Bug 39560
| Summary: | unaligned access in wihap_input_data ( wi_hostap.c ) | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Base System | Reporter: | Andrew J Forgue <ajf> | ||||
| Component: | alpha | Assignee: | freebsd-alpha (Nobody) <alpha> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Only Me | ||||||
| Priority: | Normal | ||||||
| Version: | 4.6-STABLE | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Attachments: |
|
||||||
Hi Andrew, can you please try a kernel built with the attached patch? Thanks, -robert Robert,
Sorry, It still traps however it is a little different now
Here is the trace from gdb on the core
(kgdb) where
#0 0xfffffc00003ad800 in dumpsys () at ../../kern/kern_shutdown.c:486
#1 0xfffffc00003ad348 in boot (howto=260) at
../../kern/kern_shutdown.c:316
#2 0xfffffc00003adbf0 in panic (fmt=0xfffffc000054067c "trap")
at ../../kern/kern_shutdown.c:595
#3 0xfffffc00004e163c in trap (a0=88, a1=1, a2=0, entry=2,
framep=0xfffffe0011b376a0) at ../../alpha/alpha/trap.c:551
#4 0xfffffc00004d353c in XentMM ()
#5 0xfffffc000049ccf0 in softdep_update_inodeblock
(ip=0xfffffe0001709200,
bp=0xfffffe00074ea0d0, waitfor=0) at
../../ufs/ffs/ffs_softdep.c:3813
can not access 0x1603ffc, invalid translation (invalid L3 PTE)
can not access 0x1603ffc, invalid translation (invalid L3 PTE)
I don't see how softdep_update_inodeblock can be in the wi driver,
However it did happen at exactly the same time. (When I turned
On a client)
Let me know if you want anything else.
-Andrew
> -----Original Message-----
> From: Robert Drehmel [mailto:robert@ferrari-electronic.de]
> Sent: Thursday, June 20, 2002 7:32 AM
> To: freebsd-gnats-submit@FreeBSD.org; ajf@agitated.net
> Subject: Re: alpha/39560: unaligned access in wihap_input_data (
> wi_hostap.c )
>
> Hi Andrew,
> can you please try a kernel built with the attached patch?
>
> Thanks,
> -robert
Robert, I apologize for my previous email, I derefrenced the pc register Of the next crashdump: (kgdb) l *0xfffffc00003865fc 0xfffffc00003865fc is in wi_cache_store (../../dev/wi/if_wi.c:2706). 2701 * .ip source address 2702 * .mac src 2703 * .signal, etc. 2704 */ 2705 if (sawip) 2706 sc->wi_sigcache[cache_slot].ipsrc = ip->ip_src.s_addr; 2707 bcopy( eh->ether_shost, sc->wi_sigcache[cache_slot].macsrc, 6); 2708 2709 sig = (rx_quality >> 8) & 0xFF; 2710 noise = rx_quality & 0xFF; Sorry I overlooked it. Thanks, Andrew > -----Original Message----- > From: Andrew Forgue [mailto:ajf@agitated.net] > Sent: Thursday, June 20, 2002 5:40 PM > To: 'Robert Drehmel'; 'freebsd-gnats-submit@FreeBSD.org' > Subject: RE: alpha/39560: unaligned access in wihap_input_data ( > wi_hostap.c ) > > Robert, > > Sorry, It still traps however it is a little different now > > > Here is the trace from gdb on the core > > (kgdb) where > #0 0xfffffc00003ad800 in dumpsys () at ../../kern/kern_shutdown.c:486 > #1 0xfffffc00003ad348 in boot (howto=260) at > ../../kern/kern_shutdown.c:316 > #2 0xfffffc00003adbf0 in panic (fmt=0xfffffc000054067c "trap") > at ../../kern/kern_shutdown.c:595 > #3 0xfffffc00004e163c in trap (a0=88, a1=1, a2=0, entry=2, > framep=0xfffffe0011b376a0) at ../../alpha/alpha/trap.c:551 > #4 0xfffffc00004d353c in XentMM () > #5 0xfffffc000049ccf0 in softdep_update_inodeblock > (ip=0xfffffe0001709200, > bp=0xfffffe00074ea0d0, waitfor=0) at ../../ufs/ffs/ffs_softdep.c:3813 > can not access 0x1603ffc, invalid translation (invalid L3 PTE) > can not access 0x1603ffc, invalid translation (invalid L3 PTE) > > > I don't see how softdep_update_inodeblock can