[lldb-dev] LLDB for bare-iron targets lacking support for qC /qfThreadInfo packet

Greg Clayton via lldb-dev lldb-dev at lists.llvm.org
Mon Aug 24 09:36:48 PDT 2015


I guess we can just make pid == 1 and tid == 1 in this case. As long as a stop reply is responding we should be able to debug.

> On Aug 23, 2015, at 11:51 PM, Jaydeep Patil <jaydeep.patil at imgtec.com> wrote:
> 
> Hi Greg,
> 
> The '?' packet always returns 'S05'. There is no thread information available in any of the packets.
> 
> Regards,
> Jaydeep
> 
> -----Original Message-----
> From: Greg Clayton [mailto:clayborg at gmail.com] 
> Sent: 21 August 2015 PM 09:43
> To: Jaydeep Patil
> Cc: lldb-dev at lists.llvm.org
> Subject: Re: LLDB for bare-iron targets lacking support for qC /qfThreadInfo packet
> 
> One option is to send a ? packet to get a stop reply packet back. The stop reply should contain a "thread:XXXX;" key/value pair and you could just make the pid == tid == XXXX. We currently try these in the order below:
> 
> 1 - qProcessInfo 
> 2 - qC
> 3 - qfThreadInfo
> 
> Now we could add a fourth:
> 
> 4 - ?
> 
> 
>> On Aug 21, 2015, at 12:23 AM, Jaydeep Patil <jaydeep.patil at imgtec.com> wrote:
>> 
>> Hi,
>> 
>> The gdb-stub running on a MIPS bare-iron board lacks support for qC /qfThreadInfo packet.  Once the LLDB is connected to the target it cannot proceed further as there is no process and/or thread information available from the target. However GDB is able to debug the application as it ignores the output of qC /qfThreadInfo packet if they are not supported. GDB assumes that a process/thread X is available for debugging.
>> 
>> Is there any way in LLDB to support such bare-iron targets where qC /qfThreadInfo packets are not supported?
>> 
>> Regards,
>> Jaydeep
> 



More information about the lldb-dev mailing list