Getting the LED light up... (A&M Rattler 8250PCI linux 2.6.13-AM)
Got an Analogue & Micro Rattler 8250PCI board running Linux 2.6.13-AM
I'm trying to get the debug LEDs work (LED0-2 connected to PA0-2) but havent had any success yet (How #¤@"$! hard can it be to get a LED light up...? :)
Port D bit 24 is accessible on the boards expansion connector, and I successfully tested my code to use it as gen purpose I/O (both input and output).
But when I try to write (change) port A (bit 0-2), nothing changes in the register (nor do the LED light up). And my Ethernet FCC1 freezes up!! (fs mounted by NFS so I notice)
FCC1 is using a big part of port A, but I'm quite sure I don't touch any of those bits (see code below)
I'm a Linux newbie, but I'm starting to wonder if theres any 'lock' or something by the kernel on port A?
Any ideas/tips/anything welcome!
br
Robert
PS, I have a hard time find an immap for the 8250, but since the kernel and fs compile fine on my toolchain i assume it there somewhere?
---
snips of relevant parts of my code:
(it's written and compiled as a module)
typedef struct io_port {
ulong iop_padir;
ulong iop_papar;
ulong iop_pasor;
ulong iop_paodr;
ulong iop_padat;
char res1[12];
ulong iop_pbdir;
ulong iop_pbpar;
ulong iop_pbsor;
ulong iop_pbodr;
ulong iop_pbdat;
char res2[12];
ulong iop_pcdir;
ulong iop_pcpar;
ulong iop_pcsor;
ulong iop_pcodr;
ulong iop_pcdat;
char res3[12];
ulong iop_pddir;
ulong iop_pdpar;
ulong iop_pdsor;
ulong iop_pdodr;
ulong iop_pddat;
//char res4[12];
} iop8250_t;#define IOPORT_BASE_ADDR 0x10D00 // Address to first reg in iop8250
volatile iop8250_t *iommap;
#define P0 0x8000 // P0 bit in registers
#define P1 0x4000 // P1 bit in registers
#define P2 0x2000 // P2 bit in registers
#define LED0 P2
#define LED1 P1
#define LED2 P0
iommap = (iop8250_t *)(CPM_MAP_ADDR+IOPORT_BASE_ADDR);
printk("PDIRA (0x%08X) = 0x%08X\n", &(iommap->iop_padir), iommap->iop_padir); printk("PPARA (0x%08X) = 0x%08X\n", &(iommap->iop_papar), iommap->iop_papar); printk("PSORA (0x%08X) = 0x%08X\n", &(iommap->iop_pasor), iommap->iop_pasor); printk("PDATA (0x%08X) = 0x%08X\n", &(iommap->iop_padat), iommap->iop_padat);
iommap->iop_papar &= ~LED0; // set PA2 to general I/O iommap->iop_padir |= LED0; // set PA2 as output iommap->iop_padat &= ~LED0; // Clear PA2 / LED0 on
printk("PDATA (0x%08X) = 0x%08X\n", &(iommap->iop_padat), iommap->iop_padat);
Hi Robert,
Quoted 61 lines
Got an Analogue & Micro Rattler 8250PCI board running Linux 2.6.13-AM
I'm trying to get the debug LEDs work (LED0-2 connected to PA0-2) but havent had any success yet (How #=A4@"$! hard can it be to get a LED light up...? :)
Port D bit 24 is accessible on the boards expansion connector, and I successfully tested my code to use it as gen purpose I/O (both input and output).
But when I try to write (change) port A (bit 0-2), nothing changes in the register (nor do the LED light up). And my Ethernet FCC1 freezes up!! (fs mounted by NFS so I notice)
FCC1 is using a big part of port A, but I'm quite sure I don't touch any of those bits (see code below)
I'm a Linux newbie, but I'm starting to wonder if theres any 'lock' or something by the kernel on port A?
Any ideas/tips/anything welcome!
br
RobertPS, I have a hard time find an immap for the 8250, but since the kernel and fs compile fine on my toolchain i assume it there somewhere?
---
snips of relevant parts of my code:(it's written and compiled as a module)
typedef struct io_port { ulong iop_padir; ulong iop_papar; ulong iop_pasor; ulong iop_paodr; ulong iop_padat; char res1[12]; ulong iop_pbdir; ulong iop_pbpar; ulong iop_pbsor; ulong iop_pbodr; ulong iop_pbdat; char res2[12]; ulong iop_pcdir; ulong iop_pcpar; ulong iop_pcsor; ulong iop_pcodr; ulong iop_pcdat; char res3[12]; ulong iop_pddir; ulong iop_pdpar; ulong iop_pdsor; ulong iop_pdodr; ulong iop_pddat; //char res4[12]; } iop8250_t;
Quoted 7 lines
#define
IOPORT_BASE_ADDR0x10D00// Address to first reg in iop8250volatile
iop8250_t*iommap;#define P0
0x8000// P0 bit in registers
#define P10x4000// P1 bit in registers
#define P20x2000// P2 bit in registers
Shouldn't that be
#define P0 0x80000000 // P0 bit in registers
#define P1 0x40000000 // P1 bit in registers
#define P2 0x20000000 // P2 bit in registers
?
Quoted 15 lines
#define LED0 P2
#define LED1 P1
#define LED2 P0iommap =3D (
iop8250_t*)(CPM_MAP_ADDR+IOPORT_BASE_ADDR);
printk("PDIRA (0x%08X) =3D 0x%08X\n", &(iommap->iop_padir), iommap->iop_padir);printk("PPARA (0x%08X) =3D 0x%08X\n", &(iommap->iop_papar), iommap->iop_papar);printk("PSORA (0x%08X) =3D 0x%08X\n", &(iommap->iop_pasor), iommap->iop_pasor);printk("PDATA (0x%08X) =3D 0x%08X\n", &(iommap->iop_padat), iommap->iop_padat);
Quoted 6 lines
iommap->iop_papar&=3D ~LED0; // set PA2 to general I/Oiommap->iop_padir|=3D LED0; // set PA2 as outputiommap->iop_padat&=3D ~LED0; // Clear PA2 / LED0 on
printk("PDATA (0x%08X) =3D 0x%08X\n", &(iommap->iop_padat), iommap->iop_padat);
Cheers,
Laurent Pinchart
Quoted 10 lines
Quoted 3 lines
#define P0 0x8000 // P0 bit in registers #define P1 0x4000 // P1 bit in registers #define P2 0x2000 // P2 bit in registersShouldn't that be
#define P0 0x80000000 // P0 bit in registers #define P1 0x40000000 // P1 bit in registers #define P2 0x20000000 // P2 bit in registers?
<shamed>Hmmm, next time I'll better use my calculator figuring out how many digits there is in a 32bit hex number... :P </shamed>
Many thanks for the assistance Laurent! (the LED came to life)
And you all may have a good long laugh of me now, I deserve it! :)
br
Robert