5.3->5.4 upgrade method

Vizion vizion at vizion.occoxmail.com
Thu May 26 14:31:41 PDT 2005


On Thursday 26 May 2005 14:23,  the author Kris Kennaway contributed to the 
dialogue on Re: 5.3->5.4 upgrade method:
>On Thu, May 26, 2005 at 02:09:27PM -0700, Vizion wrote:
>> On Thursday 26 May 2005 14:10,  the author Kris Kennaway contributed to
>> the
>>
>> dialogue on Re: 5.3->5.4 upgrade method:
>> >On Thu, May 26, 2005 at 01:59:15PM -0700, Vizion wrote:
>> >> On Thursday 26 May 2005 13:37,  the author Kris Kennaway contributed to
>> >> the
>> >>
>> >> dialogue on Re: 5.3->5.4 upgrade method:
>> >> >On Thu, May 26, 2005 at 03:41:38PM -0400, Louis LeBlanc wrote:
>> >> >> Ok, I've settled on upgrading to 5.4 (from 5.4 RELEASE) but I can't
>> >> >> find an explicit statement that performing such an upgrade with
>> >> >> cvsup/buildworld/buildkernel, etc. would be ok.  No kernel config
>> >> >> gotchas mentioned either.  Of course it seems implied that this
>> >> >> would be a perfectly fine way to do the upgrade.
>> >> >>
>> >> >> Has anyone else used this upgrade method going from 5.3 to 5.4?
>> >> >> Any caveats on the kernel config?
>> >> >
>> >> >Yes, it's fine as long as you precisely follow the upgrade method (see
>> >> >handbook and UPDATING).
>> >> >
>> >> >Kris
>> >>
>> >> However I thought you might find the info in this thread worth
>> >> watching:
>> >
>> >How is this at all relevant to the discussion at hand?
>> >
>> >Kris
>> >
>> >> Re: OS check fails on JDK 1.4 & FreeBSD 5.4R
>> >>  From: Vizion <vizion at vizion.occoxmail.com>
>> >>  To: freebsd-java at freebsd.org
>> >>  CC: "Scott I. Remick" <scott at sremick.net>
>> >>
>> >> On Thursday 26 May 2005 09:34, ?the author Greg Lewis contributed to
>> >> the
>> >>
>> >> dialogue on Re: OS check fails on JDK 1.4 & FreeBSD 5.4R:
>> >> >On Thu, May 26, 2005 at 12:04:36PM -0400, Scott I. Remick wrote:
>> >> >> On Thu, 26 May 2005 09:54:59 -0600, Greg Lewis wrote:
>> >> >> > That would be your problem then. ?The test expects to use -lc_r on
>> >> >> > 5.x and higher:
>> >> >> >
>> >> >> > .if ${OSVERSION} > 500000
>> >> >> > LINKIT= -lc_r
>> >> >> > .else
>> >> >> > LINKIT= -pthread
>> >> >> > .endif
>> >> >>
>> >> >> Isn't that backwards? The default threading library changed from
>> >> >> libc_r to libpthread in 5.3
>> >> >
>> >> >I don't know that its backwards, but it could probably use some help.
>> >> >-lc_r became the official way early on in 5.x, but was reverted to
>> >> >-pthread after 5.2.1. ?So it was correct for a time, but is now out of
>> >> >date. ?I suspect that the variable should just go away and
>> >> > ${PTHREAD_LIBS} should take its place.
>> >> >
>> >> >> > However, you should have libc_r on 5.4:
>> >> >>
>> >> >> ...
>> >> >>
>> >> >> > So your machine would seem to be somewhat hosed unless you've
>> >> >> > deliberately removed libc_r.
>> >> >
>> >> >[explanation of removing libc_r snipped]
>> >> >
>> >> >> So I guess I have 2 questions now:
>> >> >>
>> >> >> 1) If libpthread is the new default, replacing libc_r, and you have
>> >> >> to have all things linked to just one or the other, basically libc_r
>> >> >> has been depreciated. How come the OS version check for JDK 1.4
>> >> >> wants libc_r if the OS > 5.0?
>> >> >
>> >> >Because its old (see above).
>> >> >
>> >> >> 2) What am I to do now, since the purging of all things libc_r when
>> >> >> I upgraded to 5.3 is what ultimately fixed my system and has kept
>> >> >> things running smoothly ever since?
>> >> >
>> >> >You can do one of the following:
>> >> >
>> >> >1. Try the attached patch. ?This would be my preference since if it
>> >> > works ? I'll try and get it committed.
>> >> >2. Turn on SKIP_OS_CHECK and eliminate the test altogether.
>> >>
>> >> I have attached the Makefile.diff to this email
>> >>
>> >>
>> >> freebsd-java at freebsd.org mailing list
>> >> http://lists.freebsd.org/mailman/listinfo/freebsd-java
>> >> To unsubscribe, send any mail to "freebsd-java-unsubscribe at freebsd.org"
>>
>> It seems there was a problem re lc-r with 5.4 that Greg Lewis's diff
>> (which he is hoping to get committed) should fix.
>
>No, the problem is in a specific port and has nothing to do with a
>problem with libc_r or with the upgrade from 5.3 to 5.4, AFAICT.
>
>Kris

I think you may be right BUT nonetheless anyone upgrading might find it 
helpful to be aware of the issue

David

-- 
40 yrs navigating and computing in blue waters.
English Owner & Captain of British Registered 60' bluewater Ketch S/V Taurus.
 Currently in San Diego, CA. Sailing May bound for Europe via Panama Canal.


More information about the freebsd-questions mailing list