From: Keir Fraser Subject: Re: [patch] HVM: Expand the number of PCI interrupts from 16 to 32 Date: Mon, 09 Mar 2009 09:17:30 +0000 Message-ID: References: <20090308234803.GA31771@verge.net.au> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20090308234803.GA31771@verge.net.au> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Simon Horman , Xen-devel Cc: Yuji Shimada List-Id: xen-devel@lists.xenproject.org On 08/03/2009 23:48, "Simon Horman" wrote: > I'm not sure what the reason for this, but the tables relating > to PCI interrupts seem to be truncated to 15 entries. I am > guessing that they are just incomplete and that this has never > been an issue. This patch extends these tables to the full 32 entries. > > Tested by giving an ioemu rtl8139 device pci device 24, > which is unable to communicate with the network without this change. The rombios.c change was optimistic (you changed the bytes-per-device value, rather than the number-of-devices value). The rombios table actually already only describes 6 PCI devices, and expanding it further may risk overflowing the available space in the ROM image. Any OS we pass devices to is going to be ACPI-enabled anyway, so it doesn't matter. The DSDT changes had some bugs -- the LNK[ABCD] values were mostly wrong, and the GSI values for Device 23 Links C&D were wrong. I fixed it all up and checked it in. -- Keir