Date: Mon, 22 Nov 2010 10:29:28 -0500 From: John Baldwin <jhb@freebsd.org> To: Kostik Belousov <kostikbel@gmail.com> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org, Alexander Leidinger <netchild@freebsd.org> Subject: Re: svn commit: r215675 - head/sys/compat/linux Message-ID: <201011221029.29119.jhb@freebsd.org> In-Reply-To: <20101122125106.GZ2392@deviant.kiev.zoral.com.ua> References: <201011221242.oAMCgW9O092144@svn.freebsd.org> <20101122125106.GZ2392@deviant.kiev.zoral.com.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday, November 22, 2010 7:51:06 am Kostik Belousov wrote: > On Mon, Nov 22, 2010 at 12:42:32PM +0000, Alexander Leidinger wrote: > > Author: netchild > > Date: Mon Nov 22 12:42:32 2010 > > New Revision: 215675 > > URL: http://svn.freebsd.org/changeset/base/215675 > > > > Log: > > Do not take the process lock. The assignment to u_short inside the > > properly aligned structure is atomic on all supported architectures, and > > the thread that should see side-effect of assignment is the same thread > > that does assignment. > > > > Use a more appropriate conditional to detect the linux ABI. > > > > Suggested by: kib > > X-MFC: together with r215664 > > > > Modified: > > head/sys/compat/linux/linux_emul.c > > > > Modified: head/sys/compat/linux/linux_emul.c > > ============================================================================== > > --- head/sys/compat/linux/linux_emul.c Mon Nov 22 12:33:48 2010 (r215674) > > +++ head/sys/compat/linux/linux_emul.c Mon Nov 22 12:42:32 2010 (r215675) > > @@ -265,7 +262,8 @@ linux_proc_exec(void *arg __unused, stru > > if (__predict_false(imgp->sysent == &elf_linux_sysvec > > && p->p_sysent != &elf_linux_sysvec)) > > linux_proc_init(FIRST_THREAD_IN_PROC(p), p->p_pid, 0); > > - if (__predict_false(p->p_sysent == &elf_linux_sysvec)) > > + if (__predict_false((p->p_sysent->sv_flags & SV_ABI_MASK) == > > + SV_ABI_LINUX)) > > /* Kill threads regardless of imgp->sysent value */ > > linux_kill_threads(FIRST_THREAD_IN_PROC(p), SIGKILL); > > if (__predict_false(imgp->sysent != &elf_linux_sysvec > There are several similar comparisons around the patched one. > > I am still quite curious for the reason of all __predict() obfuscations > that are countless in the linuxolator. We are not oblidged to emulate > this aspect of Linux. I think we should only have __predict() obfuscation if there is a benchmark showing an indisputable performance gain. Otherwise it is just cruft making the code hard to read. I suspect that a Linux app calling execve() is not a sufficient critical path to warrant any of these. -- John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201011221029.29119.jhb>