| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
/gvdb/reader/corrupted/7% was failing when run with random seed
R02S2a7b9704dbb5ea704b0d724329af0fbf.
This is a fuzz test, and it turns out that this particular seed ended up
producing a file that was valid, but contained a self-referential table.
The testcase happily recursed though this table's subtable (itself) and
so on, until it ran out of stack space, causing a crash.
This bug would not impact realworld users of gvdb: these users only ever
recurse through tables a finite number of times. For dconf, subtables
are not used at all. For GSettings, each schema is a subtable, but from
within that subtable we only lookup values.
|
| | |
|
| |
|
|
| |
Drop the tests for gvdb_table_walk() which will soon be removed.
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
Clean up the Makefiles and make them as similar as possible.
Move CFLAGS to a common point of definition and stop using -I so much.
Replace the 'dbus stub' with libdconf-mock.a in tests/. Fill in some
stubs for future mock code for shm and gvdb (just to get things
compiling for now).
|
| |
|
|
|
|
|
| |
We were using a negative variable for the value of a counter designed to
bound the number of open() calls in a walk. We were decrementing it in
that case (which is fine since it would never reach zero again anyway).
It was a bit ugly, though.
|
| |
|
|
|
| |
Test the robustness of the gvdb-reader when faced with a corrupted GVDB
file.
|
| |
|
|
|
|
|
|
| |
Using g_string_append_c() causes GLib headers to appear in the test
coverage results (since it's an inline function). Avoid that.
It's probably possible to do this in a more clever way but this works
for now.
|
|
|
So far only the reader is tested.
We also include some example gvdb databases for use by the testcase.
|
