Mail Archives: djgpp-workers/1997/07/01/12:26:45
I have two comments about this:
1) Whatever the reasons for this phenomenon, it also exists in
the official port on SimTel.NET, so it's not new in bashb9.
2) Why are you so sure it's a memory leak? It can be just a
result of known deficiencies in `malloc' and friends. The fact that Bash
requests more and more core from the DPMI host doesn't necessarily mean
that it itself leaks memory (although it might).
On Tue, 1 Jul 1997, Molnar Laszlo wrote:
> Hi Workers!
>
> I think I've found a bug in the latest bash (bashb9.zip).
> When I run the file test0.sh with "sh test0.sh", I get:
>
> DPMI memory available: 22923 Kb
> DPMI swap space available: 129297 Kb
> DPMI memory available: 22911 Kb
> DPMI swap space available: 129309 Kb
> DPMI memory available: 22911 Kb
> DPMI swap space available: 129309 Kb
> DPMI memory available: 22911 Kb
> DPMI swap space available: 129309 Kb
> DPMI memory available: 22903 Kb
> DPMI swap space available: 129317 Kb
> DPMI memory available: 22887 Kb
> DPMI swap space available: 129333 Kb
> DPMI memory available: 22811 Kb
> DPMI swap space available: 129217 Kb
> DPMI memory available: 22491 Kb
> DPMI swap space available: 129025 Kb
> DPMI memory available: 21235 Kb
> DPMI swap space available: 128617 Kb
> DPMI memory available: 16235 Kb
> DPMI swap space available: 127345 Kb
> DPMI memory available: 119251 Kb
> DPMI swap space available: 0 Kb
>
> As you can see at the last step all of my 22 MB physical memory is full.
> Why is this?
>
> bye, Laszlo
>
> ps: I've found this problem in perl5.004's Configure script.
>
> Here are the 3 test files:
>
> *--test0.sh----*
> go32-v2 | grep DPMI
>
> for i in 1 2 3 4 5 6 7 8 9 10
> do
> . ./test1.sh # if you change this line to "sh ./test1.sh"
> go32-v2 | grep DPMI # then there is no problem!
> done
> *--eof---------*
>
> *--test1.sh----*
> . ./test2.sh
>
> cat >x <<ENDEND
> blah-blah
> ENDEND
> *--eof---------*
>
> *--test2.sh----*
> cat >x <<ENDEND
> blah-blah
> ENDEND
> *--eof---------*
>
>
- Raw text -