1. discuss@edk2.groups.io
  2. Messages
   Date   
Re: Issues with PatchCheck script

Sandeep Dhanvada
 

Hi Laszlo,

I am able to execute PatchCheck.py on a patch generated from git format-patch, but, could not execute this script with git commits i.e., using -1 or -2 as arguments to read git commits.

PatchCheck didn't show any Warnings or Errors on the patch file. so, i executed git send-email with my patches to devel@edk2.groups.io, but, could not see my patch set in edk2 devel mailing list, https://edk2.groups.io/g/devel/topics. do i need to do anything else to send the patch to devel mailing list ?

Thanks,
Sandeep
Build problem in UEFI Self-Certification Test 2.7B

chinghsueh.cheng@...
 

H
I need someone help that about building UEFI Self-Certification Test 2.7B problem. From website https://www.uefi.org/testtools only provide InstallX64.efi with UEFI Self-Certification Test 2.6 A. Now I need InstallX64.efi with UEFI Self-Certification Test 2.7B but no such file.
So I manual build 2.7B package but always meet the build error. Have any idea?


C:\SCTWorkspace>build -p SctPkg\UEFI\UEFI_SCT.dsc -a X64 -t VS2015x86
Build environment: Windows-10-10.0.17134
Build start time: 01:29:29, Nov.07 2020

WORKSPACE = c:\sctworkspace
ECP_SOURCE = c:\sctworkspace\edkcompatibilitypkg
EDK_SOURCE = c:\sctworkspace\edkcompatibilitypkg
EFI_SOURCE = c:\sctworkspace\edkcompatibilitypkg
EDK_TOOLS_PATH = c:\sctworkspace\basetools
EDK_TOOLS_BIN = c:\sctworkspace\basetools\bin\win32
CONF_PATH = c:\sctworkspace\conf


Architecture(s) = X64
Build target = DEBUG
Toolchain = VS2015x86

Active Platform = c:\sctworkspace\SctPkg\UEFI\UEFI_SCT.dsc

Processing meta-data .......

build.py...
c:\sctworkspace\SctPkg\TestCase\UEFI\EFI\BootServices\ImageServices\BlackBoxTest\Dependency\InvalidImage1\InvalidImage1.inf(...): error 5000: COMPONENT_TYPE is not given



- Failed -
Build end time: 01:29:39, Nov.07 2020
Build total time: 00:00:10
Re: 回复: [edk2-discuss] Refresh currently displayed form

Tim Crawford
 

I have made BZ 3053 for the VFR spec issue.

https://bugzilla.tianocore.org/show_bug.cgi?id=3053

On Wed, Nov 4, 2020 at 5:53 PM gaoliming <gaoliming@byosoft.com.cn> wrote:

Tim:
Yes. Refreshguid can be used in Question or Form. Can you help submit BZ to record the issue in VFR spec?

Thanks
Liming
-----邮件原件-----
发件人: tosubbounce+34241+0+0+0@groups.io
<tosubbounce+34241+0+0+0@groups.io> 代表 Tim Crawford
发送时间: 2020年11月5日 0:25
收件人: gaoliming <gaoliming@byosoft.com.cn>; discuss@edk2.groups.io
主题: Re: [edk2-discuss] 回复: [edk2-discuss] Refresh currently displayed
form

BootManager elements are generated dynamically. There are no questions or
actions to attach a refresh interval or refreshguid to. I don't believe a refresh
interval would produce the behavior I want; that is, to update *as soon as*
an event (plugging/pulling a USB drive) happens.

But after looking at ANTLR syntax
(BaseTools/Source/C/VfrCompile/VfrSyntax.g) I found that refreshguid can be
used at the form level. This was added some time ago [1], but is not in the
VFR specification [2]. Using this with CreateEventEx(), with refreshguid and
EventGroup being the same GUID, produces the behavior I want.

