This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFA/7.8] user breakpoint not inserted if software-single-step at same location


> I have a very simple idea, around "jump" + "always-inserted".
> 
> E.g., with, where b+ indicates a user breakpoint:
> 
>        00001 nop        <- PC
>     b+ 00002 nop        
> 
>  - enable breakpoints always inserted mode
>  - step to 00002
>  - gdb removes the sss breakpoint.
>  - due to always inserted mode, gdb does not remove b+, but
>    due to the bug, it's actually no longer planted.
>  - the b+ breakpoint should be reported to the user.
>  - now do "jump $pc".
>  - expected:
>     The breakpoint should trigger immediately again.
>  - what we get on sss targets:
>     GDB loses control, and program runs to end.
> 
> Ah, I just went ahead and tried that against my by sss-on-x86
> branch, and indeed it fails here, while it passes on
> pristine mainline / hardware stepping.

That is very clever! WDYT about pushing this new testcase now?

-- 
Joel


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]