HI All, I started to understand output of /proc/oiports and /proc/iomem. I confused to relate these two's output. Below is the kernel source for Mapping of io-port to io-mem. Could anyone please explain below code. /* We encode the physical PIO addresses (0-0xffff) into the (0- * pointer by offsetting them with a constant (0x10000) and * assuming that all the low addresses are always PIO. That means means * we can do some sanity checks on the low bits, and don't * need to just take things for granted. */ #define PIO_OFFSET 0x10000UL #define PIO_MASK 0x0ffffUL #define PIO_RESERVED 0x40000UL void __iomem *ioport_map(unsigned long port, unsigned int nr) { __iomem if (port > PIO_MASK) return NULL; return (void __iomem *) (unsigned long) (port + PIO_OFFSET); __iomem } Thanks, Prabhu
Hi, This code says that there are 0x10000 possible ioports i.e. in the range (0 - 0xFFFF). To map these into iomemory, simply an address = (0x10000 + ioport num) is used. Thus this code:
if (port > PIO_MASK) return NULL;
Checks that the ioport number is within the range of ioports.
return (void __iomem *) (unsigned long) (port + PIO_OFFSET);
And if yes, offsets it by 0x10000 and returns the resulting address that shall be used as iomemory. Thanks, Rajat Jain ________________________________________ From: kernelnewbies-bounces@kernelnewbies.org [mailto:kernelnewbies-bounces@kernelnewbies.org] On Behalf Of prabhu Sent: Wednesday, February 16, 2011 1:15 PM To: kernelnewbies Subject: iomem and ioports HI All, I started to understand output of /proc/oiports and /proc/iomem. I confused to relate these two's output. Below is the kernel source for Mapping of io-port to io-mem. Could anyone please explain below code. /* We encode the physical PIO addresses (0-0xffff) into the (0- * pointer by offsetting them with a constant (0x10000) and * assuming that all the low addresses are always PIO. That means means * we can do some sanity checks on the low bits, and don't * need to just take things for granted. */ #define PIO_OFFSET 0x10000UL #define PIO_MASK 0x0ffffUL #define PIO_RESERVED 0x40000UL void __iomem *ioport_map(unsigned long port, unsigned int nr) { __iomem if (port > PIO_MASK) return NULL; return (void __iomem *) (unsigned long) (port + PIO_OFFSET); __iomem } Thanks, Prabhu
On Wed, Feb 16, 2011 at 5:13 PM, Rajat Jain <rajatjain@juniper.net> wrote:
Hi,
This code says that there are 0x10000 possible ioports i.e. in the range (0 - 0xFFFF). To map these into iomemory, simply an address = (0x10000 + ioport num) is used. Thus this code:
if (port > PIO_MASK) return NULL;
Checks that the ioport number is within the range of ioports.
return (void __iomem *) (unsigned long) (port + PIO_OFFSET);
And if yes, offsets it by 0x10000 and returns the resulting address that shall be used as iomemory.
Some places i have seen that directly IO port is typecasted to (__iomem * ) and used as it is i.e. returned value is given to ioread and iowrite functions.Kindly explain why in some cases we dont add offset and in some cases we add?? Is it because in some board we need to memory map the port number and then read/write and in some boards we don't need to do that as we can directly access the port using arch specific functions??
Thanks,
Rajat Jain
________________________________________ From: kernelnewbies-bounces@kernelnewbies.org [mailto: kernelnewbies-bounces@kernelnewbies.org] On Behalf Of prabhu Sent: Wednesday, February 16, 2011 1:15 PM To: kernelnewbies Subject: iomem and ioports
HI All,
I started to understand output of /proc/oiports and /proc/iomem. I confused to relate these two's output.
Below is the kernel source for Mapping of io-port to io-mem. Could anyone please explain below code.
/* We encode the physical PIO addresses (0-0xffff) into the (0- * pointer by offsetting them with a constant (0x10000) and * assuming that all the low addresses are always PIO. That means means * we can do some sanity checks on the low bits, and don't * need to just take things for granted. */ #define PIO_OFFSET 0x10000UL #define PIO_MASK 0x0ffffUL #define PIO_RESERVED 0x40000UL void __iomem *ioport_map(unsigned long port, unsigned int nr) { __iomem if (port > PIO_MASK) return NULL; return (void __iomem *) (unsigned long) (port + PIO_OFFSET); __iomem }
Thanks, Prabhu
_______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org http://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
Some places i have seen that directly IO port is typecasted to (__iomem * ) and used as it is i.e. returned value is given to ioread and iowrite functions.Kindly explain why in some cases we dont add offset and in some cases we add?? Is it because in some board we need to memory map the port number and then read/write and in some boards we don't need to do that as we can directly access the port using arch specific functions?? RAJAT> I would think so.
participants (3)
-
anish singh -
prabhu -
Rajat Jain