CpuStateBuzy => CpuStateBusy
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
---
UefiCpuPkg/CpuDxe/CpuMp.c | 2 +-
UefiCpuPkg/CpuDxe/CpuMp.h | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/UefiCpuPkg/CpuDxe/CpuMp.c b/UefiCpuPkg/CpuDxe/CpuMp.
nowadays, APs are busy running when installing MP protocol in
multiple processors environment. so it whould consume much power
resource unnecessary. this patchs put APs to sleep when not procedure
to run, then via sending NMI IPI by BSP to wake up AP with hlt state.
there is a proposal several mon
Thanks.
Chen
On 03/05/2015 02:53 PM, Fan, Jeff wrote:
> Chen,
>
> I am ok with the new serial of patches. I will help to check-in the patches
> at next week.
>
> Reviewed-by: Jeff Fan
>
> Thanks!
> Jeff
> -Original Message-
> From: Chen Fan [mailto:chen.fan.f...@cn.fujitsu.com]
> Sent:
Chen,
I am ok with the new serial of patches. I will help to check-in the patches at
next week.
Reviewed-by: Jeff Fan
Thanks!
Jeff
-Original Message-
From: Chen Fan [mailto:chen.fan.f...@cn.fujitsu.com]
Sent: Thursday, March 05, 2015 2:30 PM
To: edk2-devel@lists.sourceforge.net
Cc: F
Add a new sleeping state for APs, when no procedure execution,
put AP to sleep. when need to execute procedure, only need to
wake up this AP by sent SIPI.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
---
UefiCpuPkg/CpuDxe/CpuMp.c | 61 ++
CpuState should follow the process：
Idle -> Ready -> Busy -> Finished
^ |
| |
+ - - - - - - - - - - - - +
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
---
UefiCpuPkg/CpuDxe/CpuMp.c | 9 +++-
For avoid the compiler optimizing the code, we let Parameter
and Procedure in CpuData volatile.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
Signed-off-by: Jeff Fan
---
UefiCpuPkg/CpuDxe/CpuMp.c | 4 ++--
UefiCpuPkg/CpuDxe/CpuMp.h | 4 ++--
2 files changed, 4
it works fine.
Thanks,
Chen
On 03/05/2015 02:07 PM, Fan, Jeff wrote:
> Please try the following updating to check if compiling error gone.
>
> Index: CpuMp.c
> ===
> --- CpuMp.c (revision 17009)
> +++ CpuMp.c (working copy)
> @@
Please try the following updating to check if compiling error gone.
Index: CpuMp.c
===
--- CpuMp.c (revision 17009)
+++ CpuMp.c (working copy)
@@ -1157,7 +1157,7 @@
UINTN ProcessorNumber;
CPU_DATA_BLOCK
On 03/05/2015 01:09 PM, Fan, Jeff wrote:
> Please align the volatile's position as below.
> + volatile EFI_AP_PROCEDURE Procedure;
> + volatile VOID* Parameter;
For Parameter should:
VOID* volatile Parameter;
otherwise, will cause compile error.
UefiCpuPkg/CpuDxe/CpuMp.c:1228:23: error:
Please align the volatile's position as below.
+ volatile EFI_AP_PROCEDURE Procedure;
+ volatile VOID* Parameter;
-Original Message-
From: Chen Fan [mailto:chen.fan.f...@cn.fujitsu.com]
Sent: Thursday, March 05, 2015 11:55 AM
To: edk2-devel@lists.sourceforge.net
Cc: Fan, Jeff; Ju
OK – after starting over and (more!) carefully following all the instructions
(and advice here!), I got a LOT further (probably close to the end).
BUT now I get this error:
Building ...
c:\.\myworkspace\EdkCompatibilityPkg\Compatibility\AcpiVariableHobOnSm
ramReserveHobThunk\AcpiVari
For avoid the compiler optimizing the code, we let Parameter
and Procedure in CpuData volatile.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
---
UefiCpuPkg/CpuDxe/CpuMp.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/UefiCpuPkg/CpuDxe/
CpuState should follow the process：
Idle -> Ready -> Busy -> Finished
^ |
| |
+ - - - - - - - - - - - - +
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
---
UefiCpuPkg/CpuDxe/CpuMp.c | 9 +++-
CpuStateBuzy => CpuStateBusy
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
---
UefiCpuPkg/CpuDxe/CpuMp.c | 2 +-
UefiCpuPkg/CpuDxe/CpuMp.h | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/UefiCpuPkg/CpuDxe/CpuMp.c b/UefiCpuPkg/CpuDxe/CpuMp.
Add a new sleeping state for APs, when no procedure execution,
put AP to sleep. when need to execute procedure, only need to
wake up this AP by sent SIPI.
Contributed-under: TianoCore Contribution Agreement 1.0
Signed-off-by: Chen Fan
---
UefiCpuPkg/CpuDxe/CpuMp.c | 61 ++
nowadays, APs are busy running when installing MP protocol in
multiple processors environment. so it whould consume much power
resource unnecessary. this patchs put APs to sleep when not procedure
to run, then via sending NMI IPI by BSP to wake up AP with hlt state.
there is a proposal several mon
I go it.
