Re: [ath9k-devel] [PATCH v2] ath9k: Fix panic upon attach failure

[Vasanthakumar Thiagarajan <vasanth@atheros.com>]

On Wed, Feb 18, 2009 at 01:16:03AM +0530, Luis R. Rodriguez wrote:
> On Mon, Feb 16, 2009 at 12:25 AM, Vasanthakumar Thiagarajan
> <vasanth@atheros.com> wrote:
> > [246916.338046]
> > [246916.338048] Pid: 29265, comm: insmod Not tainted (2.6.29-rc4-wl #64) 9461DUU
> > [246916.338051] EIP: 0060:[<c02ca274>] EFLAGS: 00010202 CPU: 0
> > [246916.338055] EIP is at rollback_registered+0x24/0x220
> > [246916.338057] EAX: 00000001 EBX: 00000000 ECX: 00000000 EDX: f122e8fc
> > [246916.338059] ESI: 00000000 EDI: 00000000 EBP: f6595d30 ESP: f6595d1c
> > [246916.338062]  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
> > [246916.338064] Process insmod (pid: 29265, ti=f6594000 task=f7343fe0 task.ti=f6594000)
> > [246916.338067] Stack:
> > [246916.338068]  c04a2920 22222222 f6595d48 00000000 f122f080 f6595d48 c02ca489 f122e8fc
> > [246916.338076]  f122e220 f122f080 f122e220 f6595d5c f8a03156 f122e220 f122f080 f122e220
> > [246916.338085]  f6595d80 f87359af f122f080 00002000 f874e129 f122f150 f122f080 f6290000
> > [246916.338094] Call Trace:
> > [246916.338096]  [<c02ca489>] ? unregister_netdevice+0x19/0x70
> > [246916.338100]  [<f8a03156>] ? ieee80211_unregister_hw+0x36/0xd0 [mac80211]
> > [246916.338112]  [<f87359af>] ? ath_detach+0xcf/0x250 [ath9k]
> > [246916.338127]  [<f8735d9c>] ? ath_attach+0x26c/0x740 [ath9k]
> > [246916.338139]  [<f873c33a>] ? ath_pci_probe+0x13a/0x310 [ath9k]
> > [246916.338151]  [<c0233e28>] ? _raw_spin_unlock+0x68/0x80
> > [246916.338158]  [<c023ab8e>] ? local_pci_probe+0xe/0x10
> > [246916.338162]  [<c023b8e0>] ? pci_device_probe+0x60/0x80
> > [246916.338169]  [<c029e042>] ? driver_probe_device+0x82/0x1b0
> > [246916.338174]  [<c029e1f9>] ? __driver_attach+0x89/0x90
> > [246916.338180]  [<c029d97b>] ? bus_for_each_dev+0x4b/0x70
> > [246916.338184]  [<c023b820>] ? pci_device_remove+0x0/0x40
> > [246916.338190]  [<c029ded9>] ? driver_attach+0x19/0x20
> > [246916.338193]  [<c029e170>] ? __driver_attach+0x0/0x90
> > [246916.338197]  [<c029d317>] ? bus_add_driver+0x1b7/0x230
> > [246916.338203]  [<c023b820>] ? pci_device_remove+0x0/0x40
> > [246916.338206]  [<c029e399>] ? driver_register+0x69/0x140
> > [246916.338212]  [<f859d000>] ? ath9k_init+0x0/0x54 [ath9k]
> > [246916.338221]  [<c023bb4e>] ? __pci_register_driver+0x4e/0x90
> > [246916.338225]  [<f859d000>] ? ath9k_init+0x0/0x54 [ath9k]
> > [246916.338232]  [<f859d06b>] ? ath_pci_init+0x17/0x19 [ath9k]
> > [246916.338238]  [<f859d017>] ? ath9k_init+0x17/0x54 [ath9k]
> > [246916.338245]  [<c017148e>] ? tracepoint_update_probe_range+0x7e/0xb0
> > [246916.338249]  [<c010111a>] ? do_one_initcall+0x2a/0x170
> > [246916.338252]  [<c0149f26>] ? up_read+0x16/0x30
> > [246916.338256]  [<c014aa9d>] ? __blocking_notifier_call_chain+0x4d/0x60
> > [246916.338265]  [<c0162b1a>] ? sys_init_module+0x8a/0x1c0
> > [246916.338269]  [<c022f888>] ? trace_hardirqs_on_thunk+0xc/0x10
> > [246916.338272]  [<c0103ebf>] ? sysenter_do_call+0x12/0x43
> > [246916.338276] Code: 8d bc 27 00 00 00 00 55 89 e5 56 89 c6 53 83 ec 0c a1 74 27 4a c0 85 c0 0f 85 4b 01 00 00 e8 04 7d 00 00 85 c0 0f 84 c9 01 00 00 <8b> 86 18 03 00 00 85 c0 0f 84 86 01 00 00 83 e8 01 0f 85 71 01
> > [246916.338328] EIP: [<c02ca274>] rollback_registered+0x24/0x220 SS:ESP 0068:f6595d1c
> > [246916.338335] ---[ end trace 76357c56a75ea34e ]---
> >
> > Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
> > ---
> >  drivers/net/wireless/ath9k/main.c |   25 +++++++++++++++++--------
> >  1 files changed, 17 insertions(+), 8 deletions(-)
> >
> > diff --git a/drivers/net/wireless/ath9k/main.c b/drivers/net/wireless/ath9k/main.c
> > index 787c6f7..8b2a7b8 100644
> > --- a/drivers/net/wireless/ath9k/main.c
> > +++ b/drivers/net/wireless/ath9k/main.c
> > @@ -1562,6 +1562,7 @@ bad2:
> >  bad:
> >        if (ah)
> >                ath9k_hw_detach(ah);
> > +       ath9k_exit_debug(sc);
> >
> >        return error;
> >  }
> > @@ -1570,7 +1571,7 @@ int ath_attach(u16 devid, struct ath_softc *sc)
> >  {
> >        struct ieee80211_hw *hw = sc->hw;
> >        const struct ieee80211_regdomain *regd;
> > -       int error = 0;
> > +       int error = 0, i;
> 
> Better to use unsigned int for when in use in loops, I'm told it
> performs better on some archs.

I agree signed int is useless in this case, u8 is more than enough.
But there are few more places we are doing similar things, this can
be cleaned up in a separate patch.

Vasanth