Bug#647992: chromium 14 broken with libnss3-1d from sid

[Jonathan Nieder]

Mike Hommey wrote:
> On Tue, Nov 15, 2011 at 04:36:31PM -0600, Jonathan Nieder wrote:

>> Chromium (15)'s renderer processes are all dying here:
>> 
>> 	NSSInitSingleton()
>> 	    : opencryptoki_module_(NULL),
>> 	      software_slot_(NULL),
>> 	      test_slot_(NULL),
>> 	      tpm_slot_(NULL),
>> 	      root_(NULL),
>> 	      chromeos_user_logged_in_(false) {
>> 	...
>> 	  if (nodb_init) {
>> 	    status = NSS_NoDB_Init(NULL); <---- this dies somehow
>
> This dies how? Do you have a backtrace?

No backtrace.  One can reproduce it by installing chromium-browser
from sid and running it.

Expected result: shows some page
Actual result: "Aw, snap. Something went wrong while rendering this
page".

The usual methods for debugging chromium don't work here.  The
renderer has to be forked from the zygote[1] to reproduce it, or else
chromium frustratingly just works fine, so we can't use
"--renderer-cmd-prefix='xterm -e gdb --args'" or --single-process.
Some people had success using some magic like
"--renderer-startup-dialog --allow-sandbox-debugging", but not me.
All I know is that changing the source to say

	if (nodb_init) {
	  std::cerr << "about to call NSS_NoDB_Init(NULL)\n";
	  status = NSS_NoDB_Init(NULL);
	  std::cerr << "finished NSS_NoDB_Init(NULL)\n";

causes the "about to call" line to be printed, but the "finished" line
not to.

[1] http://code.google.com/p/chromium/wiki/LinuxZygote