Avoid unnecessary allocas for indirect function arguments #44573

dotdash · 2017-09-14T20:16:22Z

The extra alloca was only necessary because it made LLVM implicitly
handle the necessary deref to get to the actual value. The same happens
for indirect arguments that have the byval attribute. But the Rust ABI
does not use the byval attribute and so we need to manually add the
deref operation to the debuginfo.

rust-highfive · 2017-09-14T20:16:33Z

r? @nikomatsakis

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

eddyb · 2017-09-14T20:46:20Z

LGTM. r? @michaelwoerister

michaelwoerister · 2017-09-15T14:23:44Z

@bors r+

Thanks @dotdash!

bors · 2017-09-15T14:23:45Z

📌 Commit ed7a7cf has been approved by michaelwoerister

Avoid unnecessary allocas for indirect function arguments The extra alloca was only necessary because it made LLVM implicitly handle the necessary deref to get to the actual value. The same happens for indirect arguments that have the byval attribute. But the Rust ABI does not use the byval attribute and so we need to manually add the deref operation to the debuginfo.

TimNN · 2017-09-17T11:23:45Z

This likely caused rollup #44649 to fail with the following error on x86_64-gnu-debug:

[00:27:41] rustc: /checkout/src/llvm/lib/CodeGen/AsmPrinter/DwarfExpression.cpp:295: void llvm::DwarfExpression::addFragmentOffset(const llvm::DIExpression*): Assertion `FragmentOffset >= OffsetInBits && "overlapping or duplicate fragments"' failed.
[00:27:41] error: Could not compile `std`.

(x86_64-gnu-debug succeeded in rollup #44652, who's only difference to #44649 was that this PR was excluded)

bors · 2017-09-17T16:52:20Z

⌛ Testing commit ed7a7cf with merge ee179583d630a07cf0009739c6edbc61ada961a8...

arielb1 · 2017-09-17T17:14:54Z

@bors r-

arielb1 · 2017-09-17T17:14:58Z

@bors retry

arielb1 · 2017-09-17T17:15:43Z

Turns out leaving LLVM assertions disabled isn't a good idea when working with LLVM.

michaelwoerister · 2017-09-18T10:16:25Z

Is there anything we can do about this? Or does this just not work in the general case?

dotdash · 2017-09-19T12:34:58Z

Looks like an LLVM bug which is only triggered by this change. The jump threading pass duplicates an dbg.declare intrinsic, causing the same fragment to be declared twice and triggering the assertion. I asked on the LLVM ML about it.

http://lists.llvm.org/pipermail/llvm-dev/2017-September/117575.html

eddyb · 2017-09-19T12:49:18Z

@dotdash It wouldn't be the first time LLVM mangles non-trivial debuginfo annotations.

michaelwoerister · 2017-09-19T13:23:21Z

@dotdash ❤️

aidanhs · 2017-09-28T01:01:09Z

For triage clarity, where is this up to? Is there a patch making its way upstream (and if so, is there a link to the LLVM phabricator so we can track it)?

dotdash · 2017-10-04T14:10:51Z

Sorry, I've been unable to get back to this earlier. Upstream patch is at https://reviews.llvm.org/D38540

dotdash · 2017-10-11T19:00:52Z

Updated to include the necessary LLVM update and added a second commit to avoid even more copies and fix a debug info oddity/bug.

bors · 2017-10-14T13:48:28Z

☔ The latest upstream changes (presumably #45162) made this pull request unmergeable. Please resolve the merge conflicts.

dotdash · 2017-10-16T18:18:28Z

Rebased and backed out the commit that was failing earlier. Should be good to go now.

eddyb · 2017-10-18T03:45:27Z

src/rustllvm/llvm-rebuild-trigger

@@ -1,4 +1,4 @@
 # If this file is modified, then llvm will be (optionally) cleaned and then rebuilt.
 # The actual contents of this file do not matter, but to trigger a change on the
 # build bots then the contents should be changed so git updates the mtime.
-2017-07-18
+2017-10-10


Do you need to change this?

There was an LLVM rebuild race.

arielb1 · 2017-10-18T10:11:30Z

r+ with the llvm rebuild trigger removed - it is not needed any more because someone else updated our LLVM.

michaelwoerister · 2017-10-18T10:51:11Z

👍

The extra alloca was only necessary because it made LLVM implicitly handle the necessary deref to get to the actual value. The same happens for indirect arguments that have the byval attribute. But the Rust ABI does not use the byval attribute and so we need to manually add the deref operation to the debuginfo.

dotdash · 2017-10-18T10:59:35Z

@bors r=arielb1

bors · 2017-10-18T10:59:35Z

📌 Commit 6bfecd4 has been approved by arielb1

bors · 2017-10-18T13:36:23Z

⌛ Testing commit 6bfecd4 with merge fc208bb...