[1]
https://github.com/tianocore/edk2/commit/e2a5feb9d6c468c0ddb0caed258
1d10585819db5
[2]
https://edk2-docs.gitbook.io/edk-ii-vfr-specification/2_vfr_description_in_bnf
/211_vfr_form_definition

Thanks,
Tim
回复: [edk2-discuss] 回复: [edk2-discuss] Refresh currently displayed form

gaoliming
 

Tim:
Yes. Refreshguid can be used in Question or Form. Can you help submit BZ to record the issue in VFR spec?

Thanks
Liming

-----邮件原件-----
发件人: tosubbounce+34241+0+0+0@groups.io
<tosubbounce+34241+0+0+0@groups.io> 代表 Tim Crawford
发送时间: 2020年11月5日 0:25
收件人: gaoliming <gaoliming@byosoft.com.cn>; discuss@edk2.groups.io
主题: Re: [edk2-discuss] 回复: [edk2-discuss] Refresh currently displayed
form

BootManager elements are generated dynamically. There are no questions or
actions to attach a refresh interval or refreshguid to. I don't believe a refresh
interval would produce the behavior I want; that is, to update *as soon as*
an event (plugging/pulling a USB drive) happens.

But after looking at ANTLR syntax
(BaseTools/Source/C/VfrCompile/VfrSyntax.g) I found that refreshguid can be
used at the form level. This was added some time ago [1], but is not in the
VFR specification [2]. Using this with CreateEventEx(), with refreshguid and
EventGroup being the same GUID, produces the behavior I want.

[1]
https://github.com/tianocore/edk2/commit/e2a5feb9d6c468c0ddb0caed258
1d10585819db5
[2]
https://edk2-docs.gitbook.io/edk-ii-vfr-specification/2_vfr_description_in_bnf
/211_vfr_form_definition

Thanks,
Tim
Re: 回复: [edk2-discuss] Refresh currently displayed form

Tim Crawford
 

BootManager elements are generated dynamically. There are no questions or actions to attach a refresh interval or refreshguid to. I don't believe a refresh interval would produce the behavior I want; that is, to update *as soon as* an event (plugging/pulling a USB drive) happens.

But after looking at ANTLR syntax (BaseTools/Source/C/VfrCompile/VfrSyntax.g) I found that refreshguid can be used at the form level. This was added some time ago [1], but is not in the VFR specification [2]. Using this with CreateEventEx(), with refreshguid and EventGroup being the same GUID, produces the behavior I want.

[1] https://github.com/tianocore/edk2/commit/e2a5feb9d6c468c0ddb0caed2581d10585819db5
[2] https://edk2-docs.gitbook.io/edk-ii-vfr-specification/2_vfr_description_in_bnf/211_vfr_form_definition

Thanks,
Tim
Re: Device specific _RST method

Igor Mammedov <imammedo@...>
 

On Wed, 4 Nov 2020 10:14:28 +0530
Kumar G <kumarg27061979@gmail.com> wrote:

Many Thanks Igor

yes, devices connected to the described controller will be enumerated by
bus specific driver.
Thanks

It should be possible to add 'Device' object under the described
controller
and then add _RST method to that. Even if Device is described in ACPI it
doesn't garantie that specific impl. of PCI/USB driver will actually take
all it describes into account.
You mean, I can have a device with a vendor id under the controller hub ?
Something like
Scope (_SB)
{
Device (PCI) {
Device(My_device) {
Name(_ADR, VENDOR_ID) <-- This is vendor id, scanned by Bus driver ??
_ADR is bus specific address, not vendor id
for ex you can check what QEMU uses to generate it
or better read a relevant spec

--- Device specific method goes here
} // End of my device

} /end of PCI
}// end of SB
with above, I am wondering, if OS driver will scan bus or this table

But I haven't seen _RST being actually used, I'd say typically PCI/USB
driver is used to reset device (port) without involving ACPI.
Yeah, this is normally i saw, where driver is taking care of reset without
asking
ACPI using specification defined bits

On Tue, 3 Nov 2020 at 16:21, Igor Mammedov <imammedo@redhat.com> wrote:

On Mon, 2 Nov 2020 23:01:18 +0530
Kumar G <kumarg27061979@gmail.com> wrote:

Hey Experts
I am looking for your help ,
how I can define device specific _RST method in acpi table in case device
is behind USB or PCIe controller.
So far, I studied, I only need to define a PCIe or USB controller in my
acpi table and all connected devices will be enumerated by OS itself.
yes, devices connected to the described controller will be enumerated by
bus specific driver.

