Re: Zero-copy transfer with kiobufs: URGENT

[mohan kumar <mohan_knb@yahoo.com>]

hi Fabrice,
  i have succesfully used the 2.2.14 kernel to
transfer all of 100 Mbps sniffed traffic from a
network. i remapped a kernel memory region to user
space (vmalloced memmory) and used it to copy data(but
still the skbuff copy was not eliminated, which i
suppose is another added advantage in ur kiobuf
model). then i used a 1 Gb file cache to store the
data and process at a slow speed. 

this sure is lousy, but is working fine for me ;-). so
i think this mmap based approach should work fine for
100 Mbps, or equivalent other wired data.

hope it helped ;-)

Mohan S

--- Fabrice Tchakountio
<ftchakou@ftchakountio.bbn.com> wrote:
> Hi Stuart -
> 
> Thanks for your excellent suggestion. I have a
> couple of comments.
> 
> * How can we deal with potentially large chunks of
> memory:
>   I thought we need vmalloc() to allocate large
> kernel chunks (e.g: 
> >= 10 MB)
> 
> * I'm concerned about memory limitations since the
> data
>   is still wired:
> 
>   you're suggesting a mmap-based approach with the
> user having the
>   possibility to free up the memory with a close()
> call. 
> 
>   In our application, we intend to store each
> transferred chunk from the
>   kernel in a ring buffer. Thus there is a potential
> risk to consume all
>   the main memory with a mmap-based approach since
> the data is still
>   wired. 
> 
>   Ideas ?
> 
>   Thanks
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> On Thu, 4 Oct 2001, Stuart Macdonald wrote:
> 
> > Hi,
> > 
> > The following works for me:
> > 
> > 1) User space calls kernel-based function which:
> > 
> >         a)  creates a Virtual Memory Area of say
> 64KB. do_mmap_pgoff()
> > returning address and then call find_vma() to     
>    get the actual vma.
> > 
> >         b) does a pvKernelMem =
> __get_free_pages(up to 128KB I believe) to
> > allocate contiguous kernel pages. Note,        
> kmalloc() does not  always
> > end up being page-aligned. Save pvKernelMem for
> kernel access to shared
> > memory.
> > 
> >         c) Map the user VMA to the kernel pages
> using: set_page_reserved()
> > on all pages then remap_page_range().
> > 
> >         d) return VMA->vm_start to the user to
> denote the beginning address
> > of the shared memory.
> > 
> >  2) Use memory. Suggest you specify "volatile" on
> variables utilizing such
> > shared memory to avoid compiler                   
>      optimization related
> > issues.
> > 
> > 3) Have a "Close" function unmap by using
> do_munmap() and free kernel pages.
> > 
> > 
> > Stuart
> > 
> > 
> 
> -
> Kernelnewbies: Help each other learn about the Linux
> kernel.
> Archive:      
> http://mail.nl.linux.org/kernelnewbies/
> IRC Channel:   irc.openprojects.net / #kernelnewbies
> Web Page:      http://www.kernelnewbies.org/


__________________________________________________
Do You Yahoo!?
NEW from Yahoo! GeoCities - quick and easy web site hosting, just $8.95/month.
http://geocities.yahoo.com/ps/info1
-
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
IRC Channel:   irc.openprojects.net / #kernelnewbies
Web Page:      http://www.kernelnewbies.org/