Junhao

FYI there is a regression in master f= or OVFM right now: https://bugzilla.tianocore.org/show_bug.cgi?id=3D= 2266 so you might be hitting that.

Thanks,

Andrew Fish

On Oct 11, 2019, at 11:13 PM, Junhao Gao &l= t;junhao.gao@intel.com> wrote:

=

-----Original Message-----
From: David= Woodhouse <dwmw2@infr= adead.org>
Sent: Friday, October 11, 2019 10:36 PM
To: Laszlo Ersek <lersek@redhat.com>; Gao, Junhao <junhao.gao@intel.com>;
devel@edk2.groups.io
Subject: Re: [edk2-devel] question about qemu+kvm+ovmf+winxp

On Fri, 2019-10-11 at 16:27 +0200, Laszlo Ersek wr= ote:
On 10/11/19 09:44, = Gao, Junhao wrote:
Hi Da= vid

      I have= found this compiled OVMF-with-csm.fd can support winxp booting
up.
     = ; OVMF-with-csm.fd path:
https://www.kraxel.org/repos/jenkins/seabios/seabios.git-csm-1.= 12.0-
33.63.g43f5df7.x86_64.rpm
  =     Then could you help to provide me the compile metho= d and base code to
reproduce this O= VMF-with-csm.fd?

Ah, good point, I should have remembered that Gerd offers a CSM bui= ld in
his repo!

So, you can easi= ly look up the build instructions, in Gerd's SeaBIOS and
OVMF= RPM "spec" files. Check out these git repositories:

https://git.kraxel.org/cgit/jenkins/seabios/
&nb= sp;https://git.kraxel.org/cgit/jenkins/edk2/

S= pecifically:

https://git.kraxel.org/cgit= /jenkins/seabios/tree/seabios.git.spec
https://git.krax= el.org/cgit/jenkins/edk2/tree/edk2.git.spec.template

Right, thanks.

Ultimately = it's as simple as:

=E2=80=A2 Build SeaBIOS wit= h CONFIG_CSM
=E2=80=A2 Drop the resulting Csm16.bin into Ovmf= Pkg/Csm/Csm16/ in the EDK2 tree
=E2=80=A2 Build EDK2 with -DC= SM_ENABLE

Thanks David, follow these steps, I can reproduce th= e OVMF-with-csm.fd.
How= ever, I find that when I choose remotes/origin/master branch, after compili= ng, qemu+kvm+ovmf+winxp can't
boot up, but when= I choose remotes/origin/UDK2018, the compiled OVMF.fd can work.
So the latest edk2 code have remo= ved CSM support? Could I apply the CSM patch to
the latest edk2 code to support my work because our project is based on th= e latest EDK2 code, thanks.
Someone else maile= d me recently to say they'd copied the Csm packages
over from= OvmfPkg to Quark and had Windows booting on Galileo boards.
=
Hm, I wonder if we should move the CSM support out of OVMF a= nd into
somewhere more generic, since it really is generic an= d not platform-
specific?

I can't make some decisions, from my = view, maybe CSM module can be
an independent pa= tch for user to apply or reject :)

