Re: [edk2] [PATCH 5/6] MdeModulePkg/VariableRuntimeDxe: factor out boot service accesses
Hi Ard, Some minor feedback added inline. On 2019/1/4 2:28, Ard Biesheuvel wrote: In preparation of providing a standalone MM based variable runtime driver, move the existing SMM driver to the new MM services table, and factor out some pieces that are specific to the traditional driver, mainly related to the use of UEFI boot services, which are not accessible to standalone MM drivers. Contributed-under: TianoCore Contribution Agreement 1.1 Signed-off-by: Ard Biesheuvel --- MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c | 18 +--- MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h | 50 + MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c | 59 -- MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.inf | 5 +- MdeModulePkg/Universal/Variable/RuntimeDxe/VariableTraditionalMm.c | 114 5 files changed, 187 insertions(+), 59 deletions(-) diff --git a/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c b/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c index 28aa2893c6f8..009d96c3a65e 100644 --- a/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c +++ b/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c @@ -21,7 +21,6 @@ WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED. #include #include #include -#include #include "Variable.h" typedef struct { @@ -419,8 +418,6 @@ MorLockInitAtEndOfDxe ( { UINTN MorSize; EFI_STATUS MorStatus; - EFI_STATUS TcgStatus; - VOID *TcgInterface; if (!mMorLockInitializationRequired) { // @@ -458,20 +455,7 @@ MorLockInitAtEndOfDxe ( // can be deduced from the absence of the TCG / TCG2 protocols, as edk2's // MOR implementation depends on (one of) those protocols. // -TcgStatus = gBS->LocateProtocol ( - , - NULL, // Registration - - ); -if (EFI_ERROR (TcgStatus)) { - TcgStatus = gBS->LocateProtocol ( - , - NULL, // Registration - - ); -} - -if (!EFI_ERROR (TcgStatus)) { +if (VariableHaveTcgProtocols ()) { // // The MOR variable originates from the platform firmware; set the MOR // Control Lock variable to report the locking capability to the OS. diff --git a/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h b/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h index 938eb5de61fa..11822575ac4d 100644 --- a/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h +++ b/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h @@ -924,4 +924,54 @@ VariableExLibAtRuntime ( VOID ); +/** + Notify the system that the SMM variable driver is ready +**/ +VOID +VariableNotifySmmReady ( + VOID + ); + +/** + Notify the system that the SMM variable write driver is ready +**/ +VOID +VariableNotifySmmWriteReady ( + VOID + ); + +/** + Variable service MM driver entry point +**/ +EFI_STATUS +EFIAPI +MmVariableServiceInitialize ( + VOID + ); + +/** + This function check if the buffer is valid per processor architecture and not overlap with SMRAM. + + @param Buffer The buffer start address to be checked. + @param Length The buffer length to be checked. + + @retval TRUE This buffer is valid per processor architecture and not overlap with SMRAM. + @retval FALSE This buffer is not valid per processor architecture or overlap with SMRAM. +**/ +BOOLEAN +VariableSmmIsBufferOutsideSmmValid ( + IN EFI_PHYSICAL_ADDRESS Buffer, + IN UINT64Length + ); + +/** + Whether the TCG or TCG2 protocols are installed in the UEFI protocol database. + This information is used by the MorLock code to infer whether an existing + MOR variable is legitimate or not. Add a line for return description? +**/ +BOOLEAN +VariableHaveTcgProtocols ( + VOID + ); + #endif diff --git a/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c b/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c index 8c53f84ff6e8..7245587052df 100644 --- a/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c +++ b/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c @@ -15,6 +15,7 @@ SmmVariableGetStatistics() should also do validation based on its own knowledge. Copyright (c) 2010 - 2016, Intel Corporation. All rights reserved. +Copyright (c) 2018, Linaro, Ltd. All rights reserved. This program and the accompanying materials are licensed and made available under the terms and conditions of the BSD License which accompanies this distribution. The full text of the license may be found at @@ -28,18 +29,15 @@ WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED. #include #include #include -#include +#include
Re: [edk2] [PATCH 5/6] MdeModulePkg/VariableRuntimeDxe: factor out boot service accesses
