Running gdb against pthread code

[hdonk]

Hi all,


I've been trying to get firefox running up in gdb, but keep hitting a brick wall with gdb. When starting up the session, I'm getting: 'warning: unable to find libthread_db matching inferior's thread


library, thread debugging will not be available'


I've built a small multi threaded test app, and got the same results. I'm assuming this is because the version of libthread_db.so that is installed from the Angstrom feeds does not match the libpthread.so we are using.


So has anyone had any luck with this, or am I reduced to rebuilding the O/S from scratch?

 

[sebt3]

So has anyone had any luck with this, or am I reduced to rebuilding the O/S from scratch?

I always had this issue on SDL apps (also using pthread). If you solve that, you're my hero

 

[hdonk]

So has anyone had any luck with this, or am I reduced to rebuilding the O/S from scratch?

I always had this issue on SDL apps (also using pthread). If you solve that, you're my hero

Good: It's not just me & is an inherent problem!


Bad: It's happening, and you've had the problem too

 

[darfgarf]

does it really need fixing? even without it you'll still get the right line of code, just won't tell you which thread is running there, whih shouldn't be needed in most cases as either only one thread can run there, or it's part of a common worker thread function, which would make it fail on random threads

 

[hdonk]

It gets worse. I've just upgraded to the latest version of gdb in the Angstrom repo, and it reports lots of CRC mismatches between the debug and normal versions of the libraries. Deffo feels time to rebuild the OS.

does it really need fixing? even without it you'll still get the right line of code, just won't tell you which thread is running there, whih shouldn't be needed in most cases as either only one thread can run there, or it's part of a common worker thread function, which would make it fail on random threads

It depends. What if the problem I'm looking for is a race condition over a mutex?


And anyway, as I posted in the previous message, it looks like there's some serious mismatches between lib versions.

 

[darfgarf]

It depends. What if the problem I'm looking for is a race condition over a mutex?

then it's a bitch to find, though you shouldn't program them in in the first place, bad coder!

 

[hdonk]

It depends. What if the problem I'm looking for is a race condition over a mutex?

then it's a bitch to find, though you shouldn't program them in in the first place, bad coder!

Oooooh, so here I am trying to fix a fuckup in Firefox, and it's my fault. Why don't you go tie a concrete bollard to your legs & fall off a bridge? Or maybe you'd get that wrong too?