Quote:
Originally Posted by pepoluan View Post
Unfortunately, lsof produces a snapshot in time only. You have to put it in a loop to capture anything.

During 1 minute, I see 2 processes accessing /dev/random. But since my loop has a 1 second delay, it's very likely that there are lots of other processes that slipped under the lsof radar, so to speak.
Yes, but this also means that such accesses happen really rarely and don't block for too much.

I still insist that in Android itself /dev/random is not used at all. You can as well check using Android cross reference:

http://androidxref.com/4.2_r1/s?n=25...system%2Ctools

This one is for Android 4.2.

The potential users are probably external applications and platform-specific libraries/applications from phone manufacturer, but this would mean they are broken, because even wpa_supplicant handles /dev/random reads in a non-blocking way (if configured to use /dev/random instead of /dev/urandom).

Btw. Blocking a single process does not block the whole system in any way. Linux uses preemptive multitasking.

Quote:
Originally Posted by pepoluan View Post
Well, they're still humans.

Remember the scandal of the missing December*? There's a proof there

*Someone messed up royally resulting in the inability of setting a contact's birthdate month to December.
Yes, but this was a trivial bug in application logic, not a core system component that causes performance degradation of the whole system, something that is unacceptable in such big project.
Follow me on twitter.
My projects: OpenFIMG | CM 7 for Spica | Kernel 3.0.x for Spica | DroidBoot (CWM-based boot menu) WiP | Other
My phones: GT-i9250 Galaxy Nexus | GT-i5700 Galaxy Spica | GT-i5800 Galaxy 3
Samsung is a registered trademark of Samsung Group, Samsung Electronics and it's listed subsidiaries.
Views expressed herein are my own and do not necessarily represent the views of my employer.