This bug report is continuation of the 346827 report. Attached simulation environment from Olly Funkster slightly modified applies and is attached. Just, please, recompile it. After running the application try to insert a BP or stop it - no reaction. Reproducible: Always
Created attachment 93863 [details] Slightly corrected example with while(1) loop
Created attachment 93879 [details] kdev-4.7.1-gdb.patch Leon, Give this patch a try and let me know if it works. Apply it to the 4.7.1 tarball. It doesn't fix the interrupt problem with my setup but I only tested with QEMU and I think it's specific to it since I've had this problem before 4.7.0, I'm building the gdbserver to try your example. This patch simply wraps the offending commit in #if statements (it's only meant for OSX) so it *should* restore the 4.7.0 behaviour on Linux.
Ok, I tested with gdbserver and everything works, including interrupt and breakpoints on the fly.
Created attachment 93903 [details] kdevelop-4.7.1-gdb.patch Cleaned up patch.
Created attachment 93904 [details] kdevelop-4.7.1-gdb-git.patch Patch for the git 4.7 branch.
Leon, does this fix your issue? Fernando, please remember to put patches up for review on reviewboard, more people will look at them then. thanks!
Yes! It was exactly I needed.
Git commit d051c4dc9b0495240eeda8a233a2c748d719ce14 by Milian Wolff, on behalf of Fernando Rodriguez. Committed on 13/08/2015 at 15:24. Pushed by mwolff into branch '4.7'. Fix interrupt of debuggee in remote debugging sessions. This fixes the insertion of breakpoints or stopping the debuggee when using gdbserver for remote debugging. NB: Code style adapted by Milian, hope I didn't break anything. M +12 -3 debuggers/gdb/gdb.cpp M +2 -0 debuggers/gdb/gdb.h http://commits.kde.org/kdevelop/d051c4dc9b0495240eeda8a233a2c748d719ce14
Thanks again Fernando! Please put your next patches up for review on reviewboard.kde.org. Cheers
I just tested this with KF5/KDevelop 5, there the patch doesn't seem to be required. The recent refactoring there seems to have fixed this among other things.
Hi Milian: It looks like it didn't apply correctly. On GDB::processLine it applied the #endif line but not the #if (Q_OS_MAC) so it won't compile. The second patch (incorrectly named kdevelop-4.7.1-gdb-git.patch) was the one meant for the git 4.7 branch, it basicly reverts the patch from bug #346827 (in case it breaks OSX) and applies the new fix.
Git commit 1715a0d59c70158a4c54d2d8b05e4181e73207b0 by Milian Wolff. Committed on 14/08/2015 at 00:46. Pushed by mwolff into branch '4.7'. Unbreak build - sorry for that. M +2 -1 debuggers/gdb/gdb.cpp http://commits.kde.org/kdevelop/1715a0d59c70158a4c54d2d8b05e4181e73207b0
On Friday, August 14, 2015 12:46:23 AM you wrote: > https://bugs.kde.org/show_bug.cgi?id=350916 > > --- Comment #12 from Milian Wolff <mail@milianw.de> --- > Git commit 1715a0d59c70158a4c54d2d8b05e4181e73207b0 by Milian Wolff. > Committed on 14/08/2015 at 00:46. > Pushed by mwolff into branch '4.7'. > > Unbreak build - sorry for that. > > M +2 -1 debuggers/gdb/gdb.cpp > > http://commits.kde.org/kdevelop/1715a0d59c70158a4c54d2d8b05e4181e73207b0 The "if (currentCmd_->cmdToSend().contains("exec-run"))" should be deleted to make sure it doesn't break anything for OSX. It's also on the -git patch.
Sorry Fernando, I seem to have messed up the apply of this patch. Can you provide me a patch against the current 4.7 head and I'll apply that one - hopefully without screwing up so badly again? Thanks, and sorry for this mess
Created attachment 94187 [details] Patch for current git branch Sure. I should've been more clear about which patch to apply.
Fernando, sorry for the late reply. Did I apply your latest patch now or is this still pending? Again, very sorry for the mess and lack of interaction from my side...