#24254: There needs to be documentation on what kernel versions the KIST Scheduler will run on ---------------------------+------------------------------------ Reporter: Dbryrtfbcbhgf | Owner: pastly Type: defect | Status: assigned Priority: Medium | Milestone: Tor: 0.3.3.x-final Component: Core Tor/Tor | Version: Severity: Normal | Resolution: Keywords: tor-sched | Actual Points: Parent ID: | Points: Reviewer: | Sponsor: ---------------------------+------------------------------------ Changes (by pastly):
* status: new => assigned * owner: (none) => pastly Comment: Thanks. I agree this should be documented. Is this the same Tor instance you are using in #24158? I suspect it is, and I suspect your Tor is still running KISTLite. Does that sound correct? I suspect that to be the case because I've now (re)looked into this. KIST support requires two things: 1. The tcp_info struct in the kernel having the right members 2. The SIOCOUTQNSD symbol being defined in a certain header As long as getsockopt has worked the same for all of time (even before the switch to putting some headers in a "uapi" directory), then the first check for KIST support will pass on kernels AT LEAST as old as 2005's linux-2.6.12-rc2, the first git commit. The second check will only pass on kernels v2.6.39 or newer. So it looks like "KIST" will only work if you have v2.6.39 or newer, which is slightly newer than your kernel. "KISTLite" should work regardless of your kernel. Hell, you don't even need to be using Linux. Now __WHERE__ should we document this? Comment in the source? Man page? Both? Neither? Looking for input from anyone here. -- Ticket URL: <https://trac.torproject.org/projects/tor/ticket/24254#comment:4> Tor Bug Tracker & Wiki <https://trac.torproject.org/> The Tor Project: anonymity online
_______________________________________________ tor-bugs mailing list tor-bugs@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs