From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chris Torek Date: Thu, 30 Apr 2009 00:06:57 +0000 Subject: Re: interrupt problem with multiple vnet devices Message-Id: <200904300006.n3U06vn17471@elf.torek.net> List-Id: References: <200904292141.n3TLfmn16267@elf.torek.net> In-Reply-To: <200904292141.n3TLfmn16267@elf.torek.net> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: sparclinux@vger.kernel.org >This is the commit that test comes from. I assume you guys >would check the commit logs for something like this, but just >in case... [snippage - will just retain commit ID] >commit 5a606b72a4309a656cd1a19ad137dc5557c4b8ea Yes, this one is in. I did (and still do) suspect a hypervisor bug. I think in this case the problem is that this test is not sufficent for the "two vnets on one vswitch" case, though, because this results in the desc->status not having IRQ_INPROGRESS set on the "other" vnet that gets a double interrupt. I'm also wondering now if it has something to do with the code we added that attempts to redistribute interrupts, so that in the case of a double interrupt, the second one goes to a different CPU than the CPU already handling the first one. Although we should still see IRQ_INPROGRESS then... Chris