Avoid unnecessary allocas for indirect function arguments The extra alloca was only necessary because it made LLVM implicitly handle the necessary deref to get to the actual value. The same happens for indirect arguments that have the byval attribute. But the Rust ABI does not use the byval attribute and so we need to manually add the deref operation to the debuginfo.

bors · 2017-10-18T18:07:17Z

☀️ Test successful - status-appveyor, status-travis
Approved by: arielb1
Pushing fc208bb to master...

dotdash requested a review from eddyb September 14, 2017 20:16

rust-highfive assigned nikomatsakis Sep 14, 2017

rust-highfive assigned michaelwoerister and unassigned nikomatsakis Sep 14, 2017

alexcrichton added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Sep 14, 2017

michaelwoerister added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Sep 15, 2017

TimNN mentioned this pull request Sep 17, 2017

Rollup of 19 pull requests #44649

Closed

arielb1 added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Sep 19, 2017

dotdash mentioned this pull request Oct 10, 2017

Backport a fix for duplicate debug data errors rust-lang/llvm#93

Merged

dotdash force-pushed the dbg_bloat branch from ed7a7cf to aeb13c1 Compare October 11, 2017 08:37

dotdash force-pushed the dbg_bloat branch from aeb13c1 to bdf4b85 Compare October 11, 2017 18:53

dotdash force-pushed the dbg_bloat branch 3 times, most recently from b18ed29 to 43c4589 Compare October 13, 2017 14:07

dotdash force-pushed the dbg_bloat branch from 43c4589 to 350698c Compare October 16, 2017 14:34

kennytm added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Oct 16, 2017

eddyb reviewed Oct 18, 2017

View reviewed changes

kennytm added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Oct 18, 2017

dotdash force-pushed the dbg_bloat branch from 350698c to 6bfecd4 Compare October 18, 2017 10:58

kennytm added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Oct 18, 2017

bors merged commit 6bfecd4 into rust-lang:master Oct 18, 2017

cuviper mentioned this pull request Nov 8, 2017

DW_TAG_formal_parameter always has DW_AT_decl_line == 1 #45010

Open

dotdash mentioned this pull request Dec 19, 2017

omitting frame pointers breaks retrieving DWARF debug info for parameter values #11906

Closed

dotdash deleted the dbg_bloat branch January 31, 2018 11:34

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Avoid unnecessary allocas for indirect function arguments #44573

Avoid unnecessary allocas for indirect function arguments #44573

dotdash commented Sep 14, 2017 •

edited

Loading

rust-highfive commented Sep 14, 2017

eddyb commented Sep 14, 2017

michaelwoerister commented Sep 15, 2017

bors commented Sep 15, 2017

TimNN commented Sep 17, 2017

bors commented Sep 17, 2017

arielb1 commented Sep 17, 2017

arielb1 commented Sep 17, 2017

arielb1 commented Sep 17, 2017

michaelwoerister commented Sep 18, 2017

dotdash commented Sep 19, 2017

eddyb commented Sep 19, 2017

michaelwoerister commented Sep 19, 2017

aidanhs commented Sep 28, 2017

dotdash commented Oct 4, 2017

dotdash commented Oct 11, 2017

bors commented Oct 14, 2017

dotdash commented Oct 16, 2017

eddyb Oct 18, 2017

arielb1 Oct 18, 2017

arielb1 commented Oct 18, 2017 •

edited

Loading

michaelwoerister commented Oct 18, 2017

dotdash commented Oct 18, 2017

bors commented Oct 18, 2017

bors commented Oct 18, 2017

bors commented Oct 18, 2017

Avoid unnecessary allocas for indirect function arguments #44573

Avoid unnecessary allocas for indirect function arguments #44573

Conversation

dotdash commented Sep 14, 2017 • edited Loading

rust-highfive commented Sep 14, 2017

eddyb commented Sep 14, 2017

michaelwoerister commented Sep 15, 2017

bors commented Sep 15, 2017

TimNN commented Sep 17, 2017

bors commented Sep 17, 2017

arielb1 commented Sep 17, 2017

arielb1 commented Sep 17, 2017

arielb1 commented Sep 17, 2017

michaelwoerister commented Sep 18, 2017

dotdash commented Sep 19, 2017

eddyb commented Sep 19, 2017

michaelwoerister commented Sep 19, 2017

aidanhs commented Sep 28, 2017

dotdash commented Oct 4, 2017

dotdash commented Oct 11, 2017

bors commented Oct 14, 2017

dotdash commented Oct 16, 2017

eddyb Oct 18, 2017

Choose a reason for hiding this comment

arielb1 Oct 18, 2017

Choose a reason for hiding this comment

arielb1 commented Oct 18, 2017 • edited Loading

michaelwoerister commented Oct 18, 2017

dotdash commented Oct 18, 2017

bors commented Oct 18, 2017

bors commented Oct 18, 2017

bors commented Oct 18, 2017

dotdash commented Sep 14, 2017 •

edited

Loading

arielb1 commented Oct 18, 2017 •

edited

Loading