data storage interrupt with if_ath(4) on PowerPC

[Justin Hibbits]

On Tue, Oct 23, 2012 at 11:37 PM, Adrian Chadd <adrian at freebsd.org> wrote:

> On 23 October 2012 20:06, Justin Hibbits <chmeeedalf at gmail.com> wrote:
> > With an Atheros 5416 card from Adrian Chadd in my PowerBook G4, I get a
> > Data Storage Interrupt after a few minutes.  It's pretty reproducible:
> >
> >  * load if_ath/if_ath_pci
> >  * ifconfig wlan0 create wlandev ath0
> >  * ifconfig wlan0 up
> >  * ifconfig wlan0 bgscan (fails with "Operation not supported")
> >  * ifconfig wlan0 list scan
> >  ... wait ...
> >  * ifconfig wland0 list scan (another just for good measure)
> >  ... wait ...
> >
> > After a few minutes it kernel panics.
>
> Ew!
>

Ew indeed.


>
> > I've attached the dcons ddb session output.  Quick synopsis:  It
> > crashes in memcpy() crossing a page boundary.  Backtrace:
>
> Is that actually a bad thing? What's a "data storage interrupt" ?
>


In this case it is a bad thing, otherwise it wouldn't panic :).  Think of
it as the equivalent of an x86 'page fault in kernel mode'.





>
> Adrian
>
> > fatal kernel trap:
> >
> >    exception       = 0x300 (data storage interrupt)
> >    virtual address = 0xd1b72000
> >    srr0            = 0x5d44b4
> >    srr1            = 0x9032
> >    lr              = 0xd1ab0fa4
> >    curthread       = 0x189fbc0
> >           pid = 0, comm = ath0 net80211 taskq
> >
> > Tracing pid 0 tid 100075 td 0x189fbc0
> > 0xe21139f4: at m_pkthdr_init+0x5c
> > 0xe2113a14: at ieee80211_send_probereq+0x14c
> > 0xe2113a74: at ieee80211_probe_curchan+0x11c
> > 0xe2113aa4: at scan_curchan+0x7c
> > 0xe2113ac4: at scan_task+0x29c
> > 0xe2113b14: at taskqueue_run_locked+0xd4
> > 0xe2113b44: at taskqueue_thread_loop+0x6c
> > 0xe2113b64: at fork_exit+0x88
> > 0xe2113b84: at fork_trampoline+0xc
> >
> > The m_pkthdr_init in the backtrace is misleading, I think.
> >
> > I can provide any more data that's needed.
> >
> >
> > - Justin
>

- Justin