On 09/16/13 23:42, Andrew Fish wrote:
>
> On Sep 16, 2013, at 1:16 PM, Laszlo Ersek wrote:
>
>> On 09/16/13 21:50, Andrew Fish wrote:
>>
>>> OVMF:
>>> Limitations: Requires QEMU, serial port debugging (not well
>>> documented), more complex setup.
>>
>> I agree that the setup is complex.
>>
>> R
Hi Feng,
I am also thinking about different ASSERTs in the driver.
If debug then ASSERT cause a hang/stop/break.
If release then ASSERT do nothing causes next operator to crash (null pointer
etc).
For example
--
UINT32
XhcReadOpReg (
IN USB_XHCI_INSTANCE *Xhc,
IN UINT32
If you look at the BDS code which creates the boot option for network, it just
gets the MAC device path and stores it into the boot option variable.
So it looks strange that the LocateDevicePath() cannot find the same MAC device
path in the system.
-Original Message-
From: Peter Jones [
[...]
> Several OVMF pages in the TianoCore wiki are somewhat outdated, but I
> have no permission to edit them. [...]
The EDK-II OVMF downloads were last built in early 2011. It would be
nice if someone uploaded something newer.
http://sourceforge.net/projects/edk2/files/OVMF/
OVMF-X64-r11337-
Still no go with EmulatorPkg package... but i can build AppPkg successfully.
Here is version information from my build system:
svn info:
Path: .
URL: https://svn.code.sf.net/p/edk2/code/trunk/edk2/EmulatorPkg
Repository Root: https://svn.code.sf.net/p/edk2/code
Repository UUID: 6f19259b-4bc3-4df7
I think I get what happens - the implementation must be holding on to the
device path protocol, the by_driver fails so we try the by_driver|exclusive
which then triggers the disconnectcontroller logic internally which isn't
protocol specific and just stops whatever it is, in this case the file
syst
On 09/17/13 09:36, Blibbet wrote:
> [...]
>> Several OVMF pages in the TianoCore wiki are somewhat outdated, but I
>> have no permission to edit them. [...]
>
> The EDK-II OVMF downloads were last built in early 2011. It would be
> nice if someone uploaded something newer.
>
> http://sourceforg
On 09/17/13 09:33, Ni, Ruiyu wrote:
> If you look at the BDS code which creates the boot option for network, it
> just gets the MAC device path and stores it into the boot option variable.
> So it looks strange that the LocateDevicePath() cannot find the same MAC
> device path in the system.
The
On 09/17/13 09:43, okosmy...@kraftway.ru wrote:
> src/Build/OvmfX64/DEBUG_GCC46/X64/MdePkg/Library/UefiDevicePathLib/UefiDevicePathLib/OUTPUT/UefiDevicePathLib.lib(UefiDevicePathLib.obj):
> In function `ConvertDeviceNodeToText':
> src/MdePkg/Library/UefiDevicePathLib/UefiDevicePathLib.c:291: *_mul
One more issue here. Look into XhcReadRuntimeReg ()
--
if (EFI_ERROR (Status)) {
DEBUG ((EFI_D_ERROR, "XhcReadRuntimeReg: Pci Io Read error - %r at %d\n",
Status, Offset));
Data = 0x;
}
return Data;
}
--
But usage of the procedure do not propose wrong return. For exa
Hi, Sergey
The Assert() in XhcReadOpReg() makes sense as it would never be false if the
XHCI controller follows XHCI spec.
For your 2nd concern, yes, we didn't care the return data for such cases.
Thanks
Feng
From: Sergey Isakov [mailto:isakov...@bk.ru]
Sent: Tuesday, September 17, 2013 16:13
On 09/17/13 08:36, Ryan Harkin wrote:
> I think FreePool(NULL) should just return without reporting an error.
I agree that it could simplify some error/exit paths.
It may not match the general "UEFI interface style" though. In my
experience only pointers to optional parameters may be NULL, and
F
Hi Sathya,
Please go through the below information and share valuable comments and
suggestions.
Setup Details :
My Intel board connected to PCIe based SCSI controller (LSISAS1064E PCI Express
to 4-Port Serial Attached SCSI Controller ) and it has three SATA
devices connected with the pcie board.
Hi!
See the patch.
I get the reason of not checking AllocateZeroPool() return value
(GetVariable() will return EFI_BUFFER_TOO_SMALL again). But check
before calling FreePool().
SetVar.c.patch
Description: Binary data
--
L
On Sep 17, 2013, at 8:59 AM, Andrew Fish wrote:
> EXCLUSIVE means kick every one else off. So for example if you were about to
> format a driver
Make that drive
> you would open the BlockIo EXCLUSIVE. So it sounds like it is functioning as
> intended. EXCLUSIVE is a massive hammer, and gene
On Sep 16, 2013, at 11:36 PM, Ryan Harkin wrote:
> Hi,
>
> Tracking down a crash I am seeing shows that I am effectively calling
> FreePool() with a NULL pointer. The system then ASSERTS and hangs.