Is there any way, I can define my device under PCIe/USb controller with
specific _RST method
It should be possible to add 'Device' object under the described controller
and then add _RST method to that. Even if Device is described in ACPI it
doesn't garantie that specific impl. of PCI/USB driver will actually take
all it describes into account.

Any sample code/pointers are welcome
But I haven't seen _RST being actually used, I'd say typically PCI/USB
driver is used to reset device (port) without involving ACPI.


Thanks
Kumar
Re: Device specific _RST method

Kumar G <kumarg27061979@...>
 

Many Thanks Igor

> yes, devices connected to the described controller will be enumerated by
> bus specific driver.

Thanks

> It should be possible to add 'Device' object under the described controller
> and then add _RST method to that. Even if Device is described in ACPI it
> doesn't garantie that specific impl. of PCI/USB driver will actually take
> all it describes into account.

You mean, I can have a device with a vendor id under the controller hub ?
 Something like
Scope (_SB)
{
 Device (PCI) {
  Device(My_device) {
   Name(_ADR, VENDOR_ID) <-- This is vendor id, scanned by Bus driver ??
   --- Device specific method goes here
  } // End of my device
  
 } /end of PCI
}// end of SB
with above, I am wondering, if OS driver will scan bus or this table

> But I haven't seen _RST being actually used, I'd say typically PCI/USB
> driver is used to reset device (port) without involving ACPI.
Yeah, this is normally i saw, where driver is taking care of reset without asking
ACPI using specification defined bits


On Tue, 3 Nov 2020 at 16:21, Igor Mammedov <imammedo@...> wrote:
On Mon, 2 Nov 2020 23:01:18 +0530
Kumar G <kumarg27061979@...> wrote:

> Hey Experts
> I am looking for your help ,
> how I can define device specific _RST method in acpi table in case device
> is behind USB or PCIe controller.
> So far, I studied, I only need to define a PCIe or USB controller in my
> acpi table and all connected devices will be enumerated by OS itself.
yes, devices connected to the described controller will be enumerated by
bus specific driver.

> Is there any way, I can define my device under PCIe/USb controller with
> specific _RST method
It should be possible to add 'Device' object under the described controller
and then add _RST method to that. Even if Device is described in ACPI it
doesn't garantie that specific impl. of PCI/USB driver will actually take
all it describes into account.

> Any sample code/pointers are welcome
But I haven't seen _RST being actually used, I'd say typically PCI/USB
driver is used to reset device (port) without involving ACPI.


> Thanks
> Kumar
Re: Device specific _RST method

Igor Mammedov <imammedo@...>
 

On Mon, 2 Nov 2020 23:01:18 +0530
Kumar G <kumarg27061979@gmail.com> wrote:

Hey Experts
I am looking for your help ,
how I can define device specific _RST method in acpi table in case device
is behind USB or PCIe controller.
So far, I studied, I only need to define a PCIe or USB controller in my
acpi table and all connected devices will be enumerated by OS itself.
yes, devices connected to the described controller will be enumerated by
bus specific driver.

Is there any way, I can define my device under PCIe/USb controller with
specific _RST method
It should be possible to add 'Device' object under the described controller
and then add _RST method to that. Even if Device is described in ACPI it
doesn't garantie that specific impl. of PCI/USB driver will actually take
all it describes into account.

Any sample code/pointers are welcome
But I haven't seen _RST being actually used, I'd say typically PCI/USB
driver is used to reset device (port) without involving ACPI.


