[INVALID] EOSM hit with the null pointer bug

Started by dfort, January 30, 2018, 05:35:56 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

dfort

All four of my EOSM's are currently showing the null pointer warning. Apparently I started doing some testing this morning with the one body that wasn't affected and now it is infected. A couple of the bodies were using firmware 2.0.3 so I downgraded to 2.0.2 and it turned out that they also have the null pointer bug.

The story of how it happened is on this post:

https://www.magiclantern.fm/forum/index.php?topic=16040.msg196547#msg196547

a1ex


dfort

Sorry, took a short break. PM'd you a link to the ROM dumps.

a1ex

All instances of 0xA5A5A5A5 (the pattern looked up by the null pointer checker) are in the same places on both ROMs.

Running changeset cad72ead39e7 (just before disabling this check on models other than 5D3) gives the warning on both ROMs.

Pretty sure it's a false alarm.

dfort

Ok--it must be me then. The build that tested out fine and didn't print out the null pointer warning is now not working and printing out the warning. All I can think of is that I wasn't in the right changeset when I started the test. That still doesn't explain the assert error on the unified build.

Anyway, thanks for the good news.

a1ex

Could not find any assert logs saved by the unified build either.

dfort

Quote from: a1ex on January 30, 2018, 07:22:34 PM
Could not find any assert logs saved by the unified build either.

Oops--looks like I had that assert log on the card when I switched over to the unified build. Having one of those days. Think I'll get some fresh air.

Thanks for all your attention, really appreciate it.