Date: Thu, 1 Feb 2007 19:37:50 -0500 From: Jung-uk Kim <jkim@FreeBSD.org> To: Divacky Roman <xdivac02@stud.fit.vutbr.cz> Cc: Perforce Change Reviews <perforce@FreeBSD.org> Subject: Re: PERFORCE change 113739 for review Message-ID: <200702011938.05616.jkim@FreeBSD.org> In-Reply-To: <200701311218.51103.jkim@FreeBSD.org> References: <200701310016.l0V0GccJ095661@repoman.freebsd.org> <20070131085906.GA28186@stud.fit.vutbr.cz> <200701311218.51103.jkim@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday 31 January 2007 12:18 pm, Jung-uk Kim wrote: > On Wednesday 31 January 2007 03:59 am, Divacky Roman wrote: > > On Wed, Jan 31, 2007 at 12:16:38AM +0000, Jung-uk Kim wrote: > > > http://perforce.freebsd.org/chv.cgi?CH=113739 > > > > > > Change 113739 by jkim@jkim_hammer on 2007/01/31 00:15:36 > > > > > > Do not reload %gs from linux_*sigcode(). > > > This seem to fix TLS on amd64, finally. :-) > > > > > > Tested on UP and QEMU. Need more testing on SMP. > > > > > > Affected files ... > > > > > > .. > > > //depot/projects/linuxolator/src/sys/amd64/linux32/linux32_loco > > >re .s#2 edit > > > > > > Differences ... > > > > > > ==== > > > //depot/projects/linuxolator/src/sys/amd64/linux32/linux32_loco > > >re .s#2 (text+ko) ==== > > > > > > @@ -11,7 +11,9 @@ > > > NON_GPROF_ENTRY(linux_sigcode) > > > call *LINUX_SIGF_HANDLER(%esp) > > > leal LINUX_SIGF_SC(%esp),%ebx /* linux scp */ > > > +#if 0 > > > movl LINUX_SC_GS(%ebx),%gs > > > +#endif > > > movl LINUX_SC_FS(%ebx),%fs > > > movl LINUX_SC_ES(%ebx),%es > > > movl LINUX_SC_DS(%ebx),%ds > > > @@ -25,7 +27,9 @@ > > > linux_rt_sigcode: > > > call *LINUX_RT_SIGF_HANDLER(%esp) > > > leal LINUX_RT_SIGF_UC(%esp),%ebx /* linux ucp */ > > > +#if 0 > > > movl LINUX_SC_GS(%ebx),%gs > > > +#endif > > > > great work! why didnt you just remove those lines but if 0 them? > > Because it's not done yet. I need more testing and clean-up before > MFP4. I'd like to test it on SMP but I don't have one. :-( I tried this on QEMU with dual-CPU emulation but it didn't look good. :-( We will have to find correct way to fix this. Jung-uk Kim
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200702011938.05616.jkim>