Sounds good. Thank you!
Best,
Yuan
On Tue, Jan 31, 2023 at 10:48 PM Wu, Hao A wrote:
> Reviewed-by: Hao A Wu
> Will update the function comment for ScsiScanCreateDevice() in ScsiBus.c
> to match the change in ScsiBus.h during merge.
>
> Best Regards,
> Hao Wu
>
> > -Original Message-
This patch cleans up comment of ScsiDisk.c.
The change can be seen at:
https://github.com/yyu/edk2/tree/comment_fix_v1
Cc: Ard Biesheuvel
Cc: Liming Gao
Cc: Hao A Wu
Cc: Ray Ni
Cc: Sivaparvathi chellaiah
Yuan Yu (1):
MdeModulePkg: ScsiDiskDxe: clean up comment in ScsiDisk.c
Comment of MdeModulePkg/Bus/Scsi/ScsiDiskDxe/ScsiDisk.c was polluted by
some previous merge. This patch clean it up.
Cc: Ard Biesheuvel
Cc: Liming Gao
Cc: Hao A Wu
Cc: Ray Ni
Cc: Sivaparvathi chellaiah
Signed-off-by: Yuan Yu
---
MdeModulePkg/Bus/Scsi/ScsiDiskDxe/ScsiDisk.c | 4
1
Hi Hao Wu,
Thank you very much for the review and the detailed advice, which is
much appreciated!
I have made the changes accordingly and sent patch v2. Please take another
look.
Thank you!
Yuan
On Wed, Jan 18, 2023 at 10:33 PM Wu, Hao A wrote:
> Thanks for the patch, inline comments
Currently DiscoverScsiDevice() returns a boolean which cannot
distinguish a "not found" situation from a real problem like
memory allocation failures.
This patch changes the return value to an EFI_STATUS so that when
memory allocation fails, it will return EFI_OUT_OF_RESOURCES.
Without this
Refactor DiscoverScsiDevice() so that it returns EFI_STATUS instead of
BOOLEAN. This will enable its callers to distinguish a "not found"
situation from real problems like memory allocation failures.
The changes can be seen at:
https://github.com/yyu/edk2/tree/scsi_bus_fix_v2
Yuan Yu (1):
Logically, the while loop that contains ScsiScanCreateDevice() should
continue regardless what the returned Status is, because the purpose of
the while loop is to scan all possible Puns in the SCSI channel.
Without this fix, some static analyzer may complain about the unused
return value.
Cc:
A while loop in SCSIBusDriverBindingStart() is supposed to scan all the
possible Puns in the SCSI channel by calling ScsiScanCreateDevice() for
each of them. Therefore, we should not abort the loop even when one of
the Puns is disconnected.
The following is one of the scenarios.
A goto following an error check may cause boot failure, because the
function that returns the status is supposed to be called for all the
possible Puns in the SCSI channel.
This patch removes the check and the goto so that the while loop will
continue to run until all devices are scanned.
The
SECSPERDAY is 86400 which exceeds the limit of a UINT16 which is 65536.
Therefore DayRemainder cannot use UINT16. This patch makes it UINT32.
Cc: Ard Biesheuvel
Cc: Jordan Justen
Cc: Laszlo Ersek
Cc: Anthony Perard
Cc: Julien Grall
Signed-off-by: Yuan Yu
---
Some value uses UINT16 which is not wide enough to hold the values that
it is supposed to hold. This series fix it by using UINT32.
The changes can be seen at:
https://github.com/yyu/edk2/tree/overflow_fix_v1
Cc: Ard Biesheuvel
Cc: Jordan Justen
Cc: Laszlo Ersek
Cc: Anthony Perard
Cc: Julien
Per UEFI Spec 2.9, EFI_HII_CONFIG_ROUTING_PROTOCOL.RouteConfig()
should return EFI_INVALID_PARAMETERS if caller passes in a NULL for
the Configuration parameter (see 35.4 EFI HII Configuration Routing
Protocol).
Add a check to return EFI_INVALID_PARAMETERS when Configuration is NULL.
Cc: Ard
Some function doesn't have certain checks to be compliant with UEFI
specification.
This series adds a check to return EFI_INVALID_PARAMETER if certain
argument is NULL.
The changes can be seen at:
https://github.com/yyu/edk2/tree/return_value_fix_v1
Cc: Ard Biesheuvel
Cc: Jordan Justen
Cc:
Introduce NetworkCfgLib which will set PcdNetworkSupport based on
"etc/networking" qemu file.
If "etc/networking" (type bool) is TRUE, then PcdNetworkSupport will be
TRUE and vice versa.
In the following patch, PcdNetworkSupport will be used to enable/disable
VirtIo net driver so that VMM will
Currently networking can only be enabled/disabled at compile time. This
patch series will add support to disable VirtIo net at runtime even if
the functionality is built into binary at compile time.
This will enable VMM to reduce attack surface without recompilation.
The changes can be seen at:
Enable/Disable VirtIo net based on the value of PcdNetworkSupport which
is controlled in NetworkCfgLib, which sets the PCD based on
"etc/networking" qemu file.
With this change, VMM can disable networking even if it is enabled at
compile time. This will allow to reduce attack surface by simply
16 matches
Mail list logo
