On Fri, 22 Jan 1999, Thomas Schenk wrote:
> >I'm fairly certain that it is the virtual to physical mapping macros that are
> >producing incorrect results. Please try changing linux/include/asm-i386/io.h
> >from:
> >
> >#define __io_virt(x) ((void *)(PAGE_OFFSET | (unsigned long)(x)))
> >#define __io_phys(x) ((unsigned long)(x) & ~PAGE_OFFSET)
> >
> >to:
> >
> >#define __io_virt(x) ((void *)((unsigned long)(x) + PAGE_OFFSET))
> >#define __io_phys(x) ((unsigned long)(x) - PAGE_OFFSET)
the + - variant just hides the real bug. If it makes a difference, then
this is clearly a bug somewhere in a driver. __io_virt(__io_virt(x))
should never be done in a driver. It would be very nice to debug the
driver bug, the bug could have other side-effects too. To debug 'invalid'
(ie. nested) uses of __io_virt() and __io_phys():
#define __io_virt(x) ({ \
void * __res; \
if (x >= PAGE_OFFSET) { \
printk("__io_virt bug in %s:%d!\n", __FILE__, __LINE__); \
__res = x; \
} else \
__res = ((void *)(PAGE_OFFSET | (unsigned long)(x))); \
__res;
})
#define __io_phys(x) ({ \
unsigned long __res; \
if (x <= PAGE_OFFSET) { \
printk("__io_phys bug in %s:%d!\n", __FILE__, __LINE__); \
__res = x; \
} else \
__res = ((unsigned long)(x) & ~PAGE_OFFSET);
__res;
})
Do these debugging variants print anything suspicious? Thanks,
-- mingo
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
