From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Message-ID: <18169.1165.5981.330400@cargo.ozlabs.ibm.com> Date: Tue, 25 Sep 2007 22:52:29 +1000 From: Paul Mackerras To: Kumar Gala Subject: Re: Please pull powerpc.git merge branch In-Reply-To: <24149ACB-D786-4415-99A8-117FAC9936E6@kernel.crashing.org> References: <18168.62378.472571.655865@cargo.ozlabs.ibm.com> <24149ACB-D786-4415-99A8-117FAC9936E6@kernel.crashing.org> Cc: PowerPC dev list , Greg KH , Linus Torvalds , Roland McGrath List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Kumar Gala writes: > Should this also go to -stable? It turns out that it's not triggerable (as an oops) from userspace in 2.6.22 and earlier. Commit 1b6610d6 of Ben H's took out #ifdef CONFIG_PPC32 around a couple of CHECK_FULL_REGS calls, which is what made it triggerable on ppc64. We have an interesting difference between ppc32 and ppc64 in our CHECK_FULL_REGS implementation - it's a BUG_ON on 64-bit but just a printk on 32-bit. So on 32-bit someone could use it to spam the logs but not to actually crash the system. So yes, I guess it should go to -stable once Linus takes it, but it's a much less serious vulnerability in 2.6.22 and earlier than I first thought (since it's only 32-bit, and just a user-triggerable printk). And yes we should consolidate the CHECK_FULL_REGS implementations as a WARN_ON with a counter to limit how many we do. Paul.