overflowing-checking for rhs of shift operators by pnkfelix · Pull Request #23536 · rust-lang/rust

pnkfelix · 2015-03-19T23:37:32Z

overflow-checking for rhs of shift operators

Subtask of #22020 (RFC 560)

rust-highfive · 2015-03-19T23:37:43Z

r? @Aatch

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

pnkfelix · 2015-03-19T23:49:00Z

cc @nikomatsakis

pnkfelix · 2015-03-19T23:53:42Z

ah, this whole sub-block is debugging code that is no longer needed; I will remove it.

pnkfelix · 2015-03-20T08:03:17Z

One of the tests in this PR uncovered #23551, so don't attempt to merge this until that has a fix (or until I revise the test to sidestep that bug).

This includes a slight refactoring of the `cast_shift_rhs` and related functions in `trans::base`, so that I can call them from much later in the compiler's control flow (so that we can clearly dilineate where automatic conversions of the RHS occur, versus where we check it). The rhs-checking and fallback-masking is generalized to 8- and 16-bit values, and the fallback-masking is turned on unconditionally. Fixrust-lang#10183. Is this a [breaking-change]? I would argue it is not; it only adds a strict definition to what was previously undefined behavior; however, there might be code that was e.g. assuming that `1_i8 << 17` yields 0. (This happens in certain contexts and at certain optimization levels.)

Note the tests have been revised to match new semantics for 8- and 16-bit values.

pnkfelix · 2015-03-20T10:29:09Z

Okay, after discovering that #23551 was a duplicate of #10183 and reading the discussion there further, I have decided to go with @glaebhoerl's advice and ensure that we do not encounter undefined behavior due to the shift's RHS.

This PR has now been updated accordingly, and is ready for review/merge.

nikomatsakis · 2015-03-20T14:37:04Z

r? @nikomatsakis

nikomatsakis · 2015-03-20T14:37:40Z

r+ -- This looks good to me. I had two questions though so not informing bors just yet (though no action is needed or requested necessarily).

pnkfelix · 2015-03-20T14:47:10Z

@nikomatsakis constant evaluation is not yet addressed.

I don't know whether we have a reliable way to force the debug-assertions off, though that sounds like a good thing to have if we do not already have it. I will investigate.

(Still, we probably can land this and leave those as to-do items. In particular, we need to deal with const-evaluation for every operation; not just shifts, and that still remains to be done.)

pnkfelix · 2015-03-21T12:27:04Z

@bors r=nikomatsakis 61ff823

bors · 2015-03-21T13:05:14Z

⌛ Testing commit 61ff823 with merge c9798cb...

bors · 2015-03-21T14:32:01Z

💔 Test failed - auto-mac-64-opt

pnkfelix · 2015-03-21T16:46:38Z

odd, thought I had done a full test ... or no, maybe I only did make check and that does not cover these pretty print tests.

…etty-rpass.

pnkfelix · 2015-03-21T16:58:52Z

(the bug I'm referring to in the commit is #20937 )

pnkfelix · 2015-03-21T17:00:22Z

@bors r=nikomatsakis 5e47c66

…atsakis overflow-checking for rhs of shift operators Subtask of rust-lang#22020 ([RFC 560](https://github.com/rust-lang/rfcs/blob/master/text/0560-integer-overflow.md))

bors · 2015-03-21T23:03:38Z

⌛ Testing commit 5e47c66 with merge 1c9b5a6...

bors · 2015-03-21T23:07:44Z

💔 Test failed - auto-mac-64-nopt-t

bors · 2015-03-23T06:26:44Z

💔 Test failed - auto-win-32-nopt-t

pnkfelix · 2015-03-23T11:51:38Z

failures:
[run-pass] run-pass/over-constrained-vregs.rs

hmm, I guess I need to pass --disable-optimize-tests to catch cases like this. (Or manually recreate the effect of that flag locally, to save myself some time...)

Update: Oh, it also needs a 32-bit build. Hmm.

pnkfelix · 2015-03-23T16:07:09Z

@bors r=nikomatsakis bb9d210

pnkfelix · 2015-03-23T20:07:05Z

@bors r=nikomatsakis bb9d210

bors · 2015-03-23T20:07:06Z

🙀 bb9d210 is not a valid commit SHA. Please try again with 97c8444.

pnkfelix · 2015-03-23T20:07:48Z

@bors r=nikomatsakis bb9d210

bors · 2015-03-23T20:07:48Z

🙀 bb9d210c99ab248e81598d70c39c3968ab9d09eb is not a valid commit SHA. Please try again with 97c8444.

pnkfelix · 2015-03-23T20:48:03Z

@bors r=nikomatsakis bb9d210

pnkfelix · 2015-03-23T20:55:57Z

@bors p=1

bors · 2015-03-23T21:07:51Z

⌛ Testing commit bb9d210 with merge 9c638f8...

bors · 2015-03-23T21:26:03Z

⛄ The build was interrupted to prioritize another pull request.

