http://d.puremagic.com/issues/show_bug.cgi?id=5790

           Summary: "Error: variable result used before set" when -release
                    -inline -O
           Product: D
           Version: D2
          Platform: x86
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: nobody@puremagic.com
        ReportedBy: jordi@rovira.cat


--- Comment #0 from Jordi Rovira i Bonet <jordi@rovira.cat> 2011-03-28 05:54:10 PDT ---
When compiling the following code with

dmd -c -release -inline -O resource.d

I get the message "Error: variable result used before set" without any line information, and compilation fails.

This happens only with the flags above. Removing any of them, or compiling in debug results in good output. I have verified that this error happens both in linux and windows, with dmd 2.052 and since many versions ago.

<Speculation>
Even GDC crashes compiling this code in release, which makes me think it is in
the front end.

<Data>
I have removed everything i could, and now the code doesn't make much sense,
but the file resource.d is:

module     bug;

import     std.c.string;

class InputStream
{
public:

    InputStream opShr(T)( out T result )
    {
        memcpy( &result, &m_data[m_pos], T.sizeof );
        m_pos += T.sizeof;

        return this;
    }

    int m_pos;
    const ubyte[] m_data;
}

struct GUID
{
    uint m_values[4];
};

struct Header
{
    int ver;
    int siz;
    GUID id;
}

void LoadResourceLegacyHeader( InputStream f, out Header header )
{
    f >> header.id;
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790


bearophile_hugs@eml.cc changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bearophile_hugs@eml.cc


--- Comment #1 from bearophile_hugs@eml.cc 2011-03-28 10:13:29 PDT ---
(In reply to comment #0)

> I have removed everything i could,

Reduced program:


struct Foo {
    int x;
    int[3] y;
}
void baz(out int[3] y2) {
    y2[0]++;
}
void spam(out Foo f) {
    baz(f.y);
}
void main() {}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790



--- Comment #2 from Jordi Rovira i Bonet <jordi@rovira.cat> 2011-03-28 10:17:14 PDT ---
(In reply to comment #1)
> (In reply to comment #0)
> 
> > I have removed everything i could,
> 
> Reduced program:
> 
> 
> struct Foo {
>     int x;
>     int[3] y;
> }
> void baz(out int[3] y2) {
>     y2[0]++;
> }
> void spam(out Foo f) {
>     baz(f.y);
> }
> void main() {}

Thanks!

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790


bearophile_hugs@eml.cc changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |critical


--- Comment #3 from bearophile_hugs@eml.cc 2011-03-28 10:18:41 PDT ---
Critical, as per Don request.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790


kennytm@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |kennytm@gmail.com


--- Comment #4 from kennytm@gmail.com 2011-03-28 10:23:03 PDT ---
Also reproducible with '-inline -noboundscheck -O'. Should be a backend issue in gother.c.

-------- 

struct H {
    double a;
    int[3] b;   // must not be the first member in a struct, and must be an
array with 3 or more members
}

void g(out int[3] t) {   // must take 'out' parameter
    t[0]=0;   // can be any mutating expression
}

void h (H* x) { // or 'out H' or 'ref H'
    g(x.b);
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790


kennytm@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Platform|x86                         |All


--- Comment #5 from kennytm@gmail.com 2011-08-02 12:50:03 PDT ---
Another test case, which is triggered only on Linux and FreeBSD.

-----------------------------------
void test5790d() {
    int x;
    int y = *(1 + &x);
}
-----------------------------------
$ dmd -O -c test5790d.d
Error: variable x used before set
-----------------------------------

See also https://github.com/D-Programming-Language/dmd/pull/243#issuecomment-1706278.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790


Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bugzilla@digitalmars.com


--- Comment #6 from Walter Bright <bugzilla@digitalmars.com> 2011-08-03 09:58:41 PDT ---
This is possibly related to:

http://d.puremagic.com/issues/show_bug.cgi?id=5239

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790


Brad Roberts <braddr@puremagic.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |braddr@puremagic.com
         AssignedTo|nobody@puremagic.com        |bugzilla@digitalmars.com


--- Comment #7 from Brad Roberts <braddr@puremagic.com> 2011-08-03 20:41:58 PDT ---
Confirmed that the reduction in comment 2 still reproduces the problem.  It reports y2 as used before set (with no line number).

The two fixes to this part of the compiler aren't sufficient to catch this one.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790


Jonas Drewsen <jdrewsen@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jdrewsen@gmail.com


--- Comment #8 from Jonas Drewsen <jdrewsen@gmail.com> 2012-01-11 14:14:30 PST ---
Additional test case using only the -O flag and no -inline or -release flags necessary.

import std.typecons;

void main(string[] args) {

    // Errors
    auto i = Tuple!(string,string)("foo", "bar")[1];

    // Ok
    auto j = Tuple!(string,string)("foo", "bar");
    auto k = j[1];
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=5790



--- Comment #9 from Kenji Hara <k.hara.pg@gmail.com> 2012-01-12 01:12:06 PST ---
(In reply to comment #8)
> Additional test case using only the -O flag and no -inline or -release flags necessary.
> 
> import std.typecons;
> 
> void main(string[] args) {
> 
>     // Errors
>     auto i = Tuple!(string,string)("foo", "bar")[1];
> 
>     // Ok
>     auto j = Tuple!(string,string)("foo", "bar");
>     auto k = j[1];
> }

This is a dup of 7263. See it.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------