On Wed, 2004-10-06 at 15:38, Jon Masters wrote: > On Wed, 6 Oct 2004 14:50:18 +0530, Thekkedath, Gopakumar > <gopakumar.thekkedath@xxxxxxx> wrote: > > > IO Ports unlike IO memory does not have a virtual to physical > > translation > > ie what ever you specify as the IO Port address in your 'IN' or 'OUT' > > assembly instructions will appear in the address lines. > > Execpt IO memory on Intel is hopefully dying over time. The preferable > way to handle device IO IMO is to memory map the device and go through > the MMU for translation. > I had read somewhere that accessing IO ports thru. MMIO is faster than thru. IO-maped-IO? any pointers as to why this would be??? wud'nt MMIO access's have the extra overhead of address translation from vir. -> phy ??? If i use PIO (x86 arch.) shud'nt the access be faster as here i have special instr. for this purpose...or is that these instr. themselves r slow...i.e they take more clk cycles to complete??? > Jon. > > -- > Kernelnewbies: Help each other learn about the Linux kernel. > Archive: http://mail.nl.linux.org/kernelnewbies/ > FAQ: http://kernelnewbies.org/faq/ > -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
