Some comments, mostly to myself :) usr/src/uts/common/sys/rctl.h No change from onnv. (fixed)
usr/src/uts/common/os/rctl.c No change from onnv. (fixed) usr/src/uts/common/sys/project.h This is more of a comment for myself. I wonder if we should think about getting rid of kpj_zoneid. If the project exists, then we know the zone exists. We could just have a pointer to the zone here. usr/src/uts/common/os/project.c 294 If you are going to create a lock ordering between project_hash_lock and kstat_install() locks, then perhaps we should move 310-316 up to 294. usr/src/uts/common/os/zone.c What is with all the calls to cpucaps_zone_remove(zone). We should talk more about these. 3993,3994,4034,4065, what is this stuff? Oh, introduced curzone macro. Ok. looks good. usr/src/uts/common/disp/thread.c 199. Reordering zone_init()/project_init()/brand_init(), yikes! :) I haven't found any problems with reordering zone_init()/project_init(). Be sure to get review from the brandz team. usr/src/uts/common/os/lwp.c 783 Reason for move of CL_EXIT not obvious. usr/src/uts/common/os/task.c 497-501 I now remember why these can't be deleted. task_create() copies the rctl set from the parent task, and task_attach() calls the set-op callbacks for each task rctl. If we removed this, the task rctl set callbacks would never get called. usr/src/cmd/zonecfg/zonecfg.c 1905 tmp -> tmp_pcap -Steve On Fri, Jan 19, 2007 at 03:06:33PM -0800, Alexander Kolbasov wrote: > If you are interested in creative ways to make your system run slower, please > take a look at the following webrev: > > http://cr.grommit.com/~akolb/webrev_56_OS/ > > and let me know what you think of it. > > - Alex Kolbasov > -- > http://blogs.sun.com/akolb > > _______________________________________________ > zones-discuss mailing list > zones-discuss@opensolaris.org _______________________________________________ zones-discuss mailing list zones-discuss@opensolaris.org