On 4/17/07, srinivas bakki <srinivas.bakki@xxxxxxxxx> wrote:
>
> Sorry if I misunderstood you. I agree with Srinivas than
> remap_pfn_range() could be your solution. However, seems like you need
> to create a VMA first (that could be created in user space via
> mmap(). You must also pass user space start address... So IMHO you
> need to coordinate between the kernel code and user space app i.e User
> space need to tell the kernel code, which VMA should be used.
Hi,
The call to mmap in the user space creates vm_area_struct with
available virtual addresses by the kernel ( file subsystem in particular) .
This vm_start, vm_end and kernel logical address must be used to call
remap_pfn_range. this maps the kernel logical addresses to the addresses
described by the vm_area_struct.
CMIIW
Regards
Srinivas Bakki
This is all correct.
But i think the problem put by Thomas is something different.
Suppose i have a kernel buffer which is mmaped into user-space.
Now inside some ioctl implementation(not in mmap), if I want to find
the corresponding user address(mmaped start address) for a kernel
buffer start address, is there any kernel function which can be used ?
OR
Is there any function available for resolving the kernel logical addr
to user virt addr translation(if mmaped) ?
This is my interpretation.
CMIIW
yogeshwar
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
