Thread: [Madwifi-devel] Kernel Oopses
Status: Beta
Brought to you by:
otaku
From: Roland B. <rb...@de...> - 2004-03-16 22:51:05
|
Running 2.6.3 with the madwifi drivers from CVS HEAD, and using a 3Com 801.11a/b/g card with X-Jack antenna, I keep on getting these oopses. I'd be happy to do more debugging and testing. Roland d0c74ac0 *pde = 00000000 Oops: 0000 [#1] CPU: 0 EIP: 0060:[<d0c74ac0>] Tainted: PF Using defaults from ksymoops -t elf32-i386 -a i386 EFLAGS: 00010297 eax: 00000000 ebx: c5fd4000 ecx: 00000000 edx: 00000001 esi: c5fd4000 edi: c5fd4000 ebp: c6a88000 esp: c6cfde90 ds: 007b es: 007b ss: 0068 Stack: c5fd4000 0000000a c6cfdf19 00000000 ffffffff c01a4326 c6cfdf19 c5fd4000 00000000 00001003 00000000 c0214ca6 c5fd4000 00000292 c021a4c4 c5fd4000 00001002 c0216313 c5fd4000 c0214b25 c5fd4000 c6cfdf2c ffffffed fffffff2 Call Trace: [<c01a4326>] vsnprintf+0x256/0x4d0 [<c0214ca6>] dev_open+0xd6/0x110 [<c021a4c4>] dev_mc_upload+0x24/0x50 [<c0216313>] dev_change_flags+0x53/0x130 [<c0214b25>] dev_load+0x25/0x70 [<c0258647>] devinet_ioctl+0x2a7/0x620 [<c025ac30>] inet_ioctl+0xc0/0x110 [<c020d265>] sock_ioctl+0xf5/0x2d0 [<c016a890>] sys_ioctl+0x100/0x280 [<c010b38b>] syscall_call+0x7/0xb Code: 0f b7 00 66 89 44 24 18 8b 86 10 08 00 00 89 34 24 89 44 24 >>EIP; d0c74ac0 <__crc_crypto_register_alg+147372/597e9e> <===== >>ebx; c5fd4000 <__crc_blk_queue_merge_bvec+587ff2/8dd501> >>esi; c5fd4000 <__crc_blk_queue_merge_bvec+587ff2/8dd501> >>edi; c5fd4000 <__crc_blk_queue_merge_bvec+587ff2/8dd501> >>ebp; c6a88000 <__crc_acpi_get_timer+75eaf1/9abd0d> >>esp; c6cfde90 <__crc_ip_queue_xmit+28c74/1ebc68> Trace; c01a4326 <vsnprintf+256/4d0> Trace; c0214ca6 <dev_open+d6/110> Trace; c021a4c4 <dev_mc_upload+24/50> Trace; c0216313 <dev_change_flags+53/130> Trace; c0214b25 <dev_load+25/70> Trace; c0258647 <devinet_ioctl+2a7/620> Trace; c025ac30 <inet_ioctl+c0/110> Trace; c020d265 <sock_ioctl+f5/2d0> Trace; c016a890 <sys_ioctl+100/280> Trace; c010b38b <syscall_call+7/b> Code; d0c74ac0 <__crc_crypto_register_alg+147372/597e9e> 00000000 <_EIP>: Code; d0c74ac0 <__crc_crypto_register_alg+147372/597e9e> <===== 0: 0f b7 00 movzwl (%eax),%eax <===== Code; d0c74ac3 <__crc_crypto_register_alg+147375/597e9e> 3: 66 89 44 24 18 mov %ax,0x18(%esp,1) Code; d0c74ac8 <__crc_crypto_register_alg+14737a/597e9e> 8: 8b 86 10 08 00 00 mov 0x810(%esi),%eax Code; d0c74ace <__crc_crypto_register_alg+147380/597e9e> e: 89 34 24 mov %esi,(%esp,1) Code; d0c74ad1 <__crc_crypto_register_alg+147383/597e9e> 11: 89 44 24 00 mov %eax,0x0(%esp,1) |
From: Mathieu L. <Mat...@so...> - 2004-03-18 15:03:24
|
On Mon, 2004-03-15 at 23:41, Roland Bauerschmidt wrote: > Running 2.6.3 with the madwifi drivers from CVS HEAD, and using a 3Com > 801.11a/b/g card with X-Jack antenna, I keep on getting these oopses. > I'd be happy to do more debugging and testing. You should try a 2.6.0 kernel: the HAL in cvs was built for 2.6.0 kernels. It does not work in other 2.6.x kernels because their build options did change. regards, Mathieu > > Roland > > d0c74ac0 > *pde = 00000000 > Oops: 0000 [#1] > CPU: 0 > EIP: 0060:[<d0c74ac0>] Tainted: PF > Using defaults from ksymoops -t elf32-i386 -a i386 > EFLAGS: 00010297 > eax: 00000000 ebx: c5fd4000 ecx: 00000000 edx: 00000001 > esi: c5fd4000 edi: c5fd4000 ebp: c6a88000 esp: c6cfde90 > ds: 007b es: 007b ss: 0068 > Stack: c5fd4000 0000000a c6cfdf19 00000000 ffffffff c01a4326 c6cfdf19 c5fd4000 > 00000000 00001003 00000000 c0214ca6 c5fd4000 00000292 c021a4c4 c5fd4000 > 00001002 c0216313 c5fd4000 c0214b25 c5fd4000 c6cfdf2c ffffffed fffffff2 > Call Trace: > [<c01a4326>] vsnprintf+0x256/0x4d0 > [<c0214ca6>] dev_open+0xd6/0x110 > [<c021a4c4>] dev_mc_upload+0x24/0x50 > [<c0216313>] dev_change_flags+0x53/0x130 > [<c0214b25>] dev_load+0x25/0x70 > [<c0258647>] devinet_ioctl+0x2a7/0x620 > [<c025ac30>] inet_ioctl+0xc0/0x110 > [<c020d265>] sock_ioctl+0xf5/0x2d0 > [<c016a890>] sys_ioctl+0x100/0x280 > [<c010b38b>] syscall_call+0x7/0xb > Code: 0f b7 00 66 89 44 24 18 8b 86 10 08 00 00 89 34 24 89 44 24 > > > >>EIP; d0c74ac0 <__crc_crypto_register_alg+147372/597e9e> <===== > > >>ebx; c5fd4000 <__crc_blk_queue_merge_bvec+587ff2/8dd501> > >>esi; c5fd4000 <__crc_blk_queue_merge_bvec+587ff2/8dd501> > >>edi; c5fd4000 <__crc_blk_queue_merge_bvec+587ff2/8dd501> > >>ebp; c6a88000 <__crc_acpi_get_timer+75eaf1/9abd0d> > >>esp; c6cfde90 <__crc_ip_queue_xmit+28c74/1ebc68> > > Trace; c01a4326 <vsnprintf+256/4d0> > Trace; c0214ca6 <dev_open+d6/110> > Trace; c021a4c4 <dev_mc_upload+24/50> > Trace; c0216313 <dev_change_flags+53/130> > Trace; c0214b25 <dev_load+25/70> > Trace; c0258647 <devinet_ioctl+2a7/620> > Trace; c025ac30 <inet_ioctl+c0/110> > Trace; c020d265 <sock_ioctl+f5/2d0> > Trace; c016a890 <sys_ioctl+100/280> > Trace; c010b38b <syscall_call+7/b> > > Code; d0c74ac0 <__crc_crypto_register_alg+147372/597e9e> > 00000000 <_EIP>: > Code; d0c74ac0 <__crc_crypto_register_alg+147372/597e9e> <===== > 0: 0f b7 00 movzwl (%eax),%eax <===== > Code; d0c74ac3 <__crc_crypto_register_alg+147375/597e9e> > 3: 66 89 44 24 18 mov %ax,0x18(%esp,1) > Code; d0c74ac8 <__crc_crypto_register_alg+14737a/597e9e> > 8: 8b 86 10 08 00 00 mov 0x810(%esi),%eax > Code; d0c74ace <__crc_crypto_register_alg+147380/597e9e> > e: 89 34 24 mov %esi,(%esp,1) > Code; d0c74ad1 <__crc_crypto_register_alg+147383/597e9e> > 11: 89 44 24 00 mov %eax,0x0(%esp,1) > > > > > ------------------------------------------------------- > This SF.Net email is sponsored by: IBM Linux Tutorials > Free Linux tutorial presented by Daniel Robbins, President and CEO of > GenToo technologies. Learn everything from fundamentals to system > administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click > _______________________________________________ > Madwifi-devel mailing list > Mad...@li... > https://lists.sourceforge.net/lists/listinfo/madwifi-devel -- Mathieu Lacage <mat...@so...> |
From: Sam L. <sa...@er...> - 2004-03-27 02:32:42
|
