Freebsd/Xen: pcifront: IRQ resource allocation failed

[sanket hase <shase@andrew.cmu.edu>]

Hi ,

I am testing pcifront on freebsd domU using following sound card
[ Cirrus Logic CS 4614/22/24 [CrystalClear SoundFusion Audio Accelerator]

Following sound driver is used for this sound card on Freebsd:
*snd_csa*-- Crystal Semiconductor CS461x/462x/4280 PCI bridge device driver

I am able to detect the card.
pci3: <PCI bus> on pcib3
csa0: <CS4280/CS4614/CS4622/CS4624/CS4630> mem
0xfafff000-0xfaffffff,0xfae00000-0xfaefffff irq 16 at device 5.0 on pci3
csa: card is Turtle Beach Santa Cruz


 But going ahead IRQ resource allocations fails, before interrupt
handler can be setup.
 I found that it fails in  "rman_reserve_resource_bound"
Because:   r = TAILQ_FIRST(&rm->rm_list)  = is still NULL , as it was
initialized.


Here is the stacktrace:
(gdb) where
#0  rman_reserve_resource_bound (rm=0xc07b3de0, start=16, end=16,
count=1, bound=0, flags=4, dev=0xc0d8f700)
    at ../../../kern/subr_rman.c:282
#1  0xc01747bc in rman_reserve_resource (rm=0xc07b3de0, start=16,
end=16, count=1, flags=4, dev=0xc0d8f700)
    at ../../../kern/subr_rman.c:491
#2  0xc035cb75 in nexus_alloc_resource (bus=0xc0db3000,
child=0xc0d8f700, type=1, rid=0xc0d8f994, start=16, end=16, count=1,
flags=6)
    at ../../../i386/i386/nexus.c:367
#3  0xc016adae in bus_generic_alloc_resource (dev=0xc0d8f800,
child=0xc0d8f700, type=1, rid=0xc0d8f994, start=16, end=16, count=1,
    flags=6) at bus_if.h:263
#4  0xc016adae in bus_generic_alloc_resource (dev=0xc0d8f500,
child=0xc0d8f700, type=1, rid=0xc0d8f994, start=16, end=16, count=1,
    flags=6) at bus_if.h:263
#5  0xc016af34 in resource_list_alloc (rl=0xc0e03604, bus=0xc0e04b00,
child=0xc0d8f700, type=1, rid=0xc0d8f994, start=16, end=16,
    count=1, flags=6) at bus_if.h:263
#6  0xc00ab3d5 in pci_alloc_resource (dev=0xc0e04b00, child=0xc0d8f700,
type=1, rid=0xc0d8f994, start=0, end=4294967295, count=1, flags=6)
    at ../../../dev/pci/pci.c:3488
#7  0xc016ad1c in bus_alloc_resource (dev=0xc0d8f700, type=1,
rid=0xc0d8f994, start=0, end=4294967295, count=1, flags=6) at bus_if.h:263
#8  0xc00b54e6 in csa_attach (dev=0xc0d8f700) at bus.h:376
#9  0xc0169ecf in device_attach (dev=0xc0d8f700) at device_if.h:178
#10 0xc016a9f0 in device_probe_and_attach (dev=0xc0d8f700) at
../../../kern/subr_bus.c:2369
#11 0xc016aad9 in bus_generic_attach (dev=0xc0e04b00) at
../../../kern/subr_bus.c:2907
#12 0xc00ae8bf in pci_attach (dev=0xc0e04b00) at ../../../dev/pci/pci.c:2611
#13 0xc0169ecf in device_attach (dev=0xc0e04b00) at device_if.h:178
#14 0xc016a9f0 in device_probe_and_attach (dev=0xc0e04b00) at
../../../kern/subr_bus.c:2369
#15 0xc016aad9 in bus_generic_attach (dev=0xc0d8f500) at
../../../kern/subr_bus.c:2907
#16 0xc033811b in xpcib_attach (dev=0xc0d8f500) at
../../../dev/xen/pcifront/pcifront.c:702
#17 0xc0169ecf in device_attach (dev=0xc0d8f500) at device_if.h:178
#18 0xc016a9f0 in device_probe_and_attach (dev=0xc0d8f500) at
../../../kern/subr_bus.c:2369
#19 0xc016aad9 in bus_generic_attach (dev=0xc0d8f800) at
../../../kern/subr_bus.c:2907
#20 0xc03383ab in xpcife_attach (dev=0xc0d8f800) at
../../../dev/xen/pcifront/pcifront.c:640
#21 0xc0169ecf in device_attach (dev=0xc0d8f800) at device_if.h:178
#22 0xc016a9f0 in device_probe_and_attach (dev=0xc0d8f800) at
../../../kern/subr_bus.c:2369
#23 0xc033792a in pcifront_backend_changed (xdev=0xc0e05000,
be_state=XenbusStateConnected) at ../../../dev/xen/pcifront/pcifront.c:346
#24 0xc032fea8 in otherend_changed (watch=0xc0e05000, vec=0xc0defa00,
len=2) at ../../../xen/xenbus/xenbus_probe.c:330
#25 0xc0330aec in xenwatch_thread (unused=0x0) at
../../../xen/xenbus/xenbus_xs.c:741
#26 0xc0129968 in fork_exit (callout=0xc0330980 <xenwatch_thread>,
arg=0x0, frame=0xc588bd38) at ../../../kern/kern_fork.c:795
#27 0xc0350e84 in fork_trampoline () at ../../../i386/xen/exception.s:240

(gdb) p rm->rm_list
$4 = {tqh_first = 0x0, tqh_last = 0xc07b3de0}
(gdb)                                                          

tqh_first is still 0 ( as it was initialzed ) which ultimately causes
the IRQ resource allocation to fail.
Am I missing anything ?

Looking forward to hear from you,

Thanks,
shase