Tags: patch


  07:05:00 pm, by The Dreamer   , 2118 words  
Categories: Software, Computer, BOINC, FreeBSD

sqlite3 SECURE_DELETE and Firefox

So a few days ago, databases/sqlite3 was updated in ports. And, in the portmaster run, I was faced with its config dialog. Think I had gone with the defaults previously, but decided to take a closer look this time. Saw that SECURE_DELETE, with the description "Overwrite deleted information with zeros". That sounds like a waste of time, I should probably turn that off.

A quick online search, I found this:

The secure_delete setting causes deleted content to be overwritten with zeros. There is a small performance penalty for this since additional I/O must occur. On the other hand, secure_delete can prevent sensitive information from lingering in unused parts of the database file after it has allegedly been deleted.

Yup, definitely just a waste of time...even says so. The OTOH, wrong. Why? Because I'm running my FreeBSD system on ZFS, which is copy-on-write. Its just spinning my wheels create a new copy of the file filled with zeros, and the old file is just unlinked somewhere intact, and then unlinking that new copy that it had filled with zeros. When just unlinking the old file achieves the same thing faster.

Of course, what happens a little while later there's an update to www/firefox in ports, where the configure fails because sqlite3 wasn't built with SQLITE_SECURE_DELETE. Well, I'm not turning on stupid for Firefox...I'm already disappointed by how slow it has become (and PGO seems to be broken again), to where chrome/chromium is now my everywhere browser. Which is working on the most part now that I don't have a Solaris workstation as part of my everywhere.

Well, its just configure that is testing for it and complaining...so there should be a way to turn it off. Hmmm, no option to do that, guess I'll have to later the configure script. Do I inject a patch into the files directory? Looks like the file is being adjusted elsewhere, though I don't see a patch in files that is working on it. Okay, its the post-patch target in the Makefile. Can I just add to that? Guess the way to do it is to change AC_MSG_ERROR to something that doesn't terminate the configure. Unfortunately I have portmaster.rc opertion "PM_DEL_BUILD_ONLY=pm_dbo" uncommented, so can't quickly look what AC_MSG_??? I could use. Find some online documentation, that describes AC_MSG_CHECKING, AC_MSG_RESULT, AC_MSG_NOTICE, AC_MSG_ERROR, AC_MSG_FAILURE, AC_MSG_WARN...first 3 are messages that aren't emitted if '--quiet' or '--silent' options are used. I don't think those options are used normally, but seems like a good idea to me. I'll use AC_MSG_NOTICE (though now that think of it, AC_MSG_RESULT is probably valid, since it was an AC_MSG_CHECKING that comes before the AC_MSG_ERROR...)

Well, AC_MSG_NOTICE is undefined. Guess the autoconf being used is different than the one I found online. AC_MSG_ERROR and AC_MSG_FAILURE cause exits, but AC_MSG_WARN writes to stderr and continues. Guess, that's what I'll have to use then.

So, I insert the change, and create quick diff so that I can reapply it as a patch for next time....


--- Makefile.orig  2013-06-03 17:45:05.000000000 -0500
+++ Makefile  2013-06-04 18:22:37.335175851 -0500
@@ -89,6 +89,7 @@
  @${REINPLACE_CMD} -e '/MOZPNG/s/=[0-9]*/=10511/' \
    -e '/^SQLITE_VERSION/s/=.*/=' \
+    -e '/with SQLITE_SECURE_DELETE/s/_ERROR/_WARN/' \

Pages: 1· 2

Now instead of subjecting some poor random forum to a long rambling thought, I will try to consolidate those things into this blog where they can be more easily ignored profess to be collected thoughts from my mind.

Latest Poopli Updaters -- http://lkc.me/poop


There are 20 years 3 months 28 days 15 hours 10 minutes and 53 seconds until the end of time.
And, it has been 4 years 8 months 29 days 22 hours 52 minutes and 3 seconds since The Doctor saved us all from the end of the World!


September 2017
Mon Tue Wed Thu Fri Sat Sun
 << <   > >>
        1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30  


  XML Feeds

Who's Online?

  • Guest Users: 1
This seal is issued to lawrencechen.net by StopTheHacker Inc.
open source blog software

hosted by
Green Web Hosting! This site hosted by DreamHost.

monitored by
Monitored by eXternalTest
SiteUptime Web Site Monitoring Service
website uptime