On Mar 18, 2004, at 7:02 AM, Mathieu Lacage wrote: > On Mon, 2004-03-15 at 23:41, Roland Bauerschmidt wrote: >> Running 2.6.3 with the madwifi drivers from CVS HEAD, and using a 3Com >> 801.11a/b/g card with X-Jack antenna, I keep on getting these oopses. >> I'd be happy to do more debugging and testing. > > You should try a 2.6.0 kernel: the HAL in cvs was built for 2.6.0 > kernels. It does not work in other 2.6.x kernels because their build > options did change. HEAD has been tested with 2.4.x and 2.6.0-2.6.3. I also tested with 2.6.4-rc2 before committing the most recent hal but don't know beyond that. The WPA branch has mostly been tested against 2.6.0 and 2.6.3 but is also known to work against various other kernel versions. The only build issue I'm aware of with the WPA branch is that it has not been updated to the latest HAL and does not include the recent led, wme, and monitor mode changes that Kevin committed. Also older 2.4 kernels that lack the kernel crypto API won't build unless you patch the kernel; I haven't decided if I'll support a 2.4 kernel w/o crypto support (WPA requires more crypto that's not included and I haven't decided how to deal with the myriad of options). Sam |
From: Denis V. <vd...@po...> - 2004-03-27 19:22:16
|
Hi. I am new to Madwifi list. I joined it because I had bad experience of having non-Atheros 11g hardware. Bad hardware, _no support from vendor_. Now I know what does it mean 'to be at mercy of firmware writers'. I joined this list in order to decide whether I shall buy and use Atheros instead. > >> Running 2.6.3 with the madwifi drivers from CVS HEAD, and using a 3Com > >> 801.11a/b/g card with X-Jack antenna, I keep on getting these oopses. > >> I'd be happy to do more debugging and testing. > > > > You should try a 2.6.0 kernel: the HAL in cvs was built for 2.6.0 > > kernels. It does not work in other 2.6.x kernels because their build > > options did change. > > HEAD has been tested with 2.4.x and 2.6.0-2.6.3. I also tested with > 2.6.4-rc2 before committing the most recent hal but don't know beyond > that. > > The WPA branch has mostly been tested against 2.6.0 and 2.6.3 but is > also known to work against various other kernel versions. The only > build issue I'm aware of with the WPA branch is that it has not been > updated to the latest HAL and does not include the recent led, wme, and > monitor mode changes that Kevin committed. Also older 2.4 kernels that > lack the kernel crypto API won't build unless you patch the kernel; I Why HAL use crypto at all? Can this be handled by GPLed part? > haven't decided if I'll support a 2.4 kernel w/o crypto support (WPA > requires more crypto that's not included and I haven't decided how to > deal with the myriad of options). Hmm. I thought Madwifi is (large GPLed part) + (smallish binary HAL dealing only with low-level stuff like channels and power levels) and HAL in binary-only only because of FCC regulations. After reading your comments I've got an impression that HAL is not small, significantly uses kernel APIs and therefore is not kernel version-independent. Larger HAL == larger chance for bugs and incompatibilities. In short, Madwifi is not 'mostly GPL', and Madwifi users will be at mercy of HAL writers. Is my impression wrong? -- vda |