Message ID | 20190502081554.5521-1-arilou@gmail.com |
---|---|
Headers | show |
Series | gdbstub: Refactor command packets handler | expand |
Jon Doron <arilou@gmail.com> writes: > This patch series refactors the old gdbstub command packets handler > with a new infrastructure which should ease extending and adding new > and missing gdb command packets. There seems to be some compiler errors and variants that this breaks the CI on: https://travis-ci.org/stsquad/qemu/builds/532410263 https://app.shippable.com/github/stsquad/qemu/runs/822/summary/console https://gitlab.com/stsquad/qemu/pipelines/61291132 You might want to consider setting up CI checks on your system and running your next series through them before posting: https://wiki.qemu.org/Testing/CI -- Alex Bennée
Jon Doron <arilou@gmail.com> writes: > This patch series refactors the old gdbstub command packets handler > with a new infrastructure which should ease extending and adding new > and missing gdb command packets. Jon, I've finished my review and things are looking pretty good. The code is a good clean-up and makes adding new features a lot easier. Thanks for the examples of extensions - they were worth it to see how this might be used although we shouldn't include them in the first merge. As they extend the gdbserver ABI we'll want to think carefully about exactly what we want to expose before we include it in master. Going forwards aside from the various comments on each patch it would be worth making sure the branch has gone through at least one CI run to make sure the non-x86 builds (and disable-tcg and other exotica) haven't been broken. It would be nice if we could extend the testing of the gdbserver. Have you been testing this with the gdb test suite or just manually? Now we have system test and linux-user binaries being built we could probably do better than the manually run tests/guest-debug/test-gdbstub.py. Finally it would be nice if we could modernise the membuf and strbuf handling with a more robust glib based approach. I understand if you don't want to do that now and I'll happily accept the patches without it but I did notice you can send the gdbserver a bit loopy if you send it some very long maint packets so it would be nice to have that a bit safer. -- Alex Bennée
Alex Bennée <alex.bennee@linaro.org> writes: > Jon Doron <arilou@gmail.com> writes: > >> This patch series refactors the old gdbstub command packets handler >> with a new infrastructure which should ease extending and adding new >> and missing gdb command packets. > > Jon, > <snip> > > Finally it would be nice if we could modernise the membuf and strbuf > handling with a more robust glib based approach. I understand if you > don't want to do that now and I'll happily accept the patches without it > but I did notice you can send the gdbserver a bit loopy if you send it > some very long maint packets so it would be nice to have that a bit > safer. Actually I had a go at this this morning and it turned out to be quite fiddly so perhaps this is something best left to a follow-up series once the re-factoring is in. -- Alex Bennée