| From: | Craig Dooley <cd5697@xxxxxxxxxx> |
| Date: | Thu, 31 Jul 2003 08:59:31 -0400 |
Yeah, sorry about that one, I wrote up the mail before checking the diary page again and realizing that you might be ripping all the iovec code for that out anyway. -Craig On Thursday 31 July 2003 04:41, Matthew Dillon wrote: > Don't worry about UIO's iovec, that one is my job. I am tentitively > planning on changing the iovec to a list of VM objects, offsets, and > extents. > > -Matt > Matthew Dillon > <dillon@xxxxxxxxxxxxx> > > :Craig Dooley wrote: > :> Theres a comment to change iov_base in struct iovec to a void * which > :> would be a pretty simple patch, just to cast it whenever doing math, but > :> should it be changed to be more forward thinking? What about systems > :> like x86 with PAE where a pointer cant address the whole memory space? > :> I dont have the FreeBSD sources on hand, but is it vm_offset_t? Is this > :> not an issue right now? Just a thought. > :> > :> -Craig > : > : For PAE, all physical address accesses are handled by u_int64_t > : pointers, just like in FreeBSD, they will be called vm_paddr_t > : and the kernel code will be modified to use it where > : appropriate. > : > : Cheers. > : > :-- > :Hiten Pandya > :hmp@xxxxxxxx > :http://hmp.serverninjas.com/ -- Craig Dooley cd5697@xxxxxxxxxx
Attachment:
pgp00007.pgp
Description: signature