Hello folks, it really looks like we should bring 8250 KGDB stuff to a certain alignment.
First, let's look at 8250.patch and what it adds to drivers/serial/serial_core.c. This implicit registration of kgdb8250 port will never actually work for the most archs because it happens after the current_port var is assigned (in 8250_kgdb.c:kgdb_init_io() ) and therefore kgdb8250_add_port will quit early. Moreover, this looks like a hack and I can't _think_ of this being accepted in mainline. The question is, why that is done that way? If this is done solely due to the reason not to make people call kgdb8250_add_port/kgdb8250_add_platform_port explicitly, then I'm absolutely against this. For some archs/SoCs it's possible to use KGDB on 8250 port without standard 8250 driver, but I guess it won't work in tis case. So, I do think it's reasonable to state that each one wanting to use kgdb over 8250-compatible serial port should call registration routine explicitly. The other idea might be to re-init the kgdb serial port at the moment more info is known (like IRQ number). This makes sense to me, but currently it doesn't work that way due to the check mentioned earlier (if (current_port != &kgdb8250_ports[i])). And the other thing is that it should not be in serial_core, it just doesn't belong there. I understand I might not be getting things right, but there's a mess that makes it very hard to understand the idea and whether the implementation conforms to it... Vitaly ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Kgdb-bugreport mailing list Kgdb-bugreport@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/kgdb-bugreport
