Re: [jamboost] A different bjam crash

From: Beman Dawes (bdawes_at_[hidden])
Date: 2003-07-26 08:35:13

At 09:05 AM 7/26/2003, Beman Dawes wrote:

>A different bjam crash this morning. Addressing exception. See call stack 

>trace below.

Usually the occasional bjam crash is not repeatable. Run it a second time 
and it doesn't fail. But this time three attempts in a row produced the 
same error with the same stack trace.

Since this was a release build (although with symbols turned on), I 
switched to a pure debug build and reran. No error.

While that would seem to point toward a compiler bug which generates bad 
code on release builds, there are other explanations. IIRC, the debug 
runtime initializes memory somewhat differently on a debug build than on a 
release build. Thus a program logic error may present differently for the 
two build types.

--Beman

>bjam.exe!file_archscan(char * archive=0x004743e8, void (void *, char *,
>int, long)* func=0x0040d960, void * closure=0x00495868) Line 255 + 0x5
>bjam.exe!timestamp(char * target=0x00465520, long * 
time=0x0148c21c) Line
>167
>bjam.exe!search(char * target=0x01d5695a, long * time=0x0148c21c, char * 
*
>another_target=0x00000000) Line 132 + 0x1a
>bjam.exe!bind_explicitly_located_target(void * xtarget=0x0148c1f4, void * 

>data=0x00000000) Line 168
>bjam.exe!hashenumerate(hash * hp=0x00002485, void (void *, void *)*
>f=0x0040c5f0, void * data=0x00000000) Line 251 + 0xa
>bjam.exe!bind_explicitly_located_targets() Line 178 + 0xd
>bjam.exe!make(int n_targets=1, const char * * targets=0x0012f6a0, int
>anyhow=0) Line 127
>bjam.exe!main(int argc=, char * * argv=, char * * arg_environ=) Line 424 
+
>
>0x11
>bjam.exe!mainCRTStartup() Line 259 + 0x12
>kernel32.dll!77e814c7() 
>ntdll.dll!77f944a8()