From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: mx.groups.io; dkim=missing; spf=pass (domain: intel.com, ip: 192.55.52.151, mailfrom: hao.a.wu@intel.com) Received: from mga17.intel.com (mga17.intel.com [192.55.52.151]) by groups.io with SMTP; Tue, 27 Aug 2019 18:40:25 -0700 X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga001.fm.intel.com ([10.253.24.23]) by fmsmga107.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 27 Aug 2019 18:40:24 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.64,439,1559545200"; d="scan'208";a="197438424" Received: from fmsmsx107.amr.corp.intel.com ([10.18.124.205]) by fmsmga001.fm.intel.com with ESMTP; 27 Aug 2019 18:40:24 -0700 Received: from shsmsx102.ccr.corp.intel.com (10.239.4.154) by fmsmsx107.amr.corp.intel.com (10.18.124.205) with Microsoft SMTP Server (TLS) id 14.3.439.0; Tue, 27 Aug 2019 18:40:24 -0700 Received: from shsmsx104.ccr.corp.intel.com ([169.254.5.112]) by shsmsx102.ccr.corp.intel.com ([169.254.2.19]) with mapi id 14.03.0439.000; Wed, 28 Aug 2019 09:40:22 +0800 From: "Wu, Hao A" To: Leif Lindholm , Laszlo Ersek , "Kinney, Michael D" CC: "devel@edk2.groups.io" , Andrew Fish , Baptiste Gerondeau , "Wang, Jian J" , "Feng, Bob C" , "Gao, Liming" Subject: Re: [PATCH 1/1] MdeModulePkg: fix !x86 builds (more) Thread-Topic: [PATCH 1/1] MdeModulePkg: fix !x86 builds (more) Thread-Index: AQHVXRpZNY7rsfvTY0GaOBDiuTsdIKcPxkXg Date: Wed, 28 Aug 2019 01:40:22 +0000 Message-ID: References: <20190827124328.9034-1-leif.lindholm@linaro.org> <6ce1988a-bd79-893e-5d8d-724b98329ab9@redhat.com> <20190827205926.GG29255@bivouac.eciton.net> In-Reply-To: <20190827205926.GG29255@bivouac.eciton.net> Accept-Language: zh-CN, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.239.127.40] MIME-Version: 1.0 Return-Path: hao.a.wu@intel.com Content-Language: en-US Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable > -----Original Message----- > From: Leif Lindholm [mailto:leif.lindholm@linaro.org] > Sent: Wednesday, August 28, 2019 4:59 AM > To: Laszlo Ersek > Cc: devel@edk2.groups.io; Andrew Fish; Kinney, Michael D; Baptiste > Gerondeau; Wang, Jian J; Wu, Hao A; Feng, Bob C; Gao, Liming > Subject: Re: [PATCH 1/1] MdeModulePkg: fix !x86 builds (more) >=20 > +Bob, Liming, >=20 > On Tue, Aug 27, 2019 at 09:26:05PM +0200, Laszlo Ersek wrote: > > Hi Leif, > > > > On 08/27/19 14:43, Leif Lindholm wrote: > > > Commit 4a1f6b85c184 > > > ("MdeModulePkg: add LockBoxNullLib for !IA32/X64 in .dsc") > > > added an ARM/AARCH64 resolution for LockBoxLib. However, this failed > > > to address the overrides provided for PEIM, DXE_DRIVER, > > > DXE_RUNTIME_DRIVER, DXE_SMM_DRIVER and UEFI_DRIVER, so any > modules > > > of those classes still failed to build. > > > > > > Break these out properly into their own LibraryClasses sections. > > > > > > Resolves BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=3D2134 > > > > > > Signed-off-by: Leif Lindholm > > > Reported-by: Baptiste Gerondeau > > > Cc: Jian J Wang > > > Cc: Hao A Wu > > > --- > > > > > > I don't understand how the above would appear to work back when I > > > submitted the previous patch but not work now, but I haven't dug > > > into it deeper. Including the x86-specific LockBoxLib in the > > > .common section is however clearly not correct. > > > > I agree with you that the present situation is not correct. > > > > According to: > > > > https://edk2-docs.gitbooks.io/edk-ii-dsc- > specification/2_dsc_overview/26_[libraryclasses]_section_processing.html > > > > the library class resolutions take effect in the following order > > (entries near the top have higher priority): > > > > > 1. associated with the INF file in the [Components] > section > > > 2. [LibraryClasses.$(Arch).$(MODULE_TYPE), > LibraryClasses.$(Arch).$(MODULE_TYPE)] > > > 3. [LibraryClasses.$(Arch).$(MODULE_TYPE)] > > > 4. [LibraryClasses.common.$(MODULE_TYPE)] > > > 5. [LibraryClasses.$(Arch)] > > > 6. [LibraryClasses.common] or [LibraryClasses] > > > > (Side comment 1: levels #2 and #3 look very similar; I think the > > difference is that #2 is supposed to be a multi-arch and/or > > multi-module-type section, while #3 is a single-arch and > > single-module-type section.) > > > > Commit 4a1f6b85c184 ("MdeModulePkg: add LockBoxNullLib for !IA32/X64 > in > > .dsc", 2019-03-27) provided a LockBoxLib resolution at level #5: >=20 > Yes. >=20 > > > [LibraryClasses.ARM, LibraryClasses.AARCH64] > > > > However, the other LockBoxLib resolutions are at level #4: > > > > > [LibraryClasses.common.PEIM] > > > [LibraryClasses.common.DXE_DRIVER] > > > [LibraryClasses.common.DXE_RUNTIME_DRIVER] > > > [LibraryClasses.common.DXE_SMM_DRIVER] > > > [LibraryClasses.common.UEFI_DRIVER] > > > > So the latter taking priority is actually specified behavior. >=20 > Hmm. That's not great. > Anyway, I stopped being lazy and did a bisect. >=20 > The culprit is > e8449e1d8e3b ("BaseTools: Decouple AutoGen Objects"), marked as > resolving https://bugzilla.tianocore.org/show_bug.cgi?id=3D1875. >=20 > This also affects SignedCapsulePkg/SignedCapsulePkg.dsc (although once > addressed, AARCH64 also needs a NULL entry added for > CompilerIntrinsicsLib. >=20 > > (Side comment 2: EBC is in the same boat, from commit cbcccd2c9d93 > > ("Update Code to pass EBC compiler", 2013-05-13): > > > > > [LibraryClasses.EBC] > > > LockBoxLib|MdeModulePkg/Library/LockBoxNullLib/LockBoxNullLib.inf > > ) > > > > As to why this breakage was not exposed right at commit 4a1f6b85c184 -- > > I have no idea. Perhaps it was hidden by a BaseTools issue that has bee= n > > fixed meanwhile. >=20 > Yes. > But it is also a fundamental change in tool behaviour introduced on 9 > August. I am really uncomfortable about this making it into the > release this week - but I also believe this is the foundation for the > multiprocess autogen. >=20 > Since you have very helpfully analyzed *what* changed ... would the > better "fix" for 2019.08 be to intentionally break the new code to > conform to the old behaviour - and then revert that patch after the > tag? >=20 > If we do that, this patch could then wait and indeed be merged as part > of the same set. >=20 > > On 08/27/19 14:43, Leif Lindholm wrote: > > > I think a fix for this issue needs to go into 2019.08, > > > > I agree the problem should be fixed in 2019.08 -- taking your word that > > commit 4a1f6b85c184 *appeared* to fix the MdeModulePkg.dsc build for > > ARM/AARCH64, we now have a regression since that commit (dated > > 2019-03-27). > > > > > but if someone has a prettier suggestion, I am not wedded to this one= . > > > > I think this is good enough. The lib class resolutions are raised to > > level #2, but they will no longer match ARM / AARCH64, so your level#5 > > addition from commit 4a1f6b85c184 will take effect. > > > > > > > > MdeModulePkg/MdeModulePkg.dsc | 16 +++++++++++++--- > > > 1 file changed, 13 insertions(+), 3 deletions(-) > > > > > > diff --git a/MdeModulePkg/MdeModulePkg.dsc > b/MdeModulePkg/MdeModulePkg.dsc > > > index 4320839abfb5..15ba96cecbed 100644 > > > --- a/MdeModulePkg/MdeModulePkg.dsc > > > +++ b/MdeModulePkg/MdeModulePkg.dsc > > > @@ -109,6 +109,8 @@ [LibraryClasses.common.PEIM] > > > HobLib|MdePkg/Library/PeiHobLib/PeiHobLib.inf > > > > MemoryAllocationLib|MdePkg/Library/PeiMemoryAllocationLib/PeiMemory > AllocationLib.inf > > > > ExtractGuidedSectionLib|MdePkg/Library/PeiExtractGuidedSectionLib/PeiEx > tractGuidedSectionLib.inf > > > + > > > +[LibraryClasses.IA32.PEIM,LibraryClasses.X64.PEIM] > > > > LockBoxLib|MdeModulePkg/Library/SmmLockBoxLib/SmmLockBoxPeiLib.inf > > > > (1) I suggest replacing "," with ", ". (That's more consistent with > > preexistent section headers in the DSC file.) Applies to the other new > > section headers too. >=20 > Yes, fair point. >=20 > > > > > > [LibraryClasses.common.DXE_CORE] > > > @@ -118,18 +120,22 @@ [LibraryClasses.common.DXE_CORE] > > > > > > [LibraryClasses.common.DXE_DRIVER] > > > HobLib|MdePkg/Library/DxeHobLib/DxeHobLib.inf > > > - > LockBoxLib|MdeModulePkg/Library/SmmLockBoxLib/SmmLockBoxDxeLib.in > f > > > > MemoryAllocationLib|MdePkg/Library/UefiMemoryAllocationLib/UefiMemo > ryAllocationLib.inf > > > > ExtractGuidedSectionLib|MdePkg/Library/DxeExtractGuidedSectionLib/DxeE > xtractGuidedSectionLib.inf > > > > CapsuleLib|MdeModulePkg/Library/DxeCapsuleLibFmp/DxeCapsuleLib.inf > > > > > > +[LibraryClasses.IA32.DXE_DRIVER,LibraryClasses.X64.DXE_DRIVER] > > > + > LockBoxLib|MdeModulePkg/Library/SmmLockBoxLib/SmmLockBoxDxeLib.in > f > > > + > > > [LibraryClasses.common.DXE_RUNTIME_DRIVER] > > > HobLib|MdePkg/Library/DxeHobLib/DxeHobLib.inf > > > > MemoryAllocationLib|MdePkg/Library/UefiMemoryAllocationLib/UefiMemo > ryAllocationLib.inf > > > > DebugLib|MdePkg/Library/UefiDebugLibConOut/UefiDebugLibConOut.inf > > > - > LockBoxLib|MdeModulePkg/Library/SmmLockBoxLib/SmmLockBoxDxeLib.in > f > > > > CapsuleLib|MdeModulePkg/Library/DxeCapsuleLibFmp/DxeRuntimeCapsule > Lib.inf > > > > > > > +[LibraryClasses.IA32.DXE_RUNTIME_DRIVER,LibraryClasses.X64.DXE_RUNTI > ME_DRIVER] > > > + > LockBoxLib|MdeModulePkg/Library/SmmLockBoxLib/SmmLockBoxDxeLib.in > f > > > + > > > [LibraryClasses.common.SMM_CORE] > > > HobLib|MdePkg/Library/DxeHobLib/DxeHobLib.inf > > > > MemoryAllocationLib|MdeModulePkg/Library/PiSmmCoreMemoryAllocatio > nLib/PiSmmCoreMemoryAllocationLib.inf > > > @@ -143,13 +149,17 @@ [LibraryClasses.common.DXE_SMM_DRIVER] > > > > MemoryAllocationLib|MdePkg/Library/SmmMemoryAllocationLib/SmmMe > moryAllocationLib.inf > > > > MmServicesTableLib|MdePkg/Library/MmServicesTableLib/MmServicesTabl > eLib.inf > > > > SmmServicesTableLib|MdePkg/Library/SmmServicesTableLib/SmmServicesT > ableLib.inf > > > + SmmMemLib|MdePkg/Library/SmmMemLib/SmmMemLib.inf > > > + > > > > +[LibraryClasses.IA32.DXE_SMM_DRIVER,LibraryClasses.X64.DXE_SMM_DRI > VER] > > > > LockBoxLib|MdeModulePkg/Library/SmmLockBoxLib/SmmLockBoxSmmLib.i > nf > > > - SmmMemLib|MdePkg/Library/SmmMemLib/SmmMemLib.inf > > > > > > > I wonder if this is really necessary. I would assume all the > > DXE_SMM_DRIVER modules are listed under > > > > [Components.IA32, Components.X64] > > > > already. But, this hunk certainly doesn't hurt. >=20 > Well, this fixes the current issue. I completely agree the file could > benefit from some overall restructuring. >=20 > > > [LibraryClasses.common.UEFI_DRIVER] > > > HobLib|MdePkg/Library/DxeHobLib/DxeHobLib.inf > > > > MemoryAllocationLib|MdePkg/Library/UefiMemoryAllocationLib/UefiMemo > ryAllocationLib.inf > > > > DebugLib|MdePkg/Library/UefiDebugLibConOut/UefiDebugLibConOut.inf > > > + > > > +[LibraryClasses.IA32.UEFI_DRIVER,LibraryClasses.X64.UEFI_DRIVER] > > > > LockBoxLib|MdeModulePkg/Library/SmmLockBoxLib/SmmLockBoxDxeLib.in > f > > > > > > [LibraryClasses.common.UEFI_APPLICATION] > > > > > > > With (1) fixed (feel free to correct that just before pushing): > > > > Reviewed-by: Laszlo Ersek >=20 > Thanks! >=20 > > Do wait for maintainer review, of course. >=20 > Of course. Per my understanding to the analysis from Leif, Laszlo and Mike, the patch will depend on the precedence of the below rules: * [LibraryClasses.common.$(MODULE_TYPE)] * [LibraryClasses.$(Arch)] So for now, we should wait for the final call on the above open, right? Best Regards, Hao Wu >=20 > Best Regards, >=20 > Leif