Arch Linux

fcpci: AVM FRITZ!Card PCI driver, revision 0.7.2
fcpci: (fcpci built on Oct 11 2007 at 06:46:53)
fcpci: -- 32 bit CAPI driver --
PCI: Enabling device 0000:00:0f.0 (0100 -> 0103)
ACPI: PCI Interrupt 0000:00:0f.0[A] -> GSI 18 (level, low) -> IRQ 18
fcpci: AVM FRITZ!Card PCI found: port 0xb400, irq 18
fcpci: Loading...
fcpci: Driver 'fcpci' attached to fcpci-stack. (152)
fcpci: Stack version 3.11-07
kcapi: Controller [001]: fcpci-b400-18 attached
kcapi: card [001] "fcpci-b400-18" ready.
fcpci: Loaded.

for me no issue and it works.
fcpci - - - - - -

First: sorry for my bad english.

I've rechecked this by reinstalling 26.23 and fcpci for this kernel.

My report with this firmware error is not relevant. I also got this error when my capi works in 2.6.22.
It seems every: capiinit reload produce this error always when the card is already active.

But my problem still exists (for which i thought i've found the reason with above firmware confusion).
Neither of my capi applications nor isdnlog and my vbox works under new kernel/fcpci.

The card/moduls load properly but i have no registered B-Channels (capiinfo)

Here some tests (first is from good(2.6.22), second is bad(2.6.23)):

------------
[root@s02 ~]# lsmod | grep -e capi -e isdn
capi 14272 0
kernelcapi 31816 2 fcpci,capi
capifs 4360 2 capi
isdn 126688 1
slhc 5760 2 ppp_generic,isdn

[root@s02 ~]# lsmod | grep -e capi -e isdn
capi 14272 0
capifs 4360 2 capi
capidrv 27540 1
kernelcapi 31944 3 capi,fcpci,capidrv
isdn 128736 6 capidrv
slhc 5760 2 ppp_generic,isdn

Here: in the bad status modul capidrv isn't loaded automatically)
-------------------------

[root@s02 ~]# capiinfo
Number of Controllers : 1
Controller 1:
Manufacturer:
CAPI Version: 2.0
Manufacturer Version: 49.23
Serial Number: 1000001
BChannels: 0
Global Options: 0x00000000
B1 protocols support: 0x00000000
B2 protocols support: 0x00000000
B3 protocols support: 0x00000000
....

[root@s02 ~]# capiinfo
Number of Controllers : 1
Controller 1:
Manufacturer: AVM GmbH
CAPI Version: 2.0
Manufacturer Version: 3.11-07 (49.23)
Serial Number: 1000001
BChannels: 2
Global Options: 0x00000039
internal controller supported
DTMF supported
Supplementary Services supported
channel allocation supported (leased lines)
B1 protocols support: 0x4000011f
64 kbit/s with HDLC framing
64 kbit/s bit-transparent operation
V.110 asynconous operation with start/stop byte framing
V.110 synconous operation with HDLC framing
T.30 modem for fax group 3
Modem asyncronous operation with start/stop byte framing
B2 protocols support: 0x00000b1b
ISO 7776 (X.75 SLP)
Transparent
LAPD with Q.921 for D channel X.25 (SAPI 16)
T.30 for fax group 3
ISO 7776 (X.75 SLP) with V.42bis compression
V.120 asyncronous mode
V.120 bit-transparent mode
B3 protocols support: 0x800000bf
Transparent
T.90NL, T.70NL, T.90
ISO 8208 (X.25 DTE-DTE)
X.25 DCE
T.30 for fax group 3
T.30 for fax group 3 with extensions
Modem
...

Yous see: In good status the bchannels and protocols are registered properly.
----------------------

[root@s02 ~]# capiinit status
1 fcpci detected fcpci-d400-21 A1 - 0xd400 0

[root@s02 ~]# capiinit status
1 fcpci running fcpci-d400-21 A1 3.11-07 0xd400 21

Here in good status the version(?) is visible and 21 at the end.
-------------------

I will recheck this another one.
Next time i update the kernel+fcpci but then i put capidrv in MODULES in rc.conf.
Maybe it's the missing module which is not autoloaded.

What says your capiinfo?

Sorry, mistake:

The test results from:
[root@s02 ~]# lsmod | grep -e capi -e isdn
are switched (bad copy&paste).
The first (with missing capidrv) is from BAD, the second is from GOD.

