Update libjpeg_turbo to 1.4.90 from https://github.com/libjpeg-turbo/

Update libjpeg_turbo to 1.4.90 from https://github.com/libjpeg-turbo/

- Updated configs files for libjpeg_turbo-1.4.90
./configure --without-mem-srcdst --without-arith-enc --without-arith-dec

- Merge tag '1.4.90'

- Did a git-merge and then resolved one conflict
  Conflicts:
	win/jsimdcfg.inc

Signed-off-by: Khasim Syed Mohammed <khasim.mohammed@linaro.org>;


Win/x64: Fix improper callee save of xmm8-xmm11
Bump TurboJPEG C API revision to 1.5
ChangeLog: Mention jpeg_crop_scanline() function
Merge branch '1.4.x'
Fix memory leak when running tjunittest -yuv
Fix v7/v8-compatible build
libjpeg API: Partial scanline decompression
Build: Make the NASM autoconf variable persistent
Use consistent/modern code formatting for dbl ptrs
Use consistent/modern code formatting for pointers
Reorder copyright messages in cjpeg/djpeg/jpegtran
Win: Display effective C/LD flags in CMake output
Java: Avoid OOM error when running 'make test'
ARM64: Avoid tbl instruction on Cortex-A53/A57
ChangeLog: Mention ARM64 Huffman & adj perf claims
Fix iOS/ARMv8 build
Fix 'make dist'; Include LICENSE.md in packages
ARM64: Avoid LD3/ST3 at run time, not compile time
ARM64 NEON SIMD implementation of Huffman encoding
Merge branch '1.4.x'
Merge branch '1.3.x' into 1.4.x
Merge branch '1.2.x' into 1.3.x
Update Android build instr. for ARMv8, PIE, etc.
Optimize ARM64 SIMD code for Cavium ThunderX
Complete the ARM64 NEON SIMD implementation
... And a lot more ...


BUG : http://crbug.com/539781

Signed-off-by: Bernhard Rosenkränzer <bero@linaro.org>;

BUG=539781
R=thakis@chromium.org
R=noel@chromium.org

[msarett, 2016-05-02 13:29:12]

msarett@google.com changed reviewers:
+ msarett@google.com

[msarett, 2016-05-02 13:29:15]

Thanks for taking this on!

"- Did a git-merge and then resolved one conflict
  Conflicts:
	win/jsimdcfg.inc"

I believe that win/jsimdcfg.inc, linux/jsimdcfg.inc, and
mac/jsimdcfg.h are all identical.  Can we match upstream
and drop the platform specific directories?

"There was conflict, have resolved it in separate patch
Copied the config.h and jconfig from public project and included this in
separate patch."

Where is this patch?  It looks like changes to config.h and
jconfig.h are included in this patch (which is fine).

"- Updated configs files for libjpeg_turbo-1.4.90
./configure --without-mem-srcdst --without-arith-enc --without-arith-dec"

This may not be the right way to generate the config files,
as chrome needs to be set up to build on many platforms
(not just the particular one that we generate the config
files for).

May be easier just to manually update jconfig.h and
jconfigint.h.

Also, does running ./configure generate a bunch of extra
files that we don't want to check in?

------------------------------------------

I realize that this is a lot to deal with...  Do you want
me to take on this patch?

https://codereview.chromium.org/1934113002/diff/1/.gitignore
File .gitignore (right):

https://codereview.chromium.org/1934113002/diff/1/.gitignore#newcode1
.gitignore:1: .DS_Store
I doubt we want to check most of this in.

https://codereview.chromium.org/1934113002/diff/1/jconfig.h
File jconfig.h (left):

https://codereview.chromium.org/1934113002/diff/1/jconfig.h#oldcode20
jconfig.h:20: #define HAVE_PROTOTYPES 1
I can confirm that HAVE_PROTOTYPES was dropped.
https://github.com/libjpeg-turbo/libjpeg-turbo/commit/bc56b754e1a6a1db9ccadf6...

https://codereview.chromium.org/1934113002/diff/1/jconfig.h
File jconfig.h (right):

https://codereview.chromium.org/1934113002/diff/1/jconfig.h#newcode28
jconfig.h:28: #define HAVE_LOCALE_H 1
This seems fine.  Here is where HAVE_LOCALE_H was added.
https://github.com/libjpeg-turbo/libjpeg-turbo/commit/ff6961f3d22060adc13cc3e...

https://codereview.chromium.org/1934113002/diff/1/jconfig.h#newcode53
jconfig.h:53: #define NEED_SYS_TYPES_H 1
Why did this change?

https://codereview.chromium.org/1934113002/diff/1/jconfigint.h
File jconfigint.h (right):

https://codereview.chromium.org/1934113002/diff/1/jconfigint.h#newcode15
jconfigint.h:15: #define SIZEOF_SIZE_T 8
Is chrome always compiled on 64-bit platforms?  Otherwise this won't be correct.

https://codereview.chromium.org/1934113002/diff/1/simd/jccolext-sse2-64.asm
File simd/jccolext-sse2-64.asm (right):

https://codereview.chromium.org/1934113002/diff/1/simd/jccolext-sse2-64.asm#n...
simd/jccolext-sse2-64.asm:2: ; jccolext.asm - colorspace conversion (64-bit
SSE2)
I believe that all of these "Adds" are actually renames.  We should be deleting
the files with the old names and modifying the gyp to build these new files.

[khasim.mohammed, 2016-05-02 14:25:51]

On 2 May 2016 at 18:59, <msarett@google.com> wrote:

> Thanks for taking this on!
>
> Thanks.


> "- Did a git-merge and then resolved one conflict
> Conflicts:
> win/jsimdcfg.inc"
>
> I believe that win/jsimdcfg.inc, linux/jsimdcfg.inc, and
> mac/jsimdcfg.h are all identical. Can we match upstream
> and drop the platform specific directories?
>
> Possible.


> "There was conflict, have resolved it in separate patch
> Copied the config.h and jconfig from public project and included this in
> separate patch."
>
> Where is this patch? It looks like changes to config.h and
> jconfig.h are included in this patch (which is fine).
>
> It's actually the win/jsimdcfg.inc which was conflicting.


> "- Updated configs files for libjpeg_turbo-1.4.90
> ./configure --without-mem-srcdst --without-arith-enc --without-arith-dec"
>
> This may not be the right way to generate the config files,
> as chrome needs to be set up to build on many platforms
> (not just the particular one that we generate the config
> files for).
>
> May be easier just to manually update jconfig.h and
> jconfigint.h.
>
> ok, but for which platform should we manually update ?

> Also, does running ./configure generate a bunch of extra
> files that we don't want to check in?
>
> Yes, it does generate extra files, but I didn't find it useful for build
or boot purposes they were information only.


> ------------------------------------------
>
> I realize that this is a lot to deal with... Do you want
> me to take on this patch?
>
> You mean, you will merge the mainline libjpeg-turbo to chromium directly ?
Yes, that would definitely help.

Let me know if anything I need to do here.


> https://codereview.chromium.org/1934113002/diff/1/.gitignore
> File .gitignore (right):
>
> https://codereview.chromium.org/1934113002/diff/1/.gitignore#newcode1
> .gitignore:1: .DS_Store
> I doubt we want to check most of this in.
>
> https://codereview.chromium.org/1934113002/diff/1/jconfig.h
> File jconfig.h (left):
>
> https://codereview.chromium.org/1934113002/diff/1/jconfig.h#oldcode20
> jconfig.h:20: #define HAVE_PROTOTYPES 1
> I can confirm that HAVE_PROTOTYPES was dropped.
>
>
https://github.com/libjpeg-turbo/libjpeg-turbo/commit/bc56b754e1a6a1db9ccadf6...
>
> https://codereview.chromium.org/1934113002/diff/1/jconfig.h
> File jconfig.h (right):
>
> https://codereview.chromium.org/1934113002/diff/1/jconfig.h#newcode28
> jconfig.h:28: #define HAVE_LOCALE_H 1
> This seems fine. Here is where HAVE_LOCALE_H was added.
>
>
https://github.com/libjpeg-turbo/libjpeg-turbo/commit/ff6961f3d22060adc13cc3e...
>
> https://codereview.chromium.org/1934113002/diff/1/jconfig.h#newcode53
> jconfig.h:53: #define NEED_SYS_TYPES_H 1
> Why did this change?
>
> https://codereview.chromium.org/1934113002/diff/1/jconfigint.h
> File jconfigint.h (right):
>
> https://codereview.chromium.org/1934113002/diff/1/jconfigint.h#newcode15
> jconfigint.h:15: #define SIZEOF_SIZE_T 8
> Is chrome always compiled on 64-bit platforms? Otherwise this won't be
> correct.
>
> https://codereview.chromium.org/1934113002/diff/1/simd/jccolext-sse2-64.asm
> File simd/jccolext-sse2-64.asm (right):
>
>
>
https://codereview.chromium.org/1934113002/diff/1/simd/jccolext-sse2-64.asm#n...
> simd/jccolext-sse2-64.asm:2: ; jccolext.asm - colorspace conversion
> (64-bit SSE2)
> I believe that all of these "Adds" are actually renames. We should be
> deleting the files with the old names and modifying the gyp to build
> these new files.
>
> https://codereview.chromium.org/1934113002/
>

Thanks.

Regards,
Khasim

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Noel Gordon, 2016-05-05 03:53:22]

We can close here I think.

[Noel Gordon, 2016-05-05 03:53:42]

Description was changed from

==========
Update libjpeg_turbo to 1.4.90 from https://github.com/libjpeg-turbo/

- Updated configs files for libjpeg_turbo-1.4.90
./configure --without-mem-srcdst --without-arith-enc --without-arith-dec

- Merge tag '1.4.90'

- Did a git-merge and then resolved one conflict
  Conflicts:
	win/jsimdcfg.inc

Signed-off-by: Khasim Syed Mohammed <khasim.mohammed@linaro.org>


Win/x64: Fix improper callee save of xmm8-xmm11
Bump TurboJPEG C API revision to 1.5
ChangeLog: Mention jpeg_crop_scanline() function
Merge branch '1.4.x'
Fix memory leak when running tjunittest -yuv
Fix v7/v8-compatible build
libjpeg API: Partial scanline decompression
Build: Make the NASM autoconf variable persistent
Use consistent/modern code formatting for dbl ptrs
Use consistent/modern code formatting for pointers
Reorder copyright messages in cjpeg/djpeg/jpegtran
Win: Display effective C/LD flags in CMake output
Java: Avoid OOM error when running 'make test'
ARM64: Avoid tbl instruction on Cortex-A53/A57
ChangeLog: Mention ARM64 Huffman & adj perf claims
Fix iOS/ARMv8 build
Fix 'make dist'; Include LICENSE.md in packages
ARM64: Avoid LD3/ST3 at run time, not compile time
ARM64 NEON SIMD implementation of Huffman encoding
Merge branch '1.4.x'
Merge branch '1.3.x' into 1.4.x
Merge branch '1.2.x' into 1.3.x
Update Android build instr. for ARMv8, PIE, etc.
Optimize ARM64 SIMD code for Cavium ThunderX
Complete the ARM64 NEON SIMD implementation
... And a lot more ...


BUG : http://crbug.com/539781

Signed-off-by: Bernhard Rosenkränzer <bero@linaro.org>;

BUG=539781
R=thakis@chromium.org
R=noel@chromium.org
==========

to

==========
Update libjpeg_turbo to 1.4.90 from https://github.com/libjpeg-turbo/

- Updated configs files for libjpeg_turbo-1.4.90
./configure --without-mem-srcdst --without-arith-enc --without-arith-dec

- Merge tag '1.4.90'

- Did a git-merge and then resolved one conflict
  Conflicts:
	win/jsimdcfg.inc

Signed-off-by: Khasim Syed Mohammed <khasim.mohammed@linaro.org>


Win/x64: Fix improper callee save of xmm8-xmm11
Bump TurboJPEG C API revision to 1.5
ChangeLog: Mention jpeg_crop_scanline() function
Merge branch '1.4.x'
Fix memory leak when running tjunittest -yuv
Fix v7/v8-compatible build
libjpeg API: Partial scanline decompression
Build: Make the NASM autoconf variable persistent
Use consistent/modern code formatting for dbl ptrs
Use consistent/modern code formatting for pointers
Reorder copyright messages in cjpeg/djpeg/jpegtran
Win: Display effective C/LD flags in CMake output
Java: Avoid OOM error when running 'make test'
ARM64: Avoid tbl instruction on Cortex-A53/A57
ChangeLog: Mention ARM64 Huffman & adj perf claims
Fix iOS/ARMv8 build
Fix 'make dist'; Include LICENSE.md in packages
ARM64: Avoid LD3/ST3 at run time, not compile time
ARM64 NEON SIMD implementation of Huffman encoding
Merge branch '1.4.x'
Merge branch '1.3.x' into 1.4.x
Merge branch '1.2.x' into 1.3.x
Update Android build instr. for ARMv8, PIE, etc.
Optimize ARM64 SIMD code for Cavium ThunderX
Complete the ARM64 NEON SIMD implementation
... And a lot more ...


BUG : http://crbug.com/539781

Signed-off-by: Bernhard Rosenkränzer <bero@linaro.org>;

BUG=539781
R=thakis@chromium.org
R=noel@chromium.org
==========

[Noel Gordon, 2016-05-05 03:53:42]

noel@chromium.org changed reviewers:
- noel@chromium.org