+++ This bug was initially created as a clone of Bug #961080 +++ I am breaking out the first two patches from bug 961080, which are uncontroversial, and landing them here. This is strictly for ease of tracking: the reviews have already happened in that bug. To solve bug 961080 in a completely safe manner, we need to determine the umask at early startup time and expose it to privileged Javascript. (The underlying system call to find out this value is not thread-safe: http://pubs.opengroup.org/onlinepubs/9699919799/functions/umask.html )

Patch 1/2. r=bsmedberg in bug 961080 comment 9

Patch 2/2. r=khuey without comment immediately after bug 961080 comment 24.

https://hg.mozilla.org/integration/mozilla-inbound/rev/284be1609a50 https://hg.mozilla.org/integration/mozilla-inbound/rev/2629257557ff Hopefully we will get the rest of bug 961080 nailed down in time to uplift to 31/ESR, but at least this will enable extensions to patch it.

Backed out in https://hg.mozilla.org/integration/mozilla-inbound/rev/8feb1c83354a for debug xpcshell bustage, e.g.: https://tbpl.mozilla.org/php/getParsedLog.php?id=38557804&tree=Mozilla-Inbound

So it turns out that "@mozilla.org/system-info;1" cannot be used off the main thread. Conveniently, the code behind OS.Constants already has logic to handle that sort of thing. Pushed to try this time: https://tbpl.mozilla.org/?tree=Try&rev=3facd5dc1839 Will re-land tomorrow if that's green.

Comment on attachment 8413310 [details] [diff] [review] reflect the umask into OS.Constants, v2b You want UINT_TO_JSVAL, I'd think. sr=me with that.

s/INT_TO_JSVAL/UINT_TO_JSVAL/ as suggested by bz. No other changes.

https://hg.mozilla.org/integration/mozilla-inbound/rev/5fe8306c6409 https://hg.mozilla.org/integration/mozilla-inbound/rev/ea4471fb806f

https://hg.mozilla.org/mozilla-central/rev/5fe8306c6409 https://hg.mozilla.org/mozilla-central/rev/ea4471fb806f