Date: Thu, 8 Jun 2006 07:03:14 +1000 From: Sam Lawrance <boris@brooknet.com.au> To: Anish Mistry <amistry@am-productions.biz> Cc: Sten Daniel S?rsdal <lists@wm-access.no>, freebsd-ports@freebsd.org Subject: Re: 6.1R: Postgresql 8.1.3 and threadsafe. Message-ID: <E02F2680-D193-4D38-B93D-FE843EA471DF@brooknet.com.au> In-Reply-To: <200606071412.45003.amistry@am-productions.biz> References: <447B76FA.1030304@wm-access.no> <20060530020354.GA96718@xor.obsecurity.org> <AA50D674-0D77-4AC0-8F48-F5A303EEF7D4@brooknet.com.au> <200606071412.45003.amistry@am-productions.biz>
next in thread | previous in thread | raw e-mail | index | archive | help
On 08/06/2006, at 4:12 AM, Anish Mistry wrote: > On Wednesday 07 June 2006 06:29, Sam Lawrance wrote: >> On 30/05/2006, at 12:03 PM, Kris Kennaway wrote: >>> On Tue, May 30, 2006 at 12:34:34AM +0200, Sten Daniel S?rsdal > wrote: >>>> Does this one go to database@ or ports@ >>>> >>>> FreeBSD 6.1 RELEASE (i386) considered threadsafe with Postgresql >>>> 8.1.3. >>>> I ran 8.1.2 without any noticable effects but after upgrading >>>> from 5.4 >>>> to 6.1 by means of making world (make buildworld; make kernel; >>>> reboot; >>>> rm -rf /usr/obj/*; make buildworld; shutdown now; mergemaster >>>> -p; make >>>> installworld; mergemaster -i; sync; reboot; make kernel; reboot) >>>> >>>> I have had segmentation faults (signal 11) of conftest whenever >>>> i try to >>>> build it from ports. My local ports and src tree is the same as >>>> the one >>>> distributed with 6.1 RELEASE (from 6.1 RELEASE CD) >>>> >>>> I built world with -O2 -pipe and kernel with -O2 -pipe. kernel >>>> is not generic but instead i run my own configuration >>>> (included). In /etc/make.conf i chose not to build games and >>>> shared docs. >>>> >>>> I'm fairly certain the memory chips are ok after a few rounds of >>>> memtest86. >>>> >>>> options CLK_USE_I8254_CALIBRATION >>>> options PQ_CACHESIZE=512 >>>> options DIRECTIO >>>> options SW_WATCHDOG >>>> options MP_WATCHDOG >>>> options ZERO_COPY_SOCKETS >>>> options AUTO_EOI_1 >>>> options AUTO_EOI_2 >>> >>> That's a lot of weird options, are you sure you need them? >>> >>> Look at what the relevant conftest.c is running, verify it >>> segfaults when you run it by hand, then post the code. >> >> Not to mention compiling the kernel with -O2 is unsupported, and >> generally considered a bad idea. > Not true. For kernel compiling -O2 has been supported since 6.0. I'm not aware of it. Got a pointer to some documentation or list archives so I can read up?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E02F2680-D193-4D38-B93D-FE843EA471DF>