Port-sparc64 archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: Using %stick where available
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hello,
On Nov 6, 2012, at 9:49 PM, Erik Fair wrote:
On Nov 6, 2012, at 18:37, Michael wrote:
Good question, I didn't do any measurements other than 'looks ok to
me'.
I think there's a bug in the code that's supposed to synchronize
the CPUs' time bases - that could cause crap you'd call 'terrible
timekeeping', like the clock appearing to jump back and forth semi-
randomly ( whenever you hop to a different CPU )
Keeping them in sync should be much easier with the time base
running at only 12MHz though.
I have a dual core HP Microserver N40L (AMD Turion II Neo) that
keeps quite good time, even when not using the HPET device, which
suggests that the amd64 port has the problem of which CPU to get
time from solved.
Well, the problem isn't exactly difficult. If it is what I think it is
it's only one extra, unnecessary settick(0) on the boot CPU which
happens when initializing the timecounter, some time after syncing
time bases on SMP boxes.
I guess I need to test the patches on my v240 ...
Please do. It's likely to have the same problem though, so as long as
it's not getting worse we're doing good.
have fun
Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
Comment: GPGTools - http://gpgtools.org
iQEVAwUBUJpOeMpnzkX8Yg2nAQLwmgf/eQL+urISvCDVbKNDZp2MOgEr0gWM2tHv
QSgKWfNsznjm5DsUqKGtWceqyu3pOnspAJCA4VwfPHR9TYx6E1uhs71aBNFBjXCV
e//ZiuiT5G3r64oMrGjD/c7tCfB9kjTBUMkkSOxMIJST7ggL2cViIC9keeMUBzfa
OWdMCggUHD/+pJjH6CfZHss9WMtOn931/I/ec8YKk4fHFtcT5OOrAgnbrKQEo2Y6
8bGrFqsPEGHjGFzSDTkZTH0nJC8CfcJ5sdsNEpr5xGnWAvWbymSMUoBI/TWfKYxH
emxbDvKa7YR6fx6E1s/kAKnzrkV+EANIy5hZYhn0atv7nPZycaTAFw==
=3U4o
-----END PGP SIGNATURE-----
Home |
Main Index |
Thread Index |
Old Index