Thanks for the comments. On Tue, Feb 24, 2009 at 5:15 PM, <johan...@sun.com> wrote: > Hi Chad, > You probably want a vm expert to take a look at this code, but I'm > happy to provide comments anyway. > > vm/as.h: > > - line 114: You probably want this member at the end of the > structure. If we end up backporting the fix, we don't want to > displace the pre-existing offsets in a struct as
Fixed. > > - line 261: It might be worthwhile to generate a separate macro > for ((struct segvn_data *)(seg)->s_data). Since you use this > construct multiple times, it might make the AS_SEG_MAP_NORESERVE > macro easier to read. I had to paste into vim before I was able to > double-check the parenthesis. Done. > > vm/vm_as.c > > - line 1039: It looks like you could be updating a_asize outsize of > a_lock. Should you be checking as_lock_held and implementing some > kind of workaround if you don't have the a_lock here? Actually, I've moved this into segvn_faultpage() where swresv is being updated. I'm also doing this as an atomic operation there, as I don't have the lock there (or at least there are some cases in which I don't have the lock there.) > > vm/vm_rm.c: > > - lines 105-135: Dumb question -> Where is this case handled now? > Nope, not a dumb question, a dumb oversight on my part. And this might completely scupper what I want to do. The only options I see for doing this are either to change how we report the virtual size of the process (i.e., ignore this case and over-report the size) or move the accounting out into every code path that involves changing the size of a file. The second definitely seems a non-starter, if for no other reason than it has the potential to introduce a performance penalty in unexpected places (e.g., how long it takes me to append to a file is proportional to how many processes have the file mmap'd.) I'm guessing the first is a non-starter, too. (I'd love to hear someone say that's not the case.) Thanks, Chad _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