On 2019/1/10 10:33, Wang, Jian J wrote: Laszlo, Regards, Jian -Original Message- From: Laszlo Ersek [mailto:ler...@redhat.com] Sent: Tuesday, January 08, 2019 11:38 PM To: Ard Biesheuvel ; edk2-devel@lists.01.org Cc: Leif Lindholm ; Kinney, Michael D ; Gao, Liming ; Wang, Jian J ; Wu, Hao A ; Jagadeesh Ujja ; Achin Gupta ; Thomas Panakamattam Abraham ; Sami Mujawar Subject: Re: [PATCH 5/6] MdeModulePkg/VariableRuntimeDxe: factor out boot service accesses On 01/03/19 19:28, Ard Biesheuvel wrote: In preparation of providing a standalone MM based variable runtime driver, move the existing SMM driver to the new MM services table, and factor out some pieces that are specific to the traditional driver, mainly related to the use of UEFI boot services, which are not accessible to standalone MM drivers. Contributed-under: TianoCore Contribution Agreement 1.1 Signed-off-by: Ard Biesheuvel --- MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c | 18 +--- MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h | 50 + MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c | 59 -- MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.inf | 5 +- MdeModulePkg/Universal/Variable/RuntimeDxe/VariableTraditionalMm.c | 114 5 files changed, 187 insertions(+), 59 deletions(-) I *vaguely* feel like we should extract the new functions to "PrivilegePolymorphic.h", rather than to "Variable.h". Please see initial commit 00663d047fc9 ("MdeModulePkg/Variable/RuntimeDxe: move SecureBootHook() decl to new header", 2017-10-10), and other commits that touched that file. I realize this is not a 100% "constructive" suggestion, and I feel appropriately bad about that. It's just that "Variable.h" has so many internals that I feel it's not a good dumping ground for these new functions. And the other header we have, looks closer in purpose. For example, MorLockInitAtEndOfDxe() is already declared in "PrivilegePolymorphic.h" (see commit f1304280435f, "MdeModulePkg/Variable/RuntimeDxe: introduce MorLockInitAtEndOfDxe() hook", 2017-10-10). Admittedly, now that we're going to have three separate builds of this driver, dedicating a separate header file to each "shared between A and B" relationship is getting a bit too complex. In retrospect, introducing "PrivilegePolymorphic.h" may not have been a "scalable" idea, after all, and I should have just dumped those functions all in "Variable.h". IOW, I think - targeting "Variable.h" now is inconsistent with earlier code, - extending "PrivilegePolymorphic.h" is also suboptimal (although still better than the previous option), - adding yet another header might be technically correct, but it would be over-engineering, - asking you to merge "PrivilegePolymorphic.h" back into "Variable.h" feels awkward, especially after I argued *for* "PrivilegePolymorphic.h" at length, when I originally introduced it. :/ Sigh. Can the variable driver maintainers comment please? (I still plan to regression-test this series, but I feel like I should force myself to at least skim the variable driver patches, beyond testing them. Because, next time I can't avoid working with this very complex driver, I wouldn't like to be *completely* lost.) I agree "PrivilegePolymorphic.h" is more appropriate place for them. Maybe Star have different opinion. At current situation, I prefer PrivilegePolymorphic.h. :) Some minor feedback will be added in another reply. Thanks, Star Thanks, Laszlo ___ edk2-devel mailing list edk2-devel@lists.01.org https://lists.01.org/mailman/listinfo/edk2-devel
Re: [edk2] [PATCH 5/6] MdeModulePkg/VariableRuntimeDxe: factor out boot service accesses
Laszlo, Regards, Jian > -Original Message- > From: Laszlo Ersek [mailto:ler...@redhat.com] > Sent: Tuesday, January 08, 2019 11:38 PM > To: Ard Biesheuvel ; edk2-devel@lists.01.org > Cc: Leif Lindholm ; Kinney, Michael D > ; Gao, Liming ; Wang, > Jian J ; Wu, Hao A ; Jagadeesh > Ujja ; Achin Gupta ; > Thomas Panakamattam Abraham ; Sami Mujawar > > Subject: Re: [PATCH 5/6] MdeModulePkg/VariableRuntimeDxe: factor out boot > service accesses > > On 01/03/19 19:28, Ard Biesheuvel wrote: > > In preparation of providing a standalone MM based variable runtime > > driver, move the existing SMM driver to the new MM services table, > > and factor out some pieces that are specific to the traditional > > driver, mainly related to the use of UEFI boot services, which are > > not accessible to standalone MM drivers. > > > > Contributed-under: TianoCore Contribution Agreement 1.1 > > Signed-off-by: Ard Biesheuvel > > --- > > MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c | 18 > +--- > > MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h | 50 > + > > MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c | 59 > -- > > MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.inf | 5 +- > > MdeModulePkg/Universal/Variable/RuntimeDxe/VariableTraditionalMm.c | > 114 > > 5 files changed, 187 insertions(+), 59 deletions(-) > > I *vaguely* feel like we should extract the new functions to > "PrivilegePolymorphic.h", rather than to "Variable.h". > > Please see initial commit 00663d047fc9 > ("MdeModulePkg/Variable/RuntimeDxe: move SecureBootHook() decl to new > header", 2017-10-10), and other commits that touched that file. > > I realize this is not a 100% "constructive" suggestion, and I feel > appropriately bad about that. It's just that "Variable.h" has so many > internals that I feel it's not a good dumping ground for these new > functions. And the other header we have, looks closer in purpose. > > For example, MorLockInitAtEndOfDxe() is already declared in > "PrivilegePolymorphic.h" (see commit f1304280435f, > "MdeModulePkg/Variable/RuntimeDxe: introduce MorLockInitAtEndOfDxe() > hook", 2017-10-10). > > Admittedly, now that we're going to have three separate builds of this > driver, dedicating a separate header file to each "shared between A and > B" relationship is getting a bit too complex. In retrospect, introducing > "PrivilegePolymorphic.h" may not have been a "scalable" idea, after all, > and I should have just dumped those functions all in "Variable.h". > > IOW, I think > - targeting "Variable.h" now is inconsistent with earlier code, > - extending "PrivilegePolymorphic.h" is also suboptimal (although still > better than the previous option), > - adding yet another header might be technically correct, but it would > be over-engineering, > - asking you to merge "PrivilegePolymorphic.h" back into "Variable.h" > feels awkward, especially after I argued *for* "PrivilegePolymorphic.h" > at length, when I originally introduced it. :/ > > Sigh. Can the variable driver maintainers comment please? > > (I still plan to regression-test this series, but I feel like I should > force myself to at least skim the variable driver patches, beyond > testing them. Because, next time I can't avoid working with this very > complex driver, I wouldn't like to be *completely* lost.) > I agree "PrivilegePolymorphic.h" is more appropriate place for them. Maybe Star have different opinion. > Thanks, > Laszlo ___ edk2-devel mailing list edk2-devel@lists.01.org https://lists.01.org/mailman/listinfo/edk2-devel
Re: [edk2] [PATCH 5/6] MdeModulePkg/VariableRuntimeDxe: factor out boot service accesses
On 01/03/19 19:28, Ard Biesheuvel wrote: > In preparation of providing a standalone MM based variable runtime > driver, move the existing SMM driver to the new MM services table, > and factor out some pieces that are specific to the traditional > driver, mainly related to the use of UEFI boot services, which are > not accessible to standalone MM drivers. > > Contributed-under: TianoCore Contribution Agreement 1.1 > Signed-off-by: Ard Biesheuvel > --- > MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c | 18 +--- > MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h | 50 > + > MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c | 59 > -- > MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.inf | 5 +- > MdeModulePkg/Universal/Variable/RuntimeDxe/VariableTraditionalMm.c | 114 > > 5 files changed, 187 insertions(+), 59 deletions(-) I *vaguely* feel like we should extract the new functions to "PrivilegePolymorphic.h", rather than to "Variable.h". Please see initial commit 00663d047fc9 ("MdeModulePkg/Variable/RuntimeDxe: move SecureBootHook() decl to new header", 2017-10-10), and other commits that touched that file. I realize this is not a 100% "constructive" suggestion, and I feel appropriately bad about that. It's just that "Variable.h" has so many internals that I feel it's not a good dumping ground for these new functions. And the other header we have, looks closer in purpose. For example, MorLockInitAtEndOfDxe() is already declared in "PrivilegePolymorphic.h" (see commit f1304280435f, "MdeModulePkg/Variable/RuntimeDxe: introduce MorLockInitAtEndOfDxe() hook", 2017-10-10). Admittedly, now that we're going to have three separate builds of this driver, dedicating a separate header file to each "shared between A and B" relationship is getting a bit too complex. In retrospect, introducing "PrivilegePolymorphic.h" may not have been a "scalable" idea, after all, and I should have just dumped those functions all in "Variable.h". IOW, I think - targeting "Variable.h" now is inconsistent with earlier code, - extending "PrivilegePolymorphic.h" is also suboptimal (although still better than the previous option), - adding yet another header might be technically correct, but it would be over-engineering, - asking you to merge "PrivilegePolymorphic.h" back into "Variable.h" feels awkward, especially after I argued *for* "PrivilegePolymorphic.h" at length, when I originally introduced it. :/ Sigh. Can the variable driver maintainers comment please? (I still plan to regression-test this series, but I feel like I should force myself to at least skim the variable driver patches, beyond testing them. Because, next time I can't avoid working with this very complex driver, I wouldn't like to be *completely* lost.) Thanks, Laszlo ___ edk2-devel mailing list edk2-devel@lists.01.org https://lists.01.org/mailman/listinfo/edk2-devel
