[broadcom43xx] KDL - PANIC: unsupport modulation type 88

[umccullough]

Tested hrev41412 (gcc4) on my Dell Inspiron B130 with broadcom43xx wifi and when it tries to connect to an open SSID, it immediately panics with:

KERN: PANIC: unsupport modulation type 88
KERN: 
KERN: Welcome to Kernel Debugging Land...
KERN: Thread 79 "ic_taskq" running on CPU 0
KERN: stack trace for thread 79 "ic_taskq"
KERN:     kernel stack: 0x80755000 to 0x80759000
KERN: frame               caller     <image>:function + offset
KERN:  0 80758c6c (+  32) 800edaa7   <kernel_x86>:arch_debug_stack_trace + 0x000f
KERN:  1 80758c8c (+  16) 8007319e   <kernel_x86> stack_trace_trampoline(void*: NULL) + 0x000b
KERN:  2 80758c9c (+  12) 800f331a   <kernel_x86>:arch_debug_call_with_fault_handler + 0x001b
KERN:  3 80758ca8 (+  48) 80073695   <kernel_x86>:debug_call_with_fault_handler + 0x0052
KERN:  4 80758cd8 (+  80) 80074646   <kernel_x86> kernel_debugger_loop(char const*:  + 0x0226
KERN:  5 80758d28 (+  64) 800748d1   <kernel_x86> kernel_debugger_internal(char const*:  + 0x0112
KERN:  6 80758d68 (+  32) 80074ab4   <kernel_x86>:panic + 0x0023
KERN:  7 80758d88 (+ 128) 80595fd8   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:bwi_encap + 0x032c
KERN:  8 80758e08 (+  80) 80597d5e   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:bwi_start + 0x0242
KERN:  9 80758e58 (+  32) 805d97a0   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:if_start + 0x0010
KERN: 10 80758e78 (+  64) 805da797   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:if_transmit + 0x0133
KERN: 11 80758eb8 (+ 112) 805bcd7c   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:ieee80211_start + 0x067c
KERN: 12 80758f28 (+  32) 805d97a0   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:if_start + 0x0010
KERN: 13 80758f48 (+  64) 805c0219   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:ieee80211_newstate_cb + 0x01cc
KERN: 14 80758f88 (+  80) 805dbad6   </boot/system/add-ons/kernel/drivers/dev/net/broadcom43xx>:tq_handle_thread + 0x008f
KERN: 15 80758fd8 (+  32) 8006616b   <kernel_x86> _create_kernel_thread_kentry() + 0x0015
KERN: 16 80758ff8 (+2139779080) 800699d8   <kernel_x86> thread_kthread_exit() + 0x0000

which I can 'co' from after several attempts - but then the dhcp request causes it to occur again.

This didn't happen with hrev40894 which I'm using with the same driver to write this bug report, so it appears to be a regression.

I will further search for the commit which broke it shortly.

[umccullough]

syslog with KDL

[umccullough]

hrev41239 succeeds while hrev41240 panics with an unhandled page fault in broadcom43xx

[umccullough]

hrev41251 fails with the same PANIC mentioned in this bug - so it was most likely the upgrade to freebsd 8.2 for the wifi drivers that caused the regression.

[korli]

Please check again with hrev41425 whether another panic happens. Thanks!

[umccullough]

Still occurs with hrev41428 (gcc4 build)

Same panic.

[pulkomandy]

The message looks pretty much like it isn't worth a panic to me. The driver may stop working, but the whole OS, seriously ?

Looking at the code, it seems valid to fail in this case. The driver in FreeBSD current seems to still do it, and the enum for modulation type has only 7 values, so 88 is clearly out of range. Now to figure out why we get this value...

[umccullough]

I think that removing the PANIC would be a suitable solution if this isn't going to be fixed again before Alpha 3.

[korli]

Could you please check again with hrev41559 or newer on trunk?

[umccullough]

It seems to be working again with hrev41581 that I just built from trunk!

Please +alpha3 and feel free close this :D

[umccullough]

Appears to be merged to alpha3 branch in hrev41594.

Closing this one :)