Thanks all of you.
Chen
On 03/05/2015 10:55 AM, Fan, Jeff wrote:
Andrew is right.
Volatile makes compiler read variable value from memory each time.
Reading from memory does not mean that it will skip the cache. It
still could get the value from the cache if cache is enabled and
Andrew is right.
Volatile makes compiler read variable value from memory each time. Reading from
memory does not mean that it will skip the cache. It still could get the value
from the cache if cache is enabled and valid. Volatile and cache are different
stores.
Thanks!
Jeff
From: Andrew Fish
> On Mar 4, 2015, at 5:34 PM, Chen Fan wrote:
>
> Hi Jeff,
>
> Thanks for your explanation, I didn't encounter any issue if leaving it
> there.
>
> I just think if use Volatile qualifier may cause the problem of
> performance. because
> access the value need to copy the address value from m
Hi Jeff,
Thanks for your explanation, I didn't encounter any issue if leaving it
there.
I just think if use Volatile qualifier may cause the problem of
performance. because
access the value need to copy the address value from main memory to cpu
cache
each time and not use the cache.
and IIRC,
Thanks for that further clarification, Michael.
Much appreciated.
Your input is of great value - so keep it up!!
Regards
Gerard
> On Mar 4, 2015, at 7:15 PM, Krau, Michael P wrote:
>
> Hello Gerard,
>
> You got it in one. The files that are included in that IP bundle are all
> tied dir
Hello Gerard,
You got it in one. The files that are included in that IP bundle are all tied
directly to the E38xx Silicon initialization and low level operation. The
MinnowBoard MAX use the E38xx processor, so that code is needed by the MAX by
induction. Though there may also be a RealTek
Daryl -
Is there any progress on providing some or all of StdLib for non-shell apps in
edk2? Which portions can be used independently now?
Tim
--
Dive into the World of Parallel Programming The Go Parallel Website, spons
Hi Michael,
Thanks for taking the time to explain all this – enjoyed the “insight”!
One thing that confuses me a little is your statement “MinnowBoard MAX requires
some Intellectual Property (IP) Binary files to build properly. ”
Personally I have no problem with that but I am curious
Hello Andrew,
Clarification question: Is this in regards to the whole of the EDK2 or the
interaction of the MinnowBoard MAX project components with the EDK2?
EDK2 itself should not be impacted by the MinnowBoard MAX project, though the
MinnowBoard MAX project is having problems on the Tip (due
Thank you Bruce,
I appreciate your prompt and well considered response.
For the de-synch issue, we are looking at a couple of possible solutions.
Your suggestion of setting up another site/mechanism for access to the
non-validated binaries is actually one of them. What we are hoping to do is
> On Mar 4, 2015, at 11:25 AM, Krau, Michael P wrote:
>
> Hello Andrew and Gerard,
>
> To respond to and to clarify the PS in Andrew’s message below:
>
> It is not the intention of the MinnowBoard MAX firmware team to complicate or
> cause issues within the UEFI Open Source Community. If
On Wed, Mar 4, 2015 at 7:25 PM, Krau, Michael P
wrote:
> Our development around MinnowBoard MAX are now centered on the Open
> Source community site. The BSD content for MinnowBoard MAX is being
> developed from the open source repository (including the platform specific
> packages – Vlv2Device
Hello Andrew and Gerard,
To respond to and to clarify the PS in Andrew’s message below:
It is not the intention of the MinnowBoard MAX firmware team to complicate or
cause issues within the UEFI Open Source Community. If the current methods
and mechanisms are failing then adjustment is necess
The ARM asm implementation of InternalMathSwapBytes64 () does
interesting things if bit 7 of operand r1 (upper 32 bits of the
input value) is set. After the recursive swap, bit 7 ends up in
the sign bit position, after which it is right shifted with sign
extension, and or'ed with the upper half of
On 4 March 2015 at 10:21, Gao, Liming wrote:
> Ard:
> Got your point. I have no other comments. Your patch is fine to me. If you
> require me to commit this change, please let me know.
>
> Reviewed-by: Liming Gao
>
Thank you.
Could you please go ahead and commit the patches?
Regards,
Ard.
Ard:
Got your point. I have no other comments. Your patch is fine to me. If you
require me to commit this change, please let me know.
Reviewed-by: Liming Gao
Thanks
Liming
-Original Message-
From: Ard Biesheuvel [mailto:ard.biesheu...@linaro.org]
Sent: Wednesday, March 04, 2015
On 02/06/15 14:52, Leif Lindholm wrote:
> Some toolchains, at least Fedora GCC, generate inline unwind tables in
> object files. These confuses GenFw to no end, leading to build failures:
> GenFw: ERROR 3000: Invalid WriteSections64(): ...
> unsupported ELF EM_AARCH64 relocation 0x105.
>
Chen,
Volatile is to tell compiler this variable maybe changed by other threads.
Compiler must access the value from memory instead of the old value in its
registers. This is different with spinlock that is to protect shared memory
between multiple threads.
For this case, CpuState may be up
35 matches
Mail list logo
