Steps to reproduce on Windows 7: 1. Enable http logging https://developer.mozilla.org/en/HTTP_Logging 2. Run Minefield Result: file contains a log string of null bytes

Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b10pre) Gecko/20110124 Firefox/4.0b10pre

I've seen several people mention this, but never been able to reproduce... Do other sorts of NSPR logging work for you? If so, is it nsHttp:5 specifically that's broken, or one of the other modules that the http logging page suggests turning on logging for?

No, it's all logging, it's not specific to nsHttp. Seems to be related to something that's no longer in the profile, creating a new profile "solves" the problem.

Hmm. Can you try to narrow that down to the guilty part of the profile?

Zipfile containing copy of profile that exhibits the problem: http://people.mozilla.org/~jdaggett/4gk51ffu.trunklatest.zip

Steps to reproduce: 1. Create a new profile 'nullbytes' 2. Options > General > Show my windows and tabs from last time 3. Enter the URL: http://www.pbs.org/wgbh/pages/frontline/are-we-safer/reporting-the-story/ 4. Quit 5. Enable http logging https://developer.mozilla.org/en/HTTP_Logging 6. Run with 'nullbytes' profile Result: logfile contains a long string of null bytes 7. Run with 'nullbytes' profile 8. Enter 'www.google.com' in the URL bar 9. Restart the browser Result: logfile is fine Not sure what's the cause on that page, maybe embedded Flash?

(In reply to comment #6) > Not sure what's the cause on that page, maybe embedded Flash? Yup, setting a stripped-down test page containing only the <object> tag of the original page causes the problem.

Honza, can you have a look at this? I'm not going to block on this unless this is something that fails in the majority of cases, which it doesn't sounds like this is. John, this isn't a regression is it, or do we know?

(In reply to comment #7) > (In reply to comment #6) > > > Not sure what's the cause on that page, maybe embedded Flash? > > Yup, setting a stripped-down test page containing only the <object> tag of the > original page causes the problem. John, could you provide the stripped page as an attachment to this bug please?

> this isn't a regression is it Assuming this is caused by bug 534764, it's a regression as of e10s, ie since we shipped OOP plugins. So not a regression from the last release, but...

Bug 534764 has never been checked in.

Sorry, meant to say "assumung this is *fixed* by bug 534764"...