www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - How to proceed with ICE?

reply =?UTF-8?B?IsOYaXZpbmQi?= <oivind.loe gmail.com> writes:
After moving to 2.062, I get the following ICE:

src/boss/core/addr.d(23): Error: Internal Compiler Error: CTFE 
literal cast(uint)PROC_INVALID
dmd: ctfeexpr.c:353: Expression* copyLiteral(Expression*): 
Assertion `0' failed.
Aborted (core dumped)

The line it fails on is here, but I have not managed to get a 
reduced testcase to fail yet..

struct Addr {
   alias uint tProcId;
   alias uint tNodeId;

   const PROC_INVALID    = 0;
   const NODE_SELF       = 0;

   tProcId m_procId = PROC_INVALID;  // << This is the line
   tNodeId m_nodeId = NODE_SELF;


  ....
}
Mar 24 2013
next sibling parent =?UTF-8?B?IsOYaXZpbmQi?= <oivind.loe gmail.com> writes:
On Sunday, 24 March 2013 at 16:01:14 UTC, Øivind wrote:

 After moving to 2.062, I get the following ICE:

 src/boss/core/addr.d(23): Error: Internal Compiler Error: CTFE 
 literal cast(uint)PROC_INVALID
 dmd: ctfeexpr.c:353: Expression* copyLiteral(Expression*): 
 Assertion `0' failed.
 Aborted (core dumped)

 The line it fails on is here, but I have not managed to get a 
 reduced testcase to fail yet..

 struct Addr {
   alias uint tProcId;
   alias uint tNodeId;

   const PROC_INVALID    = 0;
   const NODE_SELF       = 0;

   tProcId m_procId = PROC_INVALID;  // << This is the line
   tNodeId m_nodeId = NODE_SELF;


  ....
 }


Also inside the struct is the following constructor.. Commenting 
it out removes the ICE:

this(tProcId procId, tNodeId nodeId) {
     m_procId = procId;
     m_nodeId = nodeId;
   }
Mar 24 2013
prev sibling parent reply "lomereiter" <lomereiter gmail.com> writes:
DustMite (https://github.com/CyberShadow/DustMite) is an 
invaluable tool for reducing test cases like this. I have to use 
it after each frontend update =\
Mar 24 2013
parent reply =?UTF-8?B?IsOYaXZpbmQi?= <oivind.loe gmail.com> writes:
On Sunday, 24 March 2013 at 18:16:03 UTC, lomereiter wrote:

 DustMite (https://github.com/CyberShadow/DustMite) is an 
 invaluable tool for reducing test cases like this. I have to 
 use it after each frontend update =\


Thanks! After running, my project is reduced to a small set of 
files. There are now also a couple of regular compiler errors 
before the ICE (introduced by dustmite), but the ICE should not 
be there anyway.. Should I zip the remaining files with 
instructions and post them on the tracker, or do I need to take 
steps to remove the first errors?

Here is the output now:

Compiling: src/main.d -> build/debug/src/main.o
src/boss/core/addr.d(9): Error: undefined identifier procId, did 
you mean alias tProcId?
src/boss/core/annotstruct.d(3): Error: no identifier for 
declarator index
src/boss/core/annotstruct.d(3): Error: semicolon expected, not 
'EOF'
src/boss/core/annotstruct.d(3): Error: undefined identifier index
src/boss/core/annotstruct.d(3): Error: constructor addr.Addr.this 
(_error_, uint) is not callable using argument types (uint,uint)
src/boss/core/addr.d(6): Error: Internal Compiler Error: CTFE 
literal cast(uint)PROC_INVALID
dmd: ctfeexpr.c:353: Expression* copyLiteral(Expression*): 
Assertion `0' failed.
Aborted (core dumped)
make[1]: *** [build/debug/src/main.o] Error 134
Mar 24 2013
parent reply Iain Buclaw <ibuclaw ubuntu.com> writes:
On 24 March 2013 21:15, "=D8ivind" <oivind.loe gmail.com> wrote:


 On Sunday, 24 March 2013 at 18:16:03 UTC, lomereiter wrote:


 DustMite (https://github.com/**CyberShadow/DustMite<https://github.com/C=




yberShadow/DustMite>)

 is an invaluable tool for reducing test cases like this. I have to use i=




t

 after each frontend update =3D\


 Thanks! After running, my project is reduced to a small set of files.
 There are now also a couple of regular compiler errors before the ICE
 (introduced by dustmite), but the ICE should not be there anyway.. Should=


 I

 zip the remaining files with instructions and post them on the tracker, o=


r

 do I need to take steps to remove the first errors?

 Here is the output now:

 Compiling: src/main.d -> build/debug/src/main.o
 src/boss/core/addr.d(9): Error: undefined identifier procId, did you mean
 alias tProcId?
 src/boss/core/annotstruct.d(3)**: Error: no identifier for declarator
 index
 src/boss/core/annotstruct.d(3)**: Error: semicolon expected, not 'EOF'
 src/boss/core/annotstruct.d(3)**: Error: undefined identifier index
 src/boss/core/annotstruct.d(3)**: Error: constructor addr.Addr.this
 (_error_, uint) is not callable using argument types (uint,uint)
 src/boss/core/addr.d(6): Error: Internal Compiler Error: CTFE literal
 cast(uint)PROC_INVALID

 dmd: ctfeexpr.c:353: Expression* copyLiteral(Expression*): Assertion `0'
 failed.
 Aborted (core dumped)
 make[1]: *** [build/debug/src/main.o] Error 134


Put it in a new bug report please.

http://d.puremagic.com/issues/

--=20
Iain Buclaw

*(p < e ? p++ : p) =3D (c & 0x0f) + '0';
Mar 24 2013
parent =?UTF-8?B?IsOYaXZpbmQi?= <oivind.loe gmail.com> writes:
 Put it in a new bug report please.

 http://d.puremagic.com/issues/


Done:
http://d.puremagic.com/issues/show_bug.cgi?id=9847
Mar 31 2013