>
> I've noticed various seemingly "random" asserts like this before and suspect
> it may be
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Olivier Martin
---
OvmfPkg/VirtioPciDeviceDxe/VirtioPciDevice.c | 637 +
OvmfPkg/VirtioPciDeviceDxe/VirtioPciDevice.h | 246
OvmfPkg/VirtioPciDeviceDxe/VirtioPciDeviceDxe.inf | 42 ++
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Olivier Martin
---
OvmfPkg/Include/IndustryStandard/Virtio.h | 13 +
1 files changed, 13 insertions(+), 0 deletions(-)
diff --git a/OvmfPkg/Include/IndustryStandard/Virtio.h
b/OvmfPkg/Include/IndustryStandard/V
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Olivier Martin
---
.../ArmVExpressPkg/ArmFvpDxe/ArmFvpDxe.c | 49 +++-
.../ArmVExpressPkg/ArmFvpDxe/ArmFvpDxe.inf |3 +
.../ArmVExpress-RTSM-AEMv8Ax4-foundation.dsc | 10
..
Yes, but you can't say be open BY_DRIVER when something like a file system
won't let you have the block io protocol, so you either have to move to
BY_HANDLE or EXCLUSIVE. You go Exclusive, you kick it off, you go
BY_HANDLE now you're unsafe for hot-swap.
On Tue, Sep 17, 2013 at 9:16 AM, Andrew
This definition is specific to VirtIo over PCI.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Olivier Martin
---
OvmfPkg/Include/IndustryStandard/Virtio.h | 18 --
1 files changed, 0 insertions(+), 18 deletions(-)
diff --git a/OvmfPkg/Include/IndustryS
On Sep 17, 2013, at 2:55 AM, Laszlo Ersek wrote:
> On 09/17/13 08:36, Ryan Harkin wrote:
>
>> I think FreePool(NULL) should just return without reporting an error.
>
> I agree that it could simplify some error/exit paths.
>
> It may not match the general "UEFI interface style" though. In my
This patchset introduces the new protocol VIRTIO_DEVICE_PROTOCOL that abstracts
the VirtIo transport layer.
It allows to enable VirtIo on devices that do not have PCI support. These
devices might use a memory mapped view of the VirtIo controller (see Appendix
X: virtio-mmio of the Virtio spec v0
Hi all,
StrLen() function MdePkg file BaseLib/String.c will ASSERT if the string is
not DWORD aligned.
I'm very curious, what is the rationale for this ASSERT?
--
LIMITED TIME SALE - Full Year of Microsoft Training For
Clear to me, but, clearly someone didn't read it that technically and were
probably use to backspace deleting things so they implemented it that way.
A little note that cursor movement should not affect displayed characters.
I just see it as one person implements it wrong, the next may too,
especia
On Sep 17, 2013, at 2:49 PM, "David F." wrote:
> Hi,
>
> Initial testing on physical machines the backspace sending to ConOut
> correctly just back spaced with no clearing of characters. Under VirtualBox
> (OVMF) it will clear characters which is not expected. The user would need
> to sen
On Sep 17, 2013, at 2:19 PM, Binh Tran wrote:
> Hi all,
>
> StrLen() function MdePkg file BaseLib/String.c will ASSERT if the string is
> not DWORD aligned.
>
> I’m very curious, what is the rationale for this ASSERT?
The EFI ABI requires that types are naturally aligned. Since a CHAR16
QEMU 1.6 enables using KVM READONLY memory regions. There
appears to be a bug when page tables are stored in the
READONLY memory region. To work-around this, we create
page table entries in RAM. Since RAM is present in QEMU/KVM
without any initialization, this is easily accomplished
for OVMF.
Sinc
Since we no longer require flash tables to be stored uncompressed
in the flash image, we can now give extra space to the main/compressed
storage area.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jordan Justen
---
OvmfPkg/OvmfPkgX64.fdf |4 ++--
1 file changed, 2 in
Now, Transition32FlatTo64Flat calls SetCr3ForPageTables64
which is located in Ia32/PageTables64.asm.
This change is required so OVMF can replace the code in
Ia32/PageTables64.asm with code that generates page tables in
RAM.
Note: Since this change does not impact the functionality of the
current
Since we no longer building page tables in SEC C code, we no
longer need this file.
