Date: Wed, 1 Nov 2006 10:13:01 +0100 From: Divacky Roman <xdivac02@stud.fit.vutbr.cz> To: Jung-uk Kim <jkim@freebsd.org> Cc: emulation@freebsd.org Subject: Re: [kib@freebsd.org: [src] cvs commit: src/sys/amd64/linux32 linux32_sysvec.c] Message-ID: <20061101091301.GA91887@stud.fit.vutbr.cz> In-Reply-To: <200610311945.56003.jkim@FreeBSD.org> References: <20061031182625.GM1627@deviant.kiev.zoral.com.ua> <200610311558.29598.jkim@FreeBSD.org> <20061101013444.7df5fb47@Magellan.Leidinger.net> <200610311945.56003.jkim@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Oct 31, 2006 at 07:45:53PM -0500, Jung-uk Kim wrote: > On Tuesday 31 October 2006 07:34 pm, Alexander Leidinger wrote: > > Quoting Jung-uk Kim <jkim@FreeBSD.org> (Tue, 31 Oct 2006 15:58:26 > -0500): > > > On Tuesday 31 October 2006 02:36 pm, Jung-uk Kim wrote: > > > > I am running full LTP and it seems very promising so far. > > > > > > I have run LTP 20060822 against ~1 day old source + the fix. > > > Here are the results: > > > > Can you please run the 20061017 LTP? > > I cannot build LTP because gmake core dumps, i.e., TLS is missing for > amd64. Can you update your LTP tarball? Is your TLS patch for amd64 > in p4 tree functional? the state of amd64-TLS is this: its basically implemented but there's a bug that lets only first program run with TLS, all subsequent attempts to use TLS segfaults. I discussed this with davidxu@ and he told me saw something similar which was caused by reloading of %gs $somewhere. I digged through the sources and in linux32_locore.s we reload it.. I am no expert with this so I'd like to ask you for help and possible testing. can you test p4 branch (which includes amd64-tls) with the %gs/%fs reloading commented out in linux32_locore.s? and if that doesnt work look at the issue (I guess you are fluent with amd64 internals). thnx roman
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20061101091301.GA91887>