[PATCH] cpukit/score/aarch64: Use correct MAIR index

2023-08-13 Thread Kinsey Moore
The MAIR index currently assigned (1) for uncached memory segments is
not configured properly for this purpose. Instead, this switches
uncached memory segment flags to MAIR index 2 which is properly
configured for uncached inner and outer shareable domains.
---
 cpukit/score/cpu/aarch64/include/libcpu/mmu-vmsav8-64.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/cpukit/score/cpu/aarch64/include/libcpu/mmu-vmsav8-64.h 
b/cpukit/score/cpu/aarch64/include/libcpu/mmu-vmsav8-64.h
index 0d65004f88..ca9b60e6d1 100644
--- a/cpukit/score/cpu/aarch64/include/libcpu/mmu-vmsav8-64.h
+++ b/cpukit/score/cpu/aarch64/include/libcpu/mmu-vmsav8-64.h
@@ -70,7 +70,7 @@ extern "C" {
 #define AARCH64_MMU_CODE_RW_CACHED AARCH64_MMU_DATA_RW_CACHED
 
 #define AARCH64_MMU_DATA_RO \
-  ( AARCH64_MMU_FLAGS_BASE | MMU_DESC_MAIR_ATTR( 1 ) | MMU_DESC_WRITE_DISABLE )
+  ( AARCH64_MMU_FLAGS_BASE | MMU_DESC_MAIR_ATTR( 2 ) | MMU_DESC_WRITE_DISABLE )
 #define AARCH64_MMU_CODE AARCH64_MMU_DATA_RO
 #define AARCH64_MMU_CODE_RW AARCH64_MMU_DATA_RW
 
@@ -78,7 +78,7 @@ extern "C" {
 #define AARCH64_MMU_DATA_RW_CACHED \
   ( AARCH64_MMU_FLAGS_BASE | MMU_DESC_MAIR_ATTR( 3 ) )
 #define AARCH64_MMU_DATA_RW \
-  ( AARCH64_MMU_FLAGS_BASE | MMU_DESC_MAIR_ATTR( 1 ) )
+  ( AARCH64_MMU_FLAGS_BASE | MMU_DESC_MAIR_ATTR( 2 ) )
 #define AARCH64_MMU_DEVICE ( AARCH64_MMU_FLAGS_BASE | MMU_DESC_MAIR_ATTR( 0 ) )
 
 rtems_status_code aarch64_mmu_map(
-- 
2.39.2

___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel


Re: [PATCH RTEMS] spec: Add -mstrict-align to mvme2100 default build

2023-08-13 Thread Chris Johns
On 11/8/2023 1:17 pm, Vijay Kumar Banerjee wrote:
> On Wed, Aug 9, 2023 at 8:04 PM Chris Johns  wrote:
>>
>> On 10/8/2023 2:41 am, Joel Sherrill wrote:
>>> Reading the EPICS discussion, I wonder if this should be added for all
>>> motorola_powerpc BSP variants.
>>>
>>> Gedare/Chris/anyone with another board in the family? What do you think?
>>
>> Michael asked if the setting is present when building libc, libm etc in the
>> EPICS core devs meeting today and I could not answer the question?
> 
> I just built the tools for powerpc and it looks like libc, libm are
> not built with the -mstrict-align flag.

Is it a problem to not have option? I am do not know enough about this area of
the PowerPC to know.

>> Is it
>> exported in the .pc or Makefile.inc flags?
>>
> This flag is exported in ${RTEMS_BSP}.cfg file in RTEMS, which gets
> included by EPICS though CONFIG.Common.RTEMS:
> https://github.com/epics-base/epics-base/blob/7.0/configure/os/CONFIG.Common.RTEMS#L38
> 
> The `RTEMS_CUSTOM` variable is defined in Makefile.inc, and it points
> to the ${RTEMS_BSP}.cfg file

RTEMS has not needed this flag and libc and libm do not have it on so I am
confused why code in EPICS has triggering something that needs this option? I
would have thought it is needed everywhere or not at all?

Chris
___
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel