From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ml01.01.org (Postfix) with ESMTPS id 2C5611A1E02 for ; Thu, 18 Aug 2016 18:19:19 -0700 (PDT) Received: from int-mx14.intmail.prod.int.phx2.redhat.com (int-mx14.intmail.prod.int.phx2.redhat.com [10.5.11.27]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 9DDEC635DB; Fri, 19 Aug 2016 01:19:18 +0000 (UTC) Received: from lacos-laptop-7.usersys.redhat.com (ovpn-116-17.phx2.redhat.com [10.3.116.17]) by int-mx14.intmail.prod.int.phx2.redhat.com (8.14.4/8.14.4) with ESMTP id u7J1JFna022285; Thu, 18 Aug 2016 21:19:16 -0400 To: Jeff Fan , edk2-devel@ml01.01.org References: <1470128388-17960-1-git-send-email-jeff.fan@intel.com> <1470128388-17960-49-git-send-email-jeff.fan@intel.com> Cc: Michael Kinney , Feng Tian From: Laszlo Ersek Message-ID: <4f61b2b4-eeb4-8435-412f-20848347c88e@redhat.com> Date: Fri, 19 Aug 2016 03:19:15 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.2.0 MIME-Version: 1.0 In-Reply-To: <1470128388-17960-49-git-send-email-jeff.fan@intel.com> X-Scanned-By: MIMEDefang 2.68 on 10.5.11.27 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.39]); Fri, 19 Aug 2016 01:19:18 +0000 (UTC) Subject: Re: [Patch v5 48/48] UefiCpuPkg/PiSmmCpuDxeSmm: Add gEfiVariableArchProtocolGuid dependency X-BeenThere: edk2-devel@lists.01.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: EDK II Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 19 Aug 2016 01:19:19 -0000 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit On 08/02/16 10:59, Jeff Fan wrote: > PiSmmCpuDxeSmm driver's entry point will get some PCDs supported dynamic type. > In case those PCDs are set as DynamicHii type in platform DSC File, it implies > that EFI Variable Arch protocol is required. > > This fix is to add gEfiVariableArchProtocolGuid dependency on PiSmmCpuDxeSmm > driver to make sure those DynamicHii PCDs could be read correctly. > > Cc: Michael Kinney > Cc: Feng Tian > Cc: Giri P Mudusuru > Cc: Laszlo Ersek > Contributed-under: TianoCore Contribution Agreement 1.0 > Signed-off-by: Jeff Fan > Reviewed-by: Giri P Mudusuru > --- > UefiCpuPkg/PiSmmCpuDxeSmm/PiSmmCpuDxeSmm.inf | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/UefiCpuPkg/PiSmmCpuDxeSmm/PiSmmCpuDxeSmm.inf b/UefiCpuPkg/PiSmmCpuDxeSmm/PiSmmCpuDxeSmm.inf > index d7e6e07..83e3c55 100644 > --- a/UefiCpuPkg/PiSmmCpuDxeSmm/PiSmmCpuDxeSmm.inf > +++ b/UefiCpuPkg/PiSmmCpuDxeSmm/PiSmmCpuDxeSmm.inf > @@ -4,7 +4,7 @@ > # This SMM driver performs SMM initialization, deploy SMM Entry Vector, > # provides CPU specific services in SMM. > # > -# Copyright (c) 2009 - 2015, Intel Corporation. All rights reserved.
> +# Copyright (c) 2009 - 2016, Intel Corporation. All rights reserved.
> # > # This program and the accompanying materials > # are licensed and made available under the terms and conditions of the BSD License > @@ -155,7 +155,7 @@ [Pcd] > gUefiCpuPkgTokenSpaceGuid.PcdCpuSmmSyncMode ## CONSUMES > > [Depex] > - gEfiMpServiceProtocolGuid > + gEfiMpServiceProtocolGuid AND gEfiVariableArchProtocolGuid > > [UserExtensions.TianoCore."ExtraFiles"] > PiSmmCpuDxeSmmExtra.uni > This patch (commit 7503cd70fb86) breaks OVMF. From the OVMF log: Loading SMM driver at 0x0007FFDA000 EntryPoint=0x0007FFDA271 FvbServicesSmm.efi QEMU Flash: Attempting flash detection at FFE00010 QemuFlashDetected => FD behaves as ROM QemuFlashDetected => No ASSERT OvmfPkg/QemuFlashFvbServicesRuntimeDxe/QemuFlash.c(248): !_gPcd_FixedAtBuild_PcdSmmSmramRequire This symptom means that the SMM build of the QemuFlashFvbServicesRuntimeDxe driver is not actually running in SMM when it tries to access the pflash chip at startup. Therefore QEMU prevents the access, and then OVMF gives up. Here's the bisection log: git bisect start # good: [de74668f5ea713b7e91e01318f0d15d2bf0effce] MdeModulePkg/PeiCore: Fix ConverSinglePpiPointer () typo. git bisect good de74668f5ea713b7e91e01318f0d15d2bf0effce # bad: [7822a1d91d53e80525f571183a24d54488f5437f] NetworkPkg/IpSecDxe: Fix wrong IKE header "FLAG" update git bisect bad 7822a1d91d53e80525f571183a24d54488f5437f # good: [8a2d564b2a811b8dbc85f90e14a67ae4ade21201] UefiCpuPkg/MpInitLib: Sort processor by ascending order of APIC ID git bisect good 8a2d564b2a811b8dbc85f90e14a67ae4ade21201 # good: [7fadaacd50d716e8e054a94c20db56cca98e961e] UefiCpuPkg/CpuDxe: Consume MpInitLib to produce CPU MP Protocol services git bisect good 7fadaacd50d716e8e054a94c20db56cca98e961e # bad: [a012df5ec643a0c08c2b723a02919a5c9373ca74] PcAtChipsetPkg AcpiTimerLib: Wait 363 ACPI timer counts to get TSC Freq git bisect bad a012df5ec643a0c08c2b723a02919a5c9373ca74 # good: [51d4779d7bfb74bfdbb0e196846de78567127348] MdePkg/MpService.h: Fixed typo in function header to match PI spec git bisect good 51d4779d7bfb74bfdbb0e196846de78567127348 # good: [f3b91fa04adea2389c5a6d0fbd9a584d149bae09] UefiCpuPkg/CpuDxe: Fixed typo in function header to match PI spec git bisect good f3b91fa04adea2389c5a6d0fbd9a584d149bae09 # bad: [7503cd70fb864a5663edb121c9b2488b4c69e7f5] UefiCpuPkg/PiSmmCpuDxeSmm: Add gEfiVariableArchProtocolGuid dependency git bisect bad 7503cd70fb864a5663edb121c9b2488b4c69e7f5 # first bad commit: [7503cd70fb864a5663edb121c9b2488b4c69e7f5] UefiCpuPkg/PiSmmCpuDxeSmm: Add gEfiVariableArchProtocolGuid dependency I see that this patch appeared in the final, v5 version of the series, as the last patch in the series. I never got around testing v5. I request that we please revert it for now, and then we investigate the problem. >>From the time we worked on SMM enablement for OVMF, I recall that pulling in the SMM variable driver stack was tricky. The three layers (Firmware Volume Block (2), Fault Tolerant Write, and Variable) have apparent / implicit dependencies between them, but these are not actually expressed in DepExes. I don't know / don't remember why that is the case, but I recall that I had to fiddle with their inclusion order in the FDF files. Following the inclusion order of the preexistent, SMM-less variable driver stack made it all work, if I remember correctly. I don't know why those depexes are not spelled out explicitly in those drivers, but at this point I think that the new DepEx on PiSmmCpuDxeSmm.inf upsets the dispatch order, and things break. I fully agree that this should hopefully be fixed by spelling out all DepExes explicitly everywhere, and I can also imagine it is actually a bug in OVMF somewhere, but for now, until we figure it out, can we please revert the patch? The commit carries Mike's Tested-by -- I wonder what the differences are between Mike's test platform and OVMF. Thank you, Laszlo