Re: Fixed address device-shared kernel memory
Jens Axboe (axboe@image.dk)
Thu, 23 Sep 1999 00:03:11 +0200
On Wed, Sep 22 1999, Richard B. Johnson wrote:
> ptr = phys_to_vert(0xcc000);
>
> ... which is a shared memory window in kernel space.
>
> This works for about 20 seconds. Then (it appears as though)
> the kernel unmaps this virtual address, resulting in
> data observed at this address to not be the data from the
> shared-memory device.
>
> The problem with the disappearing mapping seems to correspond
> with the wakeup of kswapd or kflushd.
>
> I looked through kernel drivers and didn't get a clue as
> to how to prevent this. Many drivers, which use shared memory,
> don't even use the phys_to_vert macro. If I try that, the
> result is a panic with Unable to handle kernel paging request
> of 0x000cc000. So, the physical to virual mapping is obviously
> required in a kernel module.
>
> So, the question is; "How do I prevent the kernel from unmapping
> the physical-to-virtual translated pages?"
>
> Was I supposed to call do_mmap(); and later do_munmap(); in
> the kernel driver?
You can't just access the memory directly, you have to remap
it first.
--
* Jens Axboe <axboe@image.dk>
* Linux CD-ROM Maintainer
* http://www.kernel.dk
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/