Endless hang during stage1 when removing drop block from core::task::mod

[bstrie]

The AllowFailure struct in core::task::mod. Hangs infinitely while compiling stage1 libcore, with 0% CPU utilization from rustc.

[brson]

I can still reproduce this, but will probably work around it by using finally, which is a better pattern here than creating new types anyway.

[brson]

After that there is just one more use of drop in pipes.

[graydon]

assigning bug; change assignment if you disagree

[alexcrichton]

This looks like it was fixed in c1fdace