Problem in PCI resource allocation with MPC8245
Kenny
M9402134 at mail.ntust.edu.tw
Wed Jul 12 15:36:11 EST 2006
Hi,
I have a Freescale MPC8245 machine running Montavista Linux (kernel 2.4.20).
There are two devices connected to the PCI bus:
a USB host controller (VIA VT6212L) and a single-chip network switch (Broadcom BCM5650).
Now BCM5650 is working but VT6212L is not.
I tried to modify the PCI code but met some problems.
Here are the PCI information and boot message:
Bus 0, device 0, devfn 0 (id 1057-0006, headertype 0x00)
region 0: mask 0x00000008, now at 0x00000000
region 0: type mem, size 0 (0MB)
region 1: mask 0xfffff000, now at 0x00000000
region 1: type mem, size 4096 (4KB)
region 2: mask 0x00000008, now at 0x00000000
region 2: type mem, size 0 (0MB)
Bus 0, device 16, devfn 0 (id 14e4-5650, headertype 0x00) <-- BCM5650
region 0: mask 0xffff0004, now at 0xbfff0000
region 0: type mem, size 65536 (64KB)
region 1: mask 0x000000ff, now at 0x00000000
region 1: type I/O, size 65284 (65284B)
Bus 0, device 17, devfn 0 (id 1106-3038, headertype 0x00) <-- VT6212L (UHCI 1)
region 4: mask 0xffffffe1, now at 0x00bfffe0
region 4: type I/O, size 32 (32B)
Bus 0, device 17, devfn 1 (id 1106-3038, headertype 0x00) <-- VT6212L (UHCI 2)
region 4: mask 0xffffffe1, now at 0x00bfffc0
region 4: type I/O, size 32 (32B)
Bus 0, device 17, devfn 2 (id 1106-3104, headertype 0x00) <-- VT6212L (EHCI)
region 0: mask 0xffffff00, now at 0xbffeff00
region 0: type mem, size 256 (256B)
--
PCI: Probing PCI hardware
PCI:00:00.0 Resource 0 [00000000-ffffffff] is unassigned
PCI:00:00.0 Resource 1 [00000000-00000fff] is unassigned
PCI:00:00.0 Resource 2 [00000000-ffffffff] is unassigned
PCI: bridge rsrc 0..bfffff (100), parent c0179de4
PCI: bridge rsrc 80000000..fcffffff (200), parent c0179e00
PCI:00:10.0: Resource 0: bfff0000-bfffffff (f=204) <-- BCM5650
PCI:00:11.0: Resource 4: 00bfffe0-00bfffff (f=101) <-- VT6212L (UHCI 1)
PCI:00:11.1: Resource 4: 00bfffc0-00bfffdf (f=101) <-- VT6212L (UHCI 2)
PCI:00:11.2: Resource 0: bffeff00-bffeffff (f=200) <-- VT6212L (EHCI)
--
It seems no problem with the allocated resource regions for VT6212L, but I got "Timeout" message printed by USB driver.
So, I modify the function "mpc10x_bridge_set_resources" in "arch/ppc/kernel/mpc10x_common.c":
case MPC10X_MEM_MAP_B:
pci_init_resource(&hose->io_resource,
0x00000000,
0xffff, /* original value: 0x00bfffff, */
IORESOURCE_IO,
"PCI host bridge");
--
and the defination of MPC10X_MAPB_PCI_IO_END in "include/asm/mpc10x.h":
#define MPC10X_MAPB_PCI_IO_END (0x00010000 - 1) /* original value: (0x00c00000 - 1) */
--
After modification, the USB 1.x driver (UHCI) is working, and EHCI is still not working.
BUT I don't know why the UHCI works if I do the above modification?
The PCI information after modification is shown below:
Bus 0, device 17, function 0:
Class 0c03: PCI device 1106:3038 (rev 97).
IRQ 1.
Master Capable. Latency=128.