Thanks
Kumar
when bug 2172 & 2177 will be fixed

wenyi,xie
 

Hi，all

May I ask when bug 2172 & 2177 will be fixed, is there any plan, thank you!
https://bugzilla.tianocore.org/show_bug.cgi?id=2172
https://bugzilla.tianocore.org/show_bug.cgi?id=2177
回复: [edk2-discuss] Issues with PatchCheck script

gaoliming
 

I use git version 2.27. Do you verify the latest git version?

-----邮件原件-----
发件人: bounce+34241+423+4905953+8764802@groups.io
<bounce+34241+423+4905953+8764802@groups.io> 代表 Laszlo Ersek
发送时间: 2020年11月3日 4:49
收件人: discuss@edk2.groups.io; sandeep.dhanvada@xilinx.com
主题: Re: [edk2-discuss] Issues with PatchCheck script

On 10/28/20 12:45, Sandeep Dhanvada wrote:
Hi,

I have 2 local commits in EDK-II repo. As per the steps mentioned in
https://github.com/tianocore/tianocore.github.io/wiki/EDK-II-Development-P
rocess,
I am executing PatchCheck.py, but, the execution of this script is giving
below error. I see that the command used is "git show --pretty=%cn <%ce>,
--no-patch, --no-use-mailmap, <CommitID>, but, as git complains that there is
no --no-patch option for git show and also format for --pretty should be in
quotes. I tried on git version 1.8.3.1 and 2.1.1. Is there any specific git version
needed to execute this script ?

==========
python BaseTools/Scripts/PatchCheck.py -1
Checking git commit: HEAD
Traceback (most recent call last):
File "BaseTools/Scripts/PatchCheck.py", line 770, in <module>
sys.exit(PatchCheckApp().retval)
File "BaseTools/Scripts/PatchCheck.py", line 733, in __init__
self.process_one_arg('HEAD')
File "BaseTools/Scripts/PatchCheck.py", line 747, in process_one_arg
self.ok &= CheckOneArg(arg, self.count).ok
File "BaseTools/Scripts/PatchCheck.py", line 714, in __init__
checker = CheckGitCommits(param, max_count)
File "BaseTools/Scripts/PatchCheck.py", line 650, in __init__
self.ok &= EmailAddressCheck(email, 'Committer').ok
File "BaseTools/Scripts/PatchCheck.py", line 36, in __init__
self.error('Email address is missing!')
File "BaseTools/Scripts/PatchCheck.py", line 47, in error
print('The ' + self.description + ' email address is not valid:')
AttributeError: EmailAddressCheck instance has no attribute 'description'

==========
"--no-patch" appeared in 1.8.4, as a synonym for "-s".

Not sure why it doesn't work with 2.1.1 -- perhaps your old git instance
appears first on $PATH?

Laszlo
回复: [edk2-discuss] Refresh currently displayed form

gaoliming
 

VFR question has attribute refresh interval (Refresh interval in seconds). It will trig call back regularly.
There is one example is in edk2\MdeModulePkg\Universal\DriverSampleDxe\Vfr.vfr

You can add the logic in this question callback function to refresh the boot option. If so, the boot option can be updated.

Thanks
Liming

-----邮件原件-----
发件人: bounce+34241+421+4905953+8764802@groups.io
<bounce+34241+421+4905953+8764802@groups.io> 代表 Tomas Pilar
(tpilar)
发送时间: 2020年10月28日 22:01
收件人: discuss@edk2.groups.io; crawfxrd@gmail.com
主题: Re: [edk2-discuss] Refresh currently displayed form

Okay, I actually read the rest of your question and your problem is deeper
than that. What you need is for the device to be connected to a driver
(driver which might need to be loaded from the device option ROM), boot
manager to re-enumerate boot options, and then the page can be reloaded.
This is more difficult than what I originally thought you were trying to do.

Cheers,
Tom

On Wed, Oct 28, 2020 at 1:58 PM Tomas Pilar <tomas@nuviainc.com> wrote:

Hi,

You can add a 'refreshguid' property to a number of VFR objects. Then,
whenever you signal the event group with that guid, the object should be
refreshed by the HII Browser.

