[patch] Problem with netmos 9835 chip in IBM POS-Hardware
From: Michael Westermann
Date: Tue Feb 28 2006 - 05:19:56 EST
Hello,
I've a problem with a ibm sure pos with two Netmos 9835 chip's.
the chips have the follow pci-config
0000:00:09.0 Serial controller: NetMos Technology PCI 9835 Multi-I/O Controller (rev 01) (prog-if 02 [16550])
Subsystem: IBM: Unknown device 0299
I/O ports at 1898
I/O ports at 1890
I/O ports at 1888
I/O ports at 1880
I/O ports at 1878
I/O ports at 1850
and
0000:00:0b.0 Serial controller: NetMos Technology PCI 9835 Mult
(rev 01) (prog-if 02 [16550])
Subsystem: LSI Logic / Symbios Logic 2S (16C550 UART)
Flags: medium devsel, IRQ 11
I/O ports at 18c0
I/O ports at 18b8
I/O ports at 18b0
I/O ports at 18a8
I/O ports at 18a0
I/O ports at 1860
with dmesg can I see:
PCI: Netmos 9835 (9 parallel, 9 serial); changing class SERIAL to OTHER (use parport_serial)
but the chip have only 1 parallel and 2 serial.
A insmod parport_serial.ko produces a OOps,
ACPI: PCI Interrupt Link [LNKB] enabled at IRQ 11
ACPI: PCI Interrupt 0000:00:09.0[A] -> Link [LNKB] -> GSI 11 (level, low) -> IRQ
11
PCI parallel port detected: 9710:9735, I/O at 0x1888(0x0)
parport1: PC-style at 0x1888 [PCSPP,TRISTATE,EPP]
PCI parallel port detected: 9710:9735, I/O at 0x1898(0x1898)
PCI parallel port detected: 9710:9735, I/O at 0x1898(0x1898)
PCI parallel port detected: 9710:9735, I/O at 0x1898(0x1898)
Unable to handle kernel paging request at virtual address 9f8ba554
printing eip:
d09160d3
*pde = 00000000
Oops: 0000 [#1]
Modules linked in: parport_serial 8250_pnp rtc natsemi crc32 8250_pci 8250 seria
l_core snd_via82xx snd_ac97_codec snd_ac97_bus snd_pcm snd_timer snd_page_alloc
snd_mpu401_uart snd_rawmidi snd_seq_device snd soundcore via686a hwmon i2c_isa i
2c_core uhci_hcd usbcore parport_pc parport via_agp agpgart dm_mod
CPU: 0
EIP: 0060:[<d09160d3>] Not tainted VLI
EFLAGS: 00010a83 (2.6.15.4)
EIP is at parport_register+0x9f/0x146 [parport_serial]
eax: cfe683d4 ebx: cee400e0 ecx: 00000009 edx: 00000000
esi: 00001898 edi: cee40128 ebp: cfa52000 esp: cdec9eec
ds: 007b es: 007b ss: 0068
Process insmod (pid: 2816, threadinfo=cdec8000 task=ce9800f0)
Stack: cee400e0 00000002 00000004 00000001 cee40148 cee400e0 cfa52000 00000000
cfa52044 d09161de d09171b0 d09176a0 d09176a0 d09176c8 c01d5379 c019bc65
cfa52000 c019bc94 d09176a0 cfa52000 c019bcbe cfa52044 cfa52044 c01d52d5
Call Trace:
[<d09161de>] parport_serial_pci_probe+0x64/0xab [parport_serial]
[<c01d5379>] __driver_attach+0x0/0x3a
[<c019bc65>] pci_call_probe+0xa/0xc
[<c019bc94>] __pci_device_probe+0x2d/0x39
[<c019bcbe>] pci_device_probe+0x1e/0x30
[<c01d52d5>] driver_probe_device+0x2f/0x78
[<c01d53a2>] __driver_attach+0x29/0x3a
[<c01d4c92>] bus_for_each_dev+0x37/0x59
[<c01d53c4>] driver_attach+0x11/0x13
[<c01d5379>] __driver_attach+0x0/0x3a
[<c01d4fa9>] bus_add_driver+0x54/0x94
[<c019be62>] __pci_register_driver+0x70/0x80
[<c01259c9>] sys_init_module+0xbe/0x186
[<c0102a69>] syscall_call+0x7/0xb
Code: c7 44 24 0c 00 00 00 00 c7 44 24 08 00 00 00 00 e9 88 00 00 00 8b 54 24 10
8b 42 04 8b 52 08 8b 1c 24 83 7b 04 10 74 b8 6b c0 1c <8b> b4 28 80 01 00 00 83
fa 06 77 0c 6b c2 1c 8b 9c 28 80 01 00
OK, this problem is the subsystem_device entry 0x299 from IBM.
This is not conform to the netmos-chips and the parport_serial driver.
We have in the struct acrport_pc_pci array addr[4] and
get the index with
78: card->numports = (dev->subsystem_device & 0xf0) >> 4;
dev->subsystem_device is 0x299 (pci entry from IBM)
card->numports is 9
In the function parport_register
324: for (n = 0; n < card->numports; n++) {
325: struct parport *port;
326: int lo = card->addr[n].lo;
327: int hi = card->addr[n].hi;
n > 3 is invalid the follow is a OOps.
