drivers/isdn/hisax/bkm_a8.c, kernel 2.4.2 (Scitel Quadro)

Hi,

Please find attached a patch to fix the following problems with the Scitel Quadro ISDN card in 2.4 kernels which suddenly arised when I bought
a K7T Pro motherboard.

kernel: HiSax: Scitel port 0xcc00-0xcd00 already in use
kernel: HiSax: Card Scitel Quadro not installed !

Credits go to Roland Klabunde who told me early december:

<quote>
The Scitel [...] resource requirements are as follows:

- 1 shared interrupt for all controllers
- 1 shared port address for all controllers with a range of 128 bytes
- 1 port address for each controller with a range of 64 bytes

[...]

I've currently downloaded the ISDN stuff [...] As mentioned above, the span is *128* for pci_ioaddr1 and *64* for pci_ioaddr2 to pci_ioaddr5 [...]. What I see from the source is, that one attempts to claim a range of 256 bytes for pci_ioaddr1 to _5. That may cause the problems if the range overlaps with other boards. You may try to change the following calls in bkm_a8.c:

sct_alloc_io(pci_ioaddr1, 256) to sct_alloc_io(pci_ioaddr1, 128) sct_alloc_io(pci_ioaddr2, 256) to sct_alloc_io(pci_ioaddr1, 64) sct_alloc_io(pci_ioaddr3, 256) to sct_alloc_io(pci_ioaddr1, 64) sct_alloc_io(pci_ioaddr4, 256) to sct_alloc_io(pci_ioaddr1, 64) sct_alloc_io(pci_ioaddr5, 256) to sct_alloc_io(pci_ioaddr1, 64)

Please note the necessary changes in release_io_sct_quadro. </quote>

Too bad I went on a holiday that time and forgot all about it, untill today (shame shame shame). Anyway, this patch to 2.4.2 drivers/isdn/hisax/bkm_a8.c fixes the problem and everything runs fine again.

Ime Smits

@@ -205,9 +205,9 @@
 void
 release_io_sct_quadro(struct IsdnCardState *cs)
 {
-	release_region(cs->hw.ax.base & 0xffffffc0, 256);
+	release_region(cs->hw.ax.base & 0xffffffc0, 128);
 	if (cs->subtyp == SCT_1)
-		release_region(cs->hw.ax.plx_adr, 256);
+		release_region(cs->hw.ax.plx_adr, 64);
 }
 
 static void

@@ -403,9 +403,9 @@
 	switch(cs->subtyp) {
 		case 1:
 			cs->hw.ax.base = pci_ioaddr5 + 0x00;
-			if (sct_alloc_io(pci_ioaddr1, 256))
+			if (sct_alloc_io(pci_ioaddr1, 128))
 				return(0);
-			if (sct_alloc_io(pci_ioaddr5, 256))
+			if (sct_alloc_io(pci_ioaddr5, 64))
 				return(0);
 			/* disable all IPAC */
 			writereg(pci_ioaddr5, pci_ioaddr5 + 4,

@@ -419,17 +419,17 @@
 			break;
 		case 2:
 			cs->hw.ax.base = pci_ioaddr4 + 0x08;
-			if (sct_alloc_io(pci_ioaddr4, 256))
+			if (sct_alloc_io(pci_ioaddr4, 64))
 				return(0);
 			break;
 		case 3:
 			cs->hw.ax.base = pci_ioaddr3 + 0x10;
-			if (sct_alloc_io(pci_ioaddr3, 256))
+			if (sct_alloc_io(pci_ioaddr3, 64))
 				return(0);
 			break;
 		case 4:
 			cs->hw.ax.base = pci_ioaddr2 + 0x20;
-			if (sct_alloc_io(pci_ioaddr2, 256))
+			if (sct_alloc_io(pci_ioaddr2, 64))
 				return(0);
 			break;
 	}	

Hi all,

great to hear, that the fix solves the problem. Is there any information about, whether and when the fix will migrate into the isdn4l source pool?

Roland

----- Original Message -----
From: <[email protected]>
To: <[email protected]>
Cc: <[email protected]>; <[email protected]>; <[email protected]>
Sent: Friday, March 30, 2001 5:05 PM
Subject: [PATCH] drivers/isdn/hisax/bkm_a8.c, kernel 2.4.2 (Scitel Quadro)