[ath9k-devel] ath9k hangs on load in current git: invalid context

[ath9k-devel] ath9k hangs on load in current git: invalid context

[Pavel Roskin]

Hello!

After upgrading to the current wireless-testing.git, ath9k started
failing on the modules load.  It's a kernel with most debugging enabled,
so that may be the reason others don't see it.  However, the lock
warning is followed by a NULL pointer dereference in
ath9k_hw_intrpend().  ath5k is working on the same kernel.

Here's the relevant part of the kernel log:

ath9k 0000:0c:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17
ath9k 0000:0c:00.0: setting latency timer to 64
BUG: sleeping function called from invalid context at /home/proski/src/linux-2.6/arch/x86/mm/fault.c:1072
in_atomic(): 0, irqs_disabled(): 1, pid: 3223, name: modprobe
1 lock held by modprobe/3223:
 #0:  (&mm->mmap_sem){++++++}, at: [<ffffffff8144634b>] do_page_fault+0xfb/0x380
irq event stamp: 44288
hardirqs last  enabled at (44287): [<ffffffff8144339b>] _raw_spin_unlock_irqrestore+0x3b/0x70
hardirqs last disabled at (44288): [<ffffffff810837c7>] request_threaded_irq+0x177/0x230
softirqs last  enabled at (39570): [<ffffffff813db83f>] sk_filter+0x7f/0xe0
softirqs last disabled at (39568): [<ffffffff813db7d6>] sk_filter+0x16/0xe0
Pid: 3223, comm: modprobe Not tainted 2.6.33-rc3-wl #65
Call Trace:
 [<ffffffff8106ccb0>] ? print_irqtrace_events+0xd0/0xe0
 [<ffffffff8103266f>] __might_sleep+0xdf/0x110
 [<ffffffff81446366>] do_page_fault+0x116/0x380
 [<ffffffffa00e7770>] ? ath_isr+0x0/0x1a0 [ath9k]
 [<ffffffff814438ff>] page_fault+0x1f/0x30
 [<ffffffffa00e7770>] ? ath_isr+0x0/0x1a0 [ath9k]
 [<ffffffffa00a8d1c>] ? ath9k_hw_intrpend+0xc/0x70 [ath9k_hw]
 [<ffffffffa00e77b8>] ath_isr+0x48/0x1a0 [ath9k]
 [<ffffffffa00e7770>] ? ath_isr+0x0/0x1a0 [ath9k]
 [<ffffffffa00e7770>] ? ath_isr+0x0/0x1a0 [ath9k]
 [<ffffffff810837d3>] request_threaded_irq+0x183/0x230
 [<ffffffffa00efc5f>] ath_pci_probe+0x1cf/0x3f0 [ath9k]
 [<ffffffff811f7dc2>] local_pci_probe+0x12/0x20
 [<ffffffff811f8c20>] pci_device_probe+0x80/0xb0
 [<ffffffff812bcf4a>] ? driver_sysfs_add+0x5a/0x90
 [<ffffffff812bd089>] driver_probe_device+0x89/0x180
 [<ffffffff812bd180>] ? __driver_attach+0x0/0xa0
 [<ffffffff812bd21b>] __driver_attach+0x9b/0xa0
 [<ffffffff812bd180>] ? __driver_attach+0x0/0xa0
 [<ffffffff812bc544>] bus_for_each_dev+0x64/0x90
 [<ffffffff812bcee9>] driver_attach+0x19/0x20
 [<ffffffff812bc7eb>] bus_add_driver+0xbb/0x260
 [<ffffffff812bd50f>] driver_register+0x6f/0x130
 [<ffffffff811f8641>] __pci_register_driver+0x61/0xe0
 [<ffffffffa0077000>] ? ath9k_init+0x0/0x6f [ath9k]
 [<ffffffffa00efe9e>] ath_pci_init+0x1e/0x20 [ath9k]
 [<ffffffffa0077048>] ath9k_init+0x48/0x6f [ath9k]
 [<ffffffff810001d7>] do_one_initcall+0x37/0x190
 [<ffffffff8107da8a>] sys_init_module+0xda/0x250
 [<ffffffff810025ab>] system_call_fastpath+0x16/0x1b
BUG: unable to handle kernel NULL pointer dereference at 00000000000000f8
IP: [<ffffffffa00a8d1c>] ath9k_hw_intrpend+0xc/0x70 [ath9k_hw]
PGD 6defd067 PUD 6deb2067 PMD 0 
Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
last sysfs file: /sys/module/mac80211/initstate
CPU 1 
Pid: 3223, comm: modprobe Not tainted 2.6.33-rc3-wl #65 0DT492/Inspiron 1420                   