I've wrote the follow patch for this problem:
------------------------------------------
diff -Nur linux-2.6.15.4/drivers/parport/parport_serial.c linux-2.6.15.4.new/drivers/parport/parport_serial.c
--- linux-2.6.15.4/drivers/parport/parport_serial.c 2006-02-10 08:22:48.000000000 +0100
+++ linux-2.6.15.4.new/drivers/parport/parport_serial.c 2006-02-28 10:24:59.000000000 +0100
@@ -75,7 +75,11 @@
* and serial ports. The form is 0x00PS, where <P> is the number of
* parallel ports and <S> is the number of serial ports.
*/
- card->numports = (dev->subsystem_device & 0xf0) >> 4;
+ if (dev->subsystem_vendor == PCI_VENDOR_ID_IBM && dev->subsystem_device == PCI_DEVICE_ID_IBM_SP300_32H) {
+ card->numports = 1;
+ } else {
+ card->numports = (dev->subsystem_device & 0xf0) >> 4;
+ }
return 0;
}
@@ -321,6 +325,11 @@
card->preinit_hook (dev, card, PARPORT_IRQ_NONE, PARPORT_DMA_NONE))
return -ENODEV;
+
+ if (card->numports > 4) {
+ card->numports = 4;
+ }
+
for (n = 0; n < card->numports; n++) {
struct parport *port;
int lo = card->addr[n].lo;
diff -Nur linux-2.6.15.4/drivers/pci/quirks.c linux-2.6.15.4.new/drivers/pci/quirks.c
--- linux-2.6.15.4/drivers/pci/quirks.c 2006-02-10 08:22:48.000000000 +0100
+++ linux-2.6.15.4.new/drivers/pci/quirks.c 2006-02-28 10:17:09.000000000 +0100
@@ -1211,8 +1211,15 @@
static void __devinit quirk_netmos(struct pci_dev *dev)
{
- unsigned int num_parallel = (dev->subsystem_device & 0xf0) >> 4;
- unsigned int num_serial = dev->subsystem_device & 0xf;
+ unsigned int num_parallel;
+ unsigned int num_serial;
+ if (dev->subsystem_vendor == PCI_VENDOR_ID_IBM && dev->subsystem_device == PCI_DEVICE_ID_IBM_SP300_32H) {
+ num_parallel = 1;
+ num_serial = 2;
+ } else {
+ num_parallel = (dev->subsystem_device & 0xf0) >> 4;
+ num_serial = dev->subsystem_device & 0xf;
+ }
/*
* These Netmos parts are multiport serial devices with optional
diff -Nur linux-2.6.15.4/drivers/serial/8250_pci.c linux-2.6.15.4.new/drivers/serial/8250_pci.c
--- linux-2.6.15.4/drivers/serial/8250_pci.c 2006-02-10 08:22:48.000000000 +0100
+++ linux-2.6.15.4.new/drivers/serial/8250_pci.c 2006-02-28 10:20:01.000000000 +0100
@@ -561,7 +561,12 @@
static int __devinit pci_netmos_init(struct pci_dev *dev)
{
/* subdevice 0x00PS means <P> parallel, <S> serial */
- unsigned int num_serial = dev->subsystem_device & 0xf;
+ unsigned int num_serial;
+ if (dev->subsystem_vendor == PCI_VENDOR_ID_IBM && dev->subsystem_device == PCI_DEVICE_ID_IBM_SP300_32H) {
+ num_serial = 2;
+ } else {
+ num_serial = dev->subsystem_device & 0xf;
+ }
if (num_serial == 0)
return -ENODEV;
diff -Nur linux-2.6.15.4/include/linux/pci_ids.h linux-2.6.15.4.new/include/linux/pci_ids.h
--- linux-2.6.15.4/include/linux/pci_ids.h 2006-02-10 08:22:48.000000000 +0100
+++ linux-2.6.15.4.new/include/linux/pci_ids.h 2006-02-28 10:17:58.000000000 +0100
@@ -448,6 +448,7 @@
#define PCI_DEVICE_ID_IBM_ICOM_V2_TWO_PORTS_RVX 0x021A
#define PCI_DEVICE_ID_IBM_ICOM_V2_ONE_PORT_RVX_ONE_PORT_MDM 0x0251
#define PCI_DEVICE_ID_IBM_ICOM_FOUR_PORT_MODEL 0x252
+#define PCI_DEVICE_ID_IBM_SP300_32H 0x0299
#define PCI_VENDOR_ID_COMPEX2 0x101a /* pci.ids says "AT&T GIS (NCR)" */
#define PCI_DEVICE_ID_COMPEX2_100VG 0x0005
------------------------------------------
This work with my hardware.
I've a other problen with the port order.
The assignment between port and device are different to linux 2.4.
I've two netmos in my box:
PCI 0:9 with 1 parport and 2 serial ports
PCI 0:b without parports and 2 serial ports
The netmos chip without a parport are initialized by the 8250*
The netmos chip with parport by the parport_serial driver; the
Ports are in a wrong order:
PCI 0:9 -> /dev/ttyS[67]
PCI 0:b -> /dev/ttyS[45]
My problem is, I can't replace the Kernel without changing the userland
config files.
Michael Westermann
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/