be in the wi driver, > However it did happen at exactly the same time. (When I turned > On a client) > > Let me know if you want anything else. > -Andrew > > > -----Original Message----- > > From: Robert Drehmel [mailto:robert@ferrari-electronic.de] > > Sent: Thursday, June 20, 2002 7:32 AM > > To: freebsd-gnats-submit@FreeBSD.org; ajf@agitated.net > > Subject: Re: alpha/39560: unaligned access in wihap_input_data ( > > wi_hostap.c ) > > > > Hi Andrew, > > can you please try a kernel built with the attached patch? > > > > Thanks, > > -robert State Changed From-To: open->closed OBE: hostap is now done completely differently in 5.0 current and is both endian and alignment safe. |
While running the new hostap code for the PRISM Chipset, configuration of the card for hostap goes normal, but as soon as any network card is trying to use it as an AP, the kernel panics with the below information 0x4 (unaligned trap). Wireless Card ------------------------- wi0: <Linksys WDT11> port 0x8180-0x81bf,0x8100-0x817f mem 0x80025000-0x800253ff irq 16 at device 9.0 on pci1 wi0: interrupting at CIA irq 16 wi0: 802.11 address: 00:03:2f:00:26:01 wi0: using RF:PRISM2 MAC:HFA3841 CARD:HWB3163 rev.A wi0: Intersil Firmware: Primary 0.03.00, Station 0.08.00 Syslog Entries ----------------------------------- Jun 20 00:57:25 thunder /kernel: Jun 20 00:57:25 thunder /kernel: fatal kernel trap: Jun 20 00:57:25 thunder /kernel: Jun 20 00:57:25 thunder /kernel: trap entry = 0x4 (unaligned access fault) Jun 20 00:57:25 thunder /kernel: a0 = 0xfffffe0011b3bc02 Jun 20 00:57:25 thunder /kernel: a1 = 0x28 Jun 20 00:57:25 thunder /kernel: a2 = 0x2 Jun 20 00:57:25 thunder /kernel: pc = 0xfffffc00003895f8 Jun 20 00:57:25 thunder /kernel: ra = 0xfffffc00003819bc Jun 20 00:57:25 thunder /kernel: curproc = 0 Jun 20 00:57:25 thunder /kernel: GDB ----------------------------------- (kgdb) where #0 0xfffffc00003ad780 in dumpsys () at ../../kern/kern_shutdown.c:486 #1 0xfffffc00003ad2c8 in boot (howto=256) at ../../kern/kern_shutdown.c:316 #2 0xfffffc00003adb70 in panic (fmt=0xfffffc00005405fc "trap") at ../../kern/kern_shutdown.c:595 #3 0xfffffc00004e15bc in trap (a0=18446741874983287810, a1=40, a2=2, entry=4, framep=0xfffffe0011b3ba50) at ../../alpha/alpha/trap.c:551 #4 0xfffffc00004d3564 in XentUna () #5 0xfffffc00003819bc in wi_rxeof (sc=0xfffffe0001615120) at ../../dev/wi/if_wi.c:720 warning: Hit heuristic-fence-post without finding warning: enclosing function for address 0x4 PC Register ----------------------------------- (kgdb) l *0xfffffc00003895b8 0xfffffc00003895b8 is in wihap_data_input (../../dev/wi/wi_hostap.c:1061). 1056 struct wihap_info *whi = &sc->wi_hostap_info; 1057 struct wihap_sta_info *sta; 1058 int mcast, s; 1059 1060 /* TODS flag must be set. */ 1061 if (!(rxfrm->wi_frame_ctl & htole16(WI_FCTL_TODS))) { <-------- PC 1062 if (ifp->if_flags & IFF_DEBUG) 1063 printf("wihap_data_input: no TODS src=%6D\n", 1064 rxfrm->wi_addr2, ":"); 1065 m_freem(m); RA Register ------------------------------------ (kgdb) l *0xfffffc00003819bc 0xfffffc00003819bc is in wi_rxeof (../../dev/wi/if_wi.c:720). 715 * Give host AP code first crack at data 716 * packets. If it decides to handle it (or 717 * drop it), it will return a non-zero. 718 * Otherwise, it is destined for this host. 719 */ 720 if (wihap_data_input(sc, &rx_frame, m)) <------- RA 721 return; 722 } 723 /* Receive packet. */ 724 m_adj(m, sizeof(struct ether_header)); How-To-Repeat: Configure this card as a wireless Access Point, and attempt to use it (alpha only?).