Testing is FAR FAR FAR from complete. PCRE was able to find 2 invalid specifications that the posix API didn't complain about. Dan has fixed those. I will verify any changes with a complete relabel before submitting. -Eric On Tue, Aug 14, 2012 at 12:48 PM, Stephen Smalley <sds@xxxxxxxxxxxxx> wrote: > On Tue, 2012-08-14 at 10:01 -0400, Eric Paris wrote: >> Initial really really rough testing using pcre: >> >> Load and free DB 100 times using GLIBC: >> 8.370 seconds used by the processor. >> >> Load and free DB 100 times using pcre: >> 1.800 seconds used by the processor. >> >> This is just using PCRE to load and compile the DB. Not using an >> mmap'd precompiled version. That's next on the testing list. But >> switching to pcre is already a huge time win it seems.... > > Do you need to convert any of the regexes in file_contexts to address > the differences between PCRE and POSIX regexes? > > You should check that setfiles using glibc regex followed by setfiles > using PCRE does not change anything on the filesystem, i.e. run the > existing setfiles on all filesystems, then run the setfiles using PCRE > with -nv to report any changes that would be made without applying them. > > -- > Stephen Smalley > National Security Agency > -- 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.
