Re: Re[6]: [Simulavr-devel] simulavr + Eclipse

[Petr Hluzín]

2011/3/16 Сергей Смирнов <address@hidden>:
>> This means that simulavr is not responding to gdb at all. The message
>> 'warning: unrecognized item "timeout" in "qSupported" response' is
>> probably printed by Eclipse trying to parse error messages coming from
>> GDB. This indicates a bug in simulavr.

To be precise: It is a bug in simulavr it is not responding to GDB.
Also Eclipse is confused by word "timeout" which is either bug in
Eclipse or in GDB (e.g. badly formated MI message).
The Eclipse-GDB bug just makes the bug more painful.

You can workaround the timeout bug by adding GDB comand:
set remotetimeout 10000
(This will not cure the simulavr bug but it can make it easier to investigate.)

> I've compiled a new version of simulavr:
> simulavr-4aeff7d03ae8a3a5fe347c141d1197e1a07fe88a (last version from 
> 2011-03-06, Petr Hluzin branch)
> It burns 100% CPU and I can see the same problem with the "timeout" error.
> I can see earlier post saying that the 100% CPU burn problem is resolved:
> http://git.savannah.gnu.org/cgit/simulavr.git/commit/?id=958cc2e4f0fca31f6cb245167b3215bbea6fab3a
> Isn't it?

As the commit message says:
"Also GDB is still burning 100% waiting until GDB makes TCP connection."
So it is not perfect, yet.

Your situation is strange due GDB sending data to simulavr but
simulavr behaves as the connection has not been opened yet.

Try starting simulavr with --debug-gdb. (This will not fix the problem
but will print some diagnostic.)

How are you launching simulavr? From Eclipse?
Does the problem happen if you launch from from command line?

Or try to debug the simulavr with Visual Studio (since you built it
with it). What is it doing?
Is GdbServerSocketMingW::Connect() being periodicaly executed?
Is GdbServerSocketMingW::ReadByte() being periodicaly executed?

-- 
Petr Hluzin