Black Hat/Defcon

[foxgrrl]

I'm leaving for Las Vegas in a few hours. Should arrive late Monday night, or early Tuesday morning.

Completely unrelated, but if like me, you keep hitting this bug in FreeBSD 7.0-CURRENT-200706:

panic: kmem_malloc(131072): kmem_map too small: 332410880 total allocated
cpuid = 0
KDB: enter: panic
[thread pid 129 tid 100058 ]
Stopped at      kdb_enter+0x32: leave
db> bt
Tracing pid 129 tid 100058 td 0xc66c1200
kdb_enter(c0a1d26c,0,c0a3b084,e8c6eaec,0,...) at kdb_enter+0x32
panic(c0a3b084,20000,13d03000,e8c6eb2c,c1474088,...) at panic+0x124
kmem_malloc(c147108c,20000,2,e8c6eb7c,c08d3270,...) at kmem_malloc+0x215
page_alloc(0,20000,e8c6eb6f,2,2c6eb88,...) at page_alloc+0x27
uma_large_malloc(20000,2,c0767be7,2,d0e71228,...) at uma_large_malloc+0x50
malloc(20000,c678f060,2,e8c6ec4c,c676a709,...) at malloc+0x108
zfs_kmem_alloc(20000,2,e8c6ec4c,c67590ce,20000,...) at zfs_kmem_alloc+0x20
zio_buf_alloc(20000,d0e71228,1,0,cf17dad4,...) at zio_buf_alloc+0x19
vdev_queue_io_to_issue(e8c6ec68,0,c678c88a,13f,c6daeee4,...) at vdev_queue_io_to_issue+0x1ce
vdev_queue_io_done(d7b64228,e8c6ece4,e8c6ece4,e8c6ecf8,c6711da0,...) at vdev_queue_io_done+0xaa
vdev_geom_io_done(d7b64228,0,c6787cad,33d,48c615da,...) at vdev_geom_io_done+0x12
taskq_thread(c67d00cc,e8c6ed38,c0a19ce2,324,c66be970,...) at taskq_thread+0x140
fork_exit(c6711c60,c67d00cc,e8c6ed38) at fork_exit+0xf1
fork_trampoline() at fork_trampoline+0x8
--- trap 0, eip = 0, esp = 0xe8c6ed70, ebp = 0 ---
db> 

You can tune the following variables as a temporary work around:
kern.maxvnodes="400000" in /etc/sysctl.conf
And in /boot/loader.conf:
vfs.zfs.zil_disable="1"
vfs.zfs.prefetch_disable="1"
vfs.zfs.arc_max="83886080"
vm.kmem_size_max="536870912"

There's also a Memory modified after free panic in mtrash_ctor, called from uma_zalloc_arg, called from the malloc at zfs_kmem_alloc+0x20. I haven't had any time to look into this or report it. I was doing a zpool replace pool /dev/ad8 /dev/ad6 at the time.