Upgrade rust to 58b83e7e740108611ad1e8286ee6b44ea5cbbb0f

[Manishearth]

Just edited the hash and ran ./mach update-cargo, plus some free warning fixes.

(Tue May 5 12:55:17 2015 +0000)

[hoppipolla-critic-bot]

Critic review: https://critic.hoppipolla.co.uk/r/4921

This is an external review system which you may optionally use for the code review of your pull request.

In order to help critic track your changes, please do not make in-place history rewrites (e.g. via git rebase -i or git commit --amend) when updating this pull request.

[SimonSapin]

@bors-servo r+

[bors-servo]

📌 Commit a33aae0 has been approved by SimonSapin

[Manishearth]

@bors-servo: p=1

[jdm]

@Manishearth The libc removal will break the android build. There's a PR in the queue that does it properly.

[Manishearth]

@bors-servo: r=SimonSapin

Undid the libc thingy

[bors-servo]

📌 Commit 43a1bc8 has been approved by SimonSapin

[bors-servo]

⌛ Testing commit 43a1bc8 with merge 407fe7a...

[bors-servo]

💔 Test failed - mac1

[Manishearth]

@bors-servo: r=SimonSapin

[bors-servo]

📌 Commit 715a119 has been approved by SimonSapin

[bors-servo]

⌛ Testing commit 715a119 with merge 43d61ab...

[bors-servo]

💔 Test failed - linux1

[Manishearth]

@bors-servo: retry

[Ms2ger]

@bors-servo: r-

../../tests/reftest.rs:88:9: 88:23 error: structure `test::TestOpts` has no field named `run_benchmarks`
../../tests/reftest.rs:88         run_benchmarks: false,
                                  ^~~~~~~~~~~~~~
../../tests/reftest.rs:88:9: 88:23 help: did you mean `bench_benchmarks`?
../../tests/reftest.rs:88         run_benchmarks: false,
                                  ^~~~~~~~~~~~~~

[Manishearth]

Oh, I thought it had crashed -- I saw a purple

[Manishearth]

@bors-servo: r=SimonSapin

[bors-servo]

📌 Commit 1cb1631 has been approved by SimonSapin

[bors-servo]

⌛ Testing commit 1cb1631 with merge 05213b9...

[bors-servo]

💔 Test failed - linux1

[Ms2ger]

http://build.servo.org/builders/mac2/builds/1874/steps/shell/logs/stdio

[Manishearth]

Opening wikipedia with gdb on gives me this:

#0  0x00005555571af3d8 in style::properties::cascade (viewport_size=..., applicable_declarations=..., 
    shareable=false, parent_style=..., cached_style=...)
    at target/debug/build/style-023f74d8196076da/out/properties.rs:16628
#1  0x00005555563b4637 in layout::css::matching::LayoutNode<'ln>.PrivateMatchMethods::cascade_node_pseudo_element
    (self=0x7fffcf3fc010, layout_context=0x7fffd53fc708, parent_style=..., applicable_declarations=..., 
    style=0x7fffd644a920, applicable_declarations_cache=0x7fffce838080, new_animations_sender=0x7fffd53fc880, 
    shareable=false) at /home/manishearth/Mozilla/voser/components/layout/css/matching.rs:457
#2  0x000055555639d2cd in layout::css::matching::LayoutNode<'ln>.MatchMethods::cascade_node (self=0x7fffcf3fc010, 
    layout_context=0x7fffd53fc708, parent=..., applicable_declarations=0x7fffcf3fbd60, 
    applicable_declarations_cache=0x7fffce838080, new_animations_sender=0x7fffd53fc880)
    at /home/manishearth/Mozilla/voser/components/layout/css/matching.rs:689
#3  0x000055555636fb21 in layout::traversal::RecalcStyleForNode<'a>.PreorderDomTraversal::process (
    self=0x7fffcf3fc220, node=...) at /home/manishearth/Mozilla/voser/components/layout/traversal.rs:181
#4  0x000055555636f3de in run_parallel_helper<layout::traversal::RecalcStyleForNode> (self=0x7fffcf3fc220, 
    unsafe_node=..., proxy=0x7fffcf3fc388, top_down_func=
    {void (struct (usize, usize), struct WorkerProxy<layout::context::SharedLayoutContextWrapper, (usize, usize)> *)} 0x7fffcf3fc1c0, bottom_up_func=
    {void (struct (usize, usize), struct WorkerProxy<layout::context::SharedLayoutContextWrapper, (usize, usize)> *)} 0x7fffcf3fc1b8) at /home/manishearth/Mozilla/voser/components/layout/parallel.rs:103
#5  layout::parallel::RecalcStyleForNode<'a>.ParallelPreorderDomTraversal::run_parallel (self=0x7fffcf3fc220, 
    unsafe_node=..., proxy=0x7fffcf3fc388) at /home/manishearth/Mozilla/voser/components/layout/parallel.rs:349
#6  0x000055555637027e in layout::parallel::recalc_style (unsafe_node=..., proxy=0x7fffcf3fc388)
    at /home/manishearth/Mozilla/voser/components/layout/parallel.rs:363
#7  0x0000555556303b8b in layout::workqueue::WorkerThread<QueueData, WorkData>::start (self=0x7fffcf3fc570)
    at /home/manishearth/Mozilla/voser/components/util/workqueue.rs:162
#8  0x00005555563033dc in fnfn () at /home/manishearth/Mozilla/voser/components/util/workqueue.rs:270
---Type <return> to continue, or q <return> to quit---
#9  0x000055555630321b in fnfn () at /home/manishearth/Mozilla/voser/components/util/task.rs:16
#10 0x0000555556303111 in layout::boxed::F.FnBox<A>::call_box (self=0x7fffd44283f0, args=0)
    at /home/larsberg/rust/src/liballoc/boxed.rs:374
#11 0x00005555562d8698 in layout::boxed::Box<FnBox<A, Output = R>+ Send + 'a>.FnOnce<A>::call_once (self=..., 
    args=0) at /home/larsberg/rust/src/liballoc/boxed.rs:390
#12 0x00005555562d7da4 in fnfn () at /home/larsberg/rust/src/libstd/thread/mod.rs:346
#13 0x00005555562d7d2f in layout::rt::unwind::try::try_fn<closure> (opt_closure=0x7fffcf3fc980)
    at /home/larsberg/rust/src/libstd/rt/unwind.rs:139
#14 0x00005555562d7cca in rt::unwind::try::try_fn::h7519854721141549484 ()
#15 0x0000555557840a19 in rust_try_inner ()
#16 0x0000555557840a06 in rust_try ()
#17 0x00005555562d71c3 in layout::rt::unwind::try<closure> (f={void (())} 0x7fffcf3fc9a0)
    at /home/larsberg/rust/src/libstd/rt/unwind.rs:125
#18 0x00005555562d6f91 in fnfn () at /home/larsberg/rust/src/libstd/thread/mod.rs:346
#19 0x00005555562d89be in layout::boxed::F.FnBox<A>::call_box (self=0x7fffd442f2d0, args=0)
    at /home/larsberg/rust/src/liballoc/boxed.rs:374
#20 0x000055555783a542 in sys::thread::Thread::new::thread_start::h4913e02ff25f93a3QIv ()
#21 0x00007ffff7bc4182 in start_thread (arg=0x7fffcf3fd700) at pthread_create.c:312
#22 0x00007ffff672147d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:111

gdb says that the program received a segfault, but when run directly I get "Illegal instruction (core dumped)". O.o

[SimonSapin]

Line 16628 in the generated properties.rs is this one:

        let inherited_font_style = inherited_style.get_font();
        computed::Context {
            is_root_element: is_root_element,
            viewport_size: viewport_size,
            inherited_font_weight: inherited_font_style.font_weight,   #### Line 16628
            inherited_font_size: inherited_font_style.font_size,

In gdb, I get:

(gdb) p *inherited_style
$1 = {margin = {_ptr = {__0 = 0xec44300000000000}}, padding = {_ptr = {
      __0 = 0x7fff}}, border = {_ptr = {__0 = 0x0}}, outline = {_ptr = {
      __0 = 0x100000000}}, positionoffsets = {_ptr = {__0 = 0x0}}, box_ = {
    _ptr = {__0 = 0x0}}, inheritedbox = {_ptr = {__0 = 0x0}}, list = {_ptr = {
      __0 = 0x0}}, counters = {_ptr = {__0 = 0x0}}, background = {_ptr = {
      __0 = 0x100000000}}, color = {_ptr = {__0 = 0x0}}, font = {_ptr = {
      __0 = 0x0}}, inheritedtext = {_ptr = {__0 = 0x0}}, text = {_ptr = {
      __0 = 0x0}}, table = {_ptr = {__0 = 0x0}}, inheritedtable = {_ptr = {
      __0 = 0x0}}, pointing = {_ptr = {__0 = 0x0}}, column = {_ptr = {
      __0 = 0x0}}, effects = {_ptr = {__0 = 0x0}}, animation = {_ptr = {
      __0 = 0x0}}, shareable = false, writing_mode = {bits = 0 '\000'}, 
  root_font_size = {__0 = 0}}

All of these {_ptr = {__0 = …}} are of type Arc<_>, which is defined as:

pub struct Arc<T> {
    _ptr: NonZero<*mut ArcInner<T>>,
}

… so we have a NonZero<*mut _> which is zero. This sounds like bad news.