Cheers,
Tom

On Wed, Oct 28, 2020 at 1:04 PM Tim Crawford <crawfxrd@gmail.com>
wrote:

Hi all,

How can I refresh the currently displayed form?

The problem I'm trying to solve is that when on the BootManager page,
insertion or removal of a device does not update the form, leaving the list
of boot options in an invalid state.

Right now I'm trying to hack in an event to accomplish this, but someone
said there may be a mechanism to do this.

Thanks,
Tim
Device specific _RST method

Kumar G <kumarg27061979@...>
 

Hey Experts
I am looking for your help ,
how I can define device specific _RST method in acpi table in case device is behind USB or PCIe controller.
So far, I studied, I only need to define a PCIe or USB controller in my acpi table and all connected devices will be enumerated by OS itself.

Is there any way, I can define my device under PCIe/USb controller with specific _RST method

Any sample code/pointers are welcome

Thanks
Kumar
enable secure boot on edk2-vUDK2018

luccons@...
 

Hi guys,

I'm following some README on tianocore wiki in order to enable secure boot on edk2 version UDK2018 but I notice that I don't build AuthVariableLIb.c placed into SecurityPkg/LIbrary/AuthVariableLib folder, for me mandatory because I've placed in hardcoded way my certificates PK, KEK, DB and DBK.

Any suggestions?

Thank you and best regards,
Francesco

?
0008_enable_tpm_support.patch

Re: Newbie question about EFI_BOOT_SERVICES_MEMORY

Laszlo Ersek
 

On 10/26/20 17:08, Vestigial Bidya wrote:
Hi, this question is about the UEFI PDFs rather than the Tianocore
implementation. I'm writing my DXE Foundation, and the documentation states
" Allocate the UEFI Boot Services Table from EFI_BOOT_SERVICES_MEMORY"
Please be a *lot* more specific than this. People trying to help
shouldn't have to hunt down the issue you are reporting.

For reference, you are (most likely) talking about the Platform Init
spec v1.7, volume 2, section "9.5 Handing Control to DXE Dispatcher".

Indeed this section uses terms like EFI_BOOT_SERVICES_MEMORY and
EFI_RUNTIME_SERVICES_MEMORY. They seem like typographical errors to me.

The way the term EFI_BOOT_SERVICES_MEMORY is formatted makes it look like
a defined Protocol or EFI_GUID, but its not mentioned anywhere else in the
docs (or on Google). Is the author trying to convey just that these
allocations should be from memory that will be released after calling
EXIT_BOOT_SERVICES ? Or is EFI_BOOT_SERVICES_MEMORY an actual term in the
spec that I just can't find?
They are simple typos, in my opinion. The replacement texts should be
"EFI boot services memory" and "runtime memory". These latter terms are
repeatedly used in the UEFI spec too (v2.8).

Can you file a bug for the PI spec in the bug tracker at
<https://mantis.uefi.org>?

Thanks
Laszlo
Re: Issues with PatchCheck script

Laszlo Ersek
 

On 10/28/20 12:45, Sandeep Dhanvada wrote:
Hi,

I have 2 local commits in EDK-II repo. As per the steps mentioned in https://github.com/tianocore/tianocore.github.io/wiki/EDK-II-Development-Process,
I am executing PatchCheck.py, but, the execution of this script is giving below error. I see that the command used is "git show --pretty=%cn <%ce>, --no-patch, --no-use-mailmap, <CommitID>, but, as git complains that there is no --no-patch option for git show and also format for --pretty should be in quotes. I tried on git version 1.8.3.1 and 2.1.1. Is there any specific git version needed to execute this script ?

