Evan Huus
changed
bug 8380
| What |
Removed |
Added |
| Attachment #10097 Flags |
|
review_for_checkin-
|
Comment # 2
on bug 8380
from Evan Huus
Created attachment 10097 [details]
Demonstration fix
The attached patch (NOT to be checked in) fixes the invalid memory read by
bailing out in the case:
- we're already reassembled
- the new fragment is past the end of our conceptual buffer
- we do NOT have the PARTIAL_REASSEMBLY flag
This prevents the later code from thinking we have an overlap.
However, even with this patch we still get a large number of warnings, and I'm
not sure that bailing out is the right thing to do here. Even if we do bail, we
should potentially be throwing an exception instead of just returning?
There are a lot of interconnected pieces here that I don't fully understand.
You are receiving this mail because:
- You are watching all bug changes.