[reiserfs-list] Re: [OpenAFS-devel] more on the 2.2.18pre17 SMPcpu hog/etc.
Mon, 4 Dec 2000 19:07:00 -0500 (EST)
On Tue, 5 Dec 2000, Harald Barth wrote:
> warlord> This implies that it has to walk the directory at the beginning, so
> warlord> it doesn't matter how long it takes to walk it.
> It matters to me. Init 300Mb cache on Sun1000E (Sol7): Enough time for
> a trip to Burger King to get a banana milkshake and back:
> Nov 30 21:27:59 shake.stacken.kth.se unix: Starting AFS cache scan...
> Nov 30 21:56:07 shake.stacken.kth.se unix: found 0 non-empty cache files (0%).
Yeah, we've noticed that Solaris 7 is really horrible about this as
compared to other systems we run.
> jhutz> Not true.
> Wasn't reading afs/afs_memcache.c:afs_InitMemCache() right. However
> the (memCache+index)->data could point into any zeroed memory area
> (mmap?) of right size.
Yeah, I suppose it could. So yes, the memcache code could be rewritten
to use a single large block of memory, if you were careful to allocate
and free it in the right way.