==========
python BaseTools/Scripts/PatchCheck.py -1
Checking git commit: HEAD
Traceback (most recent call last):
File "BaseTools/Scripts/PatchCheck.py", line 770, in <module>
sys.exit(PatchCheckApp().retval)
File "BaseTools/Scripts/PatchCheck.py", line 733, in __init__
self.process_one_arg('HEAD')
File "BaseTools/Scripts/PatchCheck.py", line 747, in process_one_arg
self.ok &= CheckOneArg(arg, self.count).ok
File "BaseTools/Scripts/PatchCheck.py", line 714, in __init__
checker = CheckGitCommits(param, max_count)
File "BaseTools/Scripts/PatchCheck.py", line 650, in __init__
self.ok &= EmailAddressCheck(email, 'Committer').ok
File "BaseTools/Scripts/PatchCheck.py", line 36, in __init__
self.error('Email address is missing!')
File "BaseTools/Scripts/PatchCheck.py", line 47, in error
print('The ' + self.description + ' email address is not valid:')
AttributeError: EmailAddressCheck instance has no attribute 'description'

==========
"--no-patch" appeared in 1.8.4, as a synonym for "-s".

Not sure why it doesn't work with 2.1.1 -- perhaps your old git instance
appears first on $PATH?

Laszlo
Re: Refresh currently displayed form

Tim Crawford
 

I've added a protocol notify event for gEfiBlockIoProtocolGuid and hacked it to signal the event on protocol removal as well. edk2 doesn't seem to have a mechanism to notify about a protocol being removed, only created? This works well enough to for now since I only care about USB flash drives, although I haven't dug into the boot log to see what effect calling notify events on protocol removal has in other places.

I'm now looking at methods to refresh the form in place, so I'll grep around for code using a refreshguid as an example.
Re: Refresh currently displayed form

Tomas Pilar (tpilar)
 

Okay, I actually read the rest of your question and your problem is deeper
than that. What you need is for the device to be connected to a driver
(driver which might need to be loaded from the device option ROM), boot
manager to re-enumerate boot options, and then the page can be reloaded.
This is more difficult than what I originally thought you were trying to do.

Cheers,
Tom

On Wed, Oct 28, 2020 at 1:58 PM Tomas Pilar <tomas@nuviainc.com> wrote:

Hi,

You can add a 'refreshguid' property to a number of VFR objects. Then,
whenever you signal the event group with that guid, the object should be
refreshed by the HII Browser.

Cheers,
Tom

On Wed, Oct 28, 2020 at 1:04 PM Tim Crawford <crawfxrd@gmail.com> wrote:

Hi all,

How can I refresh the currently displayed form?

The problem I'm trying to solve is that when on the BootManager page,
insertion or removal of a device does not update the form, leaving the list
of boot options in an invalid state.

Right now I'm trying to hack in an event to accomplish this, but someone
said there may be a mechanism to do this.

Thanks,
Tim
Re: Refresh currently displayed form

Tomas Pilar (tpilar)
 

Hi,

You can add a 'refreshguid' property to a number of VFR objects. Then,
whenever you signal the event group with that guid, the object should be
refreshed by the HII Browser.

Cheers,
Tom

On Wed, Oct 28, 2020 at 1:04 PM Tim Crawford <crawfxrd@gmail.com> wrote:

Hi all,

How can I refresh the currently displayed form?

The problem I'm trying to solve is that when on the BootManager page,
insertion or removal of a device does not update the form, leaving the list
of boot options in an invalid state.

Right now I'm trying to hack in an event to accomplish this, but someone
said there may be a mechanism to do this.

Thanks,
Tim
Refresh currently displayed form

Tim Crawford
 

Hi all,

How can I refresh the currently displayed form?

The problem I'm trying to solve is that when on the BootManager page, insertion or removal of a device does not update the form, leaving the list of boot options in an invalid state.

Right now I'm trying to hack in an event to accomplish this, but someone said there may be a mechanism to do this.

Thanks,
Tim
Re: Issues with PatchCheck script

Sandeep Dhanvada
 

No. I am using Python 2.7.5.
But, in any case, the script is issuing "git show --nopatch" command and git is complaining about --no-patch
541 - 560 of 971