake bash
[command returns quickly since all caches are valid]
$ touch ../meta/classes/gettext.bbclass
$ time bitbake bash
[reparse is triggered, time is longer than above]
$ echo 'FOO = "1"' >> conf/local.conf
$ time bitbake bash
[reparse is triggered, but with a base configuration reload too]
As far as changes go, I like this one a lot, it makes memory resident bitbake
truly usable and may be the tweak we need to make it the default.
The new pyinotify dependency is covered in the previous commit.
(Bitbake rev: 0557d03c170fba8d7efe82be1b9641d0eb229213)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We need inotify support within bitbake and pyinotify provides the best
mechanism to add this. We have a few options:
a) Depend on pyinotify from the system
b) Add in our own copy
c) Only use pyinotify in cases like the memory resident server
For a), it would mean adding in dependencies, updating documentation and
generally creating churn for users as well as having implications for things
like the build-appliance recipe.
It turns out that glibc has the C functionality we need from version 2.4
onwards (2006) and that we just need a single python file for b), there
is no binary module needed. We therefore add in a copy of pyinotify 0.9.5
into the tree meaning we can depend on it simply and unconditionally.
c) is unattractive as we need fewer possible code paths, not more.
(Bitbake rev: d49004a4e247e3958a2f7ea9ffe5ec92794e1352)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If the cooker is in an error state, we shouldn't continue to try parsing.
This fixes an issue where an invalid PR server is detected when bitbake
is started and ensures bitbake exits cleanly rather than hanging.
[YOCTO #6934]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The bug has a long discussion of this. Basically, in some environments,
the exact details of which aren't understood, a Ctrl+C signal to the
UI is being transmitted to all the process children. Looking at the output
of "ps ax -O tpgid", its clear the main process is still the terminal
owner of these processes.
stty -a on a problematic system shows: "-ignbrk brkint"
and on a working system shows: "-ignbrk -brkint"
The description of brkint would suggest this is the problem, setting up
that terminal environment wasn't able to reproduce the problem though.
It was confirmed that using setsid() caused the problem to be resolved
and is probably the right thing to be doing anyway, so lets do it.
[YOCTO #6949]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently, if you reference a file url, its checksum is included in the
task hash, however if you change to a different file at a different
location, perhaps taking advantage of the FILESPATH functionality, the
system will not reparse the file in question and change its checksum to
match the new file.
To correctly handle this, the system not only needs to know if the
existing file still exists or not, but also check the existance
of every file it would have looked at when computing the original file.
We already do this in the bitbake parsing code for class inclusion. This
change uses the same technique to log the file list we looked at and
if files in these locations exist when they previously did not, to
invalidate and reparse the file.
Since data stored in the cache is flattened text, we have to use a string
form of the data and split on the ":" character which is ugly, but is
an internal detail we can improve later if a better method is found.
The cache version changes to trigger a reparse since the previous
cache data is now incompatible.
[YOCTO #7019]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
In some cases for cache purpoes we not only need to know which file
is going to be used but also which paths were considered. Add a
localpaths method which includes the history.
The core which() funciton already supports this, this just extends
the function to preserve the extra data we need. localpath becomes
just a special case of the case with history.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
A dash character is illegal in function names in sh (but not bash). Since
our shell tasks run under sh and the shell parser is sh based, EXPORT_FUNCTIONS
won't work with class names containing a dash.
We can't change sh, we can ensure the user is warned about the problem
straight away though.
[YOCTO #7006]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The get/set_taskdata functions are now part of the API of the class,
ensure they exist in the base class definition so the noop handler
works.
[YOCTO #7233]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This makes it possble to fetch Gerrit review references which are
normally stored under refs/changes.
Please disregard previous patch with the same topic.
Signed-off-by: Fredrik Svensson <fredrik.svensson@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
To support yocto on systems with python3 as default version, scripts
should use /usr/bin/env python in the shebang, as this allows the use of
a fake env to mimic python2 as default version.
This patch simply replaces occurrences of #!/usr/bin/python with
#!/usr/bin/env python and was done with this oneliner:
git grep -lE '^#!/usr/bin/python' | xargs \
sed -i 's|/usr/bin/python|/usr/bin/env python|'
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Fixed:
1) Run "bitbake recipe" in the terminal
2) Close the terminal while building
3) $ ps aux | grep bitbake-worker
There will be many processes, and they will keep the resources (e.g.,
memory), and won't exit unless kill or kill -9.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If you don't explicitly specify to use a global variable when doing an
assignment, you will be setting a local variable instead, which means
this function wasn't working at all. It explains some odd behaviour we
have seen in the layer index where event handlers were sometimes
bleeding into other contexts where they should not have been.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The shellshock patches changed the way bash functions are exported.
Unfortunately different distros used slightly different formats,
Fedora went with BASH_FUNC_XXX()=() { echo foo; } and Ubuntu went with
BASH_FUNC_foo%%=() { echo foo; }.
The former causes errors in dealing with out output from emit_env,
the functions are not exported in either case any more.
This patch handles things so the functions work as expected in either
case.
[YOCTO #6880]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
After Ctrl+C is pressed to interrupt bitbake, it loops continually, running
at 100% cpu. This patch selects on the correct file descriptors resolving
the excess cpu usage.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We need to consistently use LogMessage.INFO/WARNING/ERROR to make sure toaster knows
how to categories these rather than passing in the "raw" loglevel value
which in best case comes from python logging but worst case any value.