I/O at 0xffe0 [0xffff]. <-- I/O change to 0xffe0
Bus 0, device 17, function 1:
Class 0c03: PCI device 1106:3038 (rev 97).
IRQ 1.
Master Capable. Latency=128.
I/O at 0xffc0 [0xffdf]. <-- I/O change to 0xffc0
Bus 0, device 17, function 2:
Class 0c03: PCI device 1106:3104 (rev 99).
IRQ 1.
Master Capable. Latency=128.
Non-prefetchable 32 bit memory at 0xbffeff00 [0xbffeffff].
--
I do the modification for the PCI memory configuration similar to the modification for PCI I/O configuration:
[arch/ppc/kernel/mpc10x_common.c]
case MPC10X_MEM_MAP_B:
... snipped
pci_init_resource (&hose->mem_resources[0],
0x00000000, /* original value: 0x80000000, */
0x000fffff, /* original value: 0xfcffffff, */
IORESOURCE_MEM,
"PCI host bridge");
--
[include/asm/mpc10x.h]
#define MPC10X_MAPB_ISA_MEM_BASE 0x00000000 /* original value: 0x80000000 */
... snipped
#define MPC10X_MAPB_PCI_MEM_START 0x00000000 /* original value: 0x80000000 */
#define MPC10X_MAPB_PCI_MEM_END (0x00100000 - 1) /* original value: (0xC0000000 - 1) */
--
Surprisingly, after this modification, the USB2.0 driver (EHCI) can work.
BUT, the BCM5650 driver failed...
Here are the PCI information and boot message:
Bus 0, device 16, function 0:
Class 0280: PCI device 14e4:5650 (rev 17).
Master Capable. Latency=128.
Non-prefetchable 64 bit memory at 0xf0000 [0xfffff]. <-- BCM5650 PCI memory address change to 0xf0000
Bus 0, device 17, function 0:
Class 0c03: PCI device 1106:3038 (rev 97).
IRQ 1.
Master Capable. Latency=128.
I/O at 0xffe0 [0xffff].
Bus 0, device 17, function 1:
Class 0c03: PCI device 1106:3038 (rev 97).
IRQ 1.
Master Capable. Latency=128.
I/O at 0xffc0 [0xffdf].
Bus 0, device 17, function 2:
Class 0c03: PCI device 1106:3104 (rev 99).
IRQ 1.
Master Capable. Latency=128.
Non-prefetchable 32 bit memory at 0xeff00 [0xeffff]. <-- EHCI PCI memory address change to 0xeff00
--
PCI: bridge rsrc 0..ffff (100), parent c016fde4
PCI: bridge rsrc 0..fffff (200), parent c016fe00
PCI:00:10.0: Resource 0: 000f0000-000fffff (f=204)
PCI:00:11.0: Resource 4: 0000ffe0-0000ffff (f=101)
PCI:00:11.1: Resource 4: 0000ffc0-0000ffdf (f=101)
PCI:00:11.2: Resource 0: 000eff00-000effff (f=200)
--
And the failure message during inserting BCM5650 driver into kernel is:
__ioremap(): phys addr f0000 is RAM lr c0011db0
Oops: kernel access of bad area, sig: 11
NIP: C4881164 XER: 20000000 LR: C4887EDC SP: C3729C90 REGS: c3729be0 TRAP: 0300
Not tainted
MSR: 00001032 EE: 0 PR: 0 FP: 0 ME: 1 IR/DR: 11
DAR: 00000148, DSISR: 22000000
TASK = c3728000[109] 'insmod' Last syscall: 128
... snipped
--
I have really no idea about this problem because I'm pretty new to PowerPC and PCI architecture :'(
(So I think my modification may not reasonable)
Can anyone help me solve this problem ?
Any pointers and help would be greatly appreciated.
Thank you!
Kenny
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://ozlabs.org/pipermail/linuxppc-embedded/attachments/20060712/e54366ac/attachment.htm
More information about the Linuxppc-embedded
mailing list