[Manishearth]

I suspect rust-lang/rust#24917

[Manishearth]

If everything is zero (it isn't, it almost is) it might simply be an issue of gdb not doing line numbers correctly and the variable is being accessed before it is initialized.

We still have a segfault, though.

[SimonSapin]

I’m skeptical that inherited_style is not initialized yet, as it’s either set from the INITIAL_VALUES lazy_static which uses a lock internally for initialization, or from the parent_style parameter. But I don’t know.

[Manishearth]

I get less zeroes

$2 = {margin = {_ptr = {__0 = 0x5633e73b00002aaa}}, padding = {_ptr = {__0 = 0x5555}}, border = {_ptr = {
      __0 = 0x200000000}}, outline = {_ptr = {__0 = 0x0}}, positionoffsets = {_ptr = {__0 = 0x0}}, box_ = {
    _ptr = {__0 = 0xe626416100000000}}, inheritedbox = {_ptr = {__0 = 0xed4444d4000002d8}}, list = {_ptr = {
      __0 = 0xed4444d400002aaa}}, counters = {_ptr = {__0 = 0x1d1d1d1d00002aaa}}, background = {_ptr = {
      __0 = 0xb7761a701d1d1d1d}}, color = {_ptr = {__0 = 0x2aaa}}, font = {_ptr = {__0 = 0xed4444b000000000}}, 
  inheritedtext = {_ptr = {__0 = 0xed4444a800002aaa}}, text = {_ptr = {__0 = 0x1d1d1d1d00002aaa}}, table = {
    _ptr = {__0 = 0x1d1d1d1d1d1d1d1d}}, inheritedtable = {_ptr = {__0 = 0x1d1d1d1d1d1d1d1d}}, pointing = {_ptr = {
      __0 = 0x1d1d1d1d1d1d1d1d}}, column = {_ptr = {__0 = 0x1d1d1d1d1d1d1d1d}}, effects = {_ptr = {
      __0 = 0xed4444a81d1d1d1d}}, animation = {_ptr = {__0 = 0xb776123000002aaa}}, shareable = 170, 
  writing_mode = {Python Exception <class 'gdb.error'> Argument to arithmetic operation not a number or boolean.: 
bits = 42 '*'}, root_font_size = }

This is with rr

[Manishearth]

On the other hand,

(gdb) p *inherited_font_style 
Cannot access memory at address 0xed4444b000000010

[Manishearth]

The first time around, inherited_font_style is fine, but the next time it hits that breakpoint I get the above error.

[bors-servo]

☔ The latest upstream changes (presumably #5968) made this pull request unmergeable. Please resolve the merge conflicts.

[larsbergstrom]

Do we need to find somebody on Rust to help us debug this? I'm not sure what our next steps are, but being blocked on an update because rustc is now generating invalid code seems bad.

[Manishearth]

At the moment; It's blocking me from finalizing the sizeof plugin, but that plugin isn't a blocking one; it just makes stuff easier for njn.

So low-pri to fix, but it's going to hit us eventually, so we should get it fixed :)

[michaelwu]

diff --git a/components/script/dom/node.rs b/components/script/dom/node.rs
index 4615e0e..720a8b2 100644
--- a/components/script/dom/node.rs
+++ b/components/script/dom/node.rs
@@ -249,7 +249,7 @@ impl LayoutDataRef {
     #[inline]
     #[allow(unsafe_code)]
     pub unsafe fn borrow_unchecked(&self) -> *const Option<LayoutData> {
-        mem::transmute(&self.data_cell)
+        self.data_cell.as_unsafe_cell().get() as *const _
     }

     /// Borrows the layout data immutably. This function is *not* thread-safe.

[SimonSapin]

#6139

[Ms2ger]

FWIW, the issue was that the layout of RefCell changed to support unsized types.

[Manishearth]

Ohhhhhhhh

methinks we need to do another transmute audit

[mbrubeck]

Moved to #6149.