From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Subject: Re: [PATCH 1/3] fbdev: Add framebuffer and display update module support for pnx4008 Date: Fri, 30 Jun 2006 22:11:36 -0700 Message-ID: <20060630221136.6ddc1ef3.akpm@osdl.org> References: <44A51257.10804@gmail.com> <20060630204304.7afe6c91.akpm@osdl.org> <44A601DB.5090807@gmail.com> Reply-To: linux-fbdev-devel@lists.sourceforge.net Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from sc8-sf-mx2-b.sourceforge.net ([10.3.1.92] helo=mail.sourceforge.net) by sc8-sf-list1-new.sourceforge.net with esmtp (Exim 4.43) id 1FwXlc-0003Wu-Gh for linux-fbdev-devel@lists.sourceforge.net; Fri, 30 Jun 2006 22:11:40 -0700 Received: from smtp.osdl.org ([65.172.181.4]) by mail.sourceforge.net with esmtps (TLSv1:DES-CBC3-SHA:168) (Exim 4.44) id 1FwXlc-0006PI-4k for linux-fbdev-devel@lists.sourceforge.net; Fri, 30 Jun 2006 22:11:40 -0700 In-Reply-To: <44A601DB.5090807@gmail.com> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: linux-fbdev-devel-bounces@lists.sourceforge.net Errors-To: linux-fbdev-devel-bounces@lists.sourceforge.net To: "Antonino A. Daplas" Cc: linux-fbdev-devel@lists.sourceforge.net, gtolstolytkin@ru.mvista.com On Sat, 01 Jul 2006 13:02:19 +0800 "Antonino A. Daplas" wrote: > Andrew Morton wrote: > > On Fri, 30 Jun 2006 20:00:23 +0800 > > "Antonino A. Daplas" wrote: > > > >> +static struct fb_var_screeninfo rgbfb_var __initdata = { > > > > Is that right? We later do > > > > + info->var = rgbfb_var; > > > > but rgbfb_var is about to be unloaded from memory. I suspect this'll crash > > if the driver is statically linked into the kernel. > > > > Should be okay, as the assignment is done during init only. But we're setting info->var to point at memory which will be freed up by free_initmem(). So if anyone _uses_ info->var after free_initmem() they'll reference random memory or they'll oops. Looking at, say, drivers/video/vfb.c:vfb_probe(), I see info->var = vfb_default; vfb_default is __init and `info' has a lifetime beyond the execution of this function. It looks wrong. Bear in mind that the __init and __initdata segments do not get unloaded if the code is compiled as a module (yet). So the problems only become apparent when the driver is linked into vmlinux. Also, we'll now get section mismatch warnings from the kbuild tools. But they won't help us in this case because vfb_probe() is __init. > But it should be __devinitdata, not __initdata (or vice versa). Yes, the driver seems a bit confused about that also. Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642