On 10 September 2015 at 11:45, Gao, Liming <[email protected]> wrote:
> Thanks. I am fine to GenFw tool change.
>

Thanks Liming.

Any comments on the DebugRVA patch I sent out yesterday?

Regards,
Ard.

> -----Original Message-----
> From: Ard Biesheuvel [mailto:[email protected]]
> Sent: Thursday, September 10, 2015 5:44 PM
> To: Gao, Liming
> Cc: [email protected]; [email protected]; Liu, Yingke D; 
> [email protected]; [email protected]
> Subject: Re: [edk2] [PATCH v3 0/4] enable Properties Table mem protection for 
> ARM
>
> On 10 September 2015 at 11:32, Gao, Liming <[email protected]> wrote:
>> Ard:
>>   Does GenFw tool patch impact X86 arch?
>>
>
> No, it does not.
>
> Patch #1 just removes some dead code from Elf64Convert.c (since EM_ARM is 
> only used for 32 bit code)
>
> Patch #2 changes three things in Elf32Convert.c:
> a) it removes a special case for EM_ARM
> b) it changes the handling of some ARM/RVCT specific condition from giving an 
> error to working around it
> c) it adds a sanity check to some ARM specific relative relocations
>
> Since 2a makes it more likely that ARM/RVCT users will see the error in 2b, 
> the work around is added. Since the workaround may affect ARM relative symbol 
> references, the sanity check in 2c is added.
>
> For IA32 users, there is no change.
>
> --
> Ard.
>
>
>
>> -----Original Message-----
>> From: edk2-devel [mailto:[email protected]] On Behalf Of Ard 
>> Biesheuvel
>> Sent: Thursday, September 10, 2015 4:55 PM
>> To: [email protected]; [email protected]; Liu, Yingke D
>> Cc: [email protected]; [email protected]; Gao, Liming; Ard Biesheuvel
>> Subject: [edk2] [PATCH v3 0/4] enable Properties Table mem protection for ARM
>>
>> This series implements 4 KB section alignment for ARM, which is required for 
>> supporting the Properties Table memory protection feature on this 
>> architecture.
>>
>> Patch #1 is a simple cleanup from the copy/paste that created Elf64Convert.c
>>
>> Patch #2 changes the PE/COFF .data alignment to adhere to the global section 
>> and file alignment. Not doing so means the resulting binary violates the 
>> PE/COFF spec, but special care is needed to correctly handle ARM RVCT 
>> generated ELF binaries which rely on the misalignment of some sections being 
>> preserved.
>>
>> NOTE: this patch adds some special handling to the ELF input sections, but 
>> only to a case that was formerly flagged as an error (i.e., section is 
>> misaligned wrt its own alignment). This means the risk for regressions is 
>> minimal, except for RVCT users.
>>
>> Patch #3 switches to the new GCC linker scripts for ARM
>>
>> Patch #4 sets the 4 KB alignment linker command line option to ensure .text 
>> and .data can be represented by separate memory regions, allowing their 
>> permissions to be set to R-X and RW- respectively.
>>
>> Changes since v2:
>> - removed one additional instance of the RVCT check in #1
>> - added Michael's Tested-by to all patches except #1 (since he tested 32-bit 
>> ARM
>>   only)
>> - fixed the RVCT case in patch #2 (untested on RVCT)
>>
>> Changes since v1:
>> - rebased onto latest GCC linker script changes
>> - added Laszlo's ack to #4
>>
>> Ard Biesheuvel (4):
>>   BaseTools/GenFw: remove ARM and RVCT references from ELF64 code
>>   BaseTools/GenFw: make ARM's .data adhere to PE/COFF section alignment
>>   BaseTools/ARM: move to unified GCC linker script
>>   ArmVirtPkg: use 4 KB section alignment for ARM DXE_RUNTIME modules
>>
>>  ArmVirtPkg/ArmVirt.dsc.inc              |  3 +-
>>  BaseTools/Conf/tools_def.template       | 12 +++-
>>  BaseTools/Source/C/GenFw/Elf32Convert.c | 67 +++++++++++++-------  
>> BaseTools/Source/C/GenFw/Elf64Convert.c | 23 +------
>>  4 files changed, 56 insertions(+), 49 deletions(-)
>>
>> --
>> 1.9.1
>>
>> _______________________________________________
>> edk2-devel mailing list
>> [email protected]
>> https://lists.01.org/mailman/listinfo/edk2-devel
>> _______________________________________________
>> edk2-devel mailing list
>> [email protected]
>> https://lists.01.org/mailman/listinfo/edk2-devel
_______________________________________________
edk2-devel mailing list
[email protected]
https://lists.01.org/mailman/listinfo/edk2-devel

Reply via email to