Because the D&I caches were clean before to be disabled, the cache
lines might have got dirty during the cache maintenance operations.
This fix disables D&I caches before to clean them. The performance
drops should be minimised as invalidating the I cache is only a
couple of instruction.
Signed-off-by: Eugene Cohen <eugene@hp.com>
Reviewed-by: Olivier Martin <olivier.martin@arm.com>
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@13503 6f19259b-4bc3-4df7-8a09-765794883524
If the first call of gBS->GetMemoryMap() succeeded (could happen if the Memory Map
has changed between the two gBS->GetMemoryMap() calls) in the loop block then
gBS->ExitBootServices() was never called.
Signed-off-by: Eugene Cohen <eugene@hp.com>
Reviewed-by: Olivier Martin <olivier.martin@arm.com>
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@13500 6f19259b-4bc3-4df7-8a09-765794883524
The 'NS' bit must only be set in Secure world to define the Non-Secure region
of the Non-Secure World.
This bit must not be set in Non-Secure World.
Signed-off-by: Olivier Martin <olivier.martin@arm.com>
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@13252 6f19259b-4bc3-4df7-8a09-765794883524
Previsouly the synchronization of MpCore was using the SGI (Software
Generated Interrupt) to synchronize MpCore during the early boot.
This commit replaced this mechanism by the more appropriate SEV/WFE
instructions (Send/Wait Event instructions).
That also eases the port to a new cpu/platform.
Signed-off-by: Olivier Martin <olivier.martin@arm.com>
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@13249 6f19259b-4bc3-4df7-8a09-765794883524
ArmEnableVFP could crash on an out-of-order CPU. Adding an instruction barrier after writing to CPACR cures the problem.
Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Olivier Martin <olivier.martin@arm.com>
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@13134 6f19259b-4bc3-4df7-8a09-765794883524
Some ARM Platform components (ie: PrePei) use this constructor name to
initialize the timers (at this time there is no PE loader to call the
library constructors) when PI/UEFI is started to initialize the
PerformanceLib.
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@13041 6f19259b-4bc3-4df7-8a09-765794883524
This patch adds support to invalidate Instruction Cache to the Point of Unification (PoU).
Signed-off-by: eugenecohen
Reviewed-by: oliviermartin
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@13012 6f19259b-4bc3-4df7-8a09-765794883524
The initrd was loaded at the top of the UEFI System Memory. By consequence, if the system memory
was reduced by the Linux command line then the initrd was not part of the system memory.
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@12994 6f19259b-4bc3-4df7-8a09-765794883524
Change __aeabi_uread4 from:
ldrb r2, [r0, #1]
ldrb r1, [r0]
(...)
to:
ldrb r1, [r0]
ldrb r2, [r0, #1]
(...)
This change is a workaround to handle correctly __aeabi_uread4 on ARM
Versatile Express RTSM.
It should not have any major consequence on the other ARM platforms.
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@12481 6f19259b-4bc3-4df7-8a09-765794883524
The denomination 'Normal' was used to make reference to the 'Normal'
or 'Non Secure' or 'Non Trusted' world.
To avoid confusion, this prefix has been removed from PCDs to define
the normal world.
The PCDs explicitely related to the Secure/Trusted World continue to
have the 'Sec' prefix.
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@12416 6f19259b-4bc3-4df7-8a09-765794883524
Separate the BdsBootLinux() function into two functions for Atag and Fdt specific Linux booting
- BdsBootLinuxAtag ()
- BdsBootLinuxFdt ()
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@12408 6f19259b-4bc3-4df7-8a09-765794883524
Missing from my commit "ArmPkg/BdsLib: Move some functions used to
create/update BDS Boot Entry from ArmPlatformPkg/Bds to
ArmPkg/BdsLib"
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@12325 6f19259b-4bc3-4df7-8a09-765794883524
The previous version was only checking if the Remaining Device Path node
was a MemMap DevicePath node.
This fix also checks if the Root DevicePath node is a MemMap DevicePath node.
git-svn-id: https://edk2.svn.sourceforge.net/svnroot/edk2/trunk/edk2@12315 6f19259b-4bc3-4df7-8a09-765794883524
oliviermartin