This reverts commit r14493.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jordan Justen
---
OvmfPkg/Include/IndustryStandard/X64Paging.h | 99 --
1 file changed
This reset vector code will build page tables in RAM at address
0x8, rather than relying on page tables to be present within
the flash image.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jordan Justen
---
OvmfPkg/OvmfPkgX64.fdf |2 +-
1 file changed, 1 insertion
KVM has a bug that prevents using page tables in the ROM if the ROM
region utilizes the KVM READONLY memory feature. Therefore, we
avoid using page tables stored in the ROM.
Since OVMF doesn't require memory initialization, we just build
page table entries in RAM at 0x8 very early in the OVMF
Now for X64 we use a VTF0 ResetVector which puts the page
tables in RAM. Therefore SEC no longer needs to do this.
This reverts commit r14494.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Jordan Justen
---
OvmfPkg/Sec/SecMain.c | 144 +-
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Olivier Martin
---
OvmfPkg/Include/Library/VirtioMmioDeviceLib.h | 48
.../Library/VirtioMmioDeviceLib/VirtioMmioDevice.c | 199 ++
.../Library/VirtioMmioDeviceLib/VirtioMmioDevice.h | 147 ++
This protocol is used to abstract the VirtIo transport layer.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Olivier Martin
---
OvmfPkg/Include/Protocol/VirtioDevice.h | 284 +++
OvmfPkg/OvmfPkg.dec |1 +
2 files change
Hi,
So I'm always seeming to have to print out device paths and got tried of
copy/pasting code to print it (and without creating a whole new print
wrapper). I though others probably have to do the same types of things, so
here's a sample of adding %v to the core Print() and others functions using
On Sep 16, 2013, at 11:36 PM, Ryan Harkin wrote:
> Hi,
>
> Tracking down a crash I am seeing shows that I am effectively calling
> FreePool() with a NULL pointer. The system then ASSERTS and hangs.
>
> I've noticed various seemingly "random" asserts like this before and suspect
> it may be
EXCLUSIVE means kick every one else off. So for example if you were about to
format a driver you would open the BlockIo EXCLUSIVE. So it sounds like it is
functioning as intended. EXCLUSIVE is a massive hammer, and general is not
used, unless it is really required.
Thanks,
Andrew Fish
On Sep
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Olivier Martin
---
.../ArmVExpressPkg/ArmFvpDxe/ArmFvpDxe.c | 26 ++
.../ArmVExpressPkg/ArmFvpDxe/ArmFvpDxe.inf | 35
.../ArmVExpressPkg/ArmVExpress-RTSM-AEMv8Ax4.dsc |
Hi All,
Please go through the below information and share valuable comments and
suggestions.
*Setup Details :*
My Intel board connected to PCIe based SCSI controller (*LSISAS1064E PCI
Express to 4-Port Serial Attached SCSI Controller* ) and it has three SATA
devices connected with the pcie boar
Opps, WORD 16 bit aligned, not DWORD.
From: Binh Tran [mailto:binh_t...@phoenix.com]
Sent: Tuesday, September 17, 2013 2:19 PM
To: edk2-devel@lists.sourceforge.net
Subject: [edk2] MdePkg BaseLib/String.c StrLen() question
Hi all,
StrLen() function MdePkg file BaseLib/String.c will ASSERT if the
Hi,
Initial testing on physical machines the backspace sending to ConOut
correctly just back spaced with no clearing of characters. Under
VirtualBox (OVMF) it will clear characters which is not expected. The user
would need to send "\b \b" if they want to clear a character. Perhaps the
spec sh
The patches are also available in this tag:
https://github.com/jljusten/edk2.git ovmf-vm-vtf0-v1
On Tue, Sep 17, 2013 at 4:24 PM, Jordan Justen
wrote:
> QEMU 1.6 enables using KVM READONLY memory regions. There
> appears to be a bug when page tables are stored in the
> READONLY memory region. To
On Tue, Sep 17, 2013 at 9:54 AM, Olivier Martin wrote:
Seems like at least a small commit message body might be appropriate.
> Contributed-under: TianoCore Contribution Agreement 1.0
> Signed-off-by: Olivier Martin
> ---
> OvmfPkg/VirtioPciDeviceDxe/VirtioPciDevice.c | 637
> +++
I totally agree with you that putting function which would change h/w status
into ASSERT is not good practice.
For the case, it's only used to read a register once. It's just a double
confirmation and real judgment is done by XhcResetHC().
Thanks
Feng
From: Sergey Isakov [mailto:isakov...@bk.ru
On 17 September 2013 16:52, Andrew Fish wrote:
>
> On Sep 16, 2013, at 11:36 PM, Ryan Harkin wrote:
>
> Hi,
>
> Tracking down a crash I am seeing shows that I am effectively calling
> FreePool() with a NULL pointer. The system then ASSERTS and hangs.
>
> I've noticed various seemingly "random"
47 matches
Mail list logo