Oh shit, another sorry (i must drink more coffeee)

ALL tests are switched. Sorry for confusing you ;-)

FIRST output is from bad, SECOND from good.
Grrrrr.

hrm you don't use the capiinit daemon?

Sure, capiinit is in rc.conf DAEMONS in both situations.

All settings are the same. I only down/upgrade kernel+fcpci

and what do you load in rc.conf modules?

MODULES=(mii sis900 slhc capidrv tun)

You see, capidrv is already in MODULES.
Next minutes i try a run in 2.6.23. There i will load capidrv by hand to check for errors.
But i think the problem (bchannels) comes from fcpci.
capidrv is AFAIK only a capi "application" which register.
I need capidrv to run "old" isdnlog and vbox3 on capi.
I'am back in 10 minutes or so.
If it helps i also could login in IRC#archlinux.de for discussion?

fine on irc

Back on 2.6.23 (BAD)

/etc/rc.d/capiinit stop
Here is syslog when doing:
/etc/rc.d/capiinit start and modprobe capidrv

Nov 5 20:01:59 s02 CAPI Subsystem Rev 1.1.2.8
Nov 5 20:01:59 s02 capi20: Rev 1.1.2.7: started up with major 68 (middleware+capifs)
Nov 5 20:01:59 s02 fcpci: AVM FRITZ!Card PCI driver, revision 0.7.2
Nov 5 20:01:59 s02 fcpci: (fcpci built on Oct 11 2007 at 06:46:53)
Nov 5 20:01:59 s02 fcpci: -- 32 bit CAPI driver --
Nov 5 20:01:59 s02 fcpci: AVM FRITZ!Card PCI found: port 0xd400, irq 20
Nov 5 20:01:59 s02 fcpci: Loading...
Nov 5 20:01:59 s02 fcpci: Driver 'fcpci' attached to fcpci-stack. (152)
Nov 5 20:01:59 s02 fcpci: Stack version 3.11-07
Nov 5 20:01:59 s02 kcapi: Controller [001]: fcpci-d400-20 attached
Nov 5 20:01:59 s02 kcapi: card [001] "fcpci-d400-20" ready.
Nov 5 20:01:59 s02 fcpci: Loaded.
Nov 5 20:02:07 s02 capidrv-1: now up (0 B channels)
Nov 5 20:02:07 s02 capidrv-1: not from AVM, no d-channel trace possible ()
Nov 5 20:02:07 s02 capidrv: Rev 1.1.2.2: loaded

i tried to revert all isdn changes from .23 kernel nothing changed :(
it seems a problem of capiinit or capiinfo to get the correct parameters

Today i tried the kernel change in capi.c we discussed yesterday in irc.
Nothing better with these changes.

I've googled a litte but searches with "2.6.23 fcpci" and so brought nothing relevant infos.
I've recompiled also the fcpci modul.
Next i will try fcpci-Modules from OpenSuse, but these guys are still on kernel 2.6.21/22.

Perhaps i will be a little sassy and contact Karsten Keil directly ;-)
He could'nt do more than cutting my head...

Also i posted in german achlinux forum to see if someone other has a similar card, fcpci and enviroment to
recheck my problem.

i try now to build a .23 kernel with .22 config perhaps it's just config releated

The rpm-sources from OpenSuse have the same problem.
Modul and their patches applied/compiled fine under 2.6.23, the module loads fine as ours.
But capiinfo produce the same output without any known bchannel.

config from .22 doesn't help either :/
next error source is capi4kutils or something in kernel changed that affected isdn subsystem

update on that:
the change was in the rc1 series of .23
now only the commit must be identified

Probably I disturb ... but I have the same problem on fedora 7 and with the kernel 2.6.23.1-21
Linux server.intranet.conica.it 2.6.23.1-21.fc7 #1 SMP Thu Nov 1 21:09:24 EDT 2007 i686 i686 i386 GNU/Linux
I was able to have the fcpci camplied correctly

fcpci: module license 'Proprietary' taints kernel.
fcpci: AVM FRITZ!Card PCI driver, revision 0.7.2
fcpci: (fcpci built on Nov 6 2007 at 10:17:10)
fcpci: -- 32 bit CAPI driver --
ACPI: PCI Interrupt 0000:00:09.0[A] -> GSI 17 (level, low) -> IRQ 22
fcpci: AVM FRITZ!Card PCI found: port 0x9400, irq 22
fcpci: Loading...
fcpci: Driver 'fcpci' attached to fcpci-stack. (152)
fcpci: Stack version 3.11-07
kcapi: Controller [001]: fcpci-9400-22 attached
kcapi: card [001] "fcpci-9400-22" ready.
fcpci: Loaded.

and the data are correspnding to the lspci
00:09.0 Network controller: AVM Audiovisuelles MKTG & Computer System GmbH Fritz!PCI v2.0 ISDN (rev 02)
Subsystem: AVM Audiovisuelles MKTG & Computer System GmbH Fritz!PCI v2.0 ISDN
Control: I/O+ Mem+ BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Interrupt: pin A routed to IRQ 21
Region 0: Memory at ec800000 (32-bit, non-prefetchable) [size=32]
Region 1: I/O ports at 9400 [size=32]
Capabilities: [40] Power Management version 2
Flags: PMEClk- DSI- D1- D2+ AuxCurrent=55mA PME(D0-,D1-,D2+,D3hot+,D3cold+)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

but the capiinit give a good loading of the module but is unamble
to see the Bchannel and card data.


[root@server ~]# capiinfo
Number of Controllers : 1
Controller 1:
Manufacturer:
CAPI Version: 2.0
Manufacturer Version: 49.23
Serial Number: 1000001
BChannels: 0
Global Options: 0x00000000
B1 protocols support: 0x00000000
B2 protocols support: 0x00000000
B3 protocols support: 0x00000000

0100
0000
00000000
00000000
00000000
00000000
00000000 00000000 00000000 00000000 00000000 00000000
00000000 00000000 00000000 00000000 00000000

Supplementary services support: 0x000003ff
Hold / Retrieve
Terminal Portability
ECT
3PTY
Call Forwarding
Call Deflection
MCID
CCBS

I see the capi4linux modified for 2.6.2 in march.2007.
I hope to give an help making other debug...

http://sidux.com/PNphpBB2-viewtopic-t-6496-start-0.html
it's also broken on debian systems.
this is just a guess but i think the error is in driver.c file of this fcpci package.
it forgets the manufacturer and the protocols it can provide, all this is defined in driver.c file
im not a coder so i cannot do here anything :/

Yeah Tobias, have also seen this link (sic! from 12/06). Also some blog entries of some people.

I've posted a mail in opensuse-isdn-de, maybe Karsten could/would say something.
Any mail to AVM Linux-Support is pointless, IMHO.

I'm also not a C programmer, so i could not do something usefull.
But if you (or someone other) have a further idea, i'am willing to test it.

At the moment i could live with kernel 2.6.22, i will do sec. updates "manually" on this release.

At my point we could switch this bugreport to a waiting/researching state.
And again: for the rest of the world ;-) Thank you.

i proudly present the reason why it doesn't work anymore:
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.23.y.git;a=commit;h=b520b85a963bf7b14b9614579aff14558d7ee264

"i386: Move all simple string operations out of line"
De honest I don't understand.
but I have to recompile all the kernel?

First thought was: am i on a false link?
Ashamed i must say i understand nearly nothing.
I interpret that: in kernel it isn't anymore possible to do someting like strcpy,strcat or 'a'+'b' or something.
Cause the compile "optimize" these away....
Hmmm....

And for our (my) problem:
Is there a chance to revert this "feature" to the old behavior?
Without to destabilize the whole kernel?
Is it possible for me to have a local patch to go around this in current an further versions of 2.6.23?
Short: could i test it with old string.c and string.h from 2.6.22?

we only need the header change, the rest is not important

Oh, Tobias:
Tatata. If this **is** the reason then i think you realy earn a cookie for finding this commit...

Ok, i try it today later.
Am i right:
I've only change code in string.h and make back those changes i find in the diff Link for string.h in this commit?

ok reverting the patch and recompiling the module makes it work again, with the porblem that all other binary modules breaks and need a recompile

ok C hackers are welcome to fix fcpci source code, i can't do it.
roman and thomas i would be really happy if you get it going again without changing the kernel.

i think we will have new packages this evening, roman has already successfully hacked the i686 driver

Sorry but on fedora not work
If you can suggest same tips...
(I follow your rebuild and string.h)
Even you connot help me ...
thank a lot

our fix seems not to be complete
it works only if you replace the string.h in kernel source dir, somehow we need to tweak this

Sorry but on fedora not work
If you can suggest same tips...
(I follow your rebuild and string.h)
Even you connot help me ...
thank a lot

ok final fix goes up it works now again