Commit Graph

51 Commits

Author SHA1 Message Date
Felipe de Azevedo Piovezan
f7f869c88d [DebugInfo] Don't inherit debug location of previous instruction
Prior to this commit, when lowering SIL instructions that should are
"hidden" for the purposes of debugging, the compiler just attaches the
location of the previous instruction in the name of keeping a simpler
line table.

However, this is wrong for many reasons. One such reason is this: at the
start of a basic block, inheriting the previous debug location will
almost certainly cause the instruction to have a random location in the
code, as it will depend on whatever BB was visited previously.

Other examples can be seen in the tests affect by this commit, which
changes lowering to use Line 0 instead of the line number of the
previous instruction.

CodeView doesn't handle line 0 the same way DWARF does, so this commit
preserves the old behavior for the CodeView path.

The test changes here are effectively undoing some of the diffs from
158772c2ab.

rdar://139826231&110187845
2024-11-18 10:23:41 -08:00
Felipe de Azevedo Piovezan
a321b0afe0 [DebugInfo] Update tests to expect new LLVM debug format 2024-07-23 11:06:12 -07:00
Adrian Prantl
6c39b60edc Update tests for upstream LLVM IR assembler changes 2018-11-29 09:53:13 -08:00
Pavel Yaskevich
a1df238d47 Associate @autoclosure only with parameter types
Mangled implicit closures shouldn't have `@autoclosure` flag
associated with them, because their type is just a regular
function type.
2018-11-10 11:59:28 -08:00
Erik Eckstein
39bb14b094 change mangling prefix from $S to $s
This is the final ABI mangling prefix

rdar://problem/38471478
2018-09-19 13:55:11 -07:00
Adrian Prantl
c241ff305e Emit compiler-generated debug locations for outlined transparent functions.
In the majority of the use-cases transparent functions are inlined by
the mandatory inliner which by design drops all debug info and
pretends the inlined instructions were always part of the
caller. Since an outlined copy of the function is often still
generated, attaching debug locations to it is inconsistent and can
create the false impression that it were possible to set a breakpoint
in such a function when in reality these functions are only there for
very few edge cases.

<rdar://problem/40258813>
2018-05-17 16:38:04 -07:00
Erik Eckstein
cd3d50a5d9 ABI: Change the mangling prefix from _T0 to $S 2018-01-06 13:55:59 -08:00
Pavel Yaskevich
600b15821d [Mangling/ABI] NFC: Fix DebugInfo tests to reflect label mangling changes 2017-12-18 15:44:24 -08:00
Max Moiseev
2e67c13d15 Fixing some tests 2017-04-14 16:19:06 -07:00
Slava Pestov
94ce4c2ac3 SIL: Only give closures shared linkage if they're going to be serialized
Otherwise, we don't want them to be linkonce_odr at the LLVM level
to avoid unnecessary link-time overhead.
2017-03-31 20:26:27 -07:00
Erik Eckstein
c4a11f4c92 tests: remove the now unused option -new-mangling-for-tests 2017-03-22 11:28:43 -07:00
Erik Eckstein
1d3724666f tests: convert about 400 tests to the new mangling by using the -new-mangling-for-tests option
When the new mangling is enabled permanently, the option can be removed from the RUN command lines again.
2017-01-24 15:27:45 -08:00
Dmitri Gribenko
d175b3b66d Migrate FileCheck to %FileCheck in tests 2016-08-10 23:52:02 -07:00
John McCall
c8c41b385c Implement SE-0077: precedence group declarations.
What I've implemented here deviates from the current proposal text
in the following ways:

- I had to introduce a FunctionArrowPrecedence to capture the parsing
  of -> in expression contexts.

- I found it convenient to continue to model the assignment property
  explicitly.

- The comparison and casting operators have historically been
  non-associative; I have chosen to preserve that, since I don't
  think this proposal intended to change it.

- This uses the precedence group names and higherThan/lowerThan
  as agreed in discussion.
2016-07-26 14:04:57 -07:00
Chris Lattner
af30ae3222 Remove the last parts of the Boolean protocol, finishing up:
SE-0109: Remove the Boolean protocol.

We still love you George, even if we forgot your e.
2016-07-17 22:18:17 -07:00
Chris Lattner
8746676616 Move @noescape and @autoclosure to their new places in various tests, NFC. 2016-04-15 16:05:35 -07:00
Manav Gabhawala
7928140f79 [SE-0046] Implements consistent function parameter labels by discarding extraneous parameter names and adding _ where necessary 2016-04-06 20:21:58 -04:00
Max Moiseev
61c837209b Merge remote-tracking branch 'origin/master' into swift-3-api-guidelines 2016-02-04 16:13:39 -08:00
David Farler
8a5ed405bf Make var parameters an error for Swift 3
This finishes up revisions to SE-0003 - only var function parameters
are disallowed for Swift 3.
2016-01-30 12:39:17 -08:00
David Farler
3f635d04c7 Reinstante var bindings in refutable patterns, except function parameters.
This reverts commits: b96e06da44,
                      8f2fbdc93a,
                      93b6962478,
                      64024118f4,
                      a759ca9141,
                      3434f9642b,
                      9f33429891,
                      47c043e8a6.

This commit leaves 'var' on function parameters as a warning to be
merged into Swift 2.2. For Swift 3, this will be an error, to be
converted in a follow-up.
2016-01-29 15:27:08 -08:00
Max Moiseev
f51e708a8f Merge remote-tracking branch 'origin/master' into swift-3-api-guidelines 2016-01-04 12:25:25 -08:00
ken0nek
3ac60b13f5 Add spaces before and after closure arrow in test 2015-12-23 04:38:46 +09:00
Maxim Moiseev
2c95bb6d51 BooleanType => Boolean 2015-12-10 14:56:32 -08:00
Adrian Prantl
7821341542 Add an argument-number field to DebugValueInst and friends.
This commit adds a DebugVariable field that is shared by
- AllocBoxInst
- AllocStackInst
- DebugValueInst
- DebugValueAddrInst
Currently DebugVariable only holds the Swift argument number.

This allows us to retire several expensive heuristics in IRGen that
attempted to identify which local variables actually where arguments
and recover their relative order.

Memory footprint notes:
This commit adds a 4-byte field to 4 SILInstructin subclasses.
This was offset by 8ab1e2dd50
which removed 20 bytes from *every* SILInstruction.

Caveats:
This commit surfaces a known bug in FunctionSigantureOpts, tracked in
rdar://problem/23727705 — debug info for exploded function arguments
cannot be expressed until this is fixed.

This reapplies ed2b16dc5a with a bugfix for
generic function arrguments and an additional testcase.

<rdar://problem/21185379&22705926>
2015-12-03 13:40:35 -08:00
Adrian Prantl
2740ad6976 Temporarily Revert "Add an argument-number field to DebugValueInst and friends."
while investigating buildbot breakage.

This reverts commit ed2b16dc5a.
2015-12-02 19:10:05 -08:00
Adrian Prantl
ed2b16dc5a Add an argument-number field to DebugValueInst and friends.
This commit adds a DebugVariable field that is shared by
- AllocBoxInst
- AllocStackInst
- DebugValueInst
- DebugValueAddrInst
Currently DebugVariable only holds the Swift argument number.

This allows us to retire several expensive heuristics in IRGen that
attempted to identify which local variables actually where arguments
and recover their relative order.

Memory footprint notes:
This commit adds a 4-byte field to 4 SILInstructin subclasses.
This was offset by 8ab1e2dd50
which removed 20 bytes from *every* SILInstruction.

Caveats:
This commit surfaces a known bug in FunctionSigantureOpts, tracked in
rdar://problem/23727705 — debug info for exploded function arguments
cannot be expressed until this is fixed.

<rdar://problem/21185379&22705926>
2015-12-02 18:33:07 -08:00
David Farler
93b6962478 Warn when using 'var' bindings in function parameters
These will no longer be allowed in a future Swift release.

rdar://problem/23172698
2015-11-03 17:24:20 -08:00
Joe Groff
0cd5aa8c7c Change mangling for the Swift module from 'Ss' to 's'.
'Ss' appears in manglings tens of thousands of times in the standard library and is also incredibly frequent in other modules. This alone is enough to shrink the standard library by 59KB.

Swift SVN r32409
2015-10-02 22:39:44 +00:00
Adrian Prantl
4f11ecc1da Get rid of Int in the debug info tests and preform some general cleanup.
<rdar://problem/20856296> Extinguish 'Int' from DebugInfo tests

Swift SVN r31177
2015-08-12 18:16:43 +00:00
Dmitri Hrybenko
0fce5c7b4e tests: remove uses of println() that are not relevant to the tests
Swift SVN r28016
2015-05-01 03:35:50 +00:00
Duncan Exon Smith
c6d42db070 Adapt to MD* => DI* renaming of debug info types
Applied the upgrade script from r236120 (LLVM) and r236121 (CFE).  This is the
final step of rdar://problem/20434113.



Swift SVN r27925
2015-04-29 21:40:21 +00:00
Duncan Exon Smith
093eb3c365 DebugInfo: Hand-update testcases after LLVM r231082
Update debug info testcases after moving the new hierarchy into place in
upstream LLVM r231082.

rdar://problem/19720042



Swift SVN r25715
2015-03-03 19:15:31 +00:00
Dmitri Hrybenko
3b04d1b013 tests: reorganize tests so that they actually use the target platform
Most tests were using %swift or similar substitutions, which did not
include the target triple and SDK.  The driver was defaulting to the
host OS.  Thus, we could not run the tests when the standard library was
not built for OS X.

Swift SVN r24504
2015-01-19 06:52:49 +00:00
Adrian Prantl
d7489b7272 Update testcases for new LLVM IR assembler syntax.
Swift SVN r24428
2015-01-14 23:38:18 +00:00
Chris Lattner
696749d819 Three things:
- Make @autoclosure imply @__noescape everywhere, not just in the stdlib.
 - Change various noescape diagnostics to refer to "parameters", not "arguments" of the function.
 - Teach the noescape diagnostics to notice if the noescape problem is due to an
   @autoclosure argument that is implicitly noescape.  In this case, produce a helpful note
   that indicates that the invalid use of the noescape is due to the autoclosure implying it.



Swift SVN r24222
2015-01-06 22:40:24 +00:00
Chris Lattner
8c6b2b6b5f adjust testsuite to put @autoclosure on decls instead of types, and remove a
couple of soon-to-be-invalid cases.



Swift SVN r24074
2014-12-22 20:14:11 +00:00
Adrian Prantl
162bc8d24c Updated testcases for upstream assembler changes.
Swift SVN r23942
2014-12-15 19:39:21 +00:00
Michael Gottesman
7e39f33f98 [mangle] Include a pass id in the mangling, just to be careful.
I am starting to reuse manglings for different passes. I want to make sure that
when we reuse functions we actually get a function created by the same pass.

Swift SVN r23924
2014-12-14 10:29:11 +00:00
Michael Gottesman
45b5389854 [mangle] Add support for mangling CapturePromotion specializations and wire it up.
<rdar://problem/19216281>

Swift SVN r23923
2014-12-14 08:17:32 +00:00
Graham Batty
83f27a8af7 Revert "Mark tests that don't pass on linux as XFAIL."
This reverts commit 2711ca86de7bf6a7885ccea24219a48a590b1e95.

Swift SVN r23577
2014-11-24 17:42:13 +00:00
Graham Batty
198402dcfe Mark tests that don't pass on linux as XFAIL.
Swift SVN r23573
2014-11-24 17:40:37 +00:00
Adrian Prantl
d8551038fe Upgrade debug info tests to the new LLVM debug metadata format.
Thanks to Duncan for helping out with the upgrading!

Swift SVN r22512
2014-10-04 00:25:27 +00:00
Adrian Prantl
cf91513c55 Debug info: Emit shadow copies for by-value parameters at -O0 to ensure
that they are visibly for the entirety of the function.

rdar://problem/17538433.

Swift SVN r20505
2014-07-24 18:34:21 +00:00
Dmitri Hrybenko
d14f17beef Change 'getLogicValue()' into a property 'boolValue'; change
'getArrayBoundValue()' into a property 'arrayBoundValue'.

rdar://17156123


Swift SVN r20304
2014-07-22 12:08:10 +00:00
Chris Lattner
bc481f0fe1 implement <rdar://problem/16859927> remove the underscore in "auto_closure"
autoclosure is one work, not two.



Swift SVN r20253
2014-07-21 15:23:50 +00:00
Chris Lattner
57cd2506ff Change "operator infix" to "infix operator" for consistency with the rest of the declaration
modifiers and with the func implementations of the operators.  This resolves the rest of:
<rdar://problem/17527000> change operator declarations from "operator prefix" to "prefix operator" & make operator a keyword




Swift SVN r19931
2014-07-14 16:39:10 +00:00
Dave Abrahams
cbcf9aba21 s/LogicValueType/BooleanType/
We're moving toward using that protocol for straight-up Bool types

Swift SVN r19884
2014-07-12 18:58:18 +00:00
Dave Abrahams
6d1095f44e Protocol names end in "Type," "ible," or "able"
Mechanically add "Type" to the end of any protocol names that don't end
in "Type," "ible," or "able."  Also, drop "Type" from the end of any
associated type names, except for those of the *LiteralConvertible
protocols.

There are obvious improvements to make in some of these names, which can
be handled with separate commits.

Fixes <rdar://problem/17165920> Protocols `Integer` etc should get
uglier names.

Swift SVN r19883
2014-07-12 17:29:57 +00:00
Ted Kremenek
d075f06573 Require a minimum deployment target of iOS 7 or OSX 10.9
Implements <rdar://problem/17532113>

Swift SVN r19451
2014-07-02 06:23:38 +00:00
Ted Kremenek
fad874708e Adjust test cases.
Swift SVN r17964
2014-05-12 22:01:52 +00:00