Date: Sun, 25 Aug 2013 09:58:12 -0600 From: Warner Losh <imp@bsdimp.com> To: Tim Kientzle <tim@kientzle.com> Cc: freebsd-arm@freebsd.org Subject: Re: Pretty good RPi version? Message-ID: <2DF35AA8-01C3-4F99-848A-34BD52EA5193@bsdimp.com> In-Reply-To: <E0163ECC-386A-415E-AC97-631DD9E0C7B1@kientzle.com> References: <5218FBE2.2000907@m5p.com> <20130825.195532.59669686.shigeru@os-hackers.jp> <CAJ-VmondKrRr_oxHDRULTiwb-EL3rXrVv1shJc09%2B4auXj%2ByCA@mail.gmail.com> <E0163ECC-386A-415E-AC97-631DD9E0C7B1@kientzle.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Aug 25, 2013, at 9:34 AM, Tim Kientzle wrote: >=20 > On Aug 25, 2013, at 5:07 AM, Adrian Chadd wrote: >=20 >> Hi! >>=20 >> All - if it's unstable, please complain loudly on -arm and -current = and >> file PRs. If you have the time, please help figure out which = commit(s) >> broke things. >>=20 >> There's been plenty of "stuff is unstable" talk, but I'm not sure = it's >> percolated its way up to the people doing VM hacking on other = platforms >> (notably amd64) so they are likely unaware they've broken things. >>=20 >> It's possible they've also broken things for MIPS, which has me = worried. :( >=20 > On a similar note: FreeBSD is starting the process for branching > and releasing FreeBSD 10.0. (The whole process will take a few > months, but it means we need to get bugs isolated and fixed NOW.) >=20 > I would *love* to have a solid, stable FreeBSD/ARM for 10.0. > (If nothing else, I started tinkering with BeagleBone in late 2011 > specifically so I could move my personal email server onto one. ;-) >=20 > To get there, we need as many people as possible: > * Running the bleeding edge FreeBSD. > * Reporting problems. > * Helping to diagnose those problems. >=20 > Q: Does anyone know a good way to test userland locking? >=20 > Diane Bruce has been continuing to dig into the issue that > broke sshd a while back; the clues point to some issue in > jemalloc (possibly locking related?). Once upon a time, back in the RAS/armv4 days, I found a bug. Locks were = getting stuck in threads in my employer's command and control program. = After much debugging, it came down to the atomic operations implemented = by RAS were less than atomic. There was a one or two instruction race in = them. Usually they'd work, and you needed a very high interrupt load, = like in my employer's hardware, to see the problem reliably. It would be = useful to eliminate this possibility, even though RAS is off the table = for armv6, this is really the first release with armv6 support and the = atomics haven't been through the torture test of that process yet. Warner > Q: Has anyone tried running some or all of the FreeBSD > regression tests? >=20 > Tim >=20 > _______________________________________________ > freebsd-arm@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-arm > To unsubscribe, send any mail to "freebsd-arm-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?2DF35AA8-01C3-4F99-848A-34BD52EA5193>