On Fri, Jun 26, 2015 at 04:00:50PM +0200, Rasmus Villemoes wrote: > > I tried doing a `smatch_data/db/create_db.sh -p=kernel warns.txt` and it > > was way slower. :( I had to kill it after some hours. > > That's odd. What kind of hardware do you run this on? > The number of cores shouldn't matter for building the DB, but I'm on intel quad core with hyper threading and 32 GB RAM. How big is your smatch_db.sqlite? Mine is 11GB. > Anyway, I've redone that part, so now perl/DBI is back, but the solution > is a little ugly. See the misc_tweaks2 branch on github. I just built > the database from scratch for an allmodconfig kernel in 63 minutes, > which is almost exactly twice as fast as yesterday. If this version is > also slow for you something weird must be going on. > Btw, I changed your cut commands from: - cut -f3 -d: + cut -d: -f3- Because the original breaks if the SQL has a ':' char in it. > Rasmus > > PS: Just noticed that you used warns.txt, which suggests that you used > the result from an old build. I've changed the format of some of the > output lines in (smatch_)warns.txt, so you need to build the kernel from > scratch. It was an old build. I don't see where you changed any SQL format... I've started cherry-picking the earlier patches. > > PPS: When you say "My DB takes an hour to build", is that the > create_db.sh part, or is it the time for an entire build_kernel_data.sh > run? That's just the create_db.sh. To be honest, I'm not sure how long it takes... I just kick it off and come back later. Every day if you rebuild the DB, then it gets slightly larger and larger... regards, dan carpenter -- To unsubscribe from this list: send the line "unsubscribe smatch" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
