2.6.0-test3+sk98lin driver with hardware bug make eth unusable

2.6.0-test3+sk98lin driver with hardware bug make eth unusable

[Ricardo Galli]

I've already reported this problem to syskonnect few weeks ago (without 
success as I see). 

There is a ASIC bug in several popular motherboards (including ASUS ones) 
related to TX hardware checksum. 

For packets smaller that 56 bytes (payload), as UDP dns queries, the asic 
generates a bad checksum making the drivers unusable for "normal" Internet 
usage:

12:50:26.602458 192.168.0.10.33520 > 192.168.0.3.53: [bad udp cksum 617e!]  
55764+ [45727n] A? ponti.gallimedina.net. (39) (DF) (ttl 64, id 8705, len 67)
12:50:26.733664 192.168.0.10.33515 > 192.168.0.3.53: [bad udp cksum 5e7e!]... 
12:50:40.603124 192.168.0.10.33520 > 192.168.0.3.53: [bad udp cksum 617e!]... 
12:50:40.743436 192.168.0.10.33523 > 192.168.0.3.53: [bad udp cksum 607e!]... 
12:50:54.604568 192.168.0.10.33526 > 192.168.0.3.53: [bad udp cksum 717e!]... 
12:50:54.744493 192.168.0.10.33523 > 192.168.0.3.53: [bad udp cksum 607e!]...

The only solution is to comment out
 #define USE_SK_TX_CHECKSUM
in skge.c

Could be this an option in Kconfig?

-- 
  ricardo galli       GPG id C8114D34
  http://mnm.uib.es/~gallir/

Re: 2.6.0-test3+sk98lin driver with hardware bug make eth unusable

[Martin Schlemmer]

On Tue, 2003-08-12 at 13:01, Ricardo Galli wrote:
> I've already reported this problem to syskonnect few weeks ago (without 
> success as I see). 
> 
> There is a ASIC bug in several popular motherboards (including ASUS ones) 
> related to TX hardware checksum. 
> 
> For packets smaller that 56 bytes (payload), as UDP dns queries, the asic 
> generates a bad checksum making the drivers unusable for "normal" Internet 
> usage:
> 

<snip>

> The only solution is to comment out
>  #define USE_SK_TX_CHECKSUM
> in skge.c
> 

Known issue.

Mirko will have a look as soon as he have time.


Cheers,

-- 
Martin Schlemmer

Re: 2.6.0-test3+sk98lin driver with hardware bug make eth unusable

[Ricardo Galli]

On Tuesday 12 August 2003 14:01, Martin Schlemmer shaped the electrons to 
shout:
> On Tue, 2003-08-12 at 13:01, Ricardo Galli wrote:
> > I've already reported this problem to syskonnect few weeks ago (without
> > success as I see).
> >
> > There is a ASIC bug in several popular motherboards (including ASUS ones)
> > related to TX hardware checksum.
> >
> > For packets smaller that 56 bytes (payload), as UDP dns queries, the asic
> > generates a bad checksum making the drivers unusable for "normal"
> > Internet usage:
>
> <snip>
>
> > The only solution is to comment out
> >  #define USE_SK_TX_CHECKSUM
> > in skge.c
>
> Known issue.
>
> Mirko will have a look as soon as he have time.

Thanks, I just sent a Kconfig patch as a workaround:

http://lkml.org/lkml/2003/8/12/83

BTW, I'm testing it in a ASUS P4800 Deluxe motherboard, which has the asic 
bug:

