Skip to contentSkip to conversation

2.5.62 ioremap fails

DISCUSSION Rob Murphy 1 participant 1 message Feb 24, 2003

lore
Share & tools
lore

Rob Murphy

author
#0 lore raw

I have been trying to allocate memory past the end of the kernel in a system that has more that 1 Gb of memory and have been unsuccessful. My current machine has an athlon processor and 1.5Gb of memory.

I created a driver with the following code:

#define MAXHM 0x8000000 // 128 Mb
int __init init_module(void)

{
    void *kHighMem;

    printk(KERN_ERR"before ioremap\n");
    kHighMem = ioremap(__pa(high_memory), MAXHM);
    printk(KERN_ERR"ioremap ret: %x\n",kHighMem);

    return 0;
}
void __exit leave_mod(void) {
    iounmap(kHighMem);
    printk(KERN_ERR"iounmap passed\n");

    printk(KERN_ERR,"module successfully unloaded.\n");
}

I add the line append="mem=xx" to lilo.
If I set "xx" to around 860 or above then ioremap fails. If I set "xx" to 860 or less then ioremap is successful. Has anyone had any experience trying to allocate a contiguous memory buffer beyond the end of what the kernel manages in a system with over 1Gb of memory? Any help would be appreciated in this matter.

Regards,
Rob Murphy