Update stacker and psm crates

[mati865]

Primarily to include rust-lang/stacker#54

[rust-highfive]

r? @Mark-Simulacrum

(rust-highfive has picked a reviewer for you, use r? to override)

[mati865]

cc @nagisa

[Mark-Simulacrum]

@bors r+

[bors]

📌 Commit 7954a76 has been approved by Mark-Simulacrum

[JohnTitor]

Failed in rollup: #87714 (comment)
@bors r-
Seems we need some tweaks to build the crate.

[mati865]

Updated to include rust-lang/stacker@948e604

[klensy]

Now this updates some other crates too.

[mati865]

Rust-analyzer updated them while I was committing, fixed now.

[Mark-Simulacrum]

@bors r+

[bors]

📌 Commit e70c1ab has been approved by Mark-Simulacrum

[bors]

⌛ Testing commit e70c1ab with merge 80b7d6cf083e0a0ba5bd9c1f59ab95d7ca77692a...

[bors]

💔 Test failed - checks-actions

[Mark-Simulacrum]

@bors retry libpsm-b2b57a2c9f3f7c91.rlib(x86_windows_gnu.o) : error LNK2026: module unsafe for SAFESEH image.

Maybe spurious? Could be an actual problem. Not sure.

[nagisa]

99.9% is not spurious but im still perplexed why is it coming up now.

…

On Thu, 19 Aug 2021, 22:24 Mark Rousskov, ***@***.***> wrote: @bors <https://github.com/bors> retry libpsm-b2b57a2c9f3f7c91.rlib(x86_windows_gnu.o) : error LNK2026: module unsafe for SAFESEH image. Maybe spurious? Could be an actual problem. Not sure. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#87680 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAFFZUSSAJKJ5QAADCBPZADT5VK7FANCNFSM5BLA5VXA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

[mati865]

Definitely not spurious.
Before rust-lang/stacker@e1dc224#diff-af3494399696f0e51254641cc231eda98f371372b92fb4f453e43d7f8229b08c MSVC env would always use MSVC assembly files(even for Clang), now it uses GNU assembly for Clang.

[nagisa]

That's surprising if true – clang-cl does not understand MASM, so it couldn't possibly have compiled the MSVC-specific files. And yet with CC set to clang-cl we can't possibly be falling back to regular MSVC's cl either.

[mati865]

I have no experience with MSVC but maybe Clang-cl will automatically fallback to MASM?

[camsteffen]

@bors r- (per #88168 (comment))

[nagisa]

So. I was able to reproduce this. Firstly, this only affects i686-pc-windows-msvc as far as I can tell. Secondly, this is caused by a combination of #31579 (which seems to be "just because") and use of clang-cl for assembling (which does not appear to support outputting objects that are compatible with this flag). In particular the psm's assembly contains no exception handlers at all, so it has worked “by default” when using ml to assemble.

clang-cl does not support transparent fallback to ml when encountering masm files, so I'm still not all that sure how we managed to avoid hitting issues with this before now, though. We must have been using ml at all times in the past, at least as far as msvc target is concerned. This is confusing.

[nagisa]

I adjusted psm again and now actually tested that it works to build on i686-pc-windows-msvc with CC=clang-cl set.

[mati865]

Thank you, I have just returned today and updated the PR.

[nagisa]

@bors r=Mark-Simulacrum,nagisa

[bors]

📌 Commit f58289c has been approved by Mark-Simulacrum,nagisa

[bors]

⌛ Testing commit f58289c with merge 6cfa773...

[bors]

☀️ Test successful - checks-actions
Approved by: Mark-Simulacrum,nagisa
Pushing 6cfa773 to master...