apply the change from aws-c-auth

[TingDaoK]

submodules update:

aws-c-auth         v0.7.25 -> v0.7.27
aws-c-cal          v0.7.3 -> v0.7.4
aws-c-common       v0.9.27 ✓
aws-c-compression  v0.2.18 -> v0.2.19
aws-c-event-stream v0.4.3 ✓
aws-c-http         v0.8.8 ✓
aws-c-io           v0.14.18 ✓
aws-c-mqtt         v0.10.4 ✓
aws-c-s3           v0.6.4 ✓
aws-c-sdkutils     v0.1.19 ✓
aws-checksums      v0.1.18 ✓
aws-lc             v1.33.0 -> v1.34.2
s2n                v1.5.0 -> v1.5.1

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[TingDaoK]

added this for aws/aws-lc@6fe8dcb to work on manylinux-1-x86, it's not released yet, but we will need this once it's

[graebm]

Why doesn't AWS-LC's CMakeLists.txt handle this, vs having every consumer handle it?

[graebm]

Seems weird to include x86_64 and amd64, since those are never 32bit.
And if you take those out, it's all 32-bit architectures, so you don't need to compare pointer size

Suggested change

	if(CMAKE_SYSTEM_PROCESSOR_LOWER MATCHES "x86_64|amd64|x86|i386|i686" AND CMAKE_SIZEOF_VOID_P EQUAL 4)
	if(CMAKE_SYSTEM_PROCESSOR_LOWER MATCHES "^(x86|i386|i686)$")

[TingDaoK]

I think that aws-lc can still support machine without SSE2 with -DOPENSSL_NO_ASM to disable the assembly optimization. But, for us, we can assume we don't support 20 years old x86 CPUs.

[TingDaoK]

https://github.com/awslabs/aws-crt-python/actions/runs/10689757684/job/29632625853 line 429

-- XXXXXXXXX CMAKE_SYSTEM_PROCESSOR: x86_64

https://cmake.org/cmake/help/latest/variable/CMAKE_SYSTEM_PROCESSOR.html

this will correspond to the target architecture for the build, but this is not guaranteed. (E.g. on Windows, the host may be AMD64 even when using a MSVC cl compiler with a 32-bit target.)

So, I copied the logic from aws-lc. https://github.com/aws/aws-lc/blob/main/CMakeLists.txt#L791-L800
I'll copy the comments as well probably for clarity.

[graebm]

Oh yeah, I see in the CMAKE_SYSTEM_PROCESSOR docs that it may not be correct (emphasis mine):

In many cases, this will correspond to the target architecture for the build, but this is not guaranteed. (E.g. on Windows, the host may be AMD64 even when using a MSVC cl compiler with a 32-bit target.)

When cross-compiling, a CMAKE_TOOLCHAIN_FILE should set the CMAKE_SYSTEM_PROCESSOR variable to match target architecture that it specifies

Maybe edit the comment to be more like the docs, instead of some ramble about Docker.

Be wary that CMAKE_SYSTEM_PROCESSOR may not correspond to the target architecture when cross-compiling. (E.g. on Windows, the host may be AMD64 even when using a MSVC cl compiler with a 32-bit target). The CMAKE_TOOLCHAIN_FILE is supposed to set it correctly, but may not have.

[graebm]

or just

CMAKE_SYSTEM_PROCESSOR is supposed to match the target architecture when cross-compiling, but this is not guaranteed. See: https://cmake.org/cmake/help/v3.30/variable/CMAKE_SYSTEM_PROCESSOR.html

[TingDaoK]

aws/aws-lc#1841