Date: Sat, 18 May 1996 14:59:15 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: julian@ref.tfs.com (JULIAN Elischer) Cc: current@freebsd.org Subject: Re: cvs commit: src/sys/vm device_pager.c swap_pager.c vm_fault.c vm_glue.c vm_kern.c vm_map.c vm_meter.c vm_mmap.c vm_object.c Message-ID: <199605181959.OAA01580@dyson.iquest.net> In-Reply-To: <199605181936.MAA15411@ref.tfs.com> from "JULIAN Elischer" at May 18, 96 12:36:28 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > Significant revamp of the pmap/fork code: > > 1) PTE's and UPAGES's are NO LONGER in the process's map. > > 2) PTE's and UPAGES's reside in their own objects. > > > Does this mean that we can now do the rfork() call correctly > so that two processes can share an address space by simply incrementing > the reference count in the address-space structure? > Remember that the vmspace structure contains both the pmap and map's. And, unfortunately, the map points to the pmap... Since the PTE's and UPAGES's still have to be in the process's pmap, there is still a problem. We will have to split the map and pmap up -- or support sharing map's better. I did not remove sharing map support in the code (even though it was tempting), thinking that (sharing maps) might help solve this problem. The way that sharing maps were used in the code was broken, but the underlying support is not missing. I might experiment with sharing maps in the next few weeks, and those might be the right way to go. John
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199605181959.OAA01580>