[YOCTO 6885]
Signed-off-by: Michael Wood <michael.g.wood@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Ideally, we want the PR service to have minimal influence from
queued disk IO. sqlite tends to be paranoid about data loss and
locks/fsync calls. There is a "WAL mode" which changes the journalling
mechanism and would appear much better suited to our use case.
This patch therefore switches the database to use WAL mode. With this
change, write overhead appears significantly reduced.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The sync/commit calls are happening in the submission thread which can
race against the handler. The handler may start new transactions which
then causes the submission thread to error with "cannot start a
transaction within a transaction".
The fix is to move the calls to the correct thread.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
In the commit 'prserv: Ensure data is committed', the PR server moved to
only committing transactions to the database when the PR server is
stopped. This improves performance, but it means that if the machine
running the PR server loses power unexpectedly or if the PR server
process gets SIGKILL, the uncommitted package revision data is lost.
To fix this issue, sync the database periodically, once per 30 seconds
by default, if it has been marked as dirty. To be safe, continue to
sync the database at exit regardless of its status.
Signed-off-by: Ben Shelton <ben.shelton@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Given that the "do_package_write" task doesn't exist in OE anymore,
steal another, existing example to demonstrate the "rdeptask" flag.
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Reported-by: Laszlo Papp <lpapp@kde.org>
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Scott Rifenbark <scott.m.rifenbark@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Similar to the last shared work task signature bug, we've found another
one. Looking at the improved output of diffsigs in this case:
runtaskdeps changed from [
'autoconf_2.69.bb.do_populate_sysroot:virtual:native',
'gnu-config_20120814.bb.do_populate_sysroot:virtual:native',
'libgcc-initial_4.9.bb.do_patch:virtual:nativesdk'
] to [
'autoconf_2.69.bb.do_populate_sysroot:virtual:native',
'gcc-crosssdk-initial_4.9.bb.do_patch',
'gnu-config_20120814.bb.do_populate_sysroot:virtual:native'
]
so we can get a different task hash since libgcc sorts before gnu-config
and gcc sorts after it. We could do with a way of fixing this, the best
I can come up with is to include a single parent directory. Since
recipes are never at the top of any metadata trees I've seen, this
should suffice for now.
I'm planning to burn the concept of shared work within bitbake
and do something at the metadata level in the 1.8 timeframe as its just
too fragile as things stand and hard to fix well.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If we split variables only at whitespaces, a slipped in tab will render
a value unremovable.
Signed-off-by: Stefan Müller-Klieser <s.mueller-klieser@phytec.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Forcing strict to be a string, to avoid problems when performing comparisons
[YOCTO #6762]
Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We fix the rendering of the size field in dirinfo-related pages
by directly calling filtered_filesizeformat and not rendering
it through the template engine.
Additionally, we enable error dumping into logs if an
Exception happens.
[YOCTO #6669]
Signed-off-by: Alexandru DAMIAN <alexandru.damian@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
host address
If localhost resolves to a remote address (due to a misconfigured network),
starting the pr server will fail without useful information.
To reproduce, add '<bogus ip> localhost' to /etc/hosts and run
'bitbake -p'. The error message will be:
ERROR: Timeout while attempting to communicate with bitbake server
ERROR: Could not connect to server False:
Running 'bitbake-prserv --host=localhost --port=0 --start' will fail with:
error: [Errno 99] Cannot assign requested address
Since these errors does not show the IP address of the attempted socket
binding, this results in a lot of wasted time looking at firewall rules, etc.
This patch results in the following error message if the socket binding fails:
PR Server unable to bind to <bogus ip>:0
Signed-off-by: Konrad Scherer <Konrad.Scherer@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
hob was adding the redundant characters "\1" in SSTATE_MIRRORS variable. If
needed it is expected the user will add this instead so remove the code
that was doing this.
[YOCTO #6600]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This fixes urls of the form file://some/path/file;subdir=b. It also
adds in a couple of tests so we now tests these corner cases.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Check if the 'subdir' parameter exists and assign it to 'destdir' so that
files are copied in ${WORKDIR}/destdir. This fixes urls that are of the form
file://a;subdir=b.
Signed-off-by: Roxana Ciobanu <roxana.ciobanu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Writing a log that the filesystem isn't being monitored for inode usage just
confuses users who are not aware about the nature of inodes in their filesystem,
so don't say anything, just silently disable the monitor. In general this only
happens on filesystems which don't have a limit on inodes.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
I could apply the label "release_1.0" to a super project that contains
many sub projects. If my recipes have SRC_URI's that use that label but
grab different sub-folders, than there's a bug where the cached localfile
(tar.gz) will not be unique and reused at the wrong times.
SRC_URI = "p4://perforce::1234@//depot/SuperProject/MiniProjectAAA/...;label=release_1.0 \
p4://perforce::1234@//depot/SuperProject/MiniProjectBBB/...;label=release_1.0"
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When using options like -k, -f, -v and so on with the memory resident
server, they'd currently only be set on the initial values passed to
the original command. This ensures they now match those specified
on the commandline for the options where this makes sense.
To make this work, a command to update the options on the server side
is required so this is added.
[YOCTO #5292]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
Scott Rifenbark
Richard Purdie
Paul Eggleton
Anders Darander
Ross Burton
Richard Tollerton
Alexandru DAMIAN
Fredrik Svensson
Martin Hundebøll
Robert Yang
Michael Wood
Ben Shelton
Robert P. J. Day
Scott Rifenbark
Stefan Müller-Klieser
Alejandro Hernandez
Konrad Scherer
Roxana
Chad Nelson