On Tue, Nov 15, 2005 at 01:36:10PM +0100, Michael Koch wrote: > On Tue, Nov 15, 2005 at 02:20:52AM -0800, Steve Langasek wrote:
> > If so, can you by chance offer a sample environment that's something smaller > > than "start eclipse" as a means of reproducing this bug? c++filt should by > > all rights be a very simple program, and I'd be surprised if c++filt was > > actually hanging for any reason that wasn't a bug in the caller (i.e., gcj). > Sorry. > I wrote a small Java programm (attached) that shows the problem too. Not attached :) > To reproduce the problem do this: > gcj-4.0 -C BinutilsTest.java > gij-4.0 Binutils > gij-4.0 will hangs when trying to print the stack trace. > Any other free or non-free VM does this fine. GCJ uses the external > programs c++filt and addr2line to implement this task (no other does > this). I guess this is a binutils bug becuase when downgrading to an > older binutils (2.16.1-3). Ok, so this test case still depends on gcj/gij invoking c++filt. Better than binutils, but not a very straightforward test case to explain how c++filt's behavior is wrong... I still suspect this is a gcj/gij bug, since given the choice between "bug in a program that takes a line of input, mangles it, and prints a line of output" and "bug in a program that's opening a pipe to call another program for both input and output", I have an opinion on which is more likely :), but I guess we'll need that test case in order take a closer look and find out for sure. Thanks, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/
signature.asc
Description: Digital signature
