From mboxrd@z Thu Jan 1 00:00:00 1970 From: Marcelo Tosatti Subject: Re: PATCH: dont call exit() from pci_nic_init(), let caller handle Date: Thu, 20 Mar 2008 09:01:13 -0300 Message-ID: <20080320120113.GA9084@dmt> References: <20080320001951.GI14060@us.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Cc: kvm-devel@lists.sourceforge.net, qemu-devel@nongnu.org To: Ryan Harper Return-path: Content-Disposition: inline In-Reply-To: <20080320001951.GI14060@us.ibm.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: kvm-devel-bounces@lists.sourceforge.net Errors-To: kvm-devel-bounces@lists.sourceforge.net List-Id: kvm.vger.kernel.org On Wed, Mar 19, 2008 at 07:19:51PM -0500, Ryan Harper wrote: > While exploring the PCI hotplug code recently posted, I encountered a > situation where I don't believe the current behavior is ideal. With > hotplug, we can add additional pci-based nic devices like e1000 and > rtl8139 from the qemu monitor. If one mistakenly specifies model=ne2000 > (the ISA version), qemu just exits. If a command is run from the > monitor and specifies bogus values, I don't believe the right behavior > is to exit out of the guest entirely. The attached patch (which doesn't > apply directly against qemu-cvs since hotplug hasn't been merged) > changes pci_nic_init() to return NULL on error instead of exiting > and then I've replaced all callers to check the return value and exit(), > preserving the existing behavior, but allowing flexibility so > hotplug can do the right thing and just report the error rather than > exiting the guest. Hi Ryan, Looks good, thanks. There might still be some exit()'s lurking around due to device/cpu hot/add failure. ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/