crash and burn, but is it a fair test?

[Matthew Jacob]

Modify an HBA to fail one out of 4096 commands with CAM_SEL_TIMEOUT.

Run several shells all doing loops of "camcontrol rescan all".

Fairly soon, you get the crash and burn below.

--------------------------
  at Fisap1t bausl 0 t atrrgaept 0  l9u:n  8g
  dna2e:r <aLSlI  pINF-r01-o00t 076e0> Ficxted iDoirecnt  fAcacuelsts  
SCwShI-5 dielvicee
  idan2 :k e800.r000MnBe/sl  trmaondsefers

idpa2u: Ciomdm and= Qu e7u;e iangp iEcnab led
1dd a=2 : 0972
x60i0nMsBt r(u188c743t68i0o0 5n12  byptoe siecntotrse: r2       55=H  
603Sx/T 1187:487C0)
undfa0f aftf fifsp1f b8u0s1 80 0tdafrg9et 0 l
    4
sdta0:a c<LSIk  INF-0p1-o0i0 n076t0>e rFi       xed D i r ec t   A 
ccess  SCS=I -5 0devxi1ce0 :
0dxaf0: f8f00f.0f00fMB/fsf t7ranseferds
Qd5aa0: dCom0mand
ouefuering aEnambel edp
didnat0e:r 9    21 600M B ( 1 887 4 3 68 0=0  05x12 1b0yte s:e0ctxofrs: 
2f55H f6f3Sf/Tf 11f7f4787C)e
  15b00
code segmdea3 antt i    sp1      =b ubs a0 tarsgee t0 x0 l0u,n  l0i
Sdai3:t < LS0I IxNfFf-f01-f00f 0,760 > Ftixyed pDeirec t0 xAc1cebss
  CS     I-5 dev ice      =
odDaP3: L8 00.000,0M B/psr tersa n1s,fe rsl
  ndga3:  1C,o mdmeandf Q3ueuei2n g En0a,b lged
Bdaa3: 9n216 010M
e (p18r874o36c8e00 s5s12o rb yete fsectolrsa:g s25      5H=  6i3Sn/tTe 
r1174r8u7pCt)
  ndaa4 abtl iesp1d bus, 0  trarget 0e sluumn 9e
  0a4: <ILSIO PILNF -=01 -000
   7c60>u Fixerd Drireect Acnctess  pSrCoScI-e5s sdev    i       c=e
sd a4: 80(0g.00_0MeB/sv etnratnsfe)r


[thread pid 2 tid 100021 ]
Stopped at      daioctl+0x49:   movq    0x20(%rax),%rdi
db> bt
Tracing pid 2 tid 100021 td 0xffffff001042c700
daioctl() at daioctl+0x49
g_disk_ioctl() at g_disk_ioctl+0x72
g_dev_taste() at g_dev_taste+0x15e
g_new_provider_event() at g_new_provider_event+0x95
g_run_events() at g_run_events+0x217
g_event_procbody() at g_event_procbody+0x6c
fork_exit() at fork_exit+0x12a
fork_trampoline() at fork_trampoline+0xe