[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-03-22 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

Gleb Popov  changed:

   What|Removed |Added

 CC||arr...@freebsd.org
 Resolution|--- |FIXED
 Status|New |Closed
  Flags|maintainer-feedback?(kde@Fr |maintainer-feedback+
   |eeBSD.org)  |

--- Comment #6 from Gleb Popov  ---
Pushed, thanks.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-03-22 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

--- Comment #5 from commit-h...@freebsd.org ---
A commit in branch main references this bug:

URL:
https://cgit.FreeBSD.org/ports/commit/?id=9e3ff8f0b99e7b3074f3e83c954a0395185f63f0

commit 9e3ff8f0b99e7b3074f3e83c954a0395185f63f0
Author: Tomas Tevesz 
AuthorDate: 2024-03-22 21:16:49 +
Commit: Gleb Popov 
CommitDate: 2024-03-22 21:18:36 +

devel/qt5-script: Prevent clang from optimizing the code with undefined
behavior

PR: 277236
Obtained from:  OpenBSD ports

 devel/qt5-script/Makefile |  2 +-
 ...criptcore_JavaScriptCore_interpreter_CallFrame.h (new) | 15 +++
 2 files changed, 16 insertions(+), 1 deletion(-)

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-02-23 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

Yuri Victorovich  changed:

   What|Removed |Added

 Blocks||265989


Referenced Bugs:

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=265989
[Bug 265989] cad/qcad: core dumps on -current
-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-02-23 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

--- Comment #4 from John F. Carr  ---
Here is an example:

struct S {
  long alignment;
  int this_lowbit();
};
int S::this_lowbit() { return reinterpret_cast(this) & 1; }

The method this_lowbit always returns zero with optimization enabled.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-02-23 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

--- Comment #3 from Yuri Victorovich  ---
(In reply to John F. Carr from comment #2)

I see.

In this case this should be reported to the devel/qt5-script upstream. They
should stop using hacks and do the same thing properly.


Yuri

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-02-23 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

John F. Carr  changed:

   What|Removed |Added

 CC||j...@mit.edu

--- Comment #2 from John F. Carr  ---
My first impression is clang is within its rights to break this code.  The JS
implementation tries to store metadata bits in the low bits of a pointer type. 
The compiler is allowed to assume that the low bits of a pointer value are
zero. (Assuming the pointed-to type has alignment rules, which is normally the
case.)  LLVM itself uses a wrapper class instead of an illegal pointer value. 
The wrapper class stores the pointer+tag combination as an integer and provides
a getter method to return an untagged pointer.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-02-23 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

Yuri Victorovich  changed:

   What|Removed |Added

 CC||y...@freebsd.org

--- Comment #1 from Yuri Victorovich  ---
Should this also be reported to Clang, since it's Clang that miscompiles
devel/qt5-script.

-- 
You are receiving this mail because:
You are the assignee for the bug.

maintainer-feedback requested: [Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-02-23 Thread bugzilla-noreply
Bugzilla Automation  has asked freebsd-kde (group)
 for maintainer-feedback:
Bug 277236: devel/qt5-script: clang optimization miscompiles qtscript
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236



--- Description ---
(or whatever the order of cause and effect is)

Cf. bug #265989

The attached patch (taken from OpenBSD,
https://github.com/openbsd/ports/blob/e10f7e1a39f85873183c232462850449f777e16f/
x11/qt5/qtscript/patches/patch-src_3rdparty_javascriptcore_JavaScriptCore_inter
preter_CallFrame_h)
fixes, at least on FreeBSD 14, the problem reported with cad/qcad in bug
#265989 (and presumably other things using qtscript).


[Bug 277236] devel/qt5-script: clang optimization miscompiles qtscript

2024-02-23 Thread bugzilla-noreply
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277236

Bug ID: 277236
   Summary: devel/qt5-script: clang optimization miscompiles
qtscript
   Product: Ports & Packages
   Version: Latest
  Hardware: Any
OS: Any
Status: New
  Severity: Affects Only Me
  Priority: ---
 Component: Individual Port(s)
  Assignee: k...@freebsd.org
  Reporter: i...@extreme.hu
 Flags: maintainer-feedback?(k...@freebsd.org)
  Assignee: k...@freebsd.org

Created attachment 248689
  --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=248689=edit
devel/qt5-script diff fixing optimisation issue with Clang

(or whatever the order of cause and effect is)

Cf. bug #265989

The attached patch (taken from OpenBSD,
https://github.com/openbsd/ports/blob/e10f7e1a39f85873183c232462850449f777e16f/x11/qt5/qtscript/patches/patch-src_3rdparty_javascriptcore_JavaScriptCore_interpreter_CallFrame_h)
fixes, at least on FreeBSD 14, the problem reported with cad/qcad in bug
#265989 (and presumably other things using qtscript).

-- 
You are receiving this mail because:
You are the assignee for the bug.