What is the size of virtual memory? My understanding would be that virtual memory == paging space. In my case my system has 256MB physical and a 512MB swap partition. Thus I can have at most 768MB of memory in use at one time before I exhaust all of the memory. Since kernel memory is not pageable then the total would be 768 - kernel memory which would be avaialble for user space applications. In my test I was able to allocate 705M on the heap in a user application and do a write which works. Based on this I can conclude that the kernel could not have created a copy of my buffer before it wrote the data to disk as it would have run out of memory. So my original question is, how does the kernel do this? I was under the impression that any memory passed to the kernel had to be checked to see if it is valid and then copied to make sure that it was in physical non-paged memory. The only thing I can think of is that in this case the kernel can use the user space addresses as is and is in a position to handle page faults? Does this make sense? Thanks, Tony > > From: Timur Tabi <timur.tabi@xxxxxxxxxxx> > Date: 2005/03/22 Tue AM 10:48:52 CST > To: kernelnewbies@xxxxxxxxxxxx > Subject: Re: No copy on write system call? > > stackframe@xxxxxxxxxxx wrote: > > How is it that a write (man 2 write) can be passed a buffer > > that exceeds most of the physical and virtual memory on a > > system and work successfully? > > I don't see how you can create a buffer at all that exceeds the size of > virtual memory. > > -- > Timur Tabi > Staff Software Engineer > timur.tabi@xxxxxxxxxxx > > -- > Kernelnewbies: Help each other learn about the Linux kernel. > Archive: http://mail.nl.linux.org/kernelnewbies/ > FAQ: http://kernelnewbies.org/faq/ > > -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
