[edk2-devel] [PATCH] OvmfPkg/Bhyve: Update Bhyve following changes to OVMF
Laszlo Ersek
lersek at redhat.com
Wed Nov 11 19:57:31 UTC 2020
+Tom
On 11/11/20 04:10, Rebecca Cran wrote:
> Fix BhyveX64.dsc and BhyveX64.fdf to follow breaking
> changes in OVMF.
>
> Signed-off-by: Rebecca Cran <rebecca at bsdio.com>
> ---
> OvmfPkg/Bhyve/BhyveX64.dsc | 1 +
> OvmfPkg/Bhyve/BhyveX64.fdf | 6 ++++++
> 2 files changed, 7 insertions(+)
Ouch, I'm sorry.
>
> diff --git a/OvmfPkg/Bhyve/BhyveX64.dsc b/OvmfPkg/Bhyve/BhyveX64.dsc
> index 16d2233d77..868338b460 100644
> --- a/OvmfPkg/Bhyve/BhyveX64.dsc
> +++ b/OvmfPkg/Bhyve/BhyveX64.dsc
> @@ -225,6 +225,7 @@
>
> [LibraryClasses.common]
> BaseCryptLib|CryptoPkg/Library/BaseCryptLib/BaseCryptLib.inf
> + VmgExitLib|UefiCpuPkg/Library/VmgExitLibNull/VmgExitLibNull.inf
>
> [LibraryClasses.common.SEC]
> !ifdef $(DEBUG_ON_SERIAL_PORT)
Yep, makes sense.
> diff --git a/OvmfPkg/Bhyve/BhyveX64.fdf b/OvmfPkg/Bhyve/BhyveX64.fdf
> index 5d2586ae14..8776aaf7ac 100644
> --- a/OvmfPkg/Bhyve/BhyveX64.fdf
> +++ b/OvmfPkg/Bhyve/BhyveX64.fdf
> @@ -76,6 +76,12 @@ gUefiOvmfPkgTokenSpaceGuid.PcdOvmfLockBoxStorageBase|gUefiOvmfPkgTokenSpaceGuid.
> 0x007000|0x001000
> gEfiMdePkgTokenSpaceGuid.PcdGuidedExtractHandlerTableAddress|gUefiOvmfPkgTokenSpaceGuid.PcdGuidedExtractHandlerTableSize
>
> +0x008000|0x001000
> +gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecGhcbPageTableBase|gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecGhcbPageTableSize
> +
> +0x009000|0x002000
> +gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecGhcbBase|gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecGhcbSize
> +
> 0x010000|0x010000
> gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecPeiTempRamBase|gUefiOvmfPkgTokenSpaceGuid.PcdOvmfSecPeiTempRamSize
>
>
Hmm, this, on the other hand, makes me wonder. All four PCDs are
[PcdsFixedAtBuild], from "OvmfPkg.dec", so the platform DSC/FDF files
*should not* be required to override defaults.
....
Ah, wait, you're hitting the exact PCD value checks (%error directives)
in "OvmfPkg/ResetVector/ResetVector.nasmb".
During the SEV-ES review, I completely lost track of Bhyve consuming
"OvmfPkg/ResetVector/ResetVector.inf". Sorry about that.
So the following list of commits:
(1) 6995a1b79bab OvmfPkg: Create a GHCB page for use during Sec phase
(2) 8a2732186a53 OvmfPkg/ResetVector: Add support for a 32-bit SEV check
(3) 30937f2f98c4 OvmfPkg: Use the SEV-ES work area for the SEV-ES AP
reset vector
causes a problem for the Bhyve platform.
I don't like the "OvmfPkg/Bhyve/BhyveX64.fdf" hack as presented above,
because, while it makes the symptom go away, it causes "BhyveX64.fdf"
appear as if it had anything to do with SEV-ES -- which it doesn't.
Here's what I suggest:
* patch#1:
Subject:
OvmfPkg/Bhyve: detach ResetVector from before the SEV-ES changes
Commit message:
Commits 6995a1b79bab, 8a2732186a53 and 30937f2f98c4 modified all four
regular files under "OvmfPkg/ResetVector" with SEV-ES dependencies.
These are not relevant for Bhyve. Detach the pre-SEV-ES version of
ResetVector for Bhyve.
Composing the patch:
$ git checkout -b bhyve_reset_vector master
$ rm -r OvmfPkg/ResetVector/
$ git checkout 6995a1b79bab^ -- OvmfPkg/ResetVector/
$ mv OvmfPkg/ResetVector/ OvmfPkg/Bhyve/
$ git checkout master -- OvmfPkg/ResetVector/
# add your (C) notices to all files under OvmfPkg/Bhyve/ResetVector/
# namely "PageTables64.asm", "ResetVector.inf", "ResetVector.nasmb"
# do *not* re-generate the FILE_GUID in the new INF file (this is a
# well-known GUID, namely "gEfiFirmwareVolumeTopFileGuid")
$ git add OvmfPkg/Bhyve/ResetVector/
$ git commit
* patch#2:
Subject:
OvmfPkg/Bhyve: fix build breakage after SEV-ES changes
Commit message:
Consume the SEV-ES-independent reset vector restored in the previous
patch. Use the Null instance of VmgExitLib.
Body:
> diff --git a/OvmfPkg/Bhyve/BhyveX64.dsc b/OvmfPkg/Bhyve/BhyveX64.dsc
> index 16d2233d7788..ba79ceef5563 100644
> --- a/OvmfPkg/Bhyve/BhyveX64.dsc
> +++ b/OvmfPkg/Bhyve/BhyveX64.dsc
> @@ -225,6 +225,7 @@ [LibraryClasses]
>
> [LibraryClasses.common]
> BaseCryptLib|CryptoPkg/Library/BaseCryptLib/BaseCryptLib.inf
> + VmgExitLib|UefiCpuPkg/Library/VmgExitLibNull/VmgExitLibNull.inf
>
> [LibraryClasses.common.SEC]
> !ifdef $(DEBUG_ON_SERIAL_PORT)
> @@ -571,7 +572,7 @@ [PcdsDynamicHii]
> #
> ################################################################################
> [Components]
> - OvmfPkg/ResetVector/ResetVector.inf
> + OvmfPkg/Bhyve/ResetVector/ResetVector.inf
>
> #
> # SEC Phase modules
> diff --git a/OvmfPkg/Bhyve/BhyveX64.fdf b/OvmfPkg/Bhyve/BhyveX64.fdf
> index 5d2586ae141a..f4050c4934b7 100644
> --- a/OvmfPkg/Bhyve/BhyveX64.fdf
> +++ b/OvmfPkg/Bhyve/BhyveX64.fdf
> @@ -117,7 +117,7 @@ [FV.SECFV]
> #
> INF OvmfPkg/Sec/SecMain.inf
>
> -INF RuleOverride=RESET_VECTOR OvmfPkg/ResetVector/ResetVector.inf
> +INF RuleOverride=RESET_VECTOR OvmfPkg/Bhyve/ResetVector/ResetVector.inf
>
> ################################################################################
> [FV.PEIFV]
Optimally, these changes should have been part of the SEV-ES feature
series, but we didn't realize. Sorry about the regression!
Thanks
Laszlo
-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#67301): https://edk2.groups.io/g/devel/message/67301
Mute This Topic: https://groups.io/mt/78175658/1813853
Group Owner: devel+owner at edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [edk2-devel-archive at redhat.com]
-=-=-=-=-=-=-=-=-=-=-=-
More information about the edk2-devel-archive
mailing list