From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Vrabel Subject: Re: [Xen-devel] [PATCH] auto balloon initial domain and fix dom0_mem=X inconsistencies (v5). Date: Tue, 8 May 2012 19:12:18 +0100 Message-ID: <4FA96202.9010503@citrix.com> References: <1334596539-18172-1-git-send-email-konrad.wilk@oracle.com> <20120501163707.GA8741@phenom.dumpdata.com> <4FA27084.4030005@citrix.com> <4FA2A11E.1060907@cantab.net> <4FA2B1EF.8080900@citrix.com> <20120507184808.GA7249@phenom.dumpdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20120507184808.GA7249@phenom.dumpdata.com> Sender: linux-kernel-owner@vger.kernel.org To: Konrad Rzeszutek Wilk Cc: David Vrabel , "xen-devel@lists.xensource.com" , "linux-kernel@vger.kernel.org" , "JBeulich@suse.com" List-Id: xen-devel@lists.xenproject.org On 07/05/12 19:48, Konrad Rzeszutek Wilk wrote: > On Thu, May 03, 2012 at 05:27:27PM +0100, David Vrabel wrote: >> On 03/05/12 16:15, David Vrabel wrote: >>> >>> xen: update VA mapping when releasing memory during setup >>> >>> In xen_memory_setup(), if a page that is being released has a VA >>> mapping this must also be updated. Otherwise, the page will be not >>> released completely -- it will still be referenced in Xen and won't be >>> freed util the mapping is removed and this prevents it from being >>> reallocated at a different PFN. >>> >>> This was already being done for the ISA memory region in >>> xen_ident_map_ISA() but on many systems this was omitting a few pages >>> as many systems marked a few pages below the ISA memory region as >>> reserved in the e820 map. >>> >>> Signed-off-by: David Vrabel >>> --- >> [...] >>> --- a/arch/x86/xen/mmu.c >>> +++ b/arch/x86/xen/mmu.c >>> @@ -1929,29 +1929,6 @@ static void xen_set_fixmap(unsigned idx, phys_addr_t phys, pgprot_t prot) >>> #endif >>> } >>> >>> -void __init xen_ident_map_ISA(void) >>> -{ >>> - unsigned long pa; >>> - >>> - /* >>> - * If we're dom0, then linear map the ISA machine addresses into >>> - * the kernel's address space. >>> - */ >>> - if (!xen_initial_domain()) >>> - return; >> >> It might look like this test has gone, however the new code which >> updates the VA mapping uses the e820 map and for a domU its map will not >> have a ISA region so there's no mapping to be updated. > > What if you use e820_hole=1 and the pci=xx in the guest? Are these xl configuration options? I'm not familiar with xl. The PCI memory hole should be above 3 GiB so this hole will be will above the memory that will be initially mapped at boot. I've not managed to persuade my test box to passthrough a PCI device to a guest (using xapi as the toolstack) to confirm, though. I'll have another go tomorrow. David