RIP: 0010:[<ffffffffa00a8d1c>]  [<ffffffffa00a8d1c>] ath9k_hw_intrpend+0xc/0x70 [ath9k_hw]
RSP: 0018:ffff88006de27bb8  EFLAGS: 00010096
RAX: ffff88007f245dc0 RBX: 0000000000000000 RCX: 0000000000000000
RDX: 000000000000ad00 RSI: ffff88006def9800 RDI: 0000000000000000
RBP: ffff88006de27bc8 R08: 0000000000000001 R09: ffff88006def9800
R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000000
R13: ffffffffa00e7770 R14: 0000000000000001 R15: 0000000000000000
FS:  00007f372cc6d6f0(0000) GS:ffff880002300000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 00000000000000f8 CR3: 000000007e391000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process modprobe (pid: 3223, threadinfo ffff88006de26000, task ffff88007f245dc0)
Stack:
 ffff88006de27be8 ffff88006def9800 ffff88006de27c08 ffffffffa00e77b8
<0> ffff88007f162e00 ffffffffa00e7770 ffff88006de27c08 0000000000000011
<0> 0000000000000292 ffffffffa00e7770 ffff88006de27c68 ffffffff810837d3
Call Trace:
 [<ffffffffa00e77b8>] ath_isr+0x48/0x1a0 [ath9k]
 [<ffffffffa00e7770>] ? ath_isr+0x0/0x1a0 [ath9k]
 [<ffffffffa00e7770>] ? ath_isr+0x0/0x1a0 [ath9k]
 [<ffffffff810837d3>] request_threaded_irq+0x183/0x230
 [<ffffffffa00efc5f>] ath_pci_probe+0x1cf/0x3f0 [ath9k]
 [<ffffffff811f7dc2>] local_pci_probe+0x12/0x20
 [<ffffffff811f8c20>] pci_device_probe+0x80/0xb0
 [<ffffffff812bcf4a>] ? driver_sysfs_add+0x5a/0x90
 [<ffffffff812bd089>] driver_probe_device+0x89/0x180
 [<ffffffff812bd180>] ? __driver_attach+0x0/0xa0
 [<ffffffff812bd21b>] __driver_attach+0x9b/0xa0
 [<ffffffff812bd180>] ? __driver_attach+0x0/0xa0
 [<ffffffff812bc544>] bus_for_each_dev+0x64/0x90
 [<ffffffff812bcee9>] driver_attach+0x19/0x20
 [<ffffffff812bc7eb>] bus_add_driver+0xbb/0x260
 [<ffffffff812bd50f>] driver_register+0x6f/0x130
 [<ffffffff811f8641>] __pci_register_driver+0x61/0xe0
 [<ffffffffa0077000>] ? ath9k_init+0x0/0x6f [ath9k]
 [<ffffffffa00efe9e>] ath_pci_init+0x1e/0x20 [ath9k]
 [<ffffffffa0077048>] ath9k_init+0x48/0x6f [ath9k]
 [<ffffffff810001d7>] do_one_initcall+0x37/0x190
 [<ffffffff8107da8a>] sys_init_module+0xda/0x250
 [<ffffffff810025ab>] system_call_fastpath+0x16/0x1b
Code: f6 c1 e6 05 81 c6 1c 88 00 00 ff 10 c9 c1 e8 0f 83 e0 01 c3 66 66 2e 0f 1f 84 00 00 00 00 00 55 48 89 e5 53 48 89 fb 48 83 ec 08 <83> bf f8 00 00 00 14 74 48 48 8b 87 e0 00 00 00 be 38 40 00 00 
RIP  [<ffffffffa00a8d1c>] ath9k_hw_intrpend+0xc/0x70 [ath9k_hw]
 RSP <ffff88006de27bb8>
CR2: 00000000000000f8
---[ end trace f405f763f9bd5fdb ]---


-- 
Regards,
Pavel Roskin

[ath9k-devel] ath9k hangs on load in current git: invalid context

[Pavel Roskin]

On Wed, 2010-01-13 at 12:38 -0500, Pavel Roskin wrote:

> BUG: unable to handle kernel NULL pointer dereference at 00000000000000f8
> IP: [<ffffffffa00a8d1c>] ath9k_hw_intrpend+0xc/0x70 [ath9k_hw]

ath9k_hw_intrpend() is called with ah = NULL from ath_isr().  ath_isr()
checks if SC_OP_INVALID is set, but doesn't check if ah is NULL.

Adding debugging code shows that sc->sc_flags is 0 when it happens.
This must be caused by the recent changes in the initialization.

SC_OP_INVALID should be set in ath_pci_probe() before request_irq().

-- 
Regards,
Pavel Roskin