bors · 2015-03-23T21:38:34Z

⌛ Testing commit bb9d210 with merge b93456c...

bors · 2015-03-23T21:42:12Z

⛄ The build was interrupted to prioritize another pull request.

bors · 2015-03-23T22:43:40Z

⌛ Testing commit bb9d210 with merge 28a0b25...

overflow-checking for rhs of shift operators Subtask of #22020 ([RFC 560](https://github.com/rust-lang/rfcs/blob/master/text/0560-integer-overflow.md))

overflow-checking for rhs of shift operators Subtask of rust-lang#22020 ([RFC 560](https://github.com/rust-lang/rfcs/blob/master/text/0560-integer-overflow.md))

bors · 2015-03-24T01:19:39Z

💔 Test failed - auto-win-32-nopt-t

pnkfelix · 2015-03-24T01:22:47Z

well, here's hoping the rollup suceeds where this latest merge attempt failed...

pnkfelix · 2015-03-24T10:16:25Z

@bors retry

bors · 2015-03-24T11:16:41Z

⚡ Previous build results for 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-32-opt, auto-win-64-nopt-t, auto-win-64-opt are reusable. Rebuilding only auto-win-32-nopt-t...

bors · 2015-03-24T13:21:59Z

☀️ 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-32-nopt-t, auto-win-32-opt, auto-win-64-nopt-t, auto-win-64-opt

rust-highfive assigned Aatch Mar 19, 2015

aturon mentioned this pull request Mar 19, 2015
Tracking issue for Integer Overflow (RFC 560) #22020
Closed
15 tasks

pnkfelix mentioned this pull request Mar 19, 2015
Inaccuracies in the reference about bitshifts and overflows #23421
Closed

pnkfelix reviewed Mar 19, 2015
View reviewed changes

pnkfelix mentioned this pull request Mar 20, 2015
overflowing left shift can lead to pseudo-byte with hidden bits #23551
Closed

pnkfelix added 2 commits March 20, 2015 11:25

Test suite for overflowing shift operators.
61ff823
Note the tests have been revised to match new semantics for 8- and 16-bit values.

pnkfelix force-pushed the arith-oflo-shifts branch from 1beaee8 to 61ff823Compare March 20, 2015 10:26

rust-highfive assigned nikomatsakis and unassigned Aatch Mar 20, 2015

workaround bugs in pretty-printer so that we can pass check-stage2-pr…
5e47c66
…etty-rpass.

Fix shift-overflow in very old run-pass test.
bb9d210

pnkfelix closed this Mar 23, 2015

pnkfelix reopened this Mar 23, 2015

pnkfelix force-pushed the arith-oflo-shifts branch from 7623b75 to bb9d210Compare March 23, 2015 20:47

bors added a commit that referenced this pull request Mar 23, 2015
Auto merge of#23536- pnkfelix:arith-oflo-shifts, r=nikomatsakis
28a0b25
overflow-checking for rhs of shift operators Subtask of #22020 ([RFC 560](https://github.com/rust-lang/rfcs/blob/master/text/0560-integer-overflow.md))

bors merged commit bb9d210 into rust-lang:masterMar 24, 2015

bors mentioned this pull request Mar 24, 2015
Rolling up PRs in the queue #23654
Merged

Uh oh!

Uh oh!

Conversation

pnkfelix commented Mar 19, 2015

Uh oh!

rust-highfive commented Mar 19, 2015

Uh oh!

pnkfelix commented Mar 19, 2015

Uh oh!

pnkfelixMar 19, 2015

Choose a reason for hiding this comment

Uh oh!

pnkfelix commented Mar 20, 2015

Uh oh!

pnkfelix commented Mar 20, 2015

Uh oh!

nikomatsakis commented Mar 20, 2015

Uh oh!

nikomatsakis commented Mar 20, 2015

Uh oh!

pnkfelix commented Mar 20, 2015

Uh oh!

pnkfelix commented Mar 21, 2015

Uh oh!

bors commented Mar 21, 2015

Uh oh!

bors commented Mar 21, 2015

Uh oh!

pnkfelix commented Mar 21, 2015

Uh oh!

pnkfelix commented Mar 21, 2015

Uh oh!

pnkfelix commented Mar 21, 2015

Uh oh!

bors commented Mar 21, 2015

Uh oh!

bors commented Mar 21, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

pnkfelix commented Mar 23, 2015

Uh oh!

pnkfelix commented Mar 23, 2015

Uh oh!

pnkfelix commented Mar 23, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

pnkfelix commented Mar 23, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

pnkfelix commented Mar 23, 2015

Uh oh!

pnkfelix commented Mar 23, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

bors commented Mar 23, 2015

Uh oh!

bors commented Mar 24, 2015

Uh oh!

pnkfelix commented Mar 24, 2015

Uh oh!

pnkfelix commented Mar 24, 2015

Uh oh!

bors commented Mar 24, 2015

Uh oh!

bors commented Mar 24, 2015

Uh oh!

Reviewers

Assignees

Labels

Projects