--Apple-Mail=_F3E83584-0F81-492F-A547-24DB4A6DA9EF-- From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) by mx.groups.io with SMTP id smtpd.web11.2970.1571057984897421762 for ; Mon, 14 Oct 2019 05:59:45 -0700 Authentication-Results: mx.groups.io; dkim=missing; spf=pass (domain: redhat.com, ip: 209.132.183.28, mailfrom: lersek@redhat.com) Received: from smtp.corp.redhat.com (int-mx05.intmail.prod.int.phx2.redhat.com [10.5.11.15]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 7AAB13090FC9; Mon, 14 Oct 2019 12:59:44 +0000 (UTC) Received: from lacos-laptop-7.usersys.redhat.com (ovpn-117-254.ams2.redhat.com [10.36.117.254]) by smtp.corp.redhat.com (Postfix) with ESMTP id 8A3A95D6A7; Mon, 14 Oct 2019 12:59:43 +0000 (UTC) Subject: Re: [edk2-devel] question about qemu+kvm+ovmf+winxp To: David Woodhouse , "Gao, Junhao" , "devel@edk2.groups.io" References: <50E163A92F208D4B8CEB452055E613EB0AFE85F7@SHSMSX101.ccr.corp.intel.com> <15CC73B09CF3FDD9.17220@groups.io> <50E163A92F208D4B8CEB452055E613EB0AFE8C2B@SHSMSX101.ccr.corp.intel.com> <0d2c8aa3-0147-d297-5d03-21562e438fdf@redhat.com> <26980775fc87acc8c4d180626d853607e9f20b0e.camel@infradead.org> From: "Laszlo Ersek" Message-ID: Date: Mon, 14 Oct 2019 14:59:42 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1 MIME-Version: 1.0 In-Reply-To: <26980775fc87acc8c4d180626d853607e9f20b0e.camel@infradead.org> X-Scanned-By: MIMEDefang 2.79 on 10.5.11.15 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.43]); Mon, 14 Oct 2019 12:59:44 +0000 (UTC) Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: quoted-printable On 10/11/19 16:35, David Woodhouse wrote: > On Fri, 2019-10-11 at 16:27 +0200, Laszlo Ersek wrote: >> On 10/11/19 09:44, Gao, Junhao wrote: >>> Hi David=20 >>> >>> I have found this compiled OVMF-with-csm.fd can support winxp = booting up. >>> OVMF-with-csm.fd path: https://www.kraxel.org/repos/jenkins/se= abios/seabios.git-csm-1.12.0-33.63.g43f5df7.x86_64.rpm >>> Then could you help to provide me the compile method and base = code to reproduce this OVMF-with-csm.fd? >> >> Ah, good point, I should have remembered that Gerd offers a CSM build = in >> his repo! >> >> So, you can easily look up the build instructions, in Gerd's SeaBIOS a= nd >> OVMF RPM "spec" files. Check out these git repositories: >> >> https://git.kraxel.org/cgit/jenkins/seabios/ >> https://git.kraxel.org/cgit/jenkins/edk2/ >> >> Specifically: >> >> https://git.kraxel.org/cgit/jenkins/seabios/tree/seabios.git.spec >> https://git.kraxel.org/cgit/jenkins/edk2/tree/edk2.git.spec.template >=20 > Right, thanks. >=20 > Ultimately it's as simple as: >=20 > =E2=80=A2 Build SeaBIOS with CONFIG_CSM > =E2=80=A2 Drop the resulting Csm16.bin into OvmfPkg/Csm/Csm16/ in the = EDK2 tree > =E2=80=A2 Build EDK2 with -DCSM_ENABLE >=20 > Someone else mailed me recently to say they'd copied the Csm packages > over from OvmfPkg to Quark and had Windows booting on Galileo boards. =20 >=20 > Hm, I wonder if we should move the CSM support out of OVMF and into > somewhere more generic, since it really is generic and not platform- > specific? We moved the CSM infrastructure under OvmfPkg because IntelFrameworkPkg / IntelFrameworkModulePkg were slated for removal, and this was how we could salvage the CSM: https://bugzilla.tianocore.org/show_bug.cgi?id=3D1811 We could likely move it out of OVMF, but if the destination is still the edk2 tree -- which I would strongly prefer, as long as we support the CSM at all --, then it will take a new top level directory (CsmPkg, for example). No other package maintainer was willing to maintain the CSM, and even under OvmfPkg, I agreed to it only if you'd assume its reviewership. So now you'd likely have to introduce CsmPkg, and become its sole (or primary) maintainer. I'm 100% OK with that. (Both as OvmfPkg co-maintainer, and as one of the stewards that might have to ACK (I think?...) the creation of another top-level directory). Thanks Laszlo