[irda-users] Re: stir4210 device

[Guennadi L. <gl...@ds...>]

Hi

(sorry for not replying - was not receiving the list last September)

Just wanted to contribute, that at 
http://althaia.across.sk/~naiki/linux/index.html there's a port of the 
driver from Sigmatel for 2.4 to 2.6 by Naiki (my Czech is not too good:-), 
but I do think he was the author, as follows from 
http://www.abclinuxu.cz/forum/show/81434). After applying the patch below 
it also works quite well, apart from one thing: similar to 2.4 kernel we 
have a problem when some disturbances come while in FIr mode. We test with 
a PPP connection to another Linux system in FIr mode with 4mbps. As the 
connection gets lost, the parties switch back to SIr 9600, but even before 
that still in FIr __sometimes__ the 4210 seems to just stop reacting to 
USB requests to the bulk emdpoinds. Here is a snipplet from the usbmon 
under 2.6.15:

dd5bc9c0 503711873 S Bi:003:02 -115 2324 <
dd5bc9c0 503715872 C Bi:003:02 0 1512 = 0000000a 161f1000 ff030021 450805dc 8fd54000 40068b3a 0a010301 0a010302

(Here it still works - you see data comming from a bulk-in (Bi above) 
endpoint. I was doing an ftp get)

dd5bc940 503715873 S Bi:003:02 -115 2324 <
dd5bcac0 503715885 S Bo:003:01 -115 64 = 0000050b 90101f02 ff030021 45080034 6a544000 4006b663 0a010302 0a010301
dd5bcac0 503716874 C Bo:003:01 0 64 >
dd5bc940 503721872 C Bi:003:02 0 1512 = 0000000a 281f1001 ff030021 450805dc 8fd64000 40068b39 0a010301 0a010302

(Here it still works too, you see a bulk-in request, bulk-out 
request-response and a bulk-in response)

dd5bc9c0 503721873 S Bi:003:02 -115 2324 <
dd5bcac0 504216449 S Bo:003:01 -115 5 = 0000050b b1
dd5bcac0 504217777 C Bo:003:01 0 5 >

(Here the visible connection is broken, here, I am not sure - anybody has 
any idea who sends this 5-byte long control packets and what exactly they 
are? They are sent as long as a ppp-connection is active and idle. Use 
see status "0" above - an ACK was received from the chip)

dd5bcac0 504716446 S Bo:003:01 -115 5 = 0000050b b1
dd5bcac0 504717686 C Bo:003:01 0 5 >

dd5bcac0 505216446 S Bo:003:01 -115 5 = 0000050b b1
dd5bcac0 505367566 C Bo:003:01 -104 0

(2 more ACKs)

dd5bcac0 505716446 S Bo:003:01 -115 5 = 0000050b b1
dd5bcac0 505867475 C Bo:003:01 -104 0

(Here comes the error - error code -104 -ECONNRESET, I think, a network 
tx-timeout occurred and killed the URB, which means continuing NACKs and 
no ACK from the chip)

dd5bcac0 506216446 S Bo:003:01 -115 5 = 0000050b b1
dd5bcac0 506367384 C Bo:003:01 -104 0

dd5bcac0 506716446 S Bo:003:01 -115 5 = 0000050b b1
dd5bcac0 506867293 C Bo:003:01 -104 0

dd5bcb40 507216446 S Bo:003:01 -115 3 = 320000
dd5bcb40 507367203 C Bo:003:01 -104 0

(the stack runs firther, tries to switch to 9600 to restart discovery, 
but nothing works after this crash apart from a hardware reset / driver 
reload)

Anybody has any idea what that could be? A firmware bug?

Thanks
Guennadi
---------------------------------
Guennadi Liakhovetski, Ph.D.
DSA Daten- und Systemtechnik GmbH
Pascalstr. 28
D-52076 Aachen
Germany

diff -ur /home/gl/usr/src/kernel/stir4210-linux26/stir4210.c /home/gl/usr/src/kernel/mftd2x/stir4210-linux26/stir4210.c
--- /home/gl/usr/src/kernel/stir4210-linux26/stir4210.c	2006-01-19 15:57:36.000000000 +0100
+++ /home/gl/usr/src/kernel/mftd2x/stir4210-linux26/stir4210.c	2006-01-23 10:45:20.393211968 +0100
@@ -934,11 +934,6 @@
  	/* But IrDA-USB header is 1 byte. Jean II */
  	//skb_reserve(newskb, USB_IRDA_HEADER - 1);

-	/* Sigmatel thinks somethink different. (Naiki) */
-	if ( self->capability & SIGMATEL_4210 ) {
-		skb_reserve(newskb, USB_IRDA_SIGMATEL_HEADER);
-	}
-
  	if(docopy) {
  		/* Copy packet, so we can recycle the original */
  		memcpy(newskb->data, skb->data, urb->actual_length);
@@ -1053,7 +1048,7 @@
  	IRDA_DEBUG( 4,"%s(), Sending %u bytes of patch\n", __FUNCTION__,
  	    download_amount
  	    ) ;
-#if IRDA_IRDA_DEBUG_DUMP_XMIT
+#ifdef IRDA_IRDA_DEBUG_DUMP_XMIT
      sgtltrc_hexdump(
  	" --> Tx   ",
  	&patch_data[i],
@@ -1151,7 +1146,7 @@
                  return -EINVAL ;
              }
              pPatchFileVersion = (PPATCH_FILE_VERSION) pb ;
-#if SGTL_DBG
+#ifdef SGTL_DBG
              {
                  char szVersion[32] ;
                  memcpy(szVersion, pPatchFileVersion, sizeof(PATCH_FILE_VERSION)) ;
@@ -1192,7 +1187,7 @@
                  ((ulVersionBuild / 10) <<  4) +
                   (ulVersionBuild % 10)
                   ) ;
-#if SGTL_DBG
+#ifdef SGTL_DBG
              IRDA_DEBUG(3,"%s(), bcdVersion=0x%04X (%c%X.%X%X)\n",
  		__FUNCTION__,
                  ret,
diff -ur /home/gl/usr/src/kernel/stir4210-linux26/stir4210.h /home/gl/usr/src/kernel/mftd2x/stir4210-linux26/stir4210.h
--- /home/gl/usr/src/kernel/stir4210-linux26/stir4210.h	2005-11-09 23:05:02.000000000 +0100
+++ /home/gl/usr/src/kernel/mftd2x/stir4210-linux26/stir4210.h	2006-01-20 13:54:15.000000000 +0100
@@ -69,7 +69,7 @@

  /* Various ugly stuff to try to workaround generic problems */
  /* Send speed command in case of timeout, just for trying to get things sane */
-#define IU_BUG_KICK_TIMEOUT
+//#define IU_BUG_KICK_TIMEOUT
  /* Show the USB class descriptor */
  #undef IU_DUMP_CLASS_DESC
  /* Assume a minimum round trip latency for USB transfer (in us)...