Mail Archives: djgpp-workers/2004/12/08/20:45:04
Juan Manuel Guerrero wrote:
> On Tue, 7 Dec 2004, Frank Donahoe wrote:
>
>> I had believed that the popen.c used to build the bison.exe
>> which passed all 104 tests, and the sed which passed the
>> "eval" test differed from the version of 04-11-30 only by
>> the addition of Juan Manuel Guerrero' patch of 04-12-06.
>> This is not true. It also contains Esa A E Peuha' patch
>> of 04-11-29.
>
> The bison port has never used popen() and pclose(). The reason for
> the bison failure was that I allocated an string using strlen()
> without adding one byte more for the terminating '\0'. Later when
> this string was free()ed with djdev204 the program crashed. But
> free() of djdev203 never complained abuot this so I never noticed
> this bug. I have tested stock bison-1.875 and the test
> bison-1.875d with djdev203 and cvs sources of djdev204 running
> the testsuite with gdb. No test breaks the testsuite nor is
> skipped. Now everything seems to be fixed.
Does this mean that my nmalloc is now incorporated in 2.04? I did
put some checks in there to catch that sort of error as early as
conveniently feasible. Since allocation sizes are rounded upwards,
that error would only show up if the allocation size was a multiple
of 8.
I know I promised to finish the memalign alteration, and I did
start it, but I have just returned from the hospital with a triple
bypass. At any rate, I have done no more work on memalign, and
probably won't for at least the next month.
--
Chuck F (cbfalconer AT yahoo DOT com) (cbfalconer AT worldnet DOT att DOT net)
Available for consulting/temporary embedded and systems.
<http://cbfalconer.home.att.net> USE worldnet address!
- Raw text -