syzbot: fix ci-upstream-kasan-badwrites-root

[tarasmadan]

Describe the bug
We have a kasan-write focused instance.
It reported only 1 kasan write. https://syzkaller.appspot.com/upstream?only_manager=ci-upstream-kasan-badwrites-root
There are no fixed kasan writes found only on this manager. https://syzkaller.appspot.com/upstream/fixed?only_manager=ci-upstream-kasan-badwrites-root.

The goal was to find more KASAN: write bugs.
The reality - we have only 1 KASAN: write but reported ~20 unique bugs.
Is it a success or fail?

The options I see:

1. Do nothing.
2. Tune this syz-manager to find more KASAN: write bugs.
3. Stop this experiment.

[dvyukov]

Manager ignores are matched against kernel console output. So I suspect the "KASAN:.*Read" ignore we have is not working, since these things are on separate lines:

BUG: KASAN: slab-use-after-free in br_multicast_port_group_expired+0x4c0/0x550
Read of size 8 at addr ffff888071d6d000 by task syz.5.1232/9699

So I think the 1 unique write we got is unrelated.

But we shouldn't fix it as is as well, b/c we shouldn't produce dups by reporting same repro as 2 different KASAN bugs while trying to find KASAN writes.

[melver]

From a high-level view, the fact it's the only instance with KASAN: slab-out-of-bounds Write in hfs_bnode_read + several other unique found would indicate we should keep whatever it is contributing.

Sure, the intended result was not realized (dozens of KASAN write reports found, see Dmitry's comment above), but it did help contribute to find more unique bugs.

Would we be able to realize the same results by donating/moving the KASAN write VM resources to the normal KASAN instance(s)?

[tarasmadan]

Then the proposal is to not touch this instance until #5730 is closed.
Once #5730 is closed, we can remove ci-upstream-kasan-badwrites-root.

[dvyukov]

I think we can remove it now, it's not doing what it's supposed to be doing, and we don't plan to fix it.