On Tue, 14 Aug 2012, Eric Paris <eparis@xxxxxxxxxxxxxx> wrote: > I have code that does just that. Dan and I both wrote a version. > I'll attach it. I didn't find the speedups we were hoping for and it > didn't work correctly/completely in the face of file context > equivalencies. Although that is likely fixable. I was just looking > at all of the stem code (and wondered who wrote it but it was > pre-git). I'm surprised it made a big difference. Wouldn't the regex > code be able to return extremely quickly if it didn't match? Anyway. > I'm writing some test programs to look at all of the possibilities. At the time I wrote the code I didn't attempt to micro-benchmark it. As the use case that mattered most was a full relabel of a root filesystem (between 40,000 and 120,000 files in the common case) and the number of regexes was smaller than it is today the win was fairly obvious. >From memory the stem compression change typically improved performance by a factor of 3-5 depending on what you were doing. That was on systems like a Pentium-166 or P2-400. The performance benefits on faster systems (800MHz Athlon and better) were less important to me at the time. Prior to stem compression the relabel was entirely CPU bottlenecked on all systems. After stem compression and some new CPU releases from Intel and AMD I wrote the code to have two processes doing disk IO (one stat'ing the files and the other applying labels) so that we could have some parallelism between regex checks and seeks on disk. A theoretical regex library wouldn't be much slower than the integer checks we are doing, it could just use a 32bit integer compare to check the first 4 bytes which would compare to the integer stem check we do. Does one of the current regex libraries do such optimisations where you have a common case of a non- match on a relatively simple regex? -- My Main Blog http://etbe.coker.com.au/ My Documents Blog http://doc.coker.com.au/ -- This message was distributed to subscribers of the selinux mailing list. If you no longer wish to subscribe, send mail to majordomo@xxxxxxxxxxxxx with the words "unsubscribe selinux" without quotes as the message.