[edk2] [PATCH 5/6] MdeModulePkg/VariableRuntimeDxe: factor out boot service accesses
In preparation of providing a standalone MM based variable runtime driver, move the existing SMM driver to the new MM services table, and factor out some pieces that are specific to the traditional driver, mainly related to the use of UEFI boot services, which are not accessible to standalone MM drivers. Contributed-under: TianoCore Contribution Agreement 1.1 Signed-off-by: Ard Biesheuvel --- MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c | 18 +--- MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h | 50 + MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c | 59 -- MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.inf | 5 +- MdeModulePkg/Universal/Variable/RuntimeDxe/VariableTraditionalMm.c | 114 5 files changed, 187 insertions(+), 59 deletions(-) diff --git a/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c b/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c index 28aa2893c6f8..009d96c3a65e 100644 --- a/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c +++ b/MdeModulePkg/Universal/Variable/RuntimeDxe/TcgMorLockSmm.c @@ -21,7 +21,6 @@ WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED. #include #include #include -#include #include "Variable.h" typedef struct { @@ -419,8 +418,6 @@ MorLockInitAtEndOfDxe ( { UINTN MorSize; EFI_STATUS MorStatus; - EFI_STATUS TcgStatus; - VOID *TcgInterface; if (!mMorLockInitializationRequired) { // @@ -458,20 +455,7 @@ MorLockInitAtEndOfDxe ( // can be deduced from the absence of the TCG / TCG2 protocols, as edk2's // MOR implementation depends on (one of) those protocols. // -TcgStatus = gBS->LocateProtocol ( - , - NULL, // Registration - - ); -if (EFI_ERROR (TcgStatus)) { - TcgStatus = gBS->LocateProtocol ( - , - NULL, // Registration - - ); -} - -if (!EFI_ERROR (TcgStatus)) { +if (VariableHaveTcgProtocols ()) { // // The MOR variable originates from the platform firmware; set the MOR // Control Lock variable to report the locking capability to the OS. diff --git a/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h b/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h index 938eb5de61fa..11822575ac4d 100644 --- a/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h +++ b/MdeModulePkg/Universal/Variable/RuntimeDxe/Variable.h @@ -924,4 +924,54 @@ VariableExLibAtRuntime ( VOID ); +/** + Notify the system that the SMM variable driver is ready +**/ +VOID +VariableNotifySmmReady ( + VOID + ); + +/** + Notify the system that the SMM variable write driver is ready +**/ +VOID +VariableNotifySmmWriteReady ( + VOID + ); + +/** + Variable service MM driver entry point +**/ +EFI_STATUS +EFIAPI +MmVariableServiceInitialize ( + VOID + ); + +/** + This function check if the buffer is valid per processor architecture and not overlap with SMRAM. + + @param Buffer The buffer start address to be checked. + @param Length The buffer length to be checked. + + @retval TRUE This buffer is valid per processor architecture and not overlap with SMRAM. + @retval FALSE This buffer is not valid per processor architecture or overlap with SMRAM. +**/ +BOOLEAN +VariableSmmIsBufferOutsideSmmValid ( + IN EFI_PHYSICAL_ADDRESS Buffer, + IN UINT64Length + ); + +/** + Whether the TCG or TCG2 protocols are installed in the UEFI protocol database. + This information is used by the MorLock code to infer whether an existing + MOR variable is legitimate or not. +**/ +BOOLEAN +VariableHaveTcgProtocols ( + VOID + ); + #endif diff --git a/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c b/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c index 8c53f84ff6e8..7245587052df 100644 --- a/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c +++ b/MdeModulePkg/Universal/Variable/RuntimeDxe/VariableSmm.c @@ -15,6 +15,7 @@ SmmVariableGetStatistics() should also do validation based on its own knowledge. Copyright (c) 2010 - 2016, Intel Corporation. All rights reserved. +Copyright (c) 2018, Linaro, Ltd. All rights reserved. This program and the accompanying materials are licensed and made available under the terms and conditions of the BSD License which accompanies this distribution. The full text of the license may be found at @@ -28,18 +29,15 @@ WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED. #include #include #include -#include +#include #include -#include -#include +#include #include #include "Variable.h" extern VARIABLE_INFO_ENTRY *gVariableInfo; -EFI_HANDLE