Re: Request for testing for iwlwifi

From: Tomoaki AOKI <junchoon_at_dec.sakura.ne.jp>
Date: Sat, 27 Aug 2022 07:23:39 UTC
On Thu, 25 Aug 2022 23:39:44 +0000 (UTC)
"Bjoern A. Zeeb" <bz@FreeBSD.org> wrote:

> Hi,
> 
> could some of you please test this small patch and let me know:
> 
> https://people.freebsd.org/~bz/tmp/20220825-01-lkpi-bssid.diff
> 
> In theory this should be fine.  But it does crash firmware on one of my
> test machines reliably while other cards are okay.  If the firmware crash
> happens it should be in scan_to_auth with the backtrace after the firmware
> register dump as below.
> 
> It's a prereq for enabling some more of iwlwifi so it would be good
> to know that it's a one-off problem here rather than more people seeing
> my problem.
> 
> 
> Bjoern
> 
> 
> Sample backtrace:
> 
> #0 0xffffffff80e5f85b at linux_dump_stack+0x1b
> #1 0xffffffff8297c044 at iwl_trans_txq_send_hcmd+0x414
> #2 0xffffffff829230fe at iwl_trans_send_cmd+0xce
> #3 0xffffffff82963383 at iwl_mvm_send_cmd_status+0x53
> #4 0xffffffff8296348f at iwl_mvm_send_cmd_pdu_status+0x4f
> #5 0xffffffff8292baa6 at iwl_mvm_binding_update+0x216
> #6 0xffffffff8293d42e at __iwl_mvm_assign_vif_chanctx+0x9e
> #7 0xffffffff82939d28 at iwl_mvm_assign_vif_chanctx+0x68
> #8 0xffffffff80e59df2 at lkpi_80211_mo_assign_vif_chanctx+0x22  <<<
> #9 0xffffffff80e52dc9 at lkpi_sta_scan_to_auth+0x409            <<<
> #10 0xffffffff80e595ed at lkpi_iv_newstate+0x27d
> #11 0xffffffff80d93015 at ieee80211_newstate_cb+0x1f5
> #12 0xffffffff80c4f8da at taskqueue_run_locked+0xaa
> #13 0xffffffff80c50982 at taskqueue_thread_loop+0xc2
> #14 0xffffffff80ba2e80 at fork_exit+0x80
> #15 0xffffffff810a044e at fork_trampoline+0xe
> 
> -- 
> Bjoern A. Zeeb                                                     r15:7
> 

Hi.

Tried on main at a28f83353243e2298b5f89c0dd6705dfcbda77ea, amd64.
Went fine on first boot. (Would need some more reboots to be sure.)
I'll send you dmesg and dtrace log off-list shortly.

But unfortunately, the problem on switching between em0 and iwlwifi
back-and-forth still happened. (The problem I've already reported on
another thread.)

Thanks in advance!

-- 
Tomoaki AOKI    <junchoon@dec.sakura.ne.jp>