A newbie could not figure out the real issue so easily out of the error message below. bitbake core-image-minimal Traceback (most recent call last): File "/home/lpapp/Projects/foo/poky-denzil-7.0.1/bitbake/bin//bitbake", line 258, in <module> ret = main() File "/home/lpapp/Projects/foo/poky-denzil-7.0.1/bitbake/bin//bitbake", line 226, in main server.initServer() File "/home/lpapp/Projects/foo/poky-denzil-7.0.1/bitbake/lib/bb/server/process.py", line 245, in initServer self.event_queue = ProcessEventQueue(0) File "/usr/lib/python2.7/multiprocessing/queues.py", line 63, in __init__ self._rlock = Lock() File "/usr/lib/python2.7/multiprocessing/synchronize.py", line 147, in __init__ SemLock.__init__(self, SEMAPHORE, 1, 1) OSError: [Errno 13] Permission denied Pseudo is not present but is required, building this first before the main build Traceback (most recent call last): File "/home/lpapp/Projects/foo/poky-denzil-7.0.1/bitbake/bin//bitbake", line 258, in <module> ret = main() File "/home/lpapp/Projects/foo/poky-denzil-7.0.1/bitbake/bin//bitbake", line 226, in main server.initServer() File "/home/lpapp/Projects/foo/poky-denzil-7.0.1/bitbake/lib/bb/server/process.py", line 245, in initServer self.event_queue = ProcessEventQueue(0) File "/usr/lib/python2.7/multiprocessing/queues.py", line 63, in __init__ self._rlock = Lock() File "/usr/lib/python2.7/multiprocessing/synchronize.py", line 147, in __init__ SemLock.__init__(self, SEMAPHORE, 1, 1) OSError: [Errno 13] Permission denied DEBUG: Removed the following variables from the environment: LANGUAGE, SHLVL, SUDO_USER, BBFETCH2, USERNAME, SUDO_UID, SUDO_COMMAND, SUDO_GID, BUILDDIR, OLDPWD, MAIL, LS_COLORS, LC_COLLATE Even if this patch helps something, I think it could be fine-tuned into separate non-existent and write issue error message to be "perfect" unless there is some limitation I am unaware of: http://patchwork.openembedded.org/patch/49621/
Since it's not at all clear from the above comment, in master we do produce a sensible error message instead of the traceback pasted above. What is being asked for is enhancing that new error message so that instead of combining the "does not exist" and "not writable" checks into one, it does the checks separately and shows the appropriate error depending on which issue is present.
That is what I wrote, yes.
So looking back over this I'm fairly satisfied with the current checks - it's unlikely to get triggered other than in highly customised container setups where you should understand the implications of not providing writeable /dev/shm. Having said that we would accept a patch to improve the messaging / checks if a reasonable one were to be submitted.