[webkit-reviews] review requested: [Bug 169611] WebAssembly: store state in TLS instead of on VM : [Attachment 305326] patch

[bugzilla-daemon at webkit.org]

JF Bastien <jfbastien at apple.com> has asked  for review:
Bug 169611: WebAssembly: store state in TLS instead of on VM
https://bugs.webkit.org/show_bug.cgi?id=169611

Attachment 305326: patch

https://bugs.webkit.org/attachment.cgi?id=305326&action=review




--- Comment #33 from JF Bastien <jfbastien at apple.com> ---
Created attachment 305326

  --> https://bugs.webkit.org/attachment.cgi?id=305326&action=review

patch

I talked with ggaren@ since bmalloc has the same issue. This is an acceptable
limitation of EWS because we also have internal bots which perform the build
and tests with the machdep.h header available.

The downside is that this code isn't tested by EWS and isn't available to open
source contributors. They exercise a different, slower, code path.

We do have options to "fix" the EWS situation:
 - Use the internal SDK on EWS. This means anyone with bot access can leak the
content, we can then restrict it but... eh.
 - Copy the relevant bits to a fallback header in WebKit. This could diverge,
and copying code is ew.
 - Copy the entire header from libc releases such as [0]. This is tricky
because the license is different and we'd need to talk to a lawyer.

I'm not sure either option is worth it. For now ggaren@ agreed to just go with
the status quo that we already have.

[0]:
https://opensource.apple.com/source/Libc/Libc-583/pthreads/pthread_machdep.h