This is the mail archive of the
mailing list for the GDB project.
Re: [0/17] RFC: share minimal symbols across objfiles
> It would be helpful to know which systems have this issue. Maybe we
> could make it a gdbarch parameter, so that we don't need workarounds on
> more capable systems.
Just a wild guess: Windows, and hppa-hpux. I don't remember of any other
filesystems that lock their files. Oddly, I remember seeing some
messages on GNU/Linux machines which oddly looked like the kind of
message you would get from a locking filesystem. But I don't remember
the details, and it was racy anyway. I'm not completely sure about
hppa-hpux, because I think you can still run a program that's locked
by someone else, especially just for reading. The only thing that
you can't do, IIRC, is delete (but you can move out of the way!).
Unfortunately, I can't check anymore, because I no longer have access
to hppa-hpux machines.
I have been thinking of deprecating hppa-hpux, and maybe all of hppa
if no one steps up. That might help a bit. I would love to deprecate
Windows, but I won't even dare dream about it. Besides, it turns out
that there are much much worse platforms than Windows, even if it
often drives me crazy.
If we can't determine where and how this is needed (maybe there's some
archeological info to be dug out of CVS?), we might want to consider
the idea of doing without it. And wait to see if it breaks someone.
I know that Jerome and I have occasionaly looked at this sort of
issue, but haven't always been succesful at making a proper patch.
I can try digging things up, but it's getting late, so maybe tomorrow.