gallir@antoli:~$ lspci
00:00.0 Host bridge: Intel Corp.: Unknown device 2578 (rev 02)
00:01.0 PCI bridge: Intel Corp.: Unknown device 2579 (rev 02)
00:1d.0 USB Controller: Intel Corp.: Unknown device 24d2 (rev 02)
00:1d.1 USB Controller: Intel Corp.: Unknown device 24d4 (rev 02)
00:1d.2 USB Controller: Intel Corp.: Unknown device 24d7 (rev 02)
00:1d.3 USB Controller: Intel Corp.: Unknown device 24de (rev 02)
00:1d.7 USB Controller: Intel Corp.: Unknown device 24dd (rev 02)
00:1e.0 PCI bridge: Intel Corp. 82801BA/CA/DB PCI Bridge (rev c2)
00:1f.0 ISA bridge: Intel Corp.: Unknown device 24d0 (rev 02)
00:1f.1 IDE interface: Intel Corp.: Unknown device 24db (rev 02)
00:1f.2 IDE interface: Intel Corp.: Unknown device 24d1 (rev 02)
00:1f.3 SMBus: Intel Corp.: Unknown device 24d3 (rev 02)
00:1f.5 Multimedia audio controller: Intel Corp.: Unknown device 24d5 (rev 02)
01:00.0 VGA compatible controller: nVidia Corporation NV25 [GeForce4 Ti 4200] 
(rev a3)
02:03.0 FireWire (IEEE 1394): VIA Technologies, Inc. IEEE 1394 Host Controller 
(rev 80)
02:05.0 Ethernet controller: 3Com Corporation: Unknown device 1700 (rev 12)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
02:0c.0 Multimedia audio controller: Creative Labs SB Live! EMU10k1 (rev 07)
02:0c.1 Input device controller: Creative Labs SB Live! MIDI/Game Port (rev 
07)
02:0d.0 Ethernet controller: 3Com Corporation 3c905C-TX/TX-M [Tornado] (rev 
78)


Cheers,

-- 
  ricardo galli       GPG id C8114D34
  http://mnm.uib.es/~gallir/

Re: 2.6.0-test3+sk98lin driver with hardware bug make eth unusable

[Martin Schlemmer]

[-- Attachment #1: Type: text/plain, Size: 1146 bytes --]

On Tue, 2003-08-12 at 14:40, Ricardo Galli wrote:
> On Tuesday 12 August 2003 14:01, Martin Schlemmer shaped the electrons to 
> shout:
> > On Tue, 2003-08-12 at 13:01, Ricardo Galli wrote:
> > > I've already reported this problem to syskonnect few weeks ago (without
> > > success as I see).
> > >
> > > There is a ASIC bug in several popular motherboards (including ASUS ones)
> > > related to TX hardware checksum.
> > >
> > > For packets smaller that 56 bytes (payload), as UDP dns queries, the asic
> > > generates a bad checksum making the drivers unusable for "normal"
> > > Internet usage:
> >
> > <snip>
> >
> > > The only solution is to comment out
> > >  #define USE_SK_TX_CHECKSUM
> > > in skge.c
> >
> > Known issue.
> >
> > Mirko will have a look as soon as he have time.
> 
> Thanks, I just sent a Kconfig patch as a workaround:
> 
> http://lkml.org/lkml/2003/8/12/83
> 

Should work fine with version 6.16 of the driver (does so
here at least with a P4C800):

http://www.syskonnect.com/syskonnect/support/driver/zip/linux/sk98lin_2.6.0-test3_patch.gz


Cheers,

-- 

Martin Schlemmer




[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: 2.6.0-test3+sk98lin driver with hardware bug make eth unusable

[Martin Schlemmer]

[-- Attachment #1: Type: text/plain, Size: 1490 bytes --]

On Sat, 2003-08-23 at 20:35, Martin Schlemmer wrote:
> On Tue, 2003-08-12 at 14:40, Ricardo Galli wrote:
> > On Tuesday 12 August 2003 14:01, Martin Schlemmer shaped the electrons to 
> > shout:
> > > On Tue, 2003-08-12 at 13:01, Ricardo Galli wrote:
> > > > I've already reported this problem to syskonnect few weeks ago (without
> > > > success as I see).
> > > >
> > > > There is a ASIC bug in several popular motherboards (including ASUS ones)
> > > > related to TX hardware checksum.
> > > >
> > > > For packets smaller that 56 bytes (payload), as UDP dns queries, the asic
> > > > generates a bad checksum making the drivers unusable for "normal"
> > > > Internet usage:
> > >
> > > <snip>
> > >
> > > > The only solution is to comment out
> > > >  #define USE_SK_TX_CHECKSUM
> > > > in skge.c
> > >
> > > Known issue.
> > >
> > > Mirko will have a look as soon as he have time.
> > 
> > Thanks, I just sent a Kconfig patch as a workaround:
> > 
> > http://lkml.org/lkml/2003/8/12/83
> > 
> 
> Should work fine with version 6.16 of the driver (does so
> here at least with a P4C800):
> 
> http://www.syskonnect.com/syskonnect/support/driver/zip/linux/sk98lin_2.6.0-test3_patch.gz
> 

Hi Jeff

Any chance that we could get the sk98lin drivers updated ?  Mirko did
not respond as of yet, but the 6.16 version fixes the HW checksum issues
with most (if not all) Asus boards and maybe some others ...


Thanks,

-- 

Martin Schlemmer




[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

[PATCH 2.6.0-test4][sk98lin] sk98lin driver with hardware bug make eth unusable

[Mirko Lindner]

The newest version of the sk98lin driver with some fixes for HW Csum was
sent to the driver Maintainers today (Kernel 2.4.22 and Kernel
2.6.0-test4). Here is a list with fixes and new functions:

VERSION 6.17 (Thu Aug 26 2003 - mlindner)
New Features:
- Add: Dynamic Interrupt Moderation (DIM) port up message
Problems fixed:
- Fix: Compiler warnings (void *)
- Fix: README.txt: Editorial changes (spelling)
- Fix: install.sh: "silent mode" driver check
- Fix: Disable Half Duplex with Gigabit-Speed (Yukon). Enable Full
Duplex.
- Fix: Corrected CPU detection and compile errors on single CPU machines
Known limitations:
- None

VERSION 6.16 (Thu Aug 21 2003 - mlindner)
New Features:
- Add: Removed SkNumber and SkDoDiv
- Add: Counter output as (unsigned long long)
Problems fixed:
- Fix: Ignore ConType parameter if empty value
- Fix: Removed useless defines
- Fix: UDP and TCP HW-CSum calculation (Kernel 2.5/2.6)
- Fix: UDP and TCP Proto checks
- Fix: UDP header offset
- Fix: Build without ProcFS
- Fix: Kernel 2.6 editorial changes
- Fix: Removed useless defines
Known limitations:
- None

VERSION 6.15 (Mon Aug 11 2003 - mlindner)
New Features:
- None
Problems fixed: 
- Fix: Yukon Lite parameter input and check 
- Fix: Yukon Lite speed query
- Fix: ConType parameter check and error detection
- Fix: Text for some Config- and Release-Files
- Fix: Readme changes (Support for SK-95xx compliant cards)
- Fix: modprobe description (Bug #10891)
- Fix: Insert various fixes applied to the kernel tree
- Fix: Removed History from the driver readme
Known limitations:
- None


The newest version of the driver is available on the SysKonnect page at:
http://www.syskonnect.de/syskonnect/support/driver/htm/sk98lin.htm

Cheers,

Mirko



On Sun, 2003-08-24 at 14:27, Martin Schlemmer wrote:
> On Sat, 2003-08-23 at 20:35, Martin Schlemmer wrote:
> > On Tue, 2003-08-12 at 14:40, Ricardo Galli wrote:
> > > On Tuesday 12 August 2003 14:01, Martin Schlemmer shaped the electrons to 
> > > shout:
> > > > On Tue, 2003-08-12 at 13:01, Ricardo Galli wrote:
> > > > > I've already reported this problem to syskonnect few weeks ago (without
> > > > > success as I see).
> > > > >
> > > > > There is a ASIC bug in several popular motherboards (including ASUS ones)
> > > > > related to TX hardware checksum.
> > > > >
> > > > > For packets smaller that 56 bytes (payload), as UDP dns queries, the asic
> > > > > generates a bad checksum making the drivers unusable for "normal"
> > > > > Internet usage:
> > > >
> > > > <snip>
> > > >
> > > > > The only solution is to comment out
> > > > >  #define USE_SK_TX_CHECKSUM
> > > > > in skge.c
> > > >
> > > > Known issue.
> > > >
> > > > Mirko will have a look as soon as he have time.
> > > 
> > > Thanks, I just sent a Kconfig patch as a workaround:
> > > 
> > > http://lkml.org/lkml/2003/8/12/83
> > > 
> > 
> > Should work fine with version 6.16 of the driver (does so
> > here at least with a P4C800):
> > 
> > http://www.syskonnect.com/syskonnect/support/driver/zip/linux/sk98lin_2.6.0-test3_patch.gz
> > 
> 
> Hi Jeff
> 
> Any chance that we could get the sk98lin drivers updated ?  Mirko did
> not respond as of yet, but the 6.16 version fixes the HW checksum issues
> with most (if not all) Asus boards and maybe some others ...
> 
> 
> Thanks,

Re: [PATCH 2.6.0-test4][sk98lin] sk98lin driver with hardware bug make eth unusable

[Martin Schlemmer]

On Thu, 2003-08-28 at 12:24, Mirko Lindner wrote:
> The newest version of the sk98lin driver with some fixes for HW Csum was
> sent to the driver Maintainers today (Kernel 2.4.22 and Kernel
> 2.6.0-test4). Here is a list with fixes and new functions:
> 

Thanks, I'll test tonight and let you know if there are any
issues.


Regards,

> VERSION 6.17 (Thu Aug 26 2003 - mlindner)
> New Features:
> - Add: Dynamic Interrupt Moderation (DIM) port up message
> Problems fixed:
> - Fix: Compiler warnings (void *)
> - Fix: README.txt: Editorial changes (spelling)
> - Fix: install.sh: "silent mode" driver check
> - Fix: Disable Half Duplex with Gigabit-Speed (Yukon). Enable Full
> Duplex.
> - Fix: Corrected CPU detection and compile errors on single CPU machines
> Known limitations:
> - None
> 
> VERSION 6.16 (Thu Aug 21 2003 - mlindner)
> New Features:
> - Add: Removed SkNumber and SkDoDiv
> - Add: Counter output as (unsigned long long)
> Problems fixed:
> - Fix: Ignore ConType parameter if empty value
> - Fix: Removed useless defines
> - Fix: UDP and TCP HW-CSum calculation (Kernel 2.5/2.6)
> - Fix: UDP and TCP Proto checks
> - Fix: UDP header offset
> - Fix: Build without ProcFS
> - Fix: Kernel 2.6 editorial changes
> - Fix: Removed useless defines
> Known limitations:
> - None
> 
> VERSION 6.15 (Mon Aug 11 2003 - mlindner)
> New Features:
> - None
> Problems fixed: 
> - Fix: Yukon Lite parameter input and check 
> - Fix: Yukon Lite speed query
> - Fix: ConType parameter check and error detection
> - Fix: Text for some Config- and Release-Files
> - Fix: Readme changes (Support for SK-95xx compliant cards)
> - Fix: modprobe description (Bug #10891)
> - Fix: Insert various fixes applied to the kernel tree
> - Fix: Removed History from the driver readme
> Known limitations:
> - None
> 
> 
> The newest version of the driver is available on the SysKonnect page at:
> http://www.syskonnect.de/syskonnect/support/driver/htm/sk98lin.htm
> 
> Cheers,
> 
> Mirko
> 
> 
> 
> On Sun, 2003-08-24 at 14:27, Martin Schlemmer wrote:
> > On Sat, 2003-08-23 at 20:35, Martin Schlemmer wrote:
> > > On Tue, 2003-08-12 at 14:40, Ricardo Galli wrote:
> > > > On Tuesday 12 August 2003 14:01, Martin Schlemmer shaped the electrons to 
> > > > shout:
> > > > > On Tue, 2003-08-12 at 13:01, Ricardo Galli wrote:
> > > > > > I've already reported this problem to syskonnect few weeks ago (without
> > > > > > success as I see).
> > > > > >
> > > > > > There is a ASIC bug in several popular motherboards (including ASUS ones)
> > > > > > related to TX hardware checksum.
> > > > > >
> > > > > > For packets smaller that 56 bytes (payload), as UDP dns queries, the asic
> > > > > > generates a bad checksum making the drivers unusable for "normal"
> > > > > > Internet usage:
> > > > >
> > > > > <snip>
> > > > >
> > > > > > The only solution is to comment out
> > > > > >  #define USE_SK_TX_CHECKSUM
> > > > > > in skge.c
> > > > >
> > > > > Known issue.
> > > > >
> > > > > Mirko will have a look as soon as he have time.
> > > > 
> > > > Thanks, I just sent a Kconfig patch as a workaround:
> > > > 
> > > > http://lkml.org/lkml/2003/8/12/83
> > > > 
> > > 
> > > Should work fine with version 6.16 of the driver (does so
> > > here at least with a P4C800):
> > > 
> > > http://www.syskonnect.com/syskonnect/support/driver/zip/linux/sk98lin_2.6.0-test3_patch.gz
> > > 
> > 
> > Hi Jeff
> > 
> > Any chance that we could get the sk98lin drivers updated ?  Mirko did
> > not respond as of yet, but the 6.16 version fixes the HW checksum issues
> > with most (if not all) Asus boards and maybe some others ...
> > 
> > 
> > Thanks,
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
-- 
Martin Schlemmer

Re: [PATCH 2.6.0-test4][sk98lin] sk98lin driver with hardware bug make eth unusable

[Martin Schlemmer]

[-- Attachment #1.1: Type: text/plain, Size: 477 bytes --]

On Thu, 2003-08-28 at 12:24, Mirko Lindner wrote:
> The newest version of the sk98lin driver with some fixes for HW Csum was
> sent to the driver Maintainers today (Kernel 2.4.22 and Kernel
> 2.6.0-test4). Here is a list with fixes and new functions:
> 
> VERSION 6.17 (Thu Aug 26 2003 - mlindner)
> New Features:

I need to apply attached patch to test4-bk5 to get it to compile (on
SMP).  Other than that, it seems ok.


Thanks,

-- 

Martin Schlemmer




[-- Attachment #1.2: sk98lin-smp-fixup.patch --]
[-- Type: text/plain, Size: 401 bytes --]

--- 1/drivers/net/sk98lin/skdim.c	2003-09-03 22:59:30.741762592 +0200
+++ 2/drivers/net/sk98lin/skdim.c	2003-09-03 22:58:36.145062552 +0200
@@ -319,6 +319,7 @@ GetCurrentSystemLoad(SK_AC *pAC) {
 	unsigned int  UsedTime    = 0;
 	unsigned int  SystemLoad  = 0;
 #ifdef CONFIG_SMP
+	extern int smp_num_cpus;
 	unsigned int  SKNumCpus   = smp_num_cpus;
 #else
 	unsigned int  SKNumCpus   = 1;

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH 2.6.0-test4][sk98lin] sk98lin driver with hardware bug make eth unusable

[Mirko Lindner]

Thank you,

I´ll check this next week.

Mirko


On Thu, 2003-09-04 at 23:55, Martin Schlemmer wrote:
> On Thu, 2003-08-28 at 12:24, Mirko Lindner wrote:
> > The newest version of the sk98lin driver with some fixes for HW Csum was
> > sent to the driver Maintainers today (Kernel 2.4.22 and Kernel
> > 2.6.0-test4). Here is a list with fixes and new functions:
> > 
> > VERSION 6.17 (Thu Aug 26 2003 - mlindner)
> > New Features:
> 
> I need to apply attached patch to test4-bk5 to get it to compile (on
> SMP).  Other than that, it seems ok.
> 
> 
> Thanks,

Re: 2.6.0-test3+sk98lin driver with hardware bug make eth unusable

[Ricardo Galli]

On Saturday 23 August 2003 20:35, Martin Schlemmer shaped the electrons to shout:
...
> > > Known issue.
> > >
> > > Mirko will have a look as soon as he have time.
> >
> > Thanks, I just sent a Kconfig patch as a workaround:
> >
> > http://lkml.org/lkml/2003/8/12/83
>
> Should work fine with version 6.16 of the driver (does so
> here at least with a P4C800):
>
> http://www.syskonnect.com/syskonnect/support/driver/zip/linux/sk98lin_2.6.0-test3_patch.gz

Tested with -test4, it works nicely. Thanks.

Could this new version be included in -test5? Otherwise is useless for 
lots of users (at least I received lots of "me too").


-- 
  ricardo galli       GPG id C8114D34
  http://mnm.uib.es/~gallir/