VecDeque: Use power of two capacity even for zero sized types

[bluss]

VecDeque: Use power of two capacity even for zero sized types

VecDeque depends on using a power of two capacity. Use the largest
possible power of two capacity for ZSTs.

Fixes #28488

[rust-highfive]

r? @nikomatsakis

(rust_highfive has picked a reviewer for you, use r? to override)

[bluss]

WIP -- still running tests, they probably don't compile..

[Gankro]

r=me when you get that to compile

[bluss]

It compiled with one fix.

[Gankro]

@bors r+

[bors]

📌 Commit 838160c has been approved by Gankro

[bluss]

beta nominating as bug fix

[eddyb]

We don't have an "is power of 2" method? IIRC x & (x - 1) == 0 works.

[bluss]

@eddyb Oh, we have is_power_of_two(). Didn't know that. It uses your method.

[bluss]

Updated to use is_power_of_two

[eddyb]

@bors r+

[bors]

📌 Commit 66f5dc1 has been approved by eddyb

[bors]

⌛️ Testing commit 66f5dc1 with merge 7e25e63...

[bors]

💔 Test failed - auto-linux-64-x-android-t

[bluss]

I'm not sure why the debuginfo test failed on android, but I think it's safe to retry.

@bors retry

[bors]

⌛️ Testing commit 66f5dc1 with merge 0b5b029...

[bors]

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-x-android-t, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-win-gnu-32-nopt-t, auto-win-gnu-32-opt, auto-win-gnu-64-nopt-t, auto-win-gnu-64-opt, auto-win-msvc-32-opt, auto-win-msvc-64-opt

[alexcrichton]

The libs team decided to accept this for a backport to beta

[bluss]

@alexcrichton Has a stable patch / 1.3.1 been considered? Maybe this is not so serious after all(?). I don't know, but I like fixing bugs.

[alexcrichton]

No, but this doesn't seem serious enough to warrant the first patch release in Rust ever.

[bluss]

Well maybe it's the worst regression in stable rust ever? We don't